Apports de la Médecine de Précision à la

Neurochirurgie : Applications aux Glioblastomes

Vianney Gilard

To cite this version:

Vianney Gilard. Apports de la Médecine de Précision à la Neurochirurgie : Applications aux Glioblas-
tomes. Sciences agricoles. Normandie Université, 2021. Français. �NNT : 2021NORMR075�. �tel-
03517918�

HAL Id: tel-03517918

https://theses.hal.science/tel-03517918v1
Submitted on 8 Jan 2022

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
 THÈSE
Pour obtenir le diplôme de doctorat
Spécialité SCIENCES DE LA VIE ET DE LA SANTE

Préparée au sein de l'Université de Rouen Normandie

Αppοrts de la Μédecine de Ρrécisiοn à la Νeurοchirurgie :

Αpplicatiοns aux Gliοblastοmes
Présentée et soutenue par
VIANNEY GILARD
Thèse soutenue le 29/11/2021
devant le jury composé de
PROFESSEUR DES UNIV - PRATICIEN
M. MICHEL LEFRANC HOSP., UNIVERSITE AMIENS PICARDIE Rapporteur du jury
JULES VERNE
PROFESSEUR DES UNIV - PRATICIEN
M. ABDERRAHIM OUSSALAH Rapporteur du jury
HOSP., UNIVERSITE DE LORRAINE
MAITRE DE CONF UNIV. - PRATICIEN
MME EMELINE TABOURET Membre du jury
HOSP., Aix-Marseille Université
MAITRE DE CONF UNIV. - PRATICIEN
M. ABDELLAH TEBANI Membre du jury
HOSP., Université de Rouen Normandie
PROFESSEUR DES UNIV - PRATICIEN
M. STÉPHANE MARRET Directeur de thèse
HOSP., Université de Rouen Normandie
PROFESSEUR DES UNIV - PRATICIEN
MME SOUMEYA BEKRI Co-directeur de thèse
HOSP., Université de Rouen Normandie
Thèse dirigée par STÉPHANE MARRET et SOUMEYA BEKRI, GENOMICS AND
PERSONALIZED MEDICINE IN CANCER AND NEUROLOGICAL DISORDERS
 Remerciements

Je remercie madame le Professeur Soumeya BEKRI, pour avoir accepté de

diriger cette thèse. Je vous remercie de m’avoir fait découvrir et apprécier les
Sciences Omiques ainsi que pour vos encouragements et la bienveillance que
vous m’avez témoignée. Soyez assurée de mon profond respect.

Je remercie monsieur le Professeur Stéphane MARRET, pour avoir accepté

de diriger cette thèse. Je vous remercie de m’avoir fait découvrir la recherche
clinique et fondamentale ainsi que pour vos précieux conseils. Soyez assuré
de mon profond respect.

A mes rapporteurs, monsieur le Professeur Michel LEFRANC (Université

d’Amiens) et monsieur le Professeur Abderrahim OUSSALAH (Université
de Nancy), mes sincères remerciements pour le temps consacré à la relecture
de ce manuscrit. Je suis honoré de vous compter parmi mon jury.

A madame le Docteur Emeline TABOURET d’avoir accepté de participer à

ce jury et nous faire partager ainsi votre expertise.

A monsieur le Docteur Abdellah TEBANI pour ton soutien sans faille tout
au long de ce travail. Je te remercie pour ta patience et ta pédagogie. Nos
réunions hebdomadaires m’ont été d’un grand soutien et source d’une
inspiration permanente. Sois assuré de mon estime et de mon amitié.

Au laboratoire d’anatomopathologie du CHU de Rouen, madame le

Professeur Annie LAQUERRIERE, monsieur le Dr Florent MARGUET et
madame Carole BASSET pour leur participation active à ce travail et leurs
encouragements.

A monsieur le Docteur Bruno GONZALEZ et monsieur le Professeur Carlos

AFONSO, je vous adresse mes chaleureux remerciements pour m’avoir
ouvert les portes de vos laboratoires et pour vos encouragements
bienveillants.
Je remercie vivement madame Justine FERREY et madame Isabelle
SCHMITZ-AFONSO du laboratoire COBRA qui m’ont initié à la
métabolomique, pour leur patience à mon égard et leurs précieux conseils.

Un grand merci à toute l’équipe du laboratoire de Biochimie Métabolique du

CHU de Rouen pour leur accueil (Isabelle CORDONNIER, Hélène
DRANGUET, Loréna DUJARDIN, Céline LESUEUR, Isabelle TRAGIN,
Carine PILON, Thomas PLICHET, Sarah SNANOUDJ et Bénédicte SUDRIE-
ARNAUD) pour leur soutien tout au long de ce parcours et pour leur aide
précieuse.

A tous mes collègues de l’unité INSERM U1245 – Équipe "Genetics and

Physiopathology of Neurodevelopmental Disorders" pour leurs
encouragements et soutien dans ce parcours. Un remerciement particulier à
Carole LAGNEL, Céline LESUEUR et Michelle HAUCHECORNE qui m’ont
accueilli au laboratoire en Master 2 avec bienveillance.

A tous mes collègues du service de Neurochirurgie dont je partage le

quotidien depuis maintenant 10 ans.
 Table des Matières
Remerciements
Table des matières
Publications et communications
Liste des figures
Liste des principales abréviations

INTRODUCTION GENERALE
Préambule (p1)

CHAPITRE I : CHIRURGIE ET MEDECINE DE PRECISION

1.1 Perspectives historiques (p5)

1.1.1 La neurochirurgie dans l’histoire
1.1.2 Apports de la biologie moléculaire à la médecine
1.2 Sciences omiques (p13)
1.2.1 Génomique
1.2.2 Transcriptomique
1.2.3 Protéomique
1.2.4 Métabolomique

1.3 Article I : Precision Neurosurgery: A Path Forward.

CHAPITRE II : LES GLIOBLASTOMES

1.1 Épidémiologie (p34)

1.2 Présentation clinique (p35)
1.3 Diagnostic radiologique (p35)
1.4 Histologie et biologie moléculaire (p39)
1.5 Prise en charge chirurgicale (p41)
1.6 Thérapeutiques médicales (p42)
1.7 Nouvelles thérapies et perspectives (p45)

1.8 Article II : Diagnosis and Management of Glioblastoma: A Comprehensive

Perspective.

CHAPITRE III : TRAVAIL EXPERIMENTAL

1.1 Définition du problème (p58)

1.2 Mise en place du design expérimental (p59)
1.3 Article III : Integrative Metabolomics Reveals Deep Tissue and Systemic Metabolic
Remodeling in Glioblastoma.

CONCLUSIONS ET PERSPECTIVES (p102)

 Publications & Communications
(2018 – 2021)

Publications liées à la thèse

1. Gilard V, Derrey S, Marret S, Bekri S, Tebani A. Precision Neurosurgery: A Path Forward. Journal
of Personalized Medicine. 2021. J Pers Med.
2. Gilard V, Tebani A, Dabaj I, Laquerrière A, Fontanilles M, Derrey S, Marret S, Bekri S. Diagnosis
and Management of Glioblastoma: A Comprehensive Perspective. J Pers Med. 2021 Apr 1;11(4):258.
3. Gilard V, Ferey J, Marguet F, Fontanilles M, Ducatez F, Pilon C, Lesueur C, Pereira T, Basset C,
Schmitz-Afonso I, Di Fioré F, Laquerrière A, C Afonso, Derrey S, Marret S, Bekri S, Tebani A.
Integrative Metabolomics Reveals Deep Tissue and Systemic Metabolic Remodeling in
Glioblastoma. 2021. Cancers.
Autres publications
1. Garrido E, Metayer T, Borha A, Langlois O, Curey S, Papagiannaki C, Di Palma C, Emery E, Derrey
S, Gaberel T, Gilard V. Intracranial aneurysms in pediatric population: a two-center audit. Childs
Nerv Syst. 2021 Aug;37(8):2567-2575.
2. Rault F, Terrier L, Leclerc A, Gilard V, Emery E, Derrey S, Briant AR, Gakuba C, Gaberel T.
Decreased number of deaths related to severe traumatic brain injury in intensive care unit during
the first lockdown in Normandy: at least one positive side effect of the COVID-19 pandemic. Acta
Neurochir (Wien). 2021 Jul;163(7):1829-1836.
3. Olympios N, Gilard V, Marguet F, Clatot F, Di Fiore F, Fontanilles M. TERT Promoter Alterations
in Glioblastoma: A Systematic Review. Cancers (Basel). 2021 Mar 8;13(5):1147.
4. Dabaj I, Ferey J, Marguet F, Gilard V, Basset C, Bahri Y, Brehin AC, Vanhulle C, Leturcq F, Marret
S, Laquerrière A, Schmitz-Afonso I, Afonso C, Bekri S, Tebani A. Muscle metabolic remodelling
patterns in Duchenne muscular dystrophy revealed by ultra-high-resolution mass spectrometry
imaging.Sci Rep. 2021 Jan 21;11(1):1906.
5. Gilard V, Tebani A, Bekri S, Marret S. Intraventricular Hemorrhage in Very Preterm Infants: A
Comprehensive Review. J Clin Med. 2020. Jul 31;9(8):2447.
6. Horowitz A, Saugier-Veber P, Gilard V. Inference of Diagnostic Markers and Therapeutic Targets
From CSF Proteomics for the Treatment of Hydrocephalus. Front Cell Neurosci. 2020 Oct
21;14:576028.
7. Méreaux JL, Gilard V, Le Goff F, Chastan N, Magne N, Gerardin E, Maltête D, Lebas A, Derrey
S. Practice of stereoelectroencephalography (sEEG) in drug-resistant epilepsy: Retrospective series
with surgery and thermocoagulation outcomes. Neurochirurgie. 2020 Jun;66(3):139-143.
8. Goldenberg A, Marguet F, Gilard V, Cardine AM, Hassani A, Doz F, Radi S, Vasseur S, Bou J,
Branchaud M, Houdayer C, Baert-Desurmont S, Laquerriere A, Frebourg T. Mosaic PTEN alteration
in the neural crest during embryogenesis results in multiple nervous system hamartomas. Acta
Neuropathol Commun. 2019 Dec 3;7(1):191.
9. Gilard V, Beccaria K, Hartley JC, Blanot S, Marqué S, Bourgeois M, Puget S, Thompson D, Zerah
M, Tisdall M. Brain abscess in children, a two-centre audit: outcomes and controversies. Arch Dis
Child. 2020 Mar;105(3):288-291.
10. Terrier L, Gilard V, Marguet F, Fontanilles M, Derrey S. Stereotactic brain biopsy: evaluation of
robot-assisted procedure in 60 patients. Acta Neurochir (Wien). 2019 Mar;161(3):545-552.
11. Gilard V, Chadie A, Ferracci FX, Brasseur-Daudruy M, Proust F, Marret S, Curey S. Post
hemorrhagic hydrocephalus and neurodevelopmental outcomes in a context of neonatal
intraventricular hemorrhage: an institutional experience in 122 preterm children. BMC Pediatr. 2018
Aug 31;18(1):288.
Communications orales

1. Gilard V. Metabolomics analysis in patients with high grade glioma. Congrès annuel de la Société
française de neurochirurgie. Nice Octobre 2021.

2. Gilard V. Post hemorrhagic hydrocephalus and neurodevelopmental outcomes in a context of

neonatal intraventricular hemorrhage: an institutional experience in 122 preterm children. Congrès
annuel de la Société française de neurochirurgie. Saint Malo 2020.

Communications affichées

1. Gilard V, Ferey J, Marguet F, Fontanilles M, Ducatez F, Pilon C, Lesueur C, Pereira T, Basset C,

Schmitz-Afonso I, Di Fioré F, Laquerrière A, C Afonso, Derrey S, Marret S, Bekri S, Tebani A.
Metabolomics Reveals Deep Tissue and Systemic Metabolic Remodeling in Glioblastoma. EMBL
Symposium: Multiomics to Mechanisms: Challenges in Data Integration. September 2021.
 Liste des figures

Figure 1. Photographie d’un crâne Inca ayant subi trois trépanations et survécu à la
procédure comme en attestent les berges osseuses consolidées.

Figure 2. Photographie d’un cadre de stéréotaxie dit de Leksell permettant les biopsies et
implantations d’électrodes cérébrales.

Figure 3. Représentation des techniques de biologie moléculaire des différentes sciences

omiques.

Figure 4. Représentation schématique des différentes étapes d’une étude métabolomique.

Figure 5. IRM cérébrale en coupes axiales en séquence FLAIR (A) puis en pondération T1 avec
injection de gadolinium (B) chez un patient présentant un GBM du carrefour droit.

Figure 6. IRM cérébrale en coupes axiales en pondération T1 avec injection de gadolinium chez
différents patients présentant un GBM pour illustrer leur hétérogénéité morphologique.

Figure 7. IRM cérébrale en coupes axiales en séquence de perfusion chez un patient présentant
un GBM du carrefour droit. A noter l’augmentation du volume sanguin cérébral au sein de la
lésion.

Figure 8. IRM cérébrale avec séquences de spectrométrie révélant les pics normaux du cerveau
sain (4A), la présence de lactates et de lipides au sein de la lésion témoins de la nécrose (4B)
ainsi qu’une inversion du rapport choline/NAA (4C), caractéristique d’une lésion gliale de
haut grade.

Figure 9. Vue microscopique d’un glioblastome IDH muté.

Figure 10. Schéma représentant l’arbre diagnostique de la classification OMS 2016.

Figure 11. Visualisation peropératoire d’une exérèse chirurgicale ; vue microscopique (7A) et
sous fluorescence au 5-ALA (7B).

Figure 12. Illustration du traitement d’un patient atteint d’un glioblastome selon le protocole
STUPP.

Figure 13. Critères IRM d'évaluation de la réponse en neuro-oncologie.

 Liste des principales abréviations

ADN Acide désoxyribonucléique

ARN Acide ribonucléique
ARNm ARN messager
CCS Collision Cross Section
ESI ElectroSpray Ionisation
FT-ICR Fourrier Transform-Ion Cyclotronic Resonance
FT-MS Fourrier Transform-Mass Spectrometry
GBM Glioblastome Multiforme
HDMS High Definition Mass Spectrometry
HGMD Human Gene Mutation Database
HMDB Human Metabolome DataBase
IMS Ion Mobility Spectrometry
LC Liquid Chromatography
min Minute
MS Mass spectrometry
ms Millisecond
MS/MS Tandem Mass spectrometry
m/z Rapport masse sur charge
NMR Nuclear Magnetic Resonnance
OMIM Online Mendelian Inheritance in Man
PC Principal Component
Q Quadripole
s Second
TOF Time of Flight
tR Temps de rétention
UHPLC Ultra-High Pression Liquid Chromatography
INTRODUCTION GENERALE
 Préambule

Dans la continuité de la médecine de précision, la chirurgie de précision est un concept

émergent depuis une dizaine d’année. La chirurgie, qui dans l’histoire a été longtemps

une branche à part de la médecine, a mis du temps à s’approprier les énormes progrès

technologiques dont nous avons été témoins au cours des trois dernières décennies.

En effet, une grande partie de nos pratiques chirurgicales actuelles n'ont pas changées

depuis maintenant plusieurs décennies. Bien que nous ayons tenté de rendre les

procédures chirurgicales moins morbides en utilisant des voies d’abord mini-invasives

ou endoscopiques, lorsque cela était approprié, la technologie accompagnant ces

procédures peu invasives reste relativement rudimentaire. Ces avancées ont souvent

consisté en une « ouverture » moins large de notre patient, une plus grande prise en

compte des contraintes anatomiques naturelles ou le recours à des instruments

chirurgicaux plus adaptés et variés. Ce n'est que récemment que les techniques

d’ingénierie ont amélioré nos pratiques chirurgicales que ce soit par l’essor des robots

chirurgicaux ou des systèmes de navigation per opératoire qui permettent aux

chirurgiens d'effectuer des procédures parfois plus précises que la main humaine.

Grâce à la collaboration croissante entre l'ingénierie, l'informatique et les chirurgiens,

nous commençons à entrevoir un avenir passionnant et peut-être, une prochaine

révolution chirurgicale. L'avenir de la chirurgie tournera autour du thème central de

la chirurgie guidée par l'image, de la réalité augmentée, de la résolution des images à

la navigation et à l'intelligence artificielle, en passant par les résections guidées par des

1
données de biologie moderne. Néanmoins, un élément central mérite d’être souligné :

ces technologies sont des aides à la conception et à la réalisation de l’acte chirurgical

et elles ne sauraient se substituer à une connaissance anatomique et gestuelle

rigoureuses. Elles viennent faciliter ce geste et son enseignement. Ensuite, la chirurgie

de précision n’est pas à confondre avec la précision de l’acte chirurgical même si celui-

ci est un élément de ce progrès. Comme la médecine de précision, la chirurgie de

précision place le patient comme référentiel de lui-même plus que la procédure elle-

même, en y intégrant des données médicales et biologiques individuelles tout en

tenant compte de la grande diversité interindividuelle. Les chirurgies oncologiques,

mais pas seulement, se sont appropriées ce concept. Traditionnellement, l'évaluation

peropératoire de la résection tumorale dépendait largement de l'inspection visuelle,

de la palpation des tissus et plus récemment des données d’imagerie ou d’histologie

per ou post opératoire. L’apport de la chirurgie de précision a été par exemple de

proposer des résections assistées par un spectromètre de masse. Elle repose donc sur

de larges collections biologiques issues des sciences omiques (génomique,

transcriptomique, protéomique, métabolomique et phénomique) utilisées à l’échelle

de l’individu. Son émergence a été rendue possible par l’essor du big data et de

l’intelligence artificielle. La neurochirurgie s’est approprié relativement tôt ces

avancées technologiques et son développement coïncide avec ces bonds d’ingénierie

et de biologie. Elle a été l’une des premières chirurgies à utiliser le microscope

opératoire pour la chirurgie hypophysaire puis endocrânienne au sens large ; l’une des

premières à utiliser l’imagerie et la navigation peropératoire et les techniques mini-

2
invasives. Il est donc logique que la neurochirurgie s’enrichisse de ces données

nouvelles de biologie moléculaire issues du big data clinico-biologique pour

accompagner ce bond technologique vers la chirurgie de précision.

Cette thèse porte sur les apports et applications des sciences omiques aux pathologies

neurochirurgicales. Nous développerons les éléments diagnostiques en

métabolomiques appliqués au glioblastome, tumeur cérébrale maligne.

Les objectifs de ce travail de thèse sont définis comme suit :

1. Mise en place d’une méthodologie métabolomique non ciblée et ciblée basée sur

une stratégie analytique multidimensionnelle comportant la spectrométrie de

masse à haute et très haute résolution, la chromatographie liquide ultra-haute

performance et la mobilité ionique.

2. Mise en place de la méthodologie de prétraitement, d’analyse et d’exploitation

des données

3. Application à une pathologie neurochirurgicale qu’est le glioblastome.

Le manuscrit de cette thèse est structuré en trois parties, Neurochirurgie et chirurgie

de précision (Chapitre I), le glioblastome (Chapitre II) et le travail expérimental

(Chapitre III).

Le chapitre I comporte une introduction avec une perspective historique de la

naissance et pratique de la neurochirurgie ainsi que l’apport des techniques de biologie

moléculaire à la neurochirurgie moderne.

3
Le chapitre II comprend un état des lieux des connaissances actuelles sur l’objet

principal du projet de thèse que représente le glioblastome (basés sur une revue

publiée) et le chapitre III représente le travail expérimental. Celui-ci consiste en un

article (art 3) qui décrit les aspects méthodologiques métabolomiques appliquées au

glioblastome.

4
CHAPITRE I : CHIRURGIE ET MEDECINE DE PRECISION

1.1 Perspectives historiques

1.1.1 La neurochirurgie dans l’histoire

Des techniques et des Hommes

La neurochirurgie est considérée comme la plus ancienne des pratiques chirurgicales. Les

premières trépanations ont été réalisées à la période néolithique. En France, des traces de

trépanations humaines datant de 7000 ans avant J.C. ont été retrouvées. La civilisation pré Inca

usa de techniques de trépanation de façon importante vers 2000 ans avant J.C. et le recours à

des plaques en métal était pratiqué pour couvrir le defect osseux. Les indications de

trépanation de cette époque étaient les troubles mentaux, l’épilepsie, les céphalées, les

maladies organiques et certains traumatismes crâniens. A l’époque, ces pratiques étaient

également utilisées pour des raisons magiques ou religieuses. Un outil en bronze utilisé pour
« user » la table externe de l’os du crâne a été retrouvé en Amérique du Sud. La cocaïne issue

des feuilles de coca était utilisée comme anesthésiant local. Des stigmates de consolidation

osseuse attestaient de la survie de certains patients suite à ces procédures 1-3 (Figure 1).

Figure 1. Photographie d’un crâne Inca ayant subi trois trépanations et survécu à la

procédure comme en attestent les berges osseuses consolidées 4.

5
Hippocrate, médecin grec au IVe siècle avant J.C. considéré comme le père de la médecine

décrivit l’anatomie fonctionnelle du cerveau. Il enseignait à ses disciples que l’hémisphère

cérébral contrôlait l’hémicorps controlatéral et qu’ainsi la trépanation devait se pratiquer du

côté opposé aux symptômes. Il usait d’un trépan qui arrêtait son œuvre une fois le crâne ouvert

en préservant le cerveau sous-jacent. De telles procédures sont toujours couramment utilisées

avec des outils chirurgicaux très proches dans leur conception 1,5.

Ambroise Paré, chirurgien-barbier au XVIe siècle est l’un des fondateurs de la chirurgie

moderne. Opérant sur les champs de bataille, il décrit les techniques de cautérisation et de

ligature artérielle. Sa description de la trépanation conduit à une diminution des infections et

événements hémorragiques. Jean-Louis Petit, chirurgien et anatomiste du XVIIe siècle a décrit

l’hématome extradural avec la notion d’intervalle libre. Il décrivit également l’incision de la

dure-mère et les deux plans de cette méninge. Ces notions sont toujours enseignées à nos

étudiants actuellement. Un exemple classique de convergence entre neurochirurgie et histoire

de France est la mort d’Henri II, en 1559 6. Blessé par un fragment de lance lors d’un tournoi,

le roi de France meurt d’infection dans les jours suivant cette plaie oculaire. Ambroise Paré,

par sa description de l’autopsie du roi, décrivit le phénomène d’empyème cérébral lié à la

contamination d’une plaie.

Outre-Manche, le pionnier de la neurochirurgie moderne se nomme Sir Victor Horsley (1857-

1924). Il est chirurgien militaire et travaille au Queen Square Hospital à Londres. Appuyé par

ses collègues physiologistes, il émet l’hypothèse que l’épilepsie est une maladie du cortex

cérébral et que sa résection peut permettre de stopper le développement de l’épilepsie. Il

réalise sur des observations cliniques, les premières cortectomies cérébrales chez trois patients.
L’un des patients survécut à cette chirurgie avec une diminution de la fréquence des crises.

Cette procédure, bien qu’alimentée d’explorations d’imagerie et de physiologie, est

couramment pratiquée de nos jours 7,8.

Harvey Cushing (1869-1939), premier neurochirurgien américain à opérer des tumeurs

cérébrales, aurait pratiqué plus de 2000 interventions cérébrales. Sa description de ses voies

d’abord a permis d’améliorer la survie des patients de neurochirurgie 9. En France, les

pionniers de la neurochirurgie se nomment Clovis Vincent et Thierry de Martel. Formé à Paris,

6
Clovis Vincent fonda le service de neurochirurgie de la Pitié-Salpétrière. Neurologue de

formation, il fit progresser son art et forma de nombreux élèves de renom 10.

Malheureusement, la chirurgie porte encore le poids de son histoire. Du XIIe au XVIIIe siècle,

les chirurgiens-barbiers sont distincts des médecins. Les progrès médicaux réalisés en parallèle

de la pratique chirurgicale n’enrichissent que peu le savoir-faire chirurgical. Le principal écueil

de la neurochirurgie repose sur l’asepsie. La naissance de la neurochirurgie moderne post-

guerre a pu se développer grâce aux progrès des conditions per opératoires (asepsie et

anesthésie) et à une meilleure appréhension de la pathologie grâce aux progrès de l’imagerie

et de la neurophysiologie.

Les explorations en neurochirurgie

Très tôt, les neurochirurgiens se sont définis comme des médecins ayant une pratique

chirurgicale. Clovis Vincent disait que « La neurochirurgie, ce n'est pas seulement l'art

d'enlever les tumeurs du cerveau, mais c'est le moyen d'apprendre d'une façon précise les

fonctions du cerveau humain ». Il fut d’ailleurs le premier médecin-chef d’un service de

chirurgie. Au Canada, Wilder Penfield 11, également neurologue et neurochirurgien, décrivit

la somatotopie cérébrale. Il contribua, tout comme Clovis Vincent, élève de Babinski, à la

description des circuits de la mémoire. S’appuyant sur ses travaux, Harvey Cushing réalisa les

premières stimulations peropératoires pour explorer les fonctions des zones cérébrales

réséquées. Ces explorations historiques préfigurent le développement de la chirurgie éveillée.

L‘un des pans de la neurochirurgie moderne la plus liée à ces progrès techniques est

certainement la neurochirurgie fonctionnelle. Elle vise à ajuster certaines fonctions du névraxe

quand le traitement médicamenteux est insuffisant ou non recommandé. Classiquement

d’autres approches y sont associées telles que la stimulation cérébrale profonde, la chirurgie

de l’épilepsie, la chirurgie de la douleur et la radiochiurgie. Ses pionniers viennent d’horizons

divers. Lars Leksell (1907-1986) est physicien puis neurochirurgien suédois 12. Nous lui devons

l’un des premiers cadres de stéréotaxie (Figure 2) et l’invention de la radiochirurgie. La France

est là encore bien représentée : Jean Talaraich (1911-2007) est psychiatre puis neurochirurgien
13 . Avec Jean Bancaud, neurologue, ils rédigent le premier atlas de stéréotaxie qui reste un

ouvrage de référence et proposent un cadre de stéréotaxie qui permet de réaliser implantation

7
d’électrodes et biopsie cérébrale avec une précision incomparable. Ces techniques

chirurgicales ont été développées avant le scanner cérébral (1970’s). Les repères anatomiques

étaient basés sur des abaques statistiques issus de centaines d’observations cadavériques et de

repères indirects. Ces abaques représentent une information stockée, quantifiable, normée et

mobilisable pour assister l’acte chirurgical. Conceptuellement, ces abaques sont les bases de

données informationnelles précurseurs du big data anatomo-clinico-biologiques actuel. En

effet, les repères indirects, avant l’avènement du scanner et de l’IRM, consistaient en une

ventriculographie gazeuse (inventée par Dandy) ou l’injection sous-arachnoïdienne de

lipiodol tel que décrit par Sicard en 1921 14. Il faut imaginer une opération de neurochirurgie

avant ces imageries et l’approximation que cela pouvait représenter ainsi que la morbidité

associée. La sémiologie clinique nous permettait et permet toujours d’émettre une hypothèse
topographique d’une lésion. Une ventriculographie (injection d’air ou de lipiodol) était

réalisée puis une radiographie de crâne standard laissait entrevoir la déviation du système

ventriculaire et ainsi la localisation supposée de cette lésion. Une « ouverture » du scalp large

était réalisée puis des trous de trépan étaient réalisés. Un volet crânien était taillé à la scie Gigli
(scie à ruban glissée d’un trou de trépan à un autre). En 1972, Godfrey Hounsfield 15 se voit

décerner le prix Nobel de médecine pour la découverte du premier scanner appelé

tomographie. De manière anecdotique, Sir Hounsfield était ingénieur chez EMI, société de

musique et d’électronique qui, grâce aux ventes record des Beatles dans les années 1960’s
finança les travaux de recherche de Hounsfield. Ce fut un bouleversement technologique pour

la neurochirurgie. Pour la première fois, nous pouvions voir le cerveau de manière directe sans

ouvrir la boite crânienne. Le scanner a progressivement amélioré sa résolution et l’arrivée de

l’IRM au début des années 1990’s a permis une technique non irradiante et beaucoup plus fine.

En seulement 30 ans, l’IRM est passée de sa naissance à une pratique courante qui parait
évidente à toute une génération.

8
 Figure 2. Photographie d’un cadre de stéréotaxie dit de Leksell permettant les biopsies et

implantations d’électrodes cérébrales.

En parallèle, les progrès de la magnification optique et des techniques de neuro-endoscopie

ont amené à des progrès d’approches opératoires, de gestion de dissection ou d’hémostase

voire de nouvelles interventions (ventriculocysternostomie endoscopique par exemple) 16 .

Depuis les années 2000, la robotique s’est invitée au bloc opératoire. Des robots chirurgicaux
(pour la stéréotaxie cérébrale ou rachidienne pour le moment), des systèmes de navigation

voire de réalité augmentée sont utilisés quotidiennement. Plus récemment, cette intégration

des progrès humains et techniques vient s’enrichir des progrès de la biologie moléculaire et

contribue au développement d’une nouvelle révolution qu’est la chirurgie de précision.

9
 1.1.2. Apports de la biologie moléculaire à la médecine

La naissance de la biologie

Les philosophes grecs, voraces de curiosité, regardent avec intérêt l'éventail des créatures

vivantes, de la plante la plus humble à l'homme lui-même. Un nom grec a été inventé par un

naturaliste allemand au début du XIXe siècle pour cette étude de tous les aspects physiques

de la vie naturelle - la biologie, du bios (vie) et du logos (mot ou discours) 17,18. C'est un sujet
qui comporte des subdivisions claires, comme la botanique, la zoologie ou l'anatomie. Mais

tous s'intéressent aux organismes vivants. L’un des pères de la biologie est Alcmaeon de

Crotone 19 , médecin, astronome et philosophe pythagoricien du Ve siècle avant J.C. Il

s’intéresse aux fluides corporels et émet l’hypothèse que les caractéristiques de ces fluides

pourraient être des marqueurs de pathologies plus globales : c’est le principe même d’un

biomarqueur tel qu’il se conçoit actuellement. Il est suivi un siècle plus tard par Aristote et

Théophraste qui développent respectivement la biologie animale et végétale. Dans son

ouvrage Historia animalium, Aristote 20 recense les caractéristiques biologiques propres à

chaque animal, car selon lui, « c’est par l’observation du normal et de ses caractéristiques

intrinsèques que l’on peut décrire le pathologique ». Là encore nous verrons tel que développé

plus loin dans ce manuscrit, qu’il s’agit d’un principe des sciences Omiques de caractériser sur

le plan phénotypique une population et d’établir le pathologique et son biomarqueur par

comparaison d’un groupe témoin de sujets indemnes de la pathologie qu’on cherche à

identifier. En 158 avant J.C. la ville de Pergame compte un nouveau médecin en chef des

gladiateurs. Il est un médecin grec du nom de Galien 21. Cette nomination lui donne l'occasion

d'étudier des blessures de toutes sortes. Sa connaissance des muscles lui permet d'avertir ses

patients du résultat probable de certaines opérations. Mais c'est la pratique de la dissection

des singes et des porcs qui donne à Galien des informations détaillées pour ses traités

médicaux. Près de 100 de ces ouvrages ont survécu. Ils deviennent la base de la grande

réputation de Galien dans la médecine médiévale, incontestée jusqu'aux travaux anatomiques

de Vésale. Grâce à ses expériences, Galien est capable de renverser de nombreuses croyances

de longue date, comme la théorie selon laquelle les artères contiennent de l'air et non du sang-

le transportant vers toutes les parties du corps depuis le cœur et les poumons. Cette croyance

est basée à l'origine sur les artères des animaux morts qui semblent vides. Galien démontra

10
par ses observations des plaies de gladiateurs que les artères vivantes contiennent du sang. Il

supposa par contre que le sang va et vient du cœur dans un mouvement de flux et de reflux.

Cette théorie a prévalu dans les cercles médicaux jusqu’au XVIe siècle.

Au cours des siècles profondément chrétiens du Moyen-Âge européen, l'ambiance qui prévaut

n'est pas propice à la recherche scientifique. Dieu sait ce qui est le mieux, et il devrait le faire -

puisqu'il a tout créé. Lorsque des connaissances pratiques sont nécessaires, il existe des

autorités anciennes dont les conclusions sont acceptées sans conteste - Ptolémée dans le

domaine de l'astronomie, Galien pour les questions anatomiques. Quelques savants atypiques

s'intéressent à la recherche scientifique. Le frère franciscain du XIIIe siècle, Roger Bacon, est

l'exemple le plus souvent cité 22-24 , mais ses études portent aussi bien sur l'alchimie et

l'astrologie que sur l'optique et l'astronomie. Le scepticisme pratique requis pour la science

doit attendre la Renaissance.

Des progrès liés à l’imprimerie

En 1454, l’allemand Johannes Gutenberg invente les caractères mobiles. La mécanisation de

l’écriture permet une accélération des connaissances biologiques 25. Vers 1489, Léonard de

Vinci commence une série de dessins anatomiques 26,27. Pour la précision de l'observation, ils

sont bien en avance sur tout ce qui a été tenté auparavant. Au cours des vingt-cinq années

suivantes, il dissèque une trentaine de cadavres humains, dont beaucoup dans une morgue à
Rome - jusqu'à ce qu'en 1515, le pape Léon X lui ordonne d'arrêter. Ses dessins, au nombre de

750 environ, comprennent des études sur les structures osseuses, les muscles, les organes

internes, le cerveau et même la position du fœtus dans l'utérus. Ses études sur le cœur

suggèrent qu'il était sur le point de découvrir le concept de la circulation du sang. C'est une

coïncidence de grande valeur pour la biologie, dans laquelle l'observation est de première
importance, que le regain d'intérêt de la Renaissance pour la science coïncide avec l'invention

de l'imprimerie. Dès que les livres peuvent être publiés avec des illustrations gravées sur bois

parmi les textes imprimés, les naturalistes ont non seulement un nouveau lectorat important,

mais aussi la capacité de montrer ce qu'ils ont observé avec tant de soin. Le premier à utiliser

sérieusement cette possibilité est un botaniste, Otto Brunfels, dont les trois volumes Herbarum

vivae eicones sont publiés à Strasbourg entre 1530 et 1540 28. Un naturaliste français de cette

époque fournit un bon exemple de l'impulsion de la Renaissance à égaler et peut-être même à

11
surpasser les auteurs classiques. En 1546, Pierre Belon entreprend un voyage de deux ans

autour de la Méditerranée orientale dans le but précis de trouver et de représenter les animaux

et les plantes décrits par les écrivains anciens. A Bâle, en Suisse, Vésale publie en 1543 son

grand ouvrage - De humani corporis fabrica (La structure du corps humain) 29. Il s'agit d'un

ouvrage en sept volumes comprenant de nombreuses et magnifiques illustrations gravées sur

bois. Le livre connaît un succès immédiat, bien qu'il fasse naturellement l'envie des

traditionalistes qui suivent Galien. Les théories de Galien ont, après tout, le mérite évident de

l'ancienneté. Elles ont maintenant quelque 1400 ans. Ce souci de précision et d’exhaustivité

n’est pas sans rappeler les projets actuels que sont par exemple The Human Protein Atlas ou

le séquençage du génome humain 30,31. La continuité avec nos ancêtres est que la connaissance

biologique des éléments nous constituant nous permet de mieux appréhender et traiter les

maladies. Cela nécessite donc de visualiser l’infiniment petit.

La découverte du microscopique

Marcello Malpighi 32, professeur de médecine théorique à l'université de Bologne, a été le

pionnier de l'utilisation du microscope en biologie. Un soir de 1661, sur une colline près de

Bologne, il utilise le soleil couchant comme source de lumière, l'éclairant dans son objectif à

travers une fine section préparée d'un poumon de grenouille. Sur l'image agrandie, il est clair

que tout le sang est contenu dans de petits tubes. Malpighi devient ainsi le premier scientifique

à observer les capillaires, ces minuscules vaisseaux sanguins dans lesquels le sang circule à

travers la chair. Ils sont si fins, et si nombreux, que chacun de nos corps contient plus de 100

000 kilomètres de ces conduits microscopiques. Grâce à leur découverte, le chaînon manquant

dans la circulation sanguine de Harvey 33 a été mis en évidence. Car les capillaires sont

littéralement le lien par lequel le sang riche en oxygène des artères fournit d'abord son énergie
aux cellules du corps et trouve ensuite son chemin vers les veines pour être renvoyé au cœur.

Améliorant le travail de Malpighi avec le microscope, le chercheur néerlandais Anton van

Leeuwenhoek 34 atteint un très haut degré de précision et de clarté (certaines offrant un

grossissement de 300x). Avec de tels instruments, il est en mesure d'observer des phénomènes
jusqu'alors trop petits pour être vus 35,36 . En 1674, il est le premier scientifique à donner une

description précise des globules rouges. En 1677, il observe et représente des spermatozoïdes

12
dans le sperme d'un chien. En 1683, il fournit un dessin des animalculae (ou bactéries) présents

dans la salive et la plaque dentaire.

Les siècles suivants, l’invention du microscope a permis de perfectionner la connaissance du

vivant et notre environnement tout en étayant les classifications pour conduire à la biologie

moderne que nous connaissons. Cette histoire est encore incomplète.

1.2. Sciences omiques

Au XXe siècle, la biologie a connu des avancées extraordinaires avec la découverte de l’acide

désoxyribonucléique (ADN) par Watson et Cricks en 1953 37 . Le développement de la

Polymerase Chain Reaction (PCR) par Mullis en 1988 38 révolutionne la biologie moléculaire

et ouvre la voie aux sciences omiques. Le suffixe -omique fait référence à l’exhaustivité et le

haut débit de l’information moléculaire. Les sciences omiques peuvent être définies comme

une nouvelle branche de la biologie moléculaire qui repose sur des technologies permettant la

caractérisation et la quantification collectives de molécules à partir de différents échantillons

biologiques permettant ainsi d’appréhender la structure, fonction et dynamique d'un

organismes 39. Le terme « omics » est proposé pour la première fois par Wilkins, généticien de

formation en 1994 40. La nécessité de ces sciences nouvelles est apparue criante à l’issue du

projet génome humain (HGP) en 2001 41. Ce projet a permis une meilleure compréhension de
notre génome, mais également mis en exergue les rôles fondamentaux des autres processus de

régulation biologiques. Les sciences omiques nécessitent la gestion d’un très grand nombre de

données et sont indissociables de la bio-informatique et du big data. La Figure 3 illustre les

progrès techniques de biologie moléculaire réalisés au XXe siècle.

13
 Figure 3. Représentation des techniques de biologie moléculaire des différentes sciences

omiques. D’après Khan et al.42

Cette technologie a permis de déchiffrer la cellule, le tissu et l'organisme de manière holistique

autour d'un dogme central pour la détection des gènes (génomique), de l'ARNm

(transcriptomique), des protéines (protéomique), des métabolites (métabolomique) et des

profiles de méthylation (épigenomique) dans les échantillons. Plus récemment, la phénomique

a émergé comme nouvelle branche des sciences omiques qui s’intéresse à l’étude des
expressions phénotypiques des organismes. Ces technologies ont été explorées dans diverses

branches de la médecine et des sciences de la santé que ce soit au cours du dépistage, du

diagnostic ou du pronostic de nouvelles maladies. La recherche de biomarqueurs de ces

maladies est aussi l’un des piliers des sciences omiques 43,44. La meilleure compréhension de la
physiopathologie des maladies offre des perspectives thérapeutiques nouvelles et des

approches qui seront prédictives, préventives et personnalisées. Nous allons proposer un

aperçu des différentes technologies omiques disponibles ainsi que leurs applications

potentielles.

14
 1.2.1. Génomique

La génomique est l’étude complète de l’ensemble des informations héréditaires de chaque

organisme vivant qui sont nécessaires à son développement et à son fonctionnement 45. Le mot

"génome" est une conjonction des mots "gène" et "chromosome". Dans les années 1990,

l'ensemble du génome de Haemophilus influenzae 46 , un organisme vivant en liberté, a été

séquencé, ce qui représente une contribution importante au domaine de la génomique. L'étude

de la génomique caractérise et quantifie collectivement non seulement l'expression, mais aussi

son réseau de régulation qui lui est associé. En 1984, l’idée de décoder l’intégrer du génome

humain parait à portée de main du fait des nouvelles techniques de séquençage. Le projet du

génome humain (HGP) est lancé aux États-Unis. Le projet a recours aux méthodes de
séquençage dites par fusil de chasse (HSS) qui consiste à fragmenter l'ADN en plus petites

longueurs pour séquencer séparément chacun de ses fragments. Ensuite, les fragments sont

superposés pour recréer le génome 47 . Deux des premiers objectifs du HGP étaient la

préparation de cartes génétiques et physiologiques des génomes de l'homme et de la souris

avec des technologies améliorées de séquençage de l'ADN ; le séquençage de génomes plus

petits (levures et vers) a été pris comme modèle de test, avec un objectif projeté de transfert de

technologie. En 1999, une séquence pour le chromosome 22 humain (Chr 22) a été signalée, la

première ébauche du HGP étant détaillée en 2001. Le Chr 22 a été particulièrement choisi en
raison de sa petite taille et de son association avec diverses maladies. Trois milliards de lettres

du génome de l'Homo sapiens ont finalement été partagées dans le projet final du HGP financé

par les pouvoirs publics en 2003. Ce projet, bien que relativement complet, comportait

quelques lacunes majeures telles que des séquences discontinues. Une version plus actualisée

a finalement été complétée en 2006, qui a donné la séquence complète de Chr 1, le plus grand

chromosome du génome humain, contenant 8% du matériel génétique. Sept ans plus tard, le

génome de l'Homo neanderthalensis a été rendu public 48,49.

Le premier catalogue, connu sous le nom de "Index Marker Catalog", a détaillé la cartographie

complète avec une résolution de 10-15 cM. Le génome humain est essentiellement constitué
de 98,6 % d'ADN de ‘’rebut’’. Le HGP a permis la découverte de 1800 gènes responsables de

maladies, le premier étant associé à la maladie de Parkinson et, étonnamment, 850 sites étant

concernés par des maladies courantes. La plupart des points chauds se trouvaient à proximité

15
des séquences adjacentes plutôt qu'à l'intérieur de celles-ci. Plus de mille tests génétiques et

trois cents produits issus de la biotechnologie ont résulté des essais cliniques basés sur les

données du HGP 50,51. À ce jour, les génomes de 6070 eucaryotes, 145 357 procaryotes, 17 614
virus, 12 924 plasmides et 11 732 organites ont été séquencés et sont facilement accessibles

dans le domaine public dans la base de données sur le génome du National Center for

Biotechnology Information (NCBI). Depuis quelques années, la génomique, qui a été la

première sciences omique, a permis le développement de thérapeutique innovante en

s’inscrivant dans la médecine de précision 52,53.

1.2.2. Transcriptomique

La transcriptomique est la base la plus informative pour commencer un travail de recherche,

et avec l'avènement des nouvelles techniques à haut débit, il est devenu très facile et rapide de

générer un ensemble de données et d'informations 54. Le transcriptome constitue l'ensemble

des transcriptions présentes dans une cellule, y compris l'ARNm, les ARN non codants et les

petits ARN. La transcriptomique identifie la quantité d'ARN et la structure de la transcription

et quantifie les différents niveaux d'expression des transcriptions dans l'espace et dans le

temps au cours des différentes étapes du développement et dans des conditions

physiologiques variables. Elle fournit des informations sur la diversité, les ARN non codants

et la disposition des unités transcriptionnelles dans les régions codantes. L'analyse

transcriptomique a commencé par une technique primitive appelée EST, c'est-à-dire des

balises de séquences exprimées, suivie d'une autre technique appelée SAGE, c'est-à-dire une

analyse en série de l'expression des gènes, basée sur le séquençage de Sanger. Les techniques

EST et SAGE étaient laborieuses et permettaient de déterminer un petit ensemble de

transcriptions de manière aléatoire, donnant ainsi la moitié des informations sur le

transcriptome. Les années 1990 marquent la décennie révolutionnaire de la transcriptomique

avec l'introduction d'une innovation technologique de la technique contemporaine appelée

"microarray" 55,56 . Les microarrays analysent rapidement le transcriptome des grands

mammifères et ont été utiles pour le développement de médicaments et la recherche clinique

en analysant des milliers de gènes à partir de multiples échantillons. Le principal inconvénient

16
de cette technique est l'analyse des seules séquences connues et ne permet donc pas de détecter

les nouveaux transcrits. La dernière technologie en matière d'analyse du transcriptome est

l'ARN-Seq 57, basée sur une technologie de séquençage profond. Elle permet d'identifier le
gène et l'activité temporelle des gènes dans un génome. L'ARN-Seq in situ est une forme

avancée qui donne un aperçu d'une cellule individuelle dans un tissu fixe. De même,

l'information des gènes exprimés par une communauté microbienne est appelée

métatranscriptomique 58 . La métatranscriptomique fournit un profil fonctionnel du

microbiome dans des conditions physiologiques variables 58. Les données générées sont utiles

pour l'analyse de l'enrichissement et l'analyse phylogénétique des microbes. Plusieurs

pipelines bioinformatiques sont actuellement conçus ou en cours de réalisation pour l'analyse

de l'ensemble des données du métatranscriptome 59.

1.2.3. Protéomique

L'étude des protéines complètes d'une source et des techniques impliquées pour étudier ces

protéines et leurs interactions est le fondement de la protéomique 49. La carte tridimensionnelle

des protéines et de leurs interactions délimite leur importance et leur fonctionnement dans un

organisme. Ces études sont initiées au niveau des protéines en remontant parfois jusqu'à leurs
gènes. Le phénomène d'épissage alternatif chez les eucaryotes enrichit sélectivement la

diversité du protéome. La protéomique structurelle prend en considération la structure

tridimensionnelle des protéines, ce qui contribue à la procédure de conception rationnelle des

médicaments basée sur la structure. D'autre part, la protéomique fonctionnelle est largement
axée sur la compréhension de l'expression des protéines au niveau cellulaire, des

modifications des protéines, des interactions entre les protéines, de la signalisation et des

mécanismes de la maladie 60 . Ce domaine a pris de l'ampleur avec l'avènement de la

technologie, où différentes techniques telles que les rayons X, la RMN, la spectroscopie de

masse, la chromatographie Liquide Haute Performance (CLHP) et la PAGE bidimensionnelle

ont permis de générer d'énormes données expérimentales 61.

Il est difficile de suivre la quantité colossale de données expérimentales générées par les

différentes méthodes de détection des protéines. L'analyse effectuée par des procédures de

17
bio-informatique impliquant des algorithmes, des bases de données et des pipelines pour

l'analyse informatique permet une analyse plus rapide et plus précise en quelques jours. Grâce

aux bases de données et aux portails de ressources, la gestion, le stockage et le partage des

données ont permis aux chercheurs d'obtenir et de rassembler plus facilement les données, ce

qui accélère la recherche en protéomique. La protéomique, en combinaison avec d'autres

technologies complémentaires comme la génomique et la transcriptomique (une approche au

niveau des systèmes), a un énorme potentiel pour répondre à plusieurs questions sans réponse

en biologie 49.

1.2.4. Métabolomique

1.2.4.1. Définition

La métabolomique est définie comme l'étude des métabolites issus du métabolisme chez un

individu 49,62 . Les métabolites sont les composés organiques de petite taille participant au

métabolisme. Leur étude entre en interaction avec les autres sciences omiques bien que la
métabolomique émule directement l'activité biochimique de l'organisme et décrit donc le

mieux le phénotype moléculaire. Le métabolome d'un organisme, défini par l’ensemble des

métabolites le constituant, est complexe et dynamique, car les métabolites sont

continuellement absorbés et dégradés. Les études métabolomiques tentent de fournir un

instantané complet de l'état physiologique d'un organisme à un moment donné. De manière

générale, l'étude métabolomique peut être réalisée selon deux approches : l'approche ciblée et

l'approche non ciblée. Dans le cas de l'approche non ciblée, un certain nombre de métabolites

différents sont mesurés sans aucun biais d'échantillon, tandis que dans le cas de l'approche

ciblée, des ensembles définis de métabolites sont mesurés avec un objectif du problème à

traiter. Cependant, les étapes de ces deux approches sont communes 63,64.

1.2.4.2. Stratégies analytiques

Les études métabolomiques peuvent être divisées en cinq étapes 65,66 (Figure 4). La première

étape consiste à définir le plan de l'étude en tenant compte d'un certain nombre de facteurs

18
tels que la taille de l'échantillon, sa nature, la randomisation des analyses. Cette étape permet

de s'assurer que tous les facteurs importants sont pris en compte, notamment les métabolites

impliqués et leurs interactions potentielles qui pourraient biaiser leur analyse. La deuxième

étape est la préparation de l'échantillon, où la collecte, le stockage et la préparation de

l'échantillon ont lieu. Une attention particulière sera prise à la méthode d’extraction pour ne

pas dégrader les échantillons. De la même manière, la conservation devra s’effectuer à très

basse température (-80°C) pour ne pas dégrader les métabolites d’intérêt. Dans la troisième

étape, une technique analytique comme la spectroscopie de masse ou la RMN est utilisée pour

mesurer et quantifier les métabolites. La quatrième étape consiste à prétraiter les données pour

les analyser afin d'en extraire des inférences biologiques. La dernière étape est l'analyse des

données. Cette étape comprend l'application d'analyses statistiques aux données générées.

Figure 4. Représentation schématique des différentes étapes d’une étude métabolomique 67.

Une fois qu'une étude métabolomique est terminée, elle peut être soumise à diverses

applications, car il s'agit d'une approche très proche de la capture du phénotype d'un individu.

19
Cela en fait un outil précieux pour la recherche biomédicale. Des domaines avancés comme la

médecine personnalisée utilisent l'étude métabolomique pour le diagnostic et le pronostic

médical, utile pour l'identification de la maladie. La métabolomique est utile, car elle a la

capacité d'identifier et de caractériser différents métabolites, ce qui nous permet de mieux

comprendre les mécanismes de la maladie.

20
RÉFÉRENCES

1 Tsermoulas, G., Aidonis, A. & Flint, G. The skull of Chios: trepanation in Hippocratic
medicine. J Neurosurg 121, 328-332, doi:10.3171/2014.4.JNS131886 (2014).
2 Katsevman, G. A. & Brandmeir, N. J. Trepanation of the Outer Table as a Treatment
for Scalping Injuries: Historical Perspective and Modern Applications. World
Neurosurg 145, 301-305, doi:10.1016/j.wneu.2020.09.147 (2021).
3 Greenblatt, S. H. A history of Neurosurgery : In Its Scientific and Professional Contexts.
American Association of Neurological Surgeons edn, (1997).
4 Rifkinson-Mann, S. Cranial surgery in ancient Peru. Neurosurgery 23, 411-416,
doi:10.1227/00006123-198810000-00001 (1988).
5 Ormond, D. R. & Hadjipanayis, C. G. The history of neurosurgery and its relation to
the development and refinement of the frontotemporal craniotomy. Neurosurg Focus
36, E12, doi:10.3171/2014.2.FOCUS13548 (2014).
6 Faria, M. A., Jr. The death of Henry II of France. J Neurosurg 77, 964-969,
doi:10.3171/jns.1992.77.6.0964 (1992).
7 Ellis, H. Sir Victor Horsley: pioneer neurosurgeon, physiologist and medical politician.
Br J Hosp Med (Lond) 77, 304, doi:10.12968/hmed.2016.77.5.304 (2016).
8 Powell, M. P. Sir Victor Horsley at the birth of neurosurgery. Brain 139, 631-634,
doi:10.1093/brain/awv345 (2016).
9 Ellis, H. Harvey Cushing: father of modern neurosurgery. Br J Hosp Med (Lond) 75,
597, doi:10.12968/hmed.2014.75.10.597 (2014).
10 Brunon, J. [The origins of the French neurosurgery]. Neurochirurgie 62, 119-127,
doi:10.1016/j.neuchi.2016.01.008 (2016).
11 Leblanc, R. Wilder Penfield and the vascular hypothesis of focal epilepsy. J Neurosurg
131, 1947-1953, doi:10.3171/2018.8.JNS181990 (2019).
12 Lunsford, L. D. Lars Leksell. Notes at the side of a raconteur. Stereotact Funct
Neurosurg 67, 153-168, doi:10.1159/000099445 (1996).
13 Benabid, A. L. Jean Talairach (1911-2007). Stereotact Funct Neurosurg 86, 62-63,
doi:10.1159/000111141 (2008).
14 Steimle, R. Jean A. Sicard (1872-1929). J Neurol 260, 1946-1947, doi:10.1007/s00415-
012-6822-x (2013).
15 Petrik, V., Apok, V., Britton, J. A., Bell, B. A. & Papadopoulos, M. C. Godfrey
Hounsfield and the dawn of computed tomography. Neurosurgery 58, 780-787;
discussion 780-787, doi:10.1227/01.NEU.0000204309.91666.06 (2006).
16 Kriss, T. C. & Kriss, V. M. History of the operating microscope: from magnifying glass
to microneurosurgery. Neurosurgery 42, 899-907; discussion 907-898,
doi:10.1097/00006123-199804000-00116 (1998).
17 Baluška, F. e. & Sahi, V. P. in Plant Cell Monographs, 1 online resource (X, 290 pages
219 illustrations, 211 illustrations in color (Springer International Publishing : Imprint:
Springer,, Cham, 2018).
18 Sahi, V. P. Concepts in cell biology - history and evolution. 1st edition. edn, (Springer
Berlin Heidelberg, 2018).
19 Santacroce, L., Charitos, I. A., Topi, S. & Bottalico, L. The Alcmaeon's School of
Croton: Philosophy and Science. Open Access Maced J Med Sci 7, 500-503,
doi:10.3889/oamjms.2019.072 (2019).
20 Cortes, M. E., Rodriguez, J. L., Rodriguez, M. I., Del Rio, J. P. & Vigil, P. [Aristotle
Anniversary Year: remembering the contributions to medicine and human biology of

21
 this great polymath]. Rev Med Chil 144, 1498-1499, doi:10.4067/S0034-
98872016001100021 (2016).
21 Hajar, R. The air of history: early medicine to galen (part I). Heart Views 13, 120-128,
doi:10.4103/1995-705X.102164 (2012).
22 Sanford, F. "Roger Bacon Was Mistaken". Science 91, 384,
doi:10.1126/science.91.2364.384 (1940).
23 Sidebottom, E. Roger Bacon and the beginnings of experimental science in Britain. J R
Soc Med 106, 243-245, doi:10.1177/0141076813488482 (2013).
24 Williams, S. J. Roger Bacon and his edition of the pseudo-Aristotelian Secretum
secretorum. Speculum 69, 57-73, doi:10.2307/2864785 (1994).
25 Tompkins, R. K. The surgical journal of the future: how will it appear? Surg Today 36,
403-406, doi:10.1007/s00595-005-3171-8 (2006).
26 Jones, R. Leonardo da Vinci: anatomist. Br J Gen Pract 62, 319,
doi:10.3399/bjgp12X649241 (2012).
27 Kron, T. & Krishnan, P. Leonardo DaVinci's contributions to medical physics and
biomedical engineering: celebrating the life of a 'Polymath'. Australas Phys Eng Sci
Med 42, 403-405, doi:10.1007/s13246-019-00757-2 (2019).
28 Zunic, L., Skrbo, A. & Dobraca, A. Historical Contribution of Pharmaceutics to Botany
and Pharmacognosy Development. Mater Sociomed 29, 291-300,
doi:10.5455/msm.2017.29.291-300 (2017).
29 Toledo-Pereyra, L. H. De Humani Corporis Fabrica surgical revolution. J Invest Surg
21, 232-236, doi:10.1080/08941930802330830 (2008).
30 Ponten, F., Schwenk, J. M., Asplund, A. & Edqvist, P. H. The Human Protein Atlas as
a proteomic resource for biomarker discovery. J Intern Med 270, 428-446,
doi:10.1111/j.1365-2796.2011.02427.x (2011).
31 Uhlen, M. et al. Towards a knowledge-based Human Protein Atlas. Nat Biotechnol 28,
1248-1250, doi:10.1038/nbt1210-1248 (2010).
32 Pearce, J. M. Malpighi and the discovery of capillaries. Eur Neurol 58, 253-255,
doi:10.1159/000107974 (2007).
33 Androutsos, G., Karamanou, M. & Stefanadis, C. William Harvey (1578-1657):
discoverer of blood circulation. Hellenic J Cardiol 53, 6-9 (2012).
34 Gest, H. The discovery of microorganisms by Robert Hooke and Antoni Van
Leeuwenhoek, fellows of the Royal Society. Notes Rec R Soc Lond 58, 187-201,
doi:10.1098/rsnr.2004.0055 (2004).
35 Martins e Silva, J. From the discovery of the circulation of the blood to the first steps in
hemorheology: part 2. Rev Port Cardiol 28, 1405-1439 (2009).
36 Martins e Silva, J. From the discovery of the circulation of the blood to the first steps in
hemorheology: part 1. Rev Port Cardiol 28, 1245-1268 (2009).
37 Ghannam, J. Y., Wang, J. & Jan, A. in StatPearls (2021).
38 Schluger, N. W. & Rom, W. N. The polymerase chain reaction in the diagnosis and
evaluation of pulmonary infections. Am J Respir Crit Care Med 152, 11-16,
doi:10.1164/ajrccm.152.1.7599808 (1995).
39 in Evolution of Translational Omics: Lessons Learned and the Path Forward (eds C.
M. Micheel, S. J. Nass, & G. S. Omenn) (2012).
40 Manzoni, C. et al. Genome, transcriptome and proteome: the rise of omics data and their
integration in biomedical sciences. Briefings in bioinformatics 19, 286-302,
doi:10.1093/bib/bbw114 (2018).
41 Greenhalgh, T. The Human Genome Project. J R Soc Med 98, 545,
doi:10.1258/jrsm.98.12.545 (2005).

22
42 Khan, M. S. Multiomics for biomedical applications. Journal of applied bioanalysis 6,
97-106, doi:10.17145/jab.20.012 (2020).
43 Campanella, R. et al. Personalized and translational approach for malignant brain
tumors in the era of precision medicine: the strategic contribution of an experienced
neurosurgery laboratory in a modern neurosurgery and neuro-oncology department. J
Neurol Sci 417, 117083, doi:10.1016/j.jns.2020.117083 (2020).
44 Rai, V., Mukherjee, R., Ghosh, A. K., Routray, A. & Chakraborty, C. "Omics" in oral
cancer: New approaches for biomarker discovery. Arch Oral Biol 87, 15-34,
doi:10.1016/j.archoralbio.2017.12.003 (2018).
45 Roth, S. C. What is genomic medicine? J Med Libr Assoc 107, 442-448,
doi:10.5195/jmla.2019.604 (2019).
46 Fleischmann, R. D. et al. Whole-genome random sequencing and assembly of
Haemophilus influenzae Rd. Science 269, 496-512, doi:10.1126/science.7542800
(1995).
47 Sawicki, M. P., Samara, G., Hurwitz, M. & Passaro, E., Jr. Human Genome Project. Am
J Surg 165, 258-264, doi:10.1016/s0002-9610(05)80522-7 (1993).
48 Wilson, B. J. & Nicholls, S. G. The Human Genome Project, and recent advances in
personalized genomics. Risk Manag Healthc Policy 8, 9-20,
doi:10.2147/RMHP.S58728 (2015).
49 Arivaradarajan, P. & Misra, G. 1 online resource (XV, 148 pages 132 illustrations, 129
illustrations in color (Springer Singapore : Imprint: Springer,, Singapore, 2018).
50 Bailey, J. N., Pericak-Vance, M. A. & Haines, J. L. The impact of the human genome
project on complex disease. Genes (Basel) 5, 518-535, doi:10.3390/genes5030518
(2014).
51 Gonzaga-Jauregui, C., Lupski, J. R. & Gibbs, R. A. Human genome sequencing in
health and disease. Annu Rev Med 63, 35-61, doi:10.1146/annurev-med-051010-162644
(2012).
52 Aronson, S. J. & Rehm, H. L. Building the foundation for genomics in precision
medicine. Nature 526, 336-342, doi:10.1038/nature15816 (2015).
53 Roden, D. M. & Tyndale, R. F. Genomic medicine, precision medicine, personalized
medicine: what's in a name? Clin Pharmacol Ther 94, 169-172,
doi:10.1038/clpt.2013.101 (2013).
54 Marin de Evsikova, C., Raplee, I. D., Lockhart, J., Jaimes, G. & Evsikov, A. V. The
Transcriptomic Toolbox: Resources for Interpreting Large Gene Expression Data
within a Precision Medicine Context for Metabolic Disease Atherosclerosis. J Pers Med
9, doi:10.3390/jpm9020021 (2019).
55 Li, S., Todor, A. & Luo, R. Blood transcriptomics and metabolomics for personalized
medicine. Comput Struct Biotechnol J 14, 1-7, doi:10.1016/j.csbj.2015.10.005 (2016).
56 Lowe, R., Shirley, N., Bleackley, M., Dolan, S. & Shafee, T. Transcriptomics
technologies. PLoS Comput Biol 13, e1005457, doi:10.1371/journal.pcbi.1005457
(2017).
57 Wang, Z., Gerstein, M. & Snyder, M. RNA-Seq: a revolutionary tool for
transcriptomics. Nat Rev Genet 10, 57-63, doi:10.1038/nrg2484 (2009).
58 Aguiar-Pulido, V. et al. Metagenomics, Metatranscriptomics, and Metabolomics
Approaches for Microbiome Analysis. Evol Bioinform Online 12, 5-16,
doi:10.4137/EBO.S36436 (2016).
59 Gwak, H. J., Lee, S. J. & Rho, M. Application of computational approaches to analyze
metagenomic data. J Microbiol 59, 233-241, doi:10.1007/s12275-021-0632-8 (2021).

23
60 Hortin, G. L., Jortani, S. A., Ritchie, J. C., Jr., Valdes, R., Jr. & Chan, D. W. Proteomics:
a new diagnostic frontier. Clin Chem 52, 1218-1222,
doi:10.1373/clinchem.2006.067280 (2006).
61 Aslam, B., Basit, M., Nisar, M. A., Khurshid, M. & Rasool, M. H. Proteomics:
Technologies and Their Applications. J Chromatogr Sci 55, 182-196,
doi:10.1093/chromsci/bmw167 (2017).
62 Maramorosch, K., Shatkin, A. J. & Murphy, F. A. 1 online resource (viii, 273 p., 211 p
(Academic Press,, Oxford, UK ; Burlington, MA, 2007).
63 Padmanabhan, S. Handbook of pharmacogenomics and stratified medicines.
(Elsevier/AP, Academic Press is imprint of Elsevier, 2014).
64 Clish, C. B. Metabolomics: an emerging but powerful tool for precision medicine. Cold
Spring Harb Mol Case Stud 1, a000588, doi:10.1101/mcs.a000588 (2015).
65 Wang, J. H., Byun, J. & Pennathur, S. Analytical approaches to metabolomics and
applications to systems biology. Semin Nephrol 30, 500-511,
doi:10.1016/j.semnephrol.2010.07.007 (2010).
66 Lindon, J. C. The handbook of metabolic phenotyping. 1st edition. edn, (Elsevier, 2018).
67 Alonso, A., Marsal, S. & Julia, A. Analytical methods in untargeted metabolomics: state
of the art in 2015. Front Bioeng Biotechnol 3, 23, doi:10.3389/fbioe.2015.00023 (2015).

24
L’article I présente les grandes avancées dans l’histoire de la neurochirurgie et de la biologie notamment

avec l’arrivée des sciences omiques. Il a pour but de montrer l’évolution de la neurochirurgie au XXIe

siècle et comment cette discipline s’est appuyée sur les données de biologie moléculaire pour proposer

une chirurgie de précision. Le prisme de la neuro-oncologie a été choisi, car elle illustre au mieux

l’interdisciplinarité et l’intégration des données omiques dans la pratique de la médecine personnalisée.

Article I: V Gilard, S Derrey, S Marret, S Bekri, A Tebani. Precision Neurosurgery: A Path Forward.
Journal of Personalized Medicine. 2021;11(4):258.

25
Review 1

Precision Neurosurgery: A Path Forward 2

Vianney Gilard 1*, Stéphane Derrey 2, Stéphane Marret 3, Soumeya Bekri 4 and Abdellah Tebani 4 3

1 Normandie Univ, UNIROUEN, CHU Rouen, INSERM U1245, Department of Neurosurgery, 76000, Rouen, 4
France; vianney.gilard@chu-rouen.fr (V.G.) 5
2 Normandie Univ, UNIROUEN, CHU Rouen, INSERM U1073, Department of Neurosurgery, 76000, Rouen, 6
France; stephane.derrey@chu-rouen.fr (S.D.) 7
3 Normandie Univ, UNIROUEN, CHU Rouen, INSERM U1245, Department of Neonatology, Pediatric Intensive 8
Care, and Neuropediatrics, 76000 Rouen, France; stephane.marret@chu-rouen.fr (S.M.) 9
4 Normandie Univ, UNIROUEN, INSERM U1245, CHU Rouen, Department of Metabolic Biochemistry, 76000, 10
Rouen, France; abdellah.tebani@chu-rouen.fr (A.T.), soumeya.bekri@chu-rouen.fr (S.B.) 11
12
13
* Correspondence: vianney.gilard@chu-rouen.fr (V.G.) 14
15

16

17

Abstract: Since the inception of their profession, neurosurgeons have defined them- 18

selves as physicians with a surgical practice. Throughout time, neurosurgery always 19

Citation: Gilard et al. Precision Neu- take advantage of technological advances to provide better and safer care for patients. In 20
rosurgery: A Path Forward. J. Pers. the ongoing precision medicine surge that drives the patient-centric healthcare, neuro- 21
Med. 2021, 11, x.
surgery strives to effectively embrace this data-driven medicine era. Neuro-oncology 22
https://doi.org/10.3390/xxxxx
best illustrates this convergence between surgery and precision medicine with the ad- 23
Academic Editor: Firstname Last- vent of molecular profiling, imaging and data analytics. This convenient convergence 24
name
Received: date
paves the way for new preventive, diagnostic, prognostic and targeted therapeutic 25

Accepted: date perspectives. The prominent advances in healthcare and big data utterly challenge the 26
Published: date
medical community to deeply rethink current and future medical practice. This work 27

Publisher’s Note: MDPI stays neutral

provides a historical perspective on neurosurgery. It also discusses the impact of preci- 28

with regard to jurisdictional claims in sion medicine conceptual shifts on neurosurgery through neuro-oncology lenses. 29
published maps and institutional
30
affiliations.
Keywords: neuro-oncology; neurosurgery; precision medicine; personalized medicine; 31

omics; medical imaging 32

33
Copyright: © 2021 by the authors.
Submitted for possible open access 34
publication under the terms and
conditions of the Creative Commons
Attribution (CC BY) license
(http://creativecommons.org/licenses/ 1. Introduction 35
by/4.0/).
Neurosurgery is considered as the oldest surgical practice. It has been reported that 36
the first trepanations were performed in the Neolithic period and traces of human 37
trepanations dating back to 7000 years BC have been found with bone consolidation 38
around the drill hole, proof of survival of these first patients [1,2]. Despite its long his- 39
tory, neurosurgery has been able to rely on the great technological advances of its time. 40
In recent years, with the emergence of omics-based biological lenses, precision medi- 41
cine has emerged as one of the great advances in various fields of medicine [3,4]. Its 42

J. Pers. Med. 2021, 11, x. https://doi.org/10.3390/xxxxx www.mdpi.com/journal/jpm

 2 of 8

contribution can be found in the field of diagnosis with the development of biomarkers, but it also provides 43
prognostic elements and opens the way to new targeted therapies [5]. Neurosurgery is no exception and pre- 44
cision, or personalized medicine is present in our daily surgical practice. Precision medicine (PM) aims to 45
place the patient at the center of the healthcare pathway by integrating individual medical and biological data 46
while taking into account the great diversity between individuals. These new approaches invite us to reinvent 47
ourselves as neurosurgeons and to question our practice and profession. In line with the spirit and training of 48
our mentors, engineering, imaging and molecular biology techniques became more and more familiar com- 49
ponents of the operating room. They would not replace the fundamental principles of surgery, which are the 50
core of the operative indication, including a rigorous anatomy knowledge and flawless gestures hardly 51
earned through relentless and long practice in companionship. Nevertheless, these techniques have become 52
very useful in providing high quality and safe care to our patients. One of the fields that best illustrates the 53
contribution of new technologies and interdisciplinarity, in neurosurgery, is undoubtedly neuro-oncology. In 54
the following lines, we aim to give a historical perspective and discuss how neurosurgery is evolving towards 55
its next deep technological shift by fully embracing precision medicine particularly in the field of neu- 56
ro-oncology. 57

Neurosurgery: a short historical perspective 58

Even though the history of neurosurgery seems to begin in the Neolithic period, 5000 years later, it was the 59
pre-Inca civilizations that developed trepanning techniques. The indications of these trepanations were very 60
different from today and had a religious or magical tone [2,6]. Nevertheless, these procedures were also per- 61
formed in the context of cranial trauma, headaches, epilepsy and behavioral disorders. Later, the Incas have 62
raised to the rank of art their practice of cranial surgery. This is particularly highlighted by the diversity of 63
their metal-based instruments [7,8]. By looking at the survival rate following cranial surgery throughout his- 64
tory, Kushner et al. [2] showed that the Incas had a survival rate of 75% compared to 45% during the Ameri- 65
can Civil War. The main cause of mortality at the time was infection but the Incas applied ointments and met- 66
al prosthesis, ancestor of the current cranial vault plasties [7,8]. However, sole technical skills are not enough 67
to provide quality surgery. Understanding the brain physiology is, obviously, important to treat it more effec- 68
tively. The first Greek physicians helped to better understand these concepts. Hippocrates in the fourth cen- 69
tury BC [9,10], who considered as the father of medicine, explained in his omnia opera Hippocratis, that the 70
knowledge of physiology allowed for a better understanding of the gesture and that the technique alone was 71
not enough to guarantee high quality surgery. Galen, in the 2nd century A.D., elaborated on Hippocrates’ work. 72
In addition to perfecting anatomical knowledge through the observation of a large number of subjects and 73
taking care of injured gladiators [11]. Although ancient Greece practices, such as trepanations, were refined 74
over the centuries, endocranial surgery as such did not appeared until the 16th century. The cruelty of war al- 75
lowed significant medical leaps in the Middle Ages and the Renaissance. One of the leading figures of modern 76
surgery was undoubtedly Ambroise Paré (1510-1590) [12]. He perfected the trepanation techniques and vessel 77
ligation and introduced the notion of asepsis by using red iron cauterisation. A historical convergence be- 78
tween people’s lives and neurosurgery is the death of Henry II of Navarre, who died of an infection following 79
a craniocerebral wound during a tournament. Ambroise Paré, who performed the autopsy, described a cere- 80
bral empyema as the cause of death. Across the Channel, the pioneer of modern neurosurgery was Sir Victor 81
Horsley (1857-1924) [13,14]. He was a military surgeon working at the Queen Square Hospital in London. 82
With the support of his neurophysiologist colleagues, he hypothesized that epilepsy was a disease of the cer- 83
ebral cortex and that its resection could stop the development of epilepsy. Based on clinical observations, he 84
performed the first cerebral cortectomies on three patients. This procedure, although supported by imaging 85
and physiological investigations, is commonly performed today [6,7]. Harvey Cushing (1869-1939) [15] is the 86
first American neurosurgeon to operate on brain tumors. He has performed more than 2000 brain surgeries 87
and the description of his approaches led to improved survival of neurosurgical patients. Unfortunately, sur- 88
gery still carries the burden of its own history. From the 12th to the 18th century, barber-surgeons were dis- 89
tinct from physicians. The medical progress made in parallel with surgical practice only slightly enriched the 90
surgical know-how. The main pitfall of surgery is asepsis. As so was the case for neurosurgery. The birth of 91
modern post-war neurosurgery was possible through the progress of per operative conditions (asepsis and 92
anesthesia) and to a better understanding of the pathology and the progress in imaging and neurophysiology. 93
 3 of 8

Advances in neurosurgery: first steps towards precision surgery 94

Very early on, neurosurgeons defined themselves as physicians with a surgical practice. "Neurosurgery is 95
not only the art of removing tumors from the brain, but it is the means to learn in a precise way the functions 96
of the human brain” once said the French neurosurgeon Clovis Vincent (1879-1947) [16]. He was the first phy- 97
sician in charge of a dedicated surgical department. In Canada, Wilder Penfield (1891-1976) [17], also a neu- 98
rologist and a neurosurgeon, described cerebral somatotopy. He contributed, as did Clovis Vincent, a student 99
of Babinski (1857-1932), to the description of memory circuits. Based on his work, Harvey Cushing performed 100
the first intraoperative stimulations to explore the functions of the resected brain areas. These historic explo- 101
rations prefigured the development of awake surgery. One of the aspects of modern neurosurgery most re- 102
lated to these technical advances is certainly functional neurosurgery. It aims at adjusting certain functions of 103
the neuraxis when drug treatment is insufficient or not recommended. Classically, it includes deep brain 104
stimulation, epilepsy surgery, pain surgery and radiosurgery. The pioneers come from various backgrounds. 105
Lars Leksell (1907-1986) was a Swedish physicist and neurosurgeon [18]. He is credited with one of the first 106
stereotaxis frameworks and the invention of radiosurgery. In France, Jean Talaraich (1911-2007) was a psychi- 107
atrist and then a neurosurgeon [19]. With Jean Bancaud (1921-1993), a neurologist, they wrote the first stereo- 108
taxis atlas, which remains a reference work. They proposed a stereotaxis framework that allows for the im- 109
plantation of electrodes and brain biopsy with flawless accuracy. Surgical precision and accuracy are at the 110
core of precision medicine because it relies on the development of engineering techniques. These surgical 111
techniques were developed before the brain scanner (1970's). The anatomical landmarks were based on statis- 112
tical charts derived from hundreds of cadaveric observations and indirect landmarks. These charts represent a 113
precious stored, quantifiable, standardized and ready-to-use information to assist the surgical act. Conceptu- 114
ally, these charts are the precursor informational databases of the current anatomical, clinical and biological 115
big data. In 1972, Godfrey Hounsfield (1919-2004) ( [20] was awarded the Nobel Prize in Medicine for the dis- 116
covery of the first computerized tomography (CT) scan. It was a technological breakthrough for neurosurgery. 117
For the first time we could, safely and non-invasively, see the brain directly without opening the skull. The 118
scanner resolution progressively improved and the surge of Magnetic Resonance Imaging (MRI) in the early 119
1990's allowed for a non-irradiating cleaner image. In only three decades, MRI has gone from innovation to a 120
common practice. At the same time, progress in optical magnification and neuro-endoscopy techniques led to 121
advances in operative approaches, dissection management or hemostasis, and even new interventions (i.e., 122
endoscopic ventriculocysternostomy) [21]. At the dawn of this millennium, robotics has entered the operating 123
room. Surgical robots (for cerebral or spinal stereotaxis for the moment), navigation and even augmented re- 124
ality systems are used on a daily basis. More recently, this integration of human and technical progress has 125
been enriched by advances in molecular biology and is substantially contributing to the development of mul- 126
timodal techniques paving the way to precision surgery [22]. 127

Precision medicine in daily neuro-oncology 128

Neurosurgery has always cultivated interdisciplinarity as described above. Some of its prominent mem- 129
bers came from different backgrounds. Lars Leksell for example was a physicist by training. So was also Jean 130
Talairach. The different branches of neurosurgery were enriched by their respective backgrounds. With the 131
advent of molecular biology techniques, precision medicine is no longer an outsider of the operating room but 132
became an active component. The most striking example is certainly neurosurgical oncology. This is particu- 133
larly highlighted through glioblastoma. Indeed, its management has made significant advance [23,24]. Glio- 134
blastoma, although the most common primary malignant tumor of the central nervous system, still lacks cu- 135
rative strategies. Its prognosis remains poor, even in case of complete surgical resection with a median sur- 136
vival of 15 months [25]. One of the reasons for treatment resistance in glioblastoma is its heterogeneity. Thus, 137
the integration of molecular biology, genomics and more recently metabolomics and transcriptomics data are 138
of interest and has led to a better understanding of glioblastoma heterogeneity and biological plasticity [26,27]. 139
The digital revolution and the multimodal Big Data surge have made it possible to characterize this tumor in 140
order to better understand its genesis, its clinical heterogeneity, its functional effects and the reasons under- 141
lying its resistance to treatment [28,29]. The main goals of precision medicine are to better understand glio- 142
blastoma signaling biological networks and to define common and distinct pathways that could explain their 143
 4 of 8

clinical heterogeneity. This will allow for biomarker and prognostic discovery to open the way to targeted and 144
individualized therapies by targeting an alteration in these signaling pathways [30,31]. 145

Historically, the diagnosis of glioblastoma was based on the WHO classification originally derived from 146
histological criteria alone. The latest WHO classification of 2016 has integrated the latest molecular biology 147
data as well as prognostic elements and response to treatment related to their metabolic characteristics [32]. 148
Among the new features of the 2016 WHO classification is the mutation status (mutated or wild type) of the 149
IDH1 and IDH2 genes. It is postulated that IDH-mutated gliomas inhibit tumor suppressor oncogenes, induc- 150
ing the development of gliomas. Consequently, IDH-mutated status is an important prognostic factor for gli- 151
oblastomas [33]. The second component of classification of glial lesions is the presence or absence of a 1p/19q 152
codeletion. The 1p/19q co-deletion corresponds to the complete loss of 1p and 19q secondary to a transloca- 153
tion. It is associated with a better prognosis and a better response to chemotherapy for grades II and III glio- 154
mas [34]. Another recently described signaling pathway in glioblastoma is the telomerase-related senescence 155
escape pathways through mutation of the ATRX (alpha thalassemia/mental retardation syndrome X-linked) 156
gene or TERT (telomerase reverse transcriptase) promoter [27]. The ATRX gene was initially discovered in pa- 157
tients with X-linked mental retardation. The ATRX protein is a chromatin remodeling protein whose main 158
role is to maintain genomic stability. Thus, in case of ATRX variation, there is a telomerase lengthening al- 159
lowing glial cells to escape senescence. ATRX variation is a prognosis marker, rarely observed in IDH-wild 160
type glioblastoma. Variants in the TERT promoter or ATRX gene are present in 90% of mutated IDH gliomas 161
but are rarely associated. In recent series [35,36], it appears that TERT promoter gene mutation is associated 162
with poor outcomes in IDH wild type glioblastoma and favorable outcomes in IDH mutated glioblastoma. 163
Prognostic value of TERT promoter status is discussed due to its high frequency and confounding factors. For 164
example, the association with other molecular alteration such as O(6)-methylguanine-DNA methyltransferase 165
(MGMT) methylation is of interest. MGMT is an enzyme that repairs O6-methylguanine of DNA to guanine. 166
It protects the tumor cell against cytotoxic damage from alkylating chemotherapies, such as temozolomide 167
[37] [38]. As a consequence, methylation of the MGMT promoter leads to a decrease in MGMT expression and 168
in the repair of temozolomide-induced lesions. This status is a predictive marker for a better response to al- 169
kylating agents and an independent favorable prognostic marker in glioblastoma. Variants in genes encoding 170
histones, that are first identified in childhood glioma, are a promising avenue for a better understanding of 171
glioma signaling pathways [35]. Indeed, these variants are frequently found in various cancers and could 172
contribute to their epigenetic regulation. The heterogeneity of tumors and the large number of molecular bi- 173
ology discoveries yielded through omics highlights the need for a more personalized knowledge through in- 174
tegrative strategies to achieve precision medicine. 175

The search for a single biomarker for glioblastoma has been the focus of a sustained research for many 176
years [39]. Medical imaging and MRI spectroscopy, in particular, were main stream strategies in clinical prac- 177
tice [40]. It mainly consists in highlighting specific metabolic patterns in the tumor tissue. The development of 178
this technique required a large number of patients. The patient is, in this case, his own control because the 179
healthy brain parenchyma, usually contralateral, is used as a control tissue. Thus, MR spectroscopy in glio- 180
blastomas is characterized by a specific increase in choline/N-acetylaspartate and choline/creatinine ratios 181
[41]. In addition, an elevated peak of lactate and lipids as well as a decreased peak of myoinositol are reliable 182
information for the diagnosis of glioblastoma. With the advent of Big Data, radiological diagnosis continues to 183
progress. Recently, researchers at the University of Washington proposed an algorithm based on artificial in- 184
telligence and deep learning [42]. Their software has proven effective in diagnosing six categories of brain 185
tumours from conventional MRI sequences. This radiological marker cannot, currently, replace a histological 186
analysis. This highlights the urgent need for a sensitive blood biological signature. The use of large biobanks 187
has allowed for multi omics studies to propose biomarkers defined as a clinical, radiological and biological 188
signature allowing to define sub-groups of glioblastomas [43,44]. This refers to liquid biopsies whose aim is to 189
propose a diagnosis through a simple blood sample. Liquid biopsy has the advantage to prevent from inva- 190
sive procedure such as brain biopsy and can be repeated easily [29,45,46]. The development of such tech- 191
niques has its pitfalls even though the concept is still elegant. First, the heterogeneity of glial tumors makes it 192
difficult to find a single biomarker. Secondly, the blood-brain barrier is a powerful obstacle for the discovery 193
of a blood biomarker of an intraparenchymal processes. Finally, an ideal biomarker must allow for monitor- 194
ing of therapeutic efficacy or relapse. However, therapeutics can modify this biomarker. Thus, currently, de- 195
 5 of 8

spite sustained research, there is no cleared blood biomarker for the diagnosis and follow-up of glioblastoma 196
patients. Nevertheless, recent analyses have highlighted genes or metabolites of interest in the diagnosis of 197
glioblastoma [45,47,48]. Some of them are also prognostic indicators and might pave the way to potential tar- 198
geted therapies. 199

Towards a precision surgery 200

Despite the above-mentioned progress, are we going to see the birth of precision surgery? At the dawn the 201
current century, the main progress in neurosurgery has been based on intraoperative navigation to better lo- 202
cate the tumor from preoperative imaging [49]. Awake surgery has increased the resection rate by decreasing 203
the morbidity of the surgical procedure [50]. Nowadays, imaging is performed during surgery and allows for 204
resection assessment in real time [51]. Augmented reality techniques allow for the smooth integration of these 205
imaging data into the surgeon's microscope to directly visualize the lesion and the peri-tumor area. More re- 206
cently, intraoperative analysis of resected tissue has been made possible by the miniaturization of omics tech- 207
niques. Thus, some teams have tested the integration of a mass spectrometer in the cavitronic ultrasound as- 208
pirator [52,53]. This allows to limit the resection to the tissue invaded by the tumor. One of the difficulties of 209
this technology is to define the metabolites of interest. These are different depending on whether they concern 210
the fleshy portion of the tumor, a possible cystic portion, the contrast medium or the peri-tumoral zone. Sec- 211
ondly, the contribution of mass spectrometry cannot be free of the functional limits linked to the tumor resec- 212
tion. The surgeon's hand is often stopped by the risk of morbidity of a resection. Nevertheless, in situ meta- 213
bolic phenotyping is a definite contribution and is a complement to navigation and microscopy techniques. 214
The miniaturization of this advanced imaging technologies makes its clinical use possible and opens the door 215
to a new way of understanding the procedure guided by molecular biology data. Conceptually, surgical pre- 216
cision and technological development are only two faces of the same coin: precision surgery. One of its fun- 217
damental principles is to put the patient at the center of healthcare to take into account individual and specific 218
biological and clinical attributes [54-56]. To do so, it is necessary to differentiate the normal from the patho- 219
logical, which relies on the development and analysis of large biobanks integrating the various multimodal 220
data. Worthless to note that in the current technology and regulatory state-of-the-art, these technological 221
shifts cannot yet replace the surgical decision, which still solely guided by the surgeon experience and dexter- 222
ity. 223

Conclusion 224

Neurosurgery is considered as one of the oldest surgical disciplines. Since ancient times, it has been able to 225
embrace major technological advances for the benefit of patients. Its progress has been made possible by the 226
knowledge of anatomy, the development of radiology, microscopy, robotics and, recently, big data. Indeed, 227
the development of the data-driven medicine has considerably modified clinical practice, in neuro-oncology 228
in particular. This has allowed a better pathogenesis understanding and thus opened the way for innovative 229
personalized therapies and disease management. The future of neurosurgery, and medical practice in general, 230
lies in the smooth integration of multidisciplinary perspectives in both translational and clinical neurosurgery 231
to achieve the promise of precision surgery. 232

233
Author Contributions: Conceptualization, V.G. and A.T.; writing—original draft preparation, V.G., A.T.; writing—review 234
and editing, S.D., S.M. and S.B; supervision, S.B. and S.M. All authors have read and agreed to the published version of 235
the manuscript. 236

Funding: This research received no external funding. 237

Institutional Review Board Statement: Not applicable. 238

Informed Consent Statement: Not applicable. 239

240
Data Availability Statement: Not applicable. 241
242
Conflicts of Interest: The authors declare no conflict of interest. 243
 6 of 8

244

References 245
246
247
248
1. Greenblatt, S.H. A history of Neurosurgery : In Its Scientific and Professional Contexts, American Association of Neurological 249

Surgeons ed.; 1997; 10.1136/jnnp.64.2.284pp. 625. 250

2. Kushner, D.S.; Verano, J.W.; Titelbaum, A.R. Trepanation Procedures/Outcomes: Comparison of Prehistoric Peru with 251

Other Ancient, Medieval, and American Civil War Cranial Surgery. World Neurosurg 2018, 114, 245-251, 252

doi:10.1016/j.wneu.2018.03.143. 253

3. Khoury, M.J.; Iademarco, M.F.; Riley, W.T. Precision Public Health for the Era of Precision Medicine. Am J Prev Med 2016, 254

50, 398-401, doi:10.1016/j.amepre.2015.08.031. 255

4. Ginsburg, G.S.; Phillips, K.A. Precision Medicine: From Science To Value. Health Aff (Millwood) 2018, 37, 694-701, 256

doi:10.1377/hlthaff.2017.1624. 257

5. Sarma, A.; Calfee, C.S.; Ware, L.B. Biomarkers and Precision Medicine: State of the Art. Crit Care Clin 2020, 36, 155-165, 258

doi:10.1016/j.ccc.2019.08.012. 259

6. Katsevman, G.A.; Brandmeir, N.J. Trepanation of the Outer Table as a Treatment for Scalping Injuries: Historical 260

Perspective and Modern Applications. World Neurosurg 2021, 145, 301-305, doi:10.1016/j.wneu.2020.09.147. 261

7. Marino, R., Jr.; Gonzales-Portillo, M. Preconquest Peruvian neurosurgeons: a study of Inca and pre-Columbian 262

trephination and the art of medicine in ancient Peru. Neurosurgery 2000, 47, 940-950, 263

doi:10.1097/00006123-200010000-00028. 264

8. Rifkinson-Mann, S. Cranial surgery in ancient Peru. Neurosurgery 1988, 23, 411-416, doi:10.1227/00006123-198810000-00001. 265

9. Missios, S. Hippocrates, Galen, and the uses of trepanation in the ancient classical world. Neurosurg Focus 2007, 23, E11, 266

doi:10.3171/foc.2007.23.1.11. 267

10. Tsermoulas, G.; Aidonis, A.; Flint, G. The skull of Chios: trepanation in Hippocratic medicine. J Neurosurg 2014, 121, 268

328-332, doi:10.3171/2014.4.JNS131886. 269

11. Hajar, R. The air of history: early medicine to galen (part I). Heart Views 2012, 13, 120-128, doi:10.4103/1995-705X.102164. 270

12. Faria, M.A., Jr. The death of Henry II of France. J Neurosurg 1992, 77, 964-969, doi:10.3171/jns.1992.77.6.0964. 271

13. Ellis, H. Sir Victor Horsley: pioneer neurosurgeon, physiologist and medical politician. Br J Hosp Med (Lond) 2016, 77, 304, 272

doi:10.12968/hmed.2016.77.5.304. 273

14. Powell, M.P. Sir Victor Horsley at the birth of neurosurgery. Brain 2016, 139, 631-634, doi:10.1093/brain/awv345. 274

15. Ellis, H. Harvey Cushing: father of modern neurosurgery. Br J Hosp Med (Lond) 2014, 75, 597, 275

doi:10.12968/hmed.2014.75.10.597. 276

16. Brunon, J. [The origins of the French neurosurgery]. Neurochirurgie 2016, 62, 119-127, doi:10.1016/j.neuchi.2016.01.008. 277

17. Leblanc, R. Wilder Penfield and the vascular hypothesis of focal epilepsy. J Neurosurg 2019, 131, 1947-1953, 278

doi:10.3171/2018.8.JNS181990. 279

18. Lunsford, L.D. Lars Leksell. Notes at the side of a raconteur. Stereotact Funct Neurosurg 1996, 67, 153-168, 280

doi:10.1159/000099445. 281

19. Benabid, A.L. Jean Talairach (1911-2007). Stereotact Funct Neurosurg 2008, 86, 62-63, doi:10.1159/000111141. 282

20. Petrik, V.; Apok, V.; Britton, J.A.; Bell, B.A.; Papadopoulos, M.C. Godfrey Hounsfield and the dawn of computed 283

tomography. Neurosurgery 2006, 58, 780-787; discussion 780-787, doi:10.1227/01.NEU.0000204309.91666.06. 284

21. Shim, K.W.; Park, E.K.; Kim, D.S.; Choi, J.U. Neuroendoscopy: Current and Future Perspectives. J Korean Neurosurg Soc 285

2017, 60, 322-326, doi:10.3340/jkns.2017.0202.006. 286

 2 of 8

22. Dorr, L.D.; Deshmane, P. Precision surgery. Orthopedics 2009, 32, doi:10.3928/01477447-20090728-26. 287

23. Ene, C.I.; Holland, E.C. Personalized medicine for gliomas. Surg Neurol Int 2015, 6, S89-95, doi:10.4103/2152-7806.151351. 288

24. Jain, K.K. A Critical Overview of Targeted Therapies for Glioblastoma. Front Oncol 2018, 8, 419, 289

doi:10.3389/fonc.2018.00419. 290

25. Alexander, B.M.; Cloughesy, T.F. Adult Glioblastoma. J Clin Oncol 2017, 35, 2402-2409, doi:10.1200/JCO.2017.73.0119. 291

26. Kowalczyk, T.; Ciborowski, M.; Kisluk, J.; Kretowski, A.; Barbas, C. Mass spectrometry based proteomics and 292

metabolomics in personalized oncology. Biochim Biophys Acta Mol Basis Dis 2020, 1866, 165690, 293

doi:10.1016/j.bbadis.2020.165690. 294

27. Kristensen, B.W.; Priesterbach-Ackley, L.P.; Petersen, J.K.; Wesseling, P. Molecular pathology of tumors of the central 295

nervous system. Ann Oncol 2019, 30, 1265-1278, doi:10.1093/annonc/mdz164. 296

28. Peeken, J.C.; Goldberg, T.; Pyka, T.; Bernhofer, M.; Wiestler, B.; Kessel, K.A.; Tafti, P.D.; Nusslin, F.; Braun, A.E.; Zimmer, 297

C., et al. Combining multimodal imaging and treatment features improves machine learning-based prognostic assessment 298

in patients with glioblastoma multiforme. Cancer Med 2019, 8, 128-136, doi:10.1002/cam4.1908. 299

29. Shen, J.; Song, R.; Hodges, T.R.; Heimberger, A.B.; Zhao, H. Identification of metabolites in plasma for predicting survival 300

in glioblastoma. Mol Carcinog 2018, 57, 1078-1084, doi:10.1002/mc.22815. 301

30. Touat, M.; Idbaih, A.; Sanson, M.; Ligon, K.L. Glioblastoma targeted therapy: updated approaches from recent biological 302

insights. Ann Oncol 2017, 28, 1457-1472, doi:10.1093/annonc/mdx106. 303

31. Truman, J.P.; Garcia-Barros, M.; Obeid, L.M.; Hannun, Y.A. Evolving concepts in cancer therapy through targeting 304

sphingolipid metabolism. Biochim Biophys Acta 2014, 1841, 1174-1188, doi:10.1016/j.bbalip.2013.12.013. 305

32. Banan, R.; Hartmann, C. The new WHO 2016 classification of brain tumors-what neurosurgeons need to know. Acta 306

Neurochir (Wien) 2017, 159, 403-418, doi:10.1007/s00701-016-3062-3. 307

33. Barritault, M.; Picart, T.; Poncet, D.; Fenouil, T.; d'Hombres, A.; Gabut, M.; Guyotat, J.; Jouanneau, E.; Ameli, R.; Joubert, B., 308

et al. Avoiding New Biopsies by Identification of IDH1 and TERT Promoter Mutation in Nondiagnostic Biopsies From 309

Glioma Patients. Neurosurgery 2020, 87, E513-E519, doi:10.1093/neuros/nyaa025. 310

34. Ahmed, K.A.; Chinnaiyan, P. Applying metabolomics to understand the aggressive phenotype and identify novel 311

therapeutic targets in glioblastoma. Metabolites 2014, 4, 740-750, doi:10.3390/metabo4030740. 312

35. Aldape, K.; Zadeh, G.; Mansouri, S.; Reifenberger, G.; von Deimling, A. Glioblastoma: pathology, molecular mechanisms 313

and markers. Acta Neuropathol 2015, 129, 829-848, doi:10.1007/s00401-015-1432-1. 314

36. Binabaj, M.M.; Bahrami, A.; ShahidSales, S.; Joodi, M.; Joudi Mashhad, M.; Hassanian, S.M.; Anvari, K.; Avan, A. The 315

prognostic value of MGMT promoter methylation in glioblastoma: A meta-analysis of clinical trials. J Cell Physiol 2018, 233, 316

378-386, doi:10.1002/jcp.25896. 317

37. Hegi, M.E.; Diserens, A.C.; Gorlia, T.; Hamou, M.F.; de Tribolet, N.; Weller, M.; Kros, J.M.; Hainfellner, J.A.; Mason, W.; 318

Mariani, L., et al. MGMT gene silencing and benefit from temozolomide in glioblastoma. N Engl J Med 2005, 352, 997-1003, 319

doi:10.1056/NEJMoa043331. 320

38. Le, N.Q.K.; Do, D.T.; Chiu, F.Y.; Yapp, E.K.Y.; Yeh, H.Y.; Chen, C.Y. XGBoost Improves Classification of MGMT Promoter 321

Methylation Status in IDH1 Wildtype Glioblastoma. J Pers Med 2020, 10, doi:10.3390/jpm10030128. 322

39. Geng, F.; Guo, D. Lipid droplets, potential biomarker and metabolic target in glioblastoma. Intern Med Rev (Wash D C) 323

2017, 3, doi:10.18103/imr.v3i5.443. 324

40. Galanaud, D.; Nicoli, F.; Figarella-Branger, D.; Roche, P.; Confort-Gouny, S.; Le Fur, Y.; Cozzone, P.J. [MR spectroscopy of 325

brain tumors]. J Radiol 2006, 87, 822-832, doi:10.1016/s0221-0363(06)74090-2. 326

 3 of 8

41. Price, S.J.; Young, A.M.; Scotton, W.J.; Ching, J.; Mohsen, L.A.; Boonzaier, N.R.; Lupson, V.C.; Griffiths, J.R.; McLean, M.A.; 327

Larkin, T.J. Multimodal MRI can identify perfusion and metabolic changes in the invasive margin of glioblastomas. J Magn 328

Reson Imaging 2016, 43, 487-494, doi:10.1002/jmri.24996. 329

42. Chakrabarty, S.; Sotiras, A.; Milchenko, M.; LaMontagne, P.; Hileman, M.; Marcus, D. MRI-based Identification and 330

Classification of Major Intracranial Tumor Types Using a 3D Convolutional Neural Network: A Retrospective 331

Multi-Institutional Analysis. Radiology: Artificial Intelligence 2021, 10.1148/ryai.2021200301, doi:10.1148/ryai.2021200301. 332

43. Clavreul, A.; Soulard, G.; Lemee, J.M.; Rigot, M.; Fabbro-Peray, P.; Bauchet, L.; Figarella-Branger, D.; Menei, P.; network, 333

F.G.B. The French glioblastoma biobank (FGB): a national clinicobiological database. J Transl Med 2019, 17, 133, 334

doi:10.1186/s12967-019-1859-6. 335

44. Sasmita, A.O.; Wong, Y.P.; Ling, A.P.K. Biomarkers and therapeutic advances in glioblastoma multiforme. Asia Pac J Clin 336

Oncol 2018, 14, 40-51, doi:10.1111/ajco.12756. 337

45. Huang, J.; Weinstein, S.J.; Kitahara, C.M.; Karoly, E.D.; Sampson, J.N.; Albanes, D. A prospective study of serum 338

metabolites and glioma risk. Oncotarget 2017, 8, 70366-70377, doi:10.18632/oncotarget.19705. 339

46. Boire, A.; Brandsma, D.; Brastianos, P.K.; Le Rhun, E.; Ahluwalia, M.; Junck, L.; Glantz, M.; Groves, M.D.; Lee, E.Q.; Lin, 340

N., et al. Liquid biopsy in central nervous system metastases: a RANO review and proposals for clinical applications. 341

Neuro Oncol 2019, 21, 571-584, doi:10.1093/neuonc/noz012. 342

47. Marziali, G.; Signore, M.; Buccarelli, M.; Grande, S.; Palma, A.; Biffoni, M.; Rosi, A.; D'Alessandris, Q.G.; Martini, M.; 343

Larocca, L.M., et al. Metabolic/Proteomic Signature Defines Two Glioblastoma Subtypes With Different Clinical Outcome. 344

Sci Rep 2016, 6, 21557, doi:10.1038/srep21557. 345

48. Mitropoulos, K.; Katsila, T.; Patrinos, G.P.; Pampalakis, G. Multi-Omics for Biomarker Discovery and Target Validation in 346

Biofluids for Amyotrophic Lateral Sclerosis Diagnosis. OMICS 2018, 22, 52-64, doi:10.1089/omi.2017.0183. 347

49. Schulz, C.; Waldeck, S.; Mauer, U.M. Intraoperative image guidance in neurosurgery: development, current indications, 348

and future trends. Radiol Res Pract 2012, 2012, 197364, doi:10.1155/2012/197364. 349

50. Eseonu, C.I.; Rincon-Torroella, J.; ReFaey, K.; Lee, Y.M.; Nangiana, J.; Vivas-Buitrago, T.; Quinones-Hinojosa, A. Awake 350

Craniotomy vs Craniotomy Under General Anesthesia for Perirolandic Gliomas: Evaluating Perioperative Complications 351

and Extent of Resection. Neurosurgery 2017, 81, 481-489, doi:10.1093/neuros/nyx023. 352

51. Kiris, T.; Arica, O. Impact of a low-field intraoperative MRI on the surgical results for high-grade gliomas. Acta Neurochir 353

Suppl 2011, 109, 55-59, doi:10.1007/978-3-211-99651-5_9. 354

52. Schafer, K.C.; Balog, J.; Szaniszlo, T.; Szalay, D.; Mezey, G.; Denes, J.; Bognar, L.; Oertel, M.; Takats, Z. Real time analysis 355

of brain tissue by direct combination of ultrasonic surgical aspiration and sonic spray mass spectrometry. Anal Chem 2011, 356

83, 7729-7735, doi:10.1021/ac201251s. 357

53. Ifa, D.R.; Eberlin, L.S. Ambient Ionization Mass Spectrometry for Cancer Diagnosis and Surgical Margin Evaluation. Clin 358

Chem 2016, 62, 111-123, doi:10.1373/clinchem.2014.237172. 359

54. Deng, A.; Xiong, R.; Zeng, C. [Application of precision medicine in the field of surgery]. Nan Fang Yi Ke Da Xue Xue Bao 360

2015, 35, 1662-1665. 361

55. Zheng, F. The March Towards Precision Surgery. Am J Surg 2020, 10.1016/j.amjsurg.2020.11.020, 362

doi:10.1016/j.amjsurg.2020.11.020. 363

56. Lidsky, M.E.; D'Angelica, M.I. An outlook on precision surgery. Eur J Surg Oncol 2017, 43, 853-855, 364

doi:10.1016/j.ejso.2016.11.014. 365
366
CHAPITRE II : LES GLIOBLASTOMES

1.1 Épidémiologie

Le glioblastome est la tumeur cérébrale maligne primaire la plus fréquente chez l'adulte
1
. Au cours des 20 dernières années, une incidence croissante des glioblastomes a été observée

en raison de l'augmentation de l'espérance de vie de la population générale et d'un meilleur

accès à des outils de diagnostic plus précis tels que l'IRM 2-4. Dans un article récent basé sur un

registre prospectif des tumeurs du système nerveux central (SNC) aux États-Unis 5, les

glioblastomes représentent 54% des gliomes du cerveau avec une incidence annuelle de 3,19

pour 100 000. Le pic d'incidence se situe dans la 6ème décennie pour les glioblastomes de type

sauvage IDH et plus tôt, vers la 4ème-5ème décennie, pour les glioblastomes IDH mutés 6. Malgré

de nombreux essais thérapeutiques 7-9 et des progrès dans la gestion de ces tumeurs, la survie

médiane reste faible, environ 14 à 20 mois, avec un taux de survie à cinq ans de 5% dépendant

de l'âge au moment du diagnostic, des caractéristiques moléculaires et de la gestion 10. Nous

passons ici en revue les données actuelles concernant la présentation clinique, le diagnostic et

la prise en charge des patients présentant un glioblastome.

Les facteurs de risque d'apparition de glioblastomes sont encore inconnus et les études

traitant de cette question manquent souvent de puissance. L'exposition aux rayonnements

ionisants pour le traitement d'une tumeur maligne pendant la petite enfance (12 ans) est un

facteur de risque très rare pour l'apparition d'un gliome. Le risque de développer une tumeur

cérébrale après radiothérapie est accru si les radiations se produisent à un âge plus jeune (<5

ans) et semblent liées au volume et à la dose, mais sans seuil clairement établi 11. L'incidence

accrue des glioblastomes 12,13 soulève la question des facteurs de risque environnementaux. Une

méta-analyse a établi un lien entre l'utilisation d'un téléphone portable et l'apparition de

14
glioblastomes . Cependant, ces résultats sont discordants et ont été contestés dans d'autres

études 15. Le rôle de l'exposition au tabagisme ou à des agents cancérigènes a été étudié sans

34
qu'il y ait d'association prouvée avec le glioblastome 16,17. Dans de très rares cas (<1%), il existe

une prédisposition génétique au développement de glioblastomes chez les patients atteints des

syndromes de Lynch, de Turcot de type 1 ou de Li Fraumeni 18.

1.2 Présentation clinique

La présentation clinique dépend de la localisation et de la taille de la tumeur au moment

du diagnostic. La présentation la plus courante au moment du diagnostic est le mal de tête et/ou

la nausée dans le contexte d'une tumeur de grande taille ou d'un œdème important. Les

symptômes liés à l'hypertension intracrânienne représentent 30 % des signes cliniques, suivis

du déficit moteur (20 %), de la perte de poids et de l'état du corps (17 %), de la confusion (15

%) et du déficit visuel ou de la parole (13 %) 19. L'épilepsie n'est pas rare (15 à 20 %) et est

facilement contrôlée par des thérapies anticonvulsives. L'épilepsie est associée à de meilleurs

résultats, probablement en raison de la localisation corticale des glioblastomes présentant une

crise 20. Ces symptômes sont souvent associés et conduisent à un diagnostic dans les semaines

ou les mois qui suivent leur apparition. Les glioblastomes sont localisés dans l'espace supra-

tentorial dans plus de 85% des cas, ce qui explique les symptômes mentionnés ci-dessus, et très

rarement dans le tronc cérébral et la moelle épinière (<5% chacun) ou dans le cervelet (<3%).

Jusqu'à 25 % des glioblastomes se produisent dans le lobe frontal 21, le plus grand lobe cérébral,

et sont donc responsables de troubles de l'humeur et de l'exécution chez 15 % des patients 22.

1.3 Diagnostic radiologique

Les caractéristiques typiques de l'imagerie par résonance magnétique (IRM) des

glioblastomes sont bien connues 21,23,24. Elles consistent en des lésions intraparenchymateuses

infiltrantes et hétérogènes qui surgissent et se propagent à partir de la substance blanche.

L'atteinte du corps calleux est fréquente. Les glioblastomes sont mal circonscrits et présentent

35
une augmentation du contraste à leur marge en signe de perturbation de la barrière

hématoencéphalique (Figure 1A et B). Le centre de la lésion est hypointense sur les images

pondérées en T2 en raison de la nécrose et la lésion est entourée d'un œdème cérébral qui

apparaît hyperintense sur les images pondérées en T2 et les images FLAIR (fluid-attenuated

inversion recovery). L’analyse morphologique présente des limites, car ces lésions peuvent

avoir une présentation très hétérogène sur les séquences conventionnelles (Figure 2). Des

techniques d'IRM multimodales plus récentes, telles que les séquences de diffusion/perfusion,

ont fourni des informations supplémentaires sur les caractéristiques de la lésion elle-même et

ont permis un diagnostic plus précis (Figure 3). L'imagerie pondérée par perfusion (PWI) révèle

une augmentation du flux sanguin cérébral correspondant à la néoangiogénèse et à la

perturbation de la barrière hématoencéphalique. La spectroscopie IRM (Figure 4) dans les

glioblastomes est caractérisée par une augmentation spécifique des rapports choline/N-

acétylaspartate et choline/créatinine. En outre, un pic élevé de lactate et de lipides ainsi qu'un

pic réduit de myoinositol sont des données fiables pour le diagnostic du glioblastome 25 et sont

utiles pour distinguer les glioblastomes des métastases, des lymphomes et des abcès cérébraux.

Comme le suggèrent certains auteurs, ces explorations sont également utiles pour évaluer le

degré d'envahissement péritumoral et peuvent servir de guide pour la biopsie 26 et pour suivre

l'évolution de la maladie après le traitement.

36
Figure 1. IRM cérébrale en coupes axiales en séquence FLAIR (A) puis en pondération T1

avec injection de gadolinium (B) chez un patient présentant un GBM du carrefour droit.

Figure 2. IRM cérébrale en coupes axiales en pondération T1 avec injection de gadolinium

chez différents patients présentant un GBM pour illustrer leur hétérogénéité morphologique.

37
Figure 3. IRM cérébrale en coupes axiales en séquence de perfusion chez un patient présentant

un GBM du carrefour droit. A noter l’augmentation du volume sanguin cérébral au sein de la

lésion.

38
Figure 4 : IRM cérébrale avec séquences de spectrométrie révélant les pics normaux du cerveau

sain (4A), la présence de lactates et de lipides au sein de la lésion témoin de la nécrose (4B)

ainsi qu’une inversion du rapport choline/NAA (4C), caractéristique d’une lésion gliale de haut

grade.

1.4 Histologie et biologie moléculaire

Le diagnostic du glioblastome se fait facilement sur la base de résections chirurgicales

ou de prélèvements de biopsie. Les glioblastomes sont des gliomes de haut grade, de grade IV

selon la classification de l'Organisation mondiale de la santé (OMS) des tumeurs du système

nerveux central 27. Les glioblastomes sont composés de cellules tumorales peu différenciées,
28
souvent pléomorphes, avec une différenciation astrocytaire prédominante . Les

39
caractéristiques histopathologiques comprennent l'atypie nucléaire, le pléomorphisme

cellulaire, une forte activité mitotique, la thrombose vasculaire, la prolifération microvasculaire

et la nécrose (Figure 5).

Figure 5. Vue microscopique d’un glioblastome IDH muté.

L'immunohistochimie et les techniques moléculaires complémentaires sont désormais

couramment utilisées à des fins de diagnostic et de pronostic (Figure 6). Ces études

complémentaires consistent en la recherche de l'isocitrate déshydrogénase (IDH) et du statut

ATRX 34, la codéletion du gène 1p/19q 29, l'hyperméthylation du promoteur du gène de la O6-

méthylguanine-ADN méthyltransférase (MGMT) 30, les mutations du gène p53 31, les mutations

du TERTp et les altérations du gène du récepteur du facteur de croissance épidermique (EGFR)

32
.

40
 Figure 6. Schéma représentant l’arbre diagnostique de la classification OMS 2016

d’après Louis et al. 27.

1.5 Prise en charge chirurgicale

Dans la mesure du possible, la première étape consiste en une résection chirurgicale

macroscopique complète. Les données de la littérature suggèrent qu'une résection > 90 % de

l'augmentation du contraste de la lésion chez les patients sans comorbidité améliore le résultat
33,34
du patient au moment du diagnostic et de la récidive . La résection chirurgicale est

généralement proposée aux patients de moins de 70 ans en bon état (échelle de Karnofsky >70)
35
et ayant une tumeur accessible à l'ablation complète. Sinon, un débulquage chirurgical ou

une biopsie stéréotaxique est pratiqué pour confirmer le diagnostic avant le traitement adjuvant
36
. En raison de l'importance d'une résection complète sur la survie, des progrès ont été réalisés

41
 37,38
dans les techniques chirurgicales telles que la craniotomie en état de veille ou le

neuromonitoring 39 pour améliorer la qualité de la résection et prévenir les déficits ultérieurs.

En outre, la chirurgie guidée par fluorescence (Figure 7) a été développée pour guider la
40,41
résection avec de meilleurs résultats sur la résection et la survie sans progression . Plus

récemment, les dispositifs de résection ont évolué avec l'utilisation de la thérapie thermique
42
interstitielle au laser (LITT) , qui permet une approche percutanée moins invasive grâce à

l'insertion d'une fibre optique. La lésion thermique générée induit une nécrose tumorale 43. Une

autre nouveauté dans la résection chirurgicale des tumeurs cérébrales est la surveillance

peropératoire des métabolites de la tumeur par spectrométrie de masse. Le contenu cellulaire

est analysé, ce qui permet une délimitation moléculaire précise des marges de la tumeur et donc

une résection optimale de la tumeur 44,45.

Figure 7. Visualisation peropératoire d’une exérèse chirurgicale ; vue microscopique (7A) et

sous fluorescence au 5-ALA (7B).

1.6 Thérapeutiques médicales

La norme de soins pour les patients âgés de moins de 70 ans repose sur la radiothérapie

(RT) avec témazolomide concomitant. Il a été prouvé que ce protocole améliore la survie

42
globale dans un vaste essai randomisé de phase III 46. La radiothérapie est administrée pendant

une période de six semaines avec une dose totale de 60 grays (Figure 8). Le témazolomide est

un agent alkylant administré quotidiennement pendant la RT et ensuite, pendant six cycles de

cinq jours consécutifs par mois, un mois après la fin de la RT. L'absence de promoteur MGMT
47
hypermethylé est un facteur de pronostic de l'efficacité du témozolomide. La RT seule peut

donc être proposée. Le protocole de traitement proposé par Stupp 46 permet d'augmenter le taux

de survie moyen de 12,1 mois en utilisant la RT seule à 14,6 mois, et le taux de survie à deux

ans de 8 à 26% avec le témozolomide concomitant.

Figure 8. Illustration du traitement d’un patient atteint d’un glioblastome selon le protocole

STUPP. D’après Lemée et al 48.

Cet essai contrôlé randomisé n'a pas inclus de patients de plus de 70 ans. Dans cette

dernière population, il n'existe pas de norme de soins claire et le traitement dépend de l'état

général du patient. La RT seule (54 grays) a été proposée avec un impact positif sur la survie

(29,1 semaines contre 16,9 semaines chez les patients recevant uniquement des soins de

soutien) et sans altération de la qualité de vie 10. Ces résultats ont été validés chez des patients,

43
 49
dont le statut de performance Karnofsky (KPS) > 60. Plus récemment, certains auteurs ont

montré que la RT de courte durée plus le témozolomide était associée à une survie plus longue

(9,3 contre 7,6 mois) chez les patients plus âgés (> 65 ans). Chez les patients en mauvaise santé

générale, des soins de soutien peuvent être proposés pour préserver la qualité de vie avec une

durée d'hospitalisation la plus courte possible 49.

L'évolution de la maladie est évaluée par IRM du cerveau tous les 2 à 3 mois selon

l'évaluation de la réponse en neuro-oncologie (RANO) selon les critères suivants (Figure 9) 50.

Figure 9. Critères IRM d'évaluation de la réponse en neuro-oncologie. D’après Wen et al. 50.

Au moment de la récidive, il n'y a pas de norme de soins. Les principaux déterminants

des propositions de traitement sont l'état général du patient et les traitements administrés

précédemment. Une deuxième intervention chirurgicale peut être proposée chez les jeunes

patients dont le KPS est préservé. Il a été démontré que cette stratégie est associée à une survie

plus longue chez certains patients 51 (14 mois contre 22 mois de survie globale chez les patients

ayant subi une seconde chirurgie au moment de la récidive). Quatre-vingts pour cent des

44
patients n'ont pas eu besoin de réadaptation après une deuxième chirurgie. Chez des patients

sélectionnés, l'utilisation de plaquettes intracavitaires de carmustine (BCNU) a été proposée

lors de la première chirurgie ou en cas de récidive 52,53. L'efficacité et les effets indésirables de
54
cette thérapie restent un sujet de débat . En l'absence d'une résection possible sûre, une

chimiothérapie de deuxième ligne peut être proposée, comprenant des nitrosourées, du

témazolomide ou des médicaments antiangiogéniques tels que le bevacizumab, mais sans

résultats clairs en termes de bénéfice à ce jour et de mauvais résultats avec un taux de survie

global moyen de 6 mois à partir de la récurrence 55.

1.7 Nouvelles thérapies et perspectives

En raison de l'issue défavorable chez les patients atteints de glioblastome et de la

fréquence élevée de cette maladie, des thérapies innovantes sont testées dans différents essais
56-58
contrôlés randomisés . Avec le développement d'une meilleure compréhension des voies

moléculaires déclenchant la croissance des glioblastomes, l'approche traditionnelle de la

thérapie antitumorale est progressivement complétée par une approche plus personnalisée 59,60.

Les calendriers de traitement ont été repensés, ainsi que les médicaments eux-mêmes. Deux

inconvénients majeurs consistent en la difficulté pour la plupart des médicaments de passer à

travers la barrière hématoencéphalique (BHE) et de cibler les cellules tumorales en raison de la

présence de vaisseaux anormaux et de nécrose, ce qui empêche les médicaments d'être délivrés

à une concentration appropriée. Certaines techniques émergentes ont été proposées pour

améliorer la distribution de la thérapie antitumorale, notamment la conjugaison de médicaments

avec des protéines pour faciliter le mouvement à travers la BHE et cibler spécifiquement la

tumeur 61, l'utilisation de la délivrance par convection améliorée consistant en l'administration

62
directe à l'intérieur de la tumeur via un cathéter et l'utilisation de nanoparticules.

L'augmentation de la perméabilité de la BHE pendant l'administration de la chimiothérapie par

45
 63
des ultrasons focalisés est également testée . Entre-temps, les approches d'immunothérapie

sont connues depuis longtemps, avec des résultats prometteurs dans de nombreux cancers tels

que le mélanome, mais décevants chez les patients atteints de gliomes 64. La vaccination contre

le cancer a récemment été proposée en référence à l'activation d'une réponse immunitaire contre

les antigènes tumoraux. Ces nouvelles technologies ont été appliquées pour le traitement des
64-66
glioblastomes avec différents vecteurs . Deux modalités ont été testées : les vaccins

peptidiques ciblant les protéines EGFR, IDH1 ou de choc thermique, et les vaccins à base de

cellules consistant en l'injection de cellules modifiées ex vivo, principalement des cellules

dendritiques. Malgré des résultats encourageants dans les modèles animaux en termes de

contrôle de la maladie, la vaccination contre le cancer dans les glioblastomes n'a pas encore

prouvé son efficacité sur la survie globale dans les études de phase III. Stupp et al. ont

récemment mis au point une nouvelle modalité thérapeutique dans le traitement des

glioblastomes récurrents, consistant à délivrer localement des champs électriques de faible

intensité par l'intermédiaire d'un transducteur non invasif 57. Le dispositif (NovoTTF-100A) a

été testé dans une étude de phase III et randomisé avec une chimiothérapie active comme bras

alternatif. La survie globale a été similaire dans les deux bras (6 mois) avec moins d'effets

indésirables dans le groupe NovoTTF-100A et une meilleure qualité de vie.

46
RÉFÉRENCES

1 Bauchet, L. & Ostrom, Q. T. Epidemiology and Molecular Epidemiology. Neurosurg

Clin N Am 30, 1-16, doi:10.1016/j.nec.2018.08.010 (2019).
2 Negendank, W. G. et al. Proton magnetic resonance spectroscopy in patients with glial
tumors: a multicenter study. J Neurosurg 84, 449-458,
doi:10.3171/jns.1996.84.3.0449 (1996).
3 Morgan, L. L. The epidemiology of glioma in adults: a "state of the science" review.
Neuro Oncol 17, 623-624, doi:10.1093/neuonc/nou358 (2015).
4 Kowalczyk, T., Ciborowski, M., Kisluk, J., Kretowski, A. & Barbas, C. Mass
spectrometry based proteomics and metabolomics in personalized oncology. Biochim
Biophys Acta Mol Basis Dis 1866, 165690, doi:10.1016/j.bbadis.2020.165690 (2020).
5 Dolecek, T. A., Propp, J. M., Stroup, N. E. & Kruchko, C. CBTRUS statistical report:
primary brain and central nervous system tumors diagnosed in the United States in
2005-2009. Neuro Oncol 14 Suppl 5, v1-49, doi:10.1093/neuonc/nos218 (2012).
6 Ostrom, Q. T. et al. The epidemiology of glioma in adults: a "state of the science"
review. Neuro Oncol 16, 896-913, doi:10.1093/neuonc/nou087 (2014).
7 Lim, M., Xia, Y., Bettegowda, C. & Weller, M. Current state of immunotherapy for
glioblastoma. Nat Rev Clin Oncol 15, 422-442, doi:10.1038/s41571-018-0003-5
(2018).
8 Touat, M., Idbaih, A., Sanson, M. & Ligon, K. L. Glioblastoma targeted therapy:
updated approaches from recent biological insights. Ann Oncol 28, 1457-1472,
doi:10.1093/annonc/mdx106 (2017).
9 Malkki, H. Trial Watch: Glioblastoma vaccine therapy disappointment in Phase III
trial. Nat Rev Neurol 12, 190, doi:10.1038/nrneurol.2016.38 (2016).
10 Delgado-Lopez, P. D. & Corrales-Garcia, E. M. Survival in glioblastoma: a review on
the impact of treatment modalities. Clin Transl Oncol 18, 1062-1071,
doi:10.1007/s12094-016-1497-x (2016).
11 Connelly, J. M. & Malkin, M. G. Environmental risk factors for brain tumors. Curr
Neurol Neurosci Rep 7, 208-214, doi:10.1007/s11910-007-0032-4 (2007).
12 Philips, A., Henshaw, D. L., Lamburn, G. & O'Carroll, M. J. Brain Tumours: Rise in
Glioblastoma Multiforme Incidence in England 1995-2015 Suggests an Adverse
Environmental or Lifestyle Factor. J Environ Public Health 2018, 7910754,
doi:10.1155/2018/7910754 (2018).
13 Dobes, M. et al. Increasing incidence of glioblastoma multiforme and meningioma,
and decreasing incidence of Schwannoma (2000-2008): Findings of a multicenter
Australian study. Surg Neurol Int 2, 176, doi:10.4103/2152-7806.90696 (2011).
14 Karipidis, K. et al. Mobile phone use and incidence of brain tumour histological types,
grading or anatomical location: a population-based ecological study. BMJ Open 8,
e024489, doi:10.1136/bmjopen-2018-024489 (2018).
15 Repacholi, M. H. et al. Systematic review of wireless phone use and brain cancer and
other head tumors. Bioelectromagnetics 33, 187-206, doi:10.1002/bem.20716 (2012).
16 Benke, G. et al. Occupational solvent exposure and risk of glioma in the INTEROCC
study. Br J Cancer 117, 1246-1254, doi:10.1038/bjc.2017.285 (2017).
17 Parent, M. E. et al. Lifetime occupational exposure to metals and welding fumes, and
risk of glioma: a 7-country population-based case-control study. Environ Health 16,
90, doi:10.1186/s12940-017-0300-y (2017).

47
18 Rice, T. et al. Understanding inherited genetic risk of adult glioma - a review.
Neurooncol Pract 3, 10-16, doi:10.1093/nop/npv026 (2016).
19 Yuile, P., Dent, O., Cook, R., Biggs, M. & Little, N. Survival of glioblastoma patients
related to presenting symptoms, brain site and treatment variables. J Clin Neurosci 13,
747-751, doi:10.1016/j.jocn.2005.10.011 (2006).
20 Vecht, C. J., Kerkhof, M. & Duran-Pena, A. Seizure prognosis in brain tumors: new
insights and evidence-based management. Oncologist 19, 751-759,
doi:10.1634/theoncologist.2014-0060 (2014).
21 Yan, J. L. et al. Multimodal MRI characteristics of the glioblastoma infiltration
beyond contrast enhancement. Ther Adv Neurol Disord 12, 1756286419844664,
doi:10.1177/1756286419844664 (2019).
22 Wirsching, H. G., Galanis, E. & Weller, M. Glioblastoma. Handb Clin Neurol 134,
381-397, doi:10.1016/B978-0-12-802997-8.00023-2 (2016).
23 Peeken, J. C. et al. Combining multimodal imaging and treatment features improves
machine learning-based prognostic assessment in patients with glioblastoma
multiforme. Cancer Med 8, 128-136, doi:10.1002/cam4.1908 (2019).
24 Law, M. et al. Gliomas: predicting time to progression or survival with cerebral blood
volume measurements at dynamic susceptibility-weighted contrast-enhanced perfusion
MR imaging. Radiology 247, 490-498, doi:10.1148/radiol.2472070898 (2008).
25 Lonjon, M., Mondot, L., Lonjon, N. & Chanalet, S. [Clinical factors in glioblastoma
and neuroradiology]. Neurochirurgie 56, 449-454, doi:10.1016/j.neuchi.2010.07.016
(2010).
26 Price, S. J. et al. Multimodal MRI can identify perfusion and metabolic changes in the
invasive margin of glioblastomas. J Magn Reson Imaging 43, 487-494,
doi:10.1002/jmri.24996 (2016).
27 Louis DN, O. H., Wiestler OD, Cavenee WK. The 2016 World Health Organization
Classification of Tumors of the Central Nervous System. 4th edn, Vol. 1 408 (IARC
Publication, 2016).
28 Figarella-Branger, D., Bouvier, C., Moroch, J., Michalak, S. & Burel-Vandenbos, F.
[Morphological classification of glioblastomas]. Neurochirurgie 56, 459-463,
doi:10.1016/j.neuchi.2010.07.014 (2010).
29 Bobeff, E. J. et al. Plasma amino acids indicate glioblastoma with ATRX loss. Amino
Acids 53, 119-132, doi:10.1007/s00726-020-02931-3 (2021).
30 Zhao, Y. H. et al. The Clinical Significance of O(6)-Methylguanine-DNA
Methyltransferase Promoter Methylation Status in Adult Patients With Glioblastoma:
A Meta-analysis. Front Neurol 9, 127, doi:10.3389/fneur.2018.00127 (2018).
31 He, H. et al. Method for lipidomic analysis: p53 expression modulation of sulfatide,
ganglioside, and phospholipid composition of U87 MG glioblastoma cells. Anal Chem
79, 8423-8430, doi:10.1021/ac071413m (2007).
32 Aldape, K., Zadeh, G., Mansouri, S., Reifenberger, G. & von Deimling, A.
Glioblastoma: pathology, molecular mechanisms and markers. Acta Neuropathol 129,
829-848, doi:10.1007/s00401-015-1432-1 (2015).
33 Lacroix, M. et al. A multivariate analysis of 416 patients with glioblastoma
multiforme: prognosis, extent of resection, and survival. J Neurosurg 95, 190-198,
doi:10.3171/jns.2001.95.2.0190 (2001).
34 Brown, T. J. et al. Association of the Extent of Resection With Survival in
Glioblastoma: A Systematic Review and Meta-analysis. JAMA Oncol 2, 1460-1469,
doi:10.1001/jamaoncol.2016.1373 (2016).
35 Chambless, L. B. et al. The relative value of postoperative versus preoperative
Karnofsky Performance Scale scores as a predictor of survival after surgical resection

48
 of glioblastoma multiforme. J Neurooncol 121, 359-364, doi:10.1007/s11060-014-
1640-x (2015).
36 McGirt, M. J., Villavicencio, A. T., Bulsara, K. R. & Friedman, A. H. MRI-guided
stereotactic biopsy in the diagnosis of glioma: comparison of biopsy and surgical
resection specimen. Surgical Neurology 59, 279-283, doi:10.1016/s0090-
3019(03)00048-x (2003).
37 Eseonu, C. I. et al. Awake Craniotomy vs Craniotomy Under General Anesthesia for
Perirolandic Gliomas: Evaluating Perioperative Complications and Extent of
Resection. Neurosurgery 81, 481-489, doi:10.1093/neuros/nyx023 (2017).
38 Foster, C. H., Morone, P. J. & Cohen-Gadol, A. Awake craniotomy in glioma surgery:
is it necessary? J Neurosurg Sci 63, 162-178, doi:10.23736/S0390-5616.18.04590-3
(2019).
39 Obermueller, T. et al. Intraoperative neuromonitoring for function-guided resection
differs for supratentorial motor eloquent gliomas and metastases. BMC Neurol 15,
211, doi:10.1186/s12883-015-0476-0 (2015).
40 Senders, J. T. et al. Agents for fluorescence-guided glioma surgery: a systematic
review of preclinical and clinical results. Acta Neurochir (Wien) 159, 151-167,
doi:10.1007/s00701-016-3028-5 (2017).
41 Cho, S. S., Salinas, R. & Lee, J. Y. K. Indocyanine-Green for Fluorescence-Guided
Surgery of Brain Tumors: Evidence, Techniques, and Practical Experience. Front
Surg 6, 11, doi:10.3389/fsurg.2019.00011 (2019).
42 Kamath, A. A. et al. Glioblastoma Treated With Magnetic Resonance Imaging-Guided
Laser Interstitial Thermal Therapy: Safety, Efficacy, and Outcomes. Neurosurgery 84,
836-843, doi:10.1093/neuros/nyy375 (2019).
43 Norred, S. E. & Johnson, J. A. Magnetic resonance-guided laser induced thermal
therapy for glioblastoma multiforme: a review. Biomed Res Int 2014, 761312,
doi:10.1155/2014/761312 (2014).
44 Pirro, V. et al. Intraoperative assessment of tumor margins during glioma resection by
desorption electrospray ionization-mass spectrometry. Proc Natl Acad Sci U S A 114,
6700-6705, doi:10.1073/pnas.1706459114 (2017).
45 Santagata, S. et al. Intraoperative mass spectrometry mapping of an onco-metabolite
to guide brain tumor surgery. Proc Natl Acad Sci U S A 111, 11121-11126,
doi:10.1073/pnas.1404724111 (2014).
46 Stupp, R. et al. Radiotherapy plus concomitant and adjuvant temozolomide for
glioblastoma. N Engl J Med 352, 987-996, doi:10.1056/NEJMoa043330 (2005).
47 Binabaj, M. M. et al. The prognostic value of MGMT promoter methylation in
glioblastoma: A meta-analysis of clinical trials. J Cell Physiol 233, 378-386,
doi:10.1002/jcp.25896 (2018).
48 Lemee, J. M. et al. Characterizing the peritumoral brain zone in glioblastoma: a
multidisciplinary analysis. J Neurooncol 122, 53-61, doi:10.1007/s11060-014-1695-8
(2015).
49 Heiland, D. H. et al. One decade of glioblastoma multiforme surgery in 342 elderly
patients: what have we learned? J Neurooncol 140, 385-391, doi:10.1007/s11060-018-
2964-8 (2018).
50 Wen, P. Y. et al. Updated response assessment criteria for high-grade gliomas:
response assessment in neuro-oncology working group. J Clin Oncol 28, 1963-1972,
doi:10.1200/JCO.2009.26.3541 (2010).
51 Wann, A. et al. Outcomes after second surgery for recurrent glioblastoma: a
retrospective case-control study. J Neurooncol 137, 409-415, doi:10.1007/s11060-
017-2731-2 (2018).

49
52 Sage, W. et al. Local alkylating chemotherapy applied immediately after 5-ALA
guided resection of glioblastoma does not provide additional benefit. J Neurooncol
136, 273-280, doi:10.1007/s11060-017-2649-8 (2018).
53 Xing, W. K., Shao, C., Qi, Z. Y., Yang, C. & Wang, Z. The role of Gliadel wafers in
the treatment of newly diagnosed GBM: a meta-analysis. Drug Des Devel Ther 9,
3341-3348, doi:10.2147/DDDT.S85943 (2015).
54 Grangeon, L. et al. How safe are carmustine wafers? Rev Neurol (Paris) 174, 346-351,
doi:10.1016/j.neurol.2017.09.011 (2018).
55 Taal, W. et al. Single-agent bevacizumab or lomustine versus a combination of
bevacizumab plus lomustine in patients with recurrent glioblastoma (BELOB trial): a
randomised controlled phase 2 trial. The Lancet Oncology 15, 943-953,
doi:10.1016/s1470-2045(14)70314-6 (2014).
56 Stupp, R. et al. Effect of Tumor-Treating Fields Plus Maintenance Temozolomide vs
Maintenance Temozolomide Alone on Survival in Patients With Glioblastoma: A
Randomized Clinical Trial. JAMA 318, 2306-2316, doi:10.1001/jama.2017.18718
(2017).
57 Stupp, R. et al. NovoTTF-100A versus physician's choice chemotherapy in recurrent
glioblastoma: a randomised phase III trial of a novel treatment modality. Eur J Cancer
48, 2192-2202, doi:10.1016/j.ejca.2012.04.011 (2012).
58 Lombardi, G. et al. Regorafenib compared with lomustine in patients with relapsed
glioblastoma (REGOMA): a multicentre, open-label, randomised, controlled, phase 2
trial. Lancet Oncol 20, 110-119, doi:10.1016/S1470-2045(18)30675-2 (2019).
59 Szopa, W., Burley, T. A., Kramer-Marek, G. & Kaspera, W. Diagnostic and
Therapeutic Biomarkers in Glioblastoma: Current Status and Future Perspectives.
Biomed Res Int 2017, 8013575, doi:10.1155/2017/8013575 (2017).
60 Sasmita, A. O., Wong, Y. P. & Ling, A. P. K. Biomarkers and therapeutic advances in
glioblastoma multiforme. Asia Pac J Clin Oncol 14, 40-51, doi:10.1111/ajco.12756
(2018).
61 Harder, B. G. et al. Developments in Blood-Brain Barrier Penetrance and Drug
Repurposing for Improved Treatment of Glioblastoma. Front Oncol 8, 462,
doi:10.3389/fonc.2018.00462 (2018).
62 Parodi, A. et al. Established and Emerging Strategies for Drug Delivery Across the
Blood-Brain Barrier in Brain Cancer. Pharmaceutics 11,
doi:10.3390/pharmaceutics11050245 (2019).
63 Jahangiri, A. et al. Convection-enhanced delivery in glioblastoma: a review of
preclinical and clinical studies. J Neurosurg 126, 191-200,
doi:10.3171/2016.1.JNS151591 (2017).
64 Kong, Z., Wang, Y. & Ma, W. Vaccination in the immunotherapy of glioblastoma.
Hum Vaccin Immunother 14, 255-268, doi:10.1080/21645515.2017.1388481 (2018).
65 Kong, D. S. et al. Phase III randomized trial of autologous cytokine-induced killer cell
immunotherapy for newly diagnosed glioblastoma in Korea. Oncotarget 8, 7003-7013,
doi:10.18632/oncotarget.12273 (2017).
66 Swartz, A. M., Batich, K. A., Fecci, P. E. & Sampson, J. H. Peptide vaccines for the
treatment of glioblastoma. J Neurooncol 123, 433-440, doi:10.1007/s11060-014-1676-
y (2015).

50
L’article II présente un état des lieux des connaissances sur le glioblastome. Les différentes données

épidémiologiques et physiopathologiques y sont décrites de même que la présentation clinique et les

caractéristiques radiologiques. Une attention particulière a été donnée aux nouvelles avancées

thérapeutiques et à l’impact des sciences omiques dans l’appréhension de cette pathologie.

Article II

Gilard V, Tebani A, Dabaj I, Laquerrière A, Fontanilles M, Derrey S, et al. Diagnosis

and Management of Glioblastoma: A Comprehensive Perspective. J Pers Med.
2021;11(4):258.

51
 Journal of
Personalized
Medicine

Review
Diagnosis and Management of Glioblastoma:
A Comprehensive Perspective
Vianney Gilard 1,2, *, Abdellah Tebani 2,3 , Ivana Dabaj 3,4 , Annie Laquerrière 2,5 , Maxime Fontanilles 6 ,
Stéphane Derrey 1 , Stéphane Marret 3,4 and Soumeya Bekri 2,3

1 Department of Neurosurgery, Rouen University Hospital, 76000 Rouen, France;

stephane.Derrey@chu-rouen.fr
2 Normandie Univ, UNIROUEN, CHU Rouen, INSERM U1245, 76000 Rouen, France;
abdellah.tebani@chu-rouen.fr (A.T.); annie.laquerriere@chu-rouen.fr (A.L.);
soumeya.bekri@chu-rouen.fr (S.B.)
3 Department of Metabolic Biochemistry, Rouen University Hospital, 76000 Rouen, France;
ivana.dabaj@chu-rouen.fr (I.D.); stephane.marret@chu-rouen.fr (S.M.)
4 Department of Neonatology, Pediatric Intensive Care, and Pediatric Neurology, Rouen University Hospital,
76000 Rouen, France
5 Department of Pathology, Rouen University Hospital, 76000 Rouen, France
6 Department of Medical Oncology, Cancer Centre Henri Becquerel, 76038 Rouen, France;
maxime.fontanilles@chb.unicancer.fr
* Correspondence: vianney.gilard@chu-rouen.fr

Abstract: Glioblastoma is the most common malignant brain tumor in adults. The current manage-



ment relies on surgical resection and adjuvant radiotherapy and chemotherapy. Despite advances in


our understanding of glioblastoma onset, we are still faced with an increased incidence, an altered
Citation: Gilard, V.; Tebani, A.;
quality of life and a poor prognosis, its relapse and a median overall survival of 15 months. For the
Dabaj, I.; Laquerrière, A.; Fontanilles,
past few years, the understanding of glioblastoma physiopathology has experienced an exponential
M.; Derrey, S.; Marret, S.; Bekri, S.
Diagnosis and Management of
acceleration and yielded significant insights and new treatments perspectives. In this review, through
Glioblastoma: A Comprehensive an original R-based literature analysis, we summarize the clinical presentation, current standards of
Perspective. J. Pers. Med. 2021, 11, 258. care and outcomes in patients diagnosed with glioblastoma. We also present the recent advances
and perspectives regarding pathophysiological bases as well as new therapeutic approaches such as
https://doi.org/10.3390/jpm11040258 cancer vaccination and personalized treatments.

Academic Editors: Roberto Pallini Keywords: glioblastoma; molecular pathology; omics; pathogenesis; personalized therapies
and Quintino Giorgio D’Alessandris

Received: 9 March 2021

Accepted: 30 March 2021 1. Introduction
Published: 1 April 2021
Glioblastoma is the most frequent primary malignant brain tumor in adults [1]. Over
the last 20 years, a growing incidence of glioblastomas has been observed due to an increase
Publisher’s Note: MDPI stays neutral
in general population life expectancy and a better access to more accurate diagnostic tools
with regard to jurisdictional claims in
published maps and institutional affil-
such as MRI [2–4]. In a recent article based on a prospective registry of central nervous
iations.
system (CNS) tumors in the US [5], glioblastoma accounts for 54% of brain gliomas with
an annual incidence of 3.19 per 100,000. The peak of incidence occurs in the 6th decade
for glioblastomas isocitrate dehydrogenase (IDH) wild type and earlier, around the 4th–
5th decade, for glioblastomas IDH mutant [6]. Despite many therapeutic trials [7–9]
and advances in the management of these tumors, the median survival remains poor,
Copyright: © 2021 by the authors.
approximately 14 to 20 months, with a 5% five-year survival rate depending on age at
Licensee MDPI, Basel, Switzerland.
diagnosis, molecular characteristics and management [10]. Due to the poor outcomes of
This article is an open access article
distributed under the terms and
this pathology, there is a need for a better understanding of the pathogenesis and metabolic
conditions of the Creative Commons
pathways of glioblastomas. For a few years, advances in the different areas brought new
Attribution (CC BY) license (https://
insights and new therapeutic perspectives. Precision medicine, through the surge of omics
creativecommons.org/licenses/by/ technologies, offers new approaches for various oncology diseases [11,12]. Its main goal is
4.0/). to set a personalized perspective of the disease, taking into account individual variability

J. Pers. Med. 2021, 11, 258. https://doi.org/10.3390/jpm11040258 https://www.mdpi.com/journal/jpm

J. Pers. Med. 2021, 11, 258 2 of 14

of the patient along with his environment and predisposition. This approach requires
exhaustive collection of glioblastomas data. This work proposed an exhaustive semi-
automated review of literature regarding clinical presentation, diagnosis and management
of patients presenting with glioblastoma. Moreover, we also emphasize the novelties and
perspectives concerning this disease within the precision medicine era.

2. Materials and Methods

2.1. Literature Analysis
In this study, we performed a programmatic literature search for a more efficient and
reproducible review process using the Adjutant R package [13]. We searched for articles
related to glioblastoma, glioma physiopathology, diagnosis and treatments that were pub-
lished between January 1990 and February 2020. We used only one query (glioblastoma).
The resulting document corpus included articles metadata: PubMed IDs, year of publica-
tion, authors, article titles, article abstracts and any associated Medical Subject Heading
(MeSH) terms. Titles and abstracts were decomposed into single terms, stemmed and
filtered by Adjutant package. The term frequency inversed document frequency metrics
for each term and created a sparse Document Term Matrix (DTM) for further analysis.
t-Distributed Stochastic Neighbor Embedding (t-SNE) and hdbscan algorithms were used
to perform unsupervised clustering using DTM data. The coordinates generated by t-SNE
were used in the hdbscan algorithm to derive the topic clusters. Each cluster was then
assigned a topic by using the five most frequent terms within the cluster.

2.2. Manual Curation: Inclusion and Exclusion Criteria

Following the topic clustering step, we validated our clusters using external manual
curation assessing the correspondence between articles and cluster topics. Each sampled
article was examined and either considered acceptable for further analysis or rejected.
Inclusion criteria were topic relevance assessed by a neurosurgeon specialized in neuro-
oncology; article in English; human data; original research or clinical trial. We further
refined the corpus and cluster naming. Supplementary Table S1 contains a list of all the
articles along with their corresponding cluster.

2.3. Data Analysis

All the data analysis and visualization have been done using the R software [14].

3. Results
3.1. Literature Mining and Topic Clusters
The first-round analysis generated a document corpus of 2799 articles related to
glioblastoma published over the past 30 years (Supplementary Table S1). Using article
titles and abstracts, we derived topic clusters in an unsupervised manner, and classified
articles according to their clustering status. Articles that never formed part of a cluster were
removed from further analysis, leaving 1314 documents that formed 27 clusters. Cluster
topics were assigned using the five most frequent terms within the cluster along with a
manual curation of the included articles (Figure 1). The full list of articles and related
clusters are presented in Supplementary Table S1.
J. Pers. Med. 2021, 11, 258 3 of 14
J. Pers. Med. 2021, 11, x FOR PEER REVIEW 3 of 14

Figure 1.
Figure 1. Topic
Topicrepresentation
representationofofthe
theincluded
includedliterature
literature related
related to to glioblastoma.
glioblastoma. TheThe figure
figure high-
highlights
lights twenty-seven
twenty-seven clusters.
clusters.

3.2.
3.2. Pathogenesis
Pathogenesis of
of Glioblastoma
Glioblastoma
3.2.1. Risk Factors for High Grade Gliomas
3.2.1. Risk Factors for High Grade Gliomas
Risk factors for glioblastoma onset are still unknown and studies dealing with this
Risk factors for glioblastoma onset are still unknown and studies dealing with this
question frequently lack power. The exposure to ionizing radiation for the treatment of
question frequently lack power. The exposure to ionizing radiation for the treatment of
malignancy during infancy [15] is a very rare risk factor for the onset of glioma. The risk
malignancy during infancy [15] is a very rare risk factor for the onset of glioma. The risk
of developing a brain tumor after radiotherapy is increased if the radiation occurs at a
of developing a brain tumor after radiotherapy is increased if the radiation occurs at a
younger age (<5 years) and seems volume- and dose-related but with no clear reported
younger age (<5 years) and seems volume- and dose-related but with no clear reported
threshold [16,17]. The increased incidence of glioblastoma [18,19] raises the issue of envi-
threshold [16,17]. The increased incidence of glioblastoma [18,19] raises the issue of envi-
ronmental risk factors. The association between the use of mobile phone and the occurrence
ronmental risk factors. The association between the use of mobile phone and the occur-
of glioblastoma has been reported in a meta-analysis [20]. However, these results are dis-
rence of glioblastoma has been reported in a meta-analysis [20]. However, these results
cordant and have been challenged in other studies [21,22]. The role of exposure to smoking
arecarcinogenic
or discordant and have
agents hasbeen
beenchallenged
studied with in no
other studies
proven [21,22]. The
association role of exposure
to glioblastoma to
[23,24].
smoking or carcinogenic agents has been studied with no proven association to
In very few cases (<1%), there is a genetic predisposition to glioblastoma development in glioblas-
toma [23,24].
patients In veryTurcot
with Lynch, few cases
type 1(<1%), there is a syndromes
or Li Fraumeni genetic predisposition
[25]. to glioblastoma
development in patients with Lynch, Turcot type 1 or Li Fraumeni syndromes [25].
3.2.2. Clinical Presentation
3.2.2.Clinical
Clinicalpresentation
Presentationdepends on the tumor location and size at diagnosis. The most
commonClinical presentation
presentation depends on
at diagnosis is athe tumor location
headache and/orand size in
nausea at adiagnosis.
context of The most
a large
common
tumor or presentation at diagnosis
significant edema. is a headache
Symptoms related toand/or nauseahypertension
intracranial in a context represent
of a large
tumor
30% of or significant
clinical signs edema.
followed Symptoms
by motorrelated
deficit to intracranial
(20%), hypertension
loss of body weight andrepresent 30%
condition
of clinical signs followed by motor deficit (20%), loss of body weight and
(17%), confusion (15%) and visual or speech deficit (13%) [26]. Epilepsy is not uncommon condition (17%),
confusionand
(15–20%) (15%) andcontrolled
easily visual or speech deficit (13%)therapies.
by anticonvulsant [26]. Epilepsy is not
Epilepsy is uncommon
associated with(15–
20%) and
better easily controlled
outcomes probably due by anticonvulsant
to the cortical therapies.
location ofEpilepsy is associated
glioblastomas with better
presenting with
outcomes[27].
seizures probably
These due to the cortical
symptoms are oftenlocation of glioblastomas
associated and lead to apresenting
diagnosis with
in theseizures
weeks
or months
[27]. following their
These symptoms onset.
are often An overview
associated and leadof to
clinical presentation
a diagnosis at diagnosis
in the weeks or months is
summarized
following their in onset.
FigureAn 2. Glioblastomas are located
overview of clinical in the supra-tentorial
presentation space in more
at diagnosis is summarized in
than 85% of the cases explaining the above-mentioned symptoms and very
Figure 2. Glioblastomas are located in the supra-tentorial space in more than 85% of the infrequently
in the explaining
cases brainstem the andabove-mentioned
spinal cord (<5%symptoms
each) or inandthevery
cerebellum (<3%).
infrequently in Up
the to 25% of
brainstem
J. Pers. Med. 2021, 11, x FOR PEER REVIEW 4 of 14
J. Pers. Med. 2021, 11, 258 4 of 14

and spinal cord (<5% each) or in the cerebellum (<3%). Up to 25% of glioblastomas occur
in glioblastomas
the frontal lobeoccur
[28], in
thethe
largest cerebral
frontal lobe,the
lobe [28], and therefore
largest are responsible
cerebral for moodare
lobe, and therefore
and executive for
responsible disabilities
mood and in 15% of patients
executive [26]. in 15% of patients [26].
disabilities

Figure 2. Overview of the main reported clinical features in glioblastomas.

Figure 2. Overview of the main reported clinical features in glioblastomas.
3.2.3. Radiological Characteristics
Typical magnetic
3.2.3. Radiological resonance imaging (MRI) characteristics of glioblastomas are well
Characteristics
known [2,28–31]. They consist in infiltrative, heterogeneous intraparenchymal lesions
Typical magnetic resonance imaging (MRI) characteristics of glioblastomas are well
which arise and spread from the white matter. Corpus callosum involvement is common.
known [2,28–31]. They consist in infiltrative, heterogeneous intraparenchymal lesions
Glioblastomas are poorly circumscribed and display contrast enhancement at their margin
which arise and spread from the white matter. Corpus callosum involvement is common.
as a sign of blood–brain barrier disruption. The center of the lesion is hypointense on
Glioblastomas are poorly circumscribed and display contrast enhancement at their mar-
T1-weighted images due to necrosis and the lesion is surrounded by brain edema which
gin as a sign of blood–brain barrier disruption. The center of the lesion is hypointense on
appears hyperintense on T2-weighted and fluid-attenuated inversion recovery (FLAIR) im-
T1-weighted images due to necrosis and the lesion is surrounded by brain edema which
ages. Diffusion-weighted images and apparent diffusion coefficients can provide valuable
appears hyperintense on T2-weighted and fluid-attenuated inversion recovery (FLAIR)
information concerning the suspected degree of malignancy of astrocytic tumors. More
images. Diffusion-weighted images and apparent diffusion coefficients can provide valu-
recent multimodal MRI techniques such as diffusion/perfusion sequences have provided
able information concerning the suspected degree of malignancy of astrocytic tumors.
supplementary information about the characteristics of the lesion itself and have enabled a
More
morerecent multimodal
accurate diagnosis.MRI techniques
Perfusion such as
weighted diffusion/perfusion
imaging (PWI) revealssequences
an increasehave
in thepro-
cere-
vided
bral supplementary information
blood flow corresponding about the characteristics
to neoangiogenesis of the lesion
and blood–brain itself
barrier and have
disruption. MR
enabled a more in
spectroscopy accurate diagnosis.
glioblastomas Perfusion weighted
is characterized imagingin(PWI)
by an increase reveals an increase
choline/N-acetylaspartate
in and
the cerebral blood flow corresponding
choline/creatinine to neoangiogenesis
ratios. Nevertheless, and blood–brain
these characteristics barrieractivity
reflect cellular dis-
ruption. MR spectroscopy in glioblastomas is characterized by an increase
and are not, when isolated, specific enough to diagnose glioblastoma. In addition, an in choline/N-
acetylaspartate
elevated peakand choline/creatinine
of lactate and lipids as ratios.
well as Nevertheless,
a decreased peak these characteristics
of myoinositol arereflect
reliable
cellular activity
data for and are of
the diagnosis not, when isolated,
glioblastoma specific
[30,32] and are enough totodiagnose
helpful glioblastoma.
discriminate In
glioblastomas
addition, an elevated peak of lactate and lipids as well as a decreased
from metastasis, lymphoma and brain abscess. As suggested by some authors, these explo-peak of myoinositol
arerations
reliable data
are alsofor the diagnosis
useful to assess theof glioblastoma [30,32] and
peritumoral degree are helpful
of invasion and to
maydiscriminate
be used as a
glioblastomas from metastasis, lymphoma and brain abscess.
guidance for biopsy [33,34] and to monitor disease evolution after treatment. As suggested by some au-
thors, these explorations are also useful to assess the peritumoral degree of invasion and
may3.2.4. Basicas
be used and Molecularfor
a guidance Pathology
biopsy [33,34] and to monitor disease evolution after treat-
ment. The diagnosis of glioblastoma is easily made on surgical resections or biopsy samples.
Glioblastomas are high grade gliomas, grade IV according to the World Health Organi-
3.2.4. Basic
zation and Molecular
(WHO) Pathology
classification of the central nervous system tumors [35,36]. Glioblastoma
is composed
The diagnosis of poorly differentiated,
of glioblastoma often
is easily pleomorphic
made tumor
on surgical cells with
resections predominant
or biopsy sam-
astrocytic
ples. differentiation
Glioblastomas are high[37].
gradeHistopathological
gliomas, grade IV features include
according nuclear
to the World atypia,
Health cellular
Or-
pleomorphism, high mitotic activity, vascular thrombosis, microvascular proliferation and
necrosis [28]. Since the 2016 WHO classification of the central nervous system tumors [36]
J. Pers. Med. 2021, 11, 258 5 of 14

and advances in immunohistochemistry, glioblastomas are now defined by their Isoci-

trate dehydrogenase (IDH) status dividing this entity into glioblastoma IDH-mutant or
IDH-wild type. The latter is, by far, the most frequent accounting for 90% of cases and is
predominant in patients over 55 years of age. Glioblastomas IDH-mutant (10% of cases) is
predominant in younger patients, often results from the transformation of a lower-grade
glioma and is associated with longer median survival [38]. Complementary immunohisto-
chemistry and molecular techniques are now routinely used for diagnostics and prognostic
purposes. (IDH) alterations as well as hypermethylation of the O6 -methylguanine-DNA
methyltransferase (MGMT) gene promoter [39] are predictive of longer survival contrary to
Telomerase Reverse Transcriptase promoter (TERTp) variants and chromosome 10 deletion
which are poorer prognostic factors. Gain of function variants in the P53 gene [40] and epi-
dermal growth factor receptor (EGFR) gene alterations [41] still are of uncertain prognostic
significance. MGMT is involved in DNA repair and its expression is associated with drug
resistance including temozolamide, the most frequent first-line chemotherapy used in a
context of glioblastoma. As a consequence, hypermethylation of the MGMT gene promoter
is a valuable predictive marker and is associated with longer overall survival and longer
progression-free survival [42]. More recently, it has been proven [43] that variants in the
tumor suppressor gene Phosphatase and tensin homolog (PTEN) or the loss of chromosome
10 is involved in the glioblastoma oncogenesis.

3.3. Current Management

3.3.1. Surgical Procedure
Whenever possible, the first step consists in complete macroscopic surgical resection.
Literature data suggest that a resection > 90% of the contrast enhancement of the lesion in
patients with no comorbidities improves the patient outcome at the time of diagnosis and
recurrence [44–46]. Surgical resection is usually proposed to patients under the age of 70 in
good condition (Karnofsky scale > 70) [46] and a tumor accessible to complete removal.
Otherwise, surgical debulking or stereotactic biopsy is performed to confirm the diagnosis
before adjuvant therapy [47,48]. Due to the importance of a complete resection on survival,
advances in surgical techniques have been made [49,50] such as awake craniotomy or
neuromonitoring [51] to improve the resection quality and prevent subsequent deficits.
Furthermore, fluorescence-guided surgery has been developed to guide the resection with
better outcomes on resection and progression-free survival [52–54]. More recently, resection
devices have evolved with the use of Laser-Interstitial Thermal Therapy (LITT) [55], which
provides a less invasive, percutaneous approach through the insertion of an optical fiber.
The generated thermal injury induces tumor necrosis [55–57]. Another novelty in the
surgical resection of brain tumors is mass spectrometry-based intraoperative monitoring
of tumor metabolites. Cell content is analyzed allowing for an accurate and molecular
delineation of tumor margins and thus for an optimal tumor resection [58,59]. A summary
of clinical presentations, radiology, biology and treatments together with their interactions
is illustrated in Figure 3.
ers. Med. 2021, 11, x FOR PEER REVIEW 6 of
J. Pers. Med. 2021, 11, 258 6 of 14

Figure 3. Integrative visualization summary of main clinical symptoms and signs and their interactions with radiology,
Figure 3. Integrative visualization
biology and treatment features.summary ofannotation
The box and main clinical symptoms
sizes are and
proportional signs
to the itemand their Clinical:
frequency. interactions with radiology,
blue, Biology:
biology and treatment
purple, features.
Radiology: The box and
green, Treatment: annotation sizes are proportional to the item frequency. Clinical: blue, Biol-
orange.
ogy: purple, Radiology: green, Treatment: orange.
3.3.2. Medical Treatment of Glioblastomas
3.3.2. Medical Treatment
The standard offor
of care Glioblastomas
patients aged less than 70 relies on radiotherapy (RT) and
adjuvant temozolomide. This protocol improved the overall survival in a large randomized
The
phasestandard of care
III trial [60]. for patients
Radiotherapy aged
is given for less than 70
a six-week relies
period withona total
radiotherapy
dose of 60 (RT) a
adjuvant
grays.temozolomide.
Temozolomide isThis protocolagent
an alkylating improved
administeredthe overall survival
daily during the RTinanda large
then, rando
ized phase III trial
for six cycles [60].
of five Radiotherapy
consecutive days per is month,
given onefor month
a six-week
after theperiod
end of with
the RT.a The
total dose
absence of hypermethylated MGMT promoter [61] is a negative prognostic
60 grays. Temozolomide is an alkylating agent administered daily during the RT and the and predictive
factor of temozolomide efficiency. The treatment protocol proposed by Stupp allows for
for six cycles of five consecutive days per month, one month after the end of the RT. T
increasing the average survival rate from 12.1 months using RT alone to 14.6 months,
absenceandof thehypermethylated
two-year survival rate MGMT
from 8promoter
to 26% with [61] is a negative
concomitant prognostic
temozolomide [60].and
Thispredicti
factorrandomized
of temozolomide
controlled efficiency. The treatment
trial did not include patients olderprotocol
than 70.proposed
In the latterby Stupp allows
population,
the standard
increasing of care is
the average based onrate
survival hypofractionated
from 12.1 months radiotherapy
usingand RT temozolomide
alone to 14.6 [62] months, a
whenever feasible, but the treatment depends on the patient’s general condition. RT alone
the two-year survival rate from 8 to 26% with concomitant temozolomide [60]. This ra
(54 grays) has been proposed with a positive impact on survival (29.1 weeks compared to
domized controlled
16.9 weeks trialwith
in patients didsupportive
not include patients
care alone) andolder
with nothan 70. Inofthe
alteration latter populatio
the quality of
the standard of care
life [63]. These is based
results have beenonvalidated
hypofractionated
in patients with radiotherapy andperformance
a Karnofsky [64] temozolomide [6
whenever
status feasible,
(KPS) > 60. butMore
the treatment
recently, somedepends
authorson have theshown
patient’s general condition.
that short-course RT plus RT alo
temozolomide
(54 grays) has been was associated with
proposed with longer survival
a positive (9.3 versus
impact 7.6 months)
on survival in older
(29.1 weekspatients
compared
16.9 weeks in patients with supportive care alone) and with no alteration of the quality
life [63]. These results have been validated in patients with a Karnofsky [64] performan
status (KPS) > 60. More recently, some authors have shown that short-course RT plus
J. Pers. Med. 2021, 11, 258 7 of 14

(>65 years). Malmstrom et al. [65] randomized patients aged 60 years and older presenting
with a glioblastoma to assess the optimal palliative treatment. The conclusion of the trial
was that radiotherapy alone is associated with poor outcomes. On the contrary, both
temozolamide and hypofractionated radiotherapy appeared as standards of care especially
in patients with methylation of the MGMT gene promoter. In patients with poor general
health, supportive care may be proposed to preserve the quality of life with the shortest
length of hospital stay if possible [62].
The disease progression is evaluated using brain MRI every 2 to 3 months according to
the response assessment in Neuro-oncology (RANO) criteria [66]. At the time of recurrence,
there is no standard of care. The main determinants for treatment proposals are the patients’
general condition and treatments previously administered [67]. A second surgery can be
proposed in young patients with preserved KPS. This strategy has been shown to be
associated with longer survival in selected patients [68] (14 months versus 22 months
of overall survival in patients with second surgery at recurrence). Eighty percent of
patients did not need rehabilitation after a second surgery. In selected patients, the use
of intracavitary carmustine wafers (BCNU) has been proposed at initial surgery or in
case of recurrence [69]. Efficiency and adverse effects of this therapy remains a matter
of debate [70,71]. In the absence of safe possible resection, a second line chemotherapy
can be proposed including nitrosoureas, temozolomide or antiangiogenic drugs such as
bevacizumab but with no clear results in terms of benefit to date [72] and poor outcomes
with an average overall survival rate of 6 months from recurrence [73]. Due to the poor
prognosis of glioblastomas, there is an urgent need for new therapies. The REGOMA trial
(Regorafenib in Relapsed Glioblastoma) [74] is a phase II randomized trial. The aim was to
propose the use regorafenib, an inhibitor of angiogenic and oncogenic receptor of tyrosine
kinases at the recurrence of glioblastomas. The results were rather positive on overall
survival and with few side effects. A phase III trial is expected.

3.4. Current Research and Perspectives

3.4.1. Omics Approaches
Over the last decade, precision medicine, also through omics approaches (Figure 4),
has offered new insights in the diagnosis and management of glioblastomas [75,76]. In the
past few years, radiomics brought significant insights in the characterization and predictive
models of glioblastomas [77,78]. Radiomics is based on the extraction of a large amount
of data from medical images. Radiomics is then enriched with clinical, genomics and
proteomics data to establish new diagnosis and prognosis criteria to enhance treatment
efficiency. Other promising areas in the fight against glioblastoma are the genomics and
transcriptomics approaches. The emergence of big data in the precision medicine offered
new therapeutic perspectives. Through ambitious projects such as the human genome
project [79], genomics helped us in improving the understanding of glioblastoma. In the
continuation of the project, RNA sequencing-based genes proposed genomics signature
of life expectancy in patients with glioblastomas [80]. Transcriptomics and other omics
technics proposed prognostic tools for the comprehension of the disease [81,82]. Along with
these techniques, liquid chromatography and mass spectrometry analyses from different
samples such as CSF (Cerebrospinal fluid), urine, blood or glioblastoma cell lines [83]
have provided a comprehensive view of the altered metabolic pathways in patients with
glioblastoma [84,85]. More recently, special attention has been given to alterations of lipid
metabolism in glioblastomas. Based on omics human studies, Guo et al. have described
a decrease of 90% of lipid levels in tumor tissue except for phosphatidylcholine and
cholesterol ester levels which appeared high in glioma tissue while they are absent in
normal brain tissue [86]. These data may emphasize the key role of certain lipids in the
glioblastoma metabolism to facilitate tumor growth. Moreover, the metabolic signature of
brain tumors in the plasma is of interest for the grading and prognosis of these diseases.
It has been shown that the plasma level of metabolites of interest can help to define the
grading of brain glioma and to provide prognostic information in patients with a similar
J. Pers. Med. 2021, 11, 258 8 of 14

glioma grade [87]. Nevertheless, data are still rare for these emerging approaches; however,
the promise of precision medicine and the surge of multimodal data-driven strategies can
provide valuable tools for the development of biomarkers and innovative therapies in
glioblastomas [88].

3.4.2. Novel Therapies for Glioblastomas

Due to the adverse outcome in patients with glioblastoma and the high frequency
of this disease, innovative therapies are being tested in different randomized controlled
trials [73,89,90]. With the development of a better understanding of molecular pathways
triggering glioblastoma growth [27–29], the traditional approach of antitumor therapy is
being progressively complemented by a more personalized approach [91,92]. Treatment
schedules have been rethought as well as the drugs themselves. Two major drawbacks
consist in the difficulty for most of the drugs is to pass through the blood–brain barrier
(BBB) and to target tumor cells due to the presence of abnormal vessels and necrosis, which
hampers drugs being delivered at a suitable concentration. Some emerging techniques have
been proposed to improve the distribution of antitumor therapy, notably the conjugation
of drugs with protein to facilitate the movement across the BBB and specifically target the
tumor [93], the use of convection-enhanced delivery consisting in the direct administration
inside the tumor via a catheter [94] and the use of nanoparticles. The increase in BBB
permeability during chemotherapy administration via focused ultrasounds is being also
tested [95]. Meanwhile, immunotherapy approaches are known for a long time, with
promising results in many cancers such as melanoma but deceiving results in patients with
gliomas [96]. Cancer vaccination has recently been proposed referring to the activation of
an immune response against tumor antigens. These new technologies have been applied
for glioblastoma treatment with different vectors [97–99]. Two modalities have been tested:
peptide vaccines targeting EGFR, IDH1 or heat shock proteins, and cell-based vaccines
consisting in the injection of ex vivo modified cells, mainly dendritic cells [100]. Despite
encouraging results in animal models in terms of disease control [101], cancer vaccination
in glioblastomas has not yet proven its efficacy on overall survival in phase III studies [9].
Recently, the role of tumor-associated macrophages (TAM) has been highlighted in the
genesis and resistance to treatment of glioblastoma cells [102–104]. Landry et al. [102]
showed that the TAM located in the core have different characteristics and metabolic
pathways compared to those located in the periphery of the glioblastoma. For these
reasons, they reaffirm the need for a multi-targeted approach through a modulation of
the TAM. Furthermore, tumor-associated neutrophils (TAN) are found to be involved in
necrosis onset in glioblastoma patients [105,106]. In this context, the mechanism of necrosis
could be a neutrophil-mediated ferroptosis. The latter could have a pro-tumorigenic
role [105]. Thus, targeted therapies are potential novel therapies for glioblastomas to
prevent TAN recruitment.
Stupp et al. recently developed a new therapeutic modality in the treatment of
recurrent glioblastomas consisting in the local delivery of low-intensity electric fields via a
non-invasive transducer [73]. The device (NovoTTF-100A) was tested in a phase III study
and randomized with active chemotherapy as an alternative arm. The overall survival was
similar in both arms (6 months) with fewer adverse events in the NovoTTF-100A group
and a better quality of life. For this reason, the tumor-treating field is considered as a
standard of care in some guidelines [107]. Moreover, in the last years, the association of
molecularly targeted drugs such as tyrosine kinase administration or others combined with
X-rays to decrease radioresistance due to hypoxia showed encouraging results [108–110].
d. 2021, 11, x FOR PEER REVIEW 9 of 14
J. Pers. Med. 2021, 11, 258 9 of 14

Figure 4. Overview of the main driving omics technologies and therapeutic perspectives for glioblastoma in the precision
Figure 4. Overview of the main driving omics technologies and therapeutic perspectives for glio-
medicineblastoma
era. This in
figure has been created
the precision with
medicine era.BioRender.com
This figure has(accessed on 9with
been created March 2021).
BioRender.com.

4. Conclusions 4. Conclusions
The authors are The
aware authors
of the are aware
limits of thesuch
of using limits of using topic
automatic such automatic topic literature search
literature search
tools. This limitstools.
mightThis limits might
be related be related
to recency to recency
or coverage or coverage
scope of suchscopetools of such
that tools that might lead
might
lead to some key to some key
literature literature
outputs beingoutputs
missing.being
Thismissing.
highlightsThis
thehighlights
importance the
ofimportance
man- of manual
curation combined with the use of multiple tools to consistently
ual curation combined with the use of multiple tools to consistently cover the area of in- cover the area of interest.
terest. In conclusion, despite a better understanding of the molecular pathways leading
to glioblastoma development and growth, outcomes
In conclusion, despite a better understanding of the molecular pathways leading toremain poor in terms of survival.
Many trialsand
glioblastoma development aregrowth,
currently in progress
outcomes withpoor
remain newin therapeutic approaches
terms of survival. Many to go from a global
trials are currently in progress with new therapeutic approaches to go from a global ap- approaches is
approach to a more personalized approach. The main objective of these
proach to a moretopersonalized
deliver the chosen drugThe
approach. inside
maintheobjective
tumor and of to adapt
these its concentration
approaches is to to the tumor
deliver the chosencharacteristics.
drug inside theThen,tumortheandmodality
to adaptof itsdrug delivery to
concentration has become
the as challenging as the
tumor char-
drug itself.
acteristics. Then, the modality of drug delivery has become as challenging as the drug
itself.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10
.3390/jpm11040258/s1, Table S1: Articles and clusters related to glioblastoma published in the past
Supplementary Materials: The following are available online at www.mdpi.com/xxx/s1, Table S1:
30 years.
Articles and clusters related to glioblastoma published in the past 30 years.
Author Contributions: Conceptualization, V.G.; software, A.T.; formal analysis, A.T.; data curation,
Author Contributions: Conceptualization, V.G.; software, A.T.; formal analysis, A.T.; data curation,
V.G., S.B. and S.M.; writing—original draft preparation, V.G. and S.M.; writing—review and editing,
V.G., S.B., S.M.; writing—original draft preparation, V.G., S.M.; writing—review and editing, A.T.,
I.D., A.L., M.F., S.D. andI.D.,
A.T., S.B.;A.L., M.F., S.D. A.T.;
visualization, and S.B.; visualization,
supervision, A.T.;S.M.
S.B. and supervision, S.B.
All authors andread
have S.M. All authors have
read and agreed to the published
and agreed to the published version of the manuscript. version of the manuscript.
Funding:
Funding: This research Thisno
received research received
external no external funding.
funding.
Institutional
Institutional Review Review Not
Board Statement: Board Statement: Not applicable.
applicable.
Informed ConsentInformed Consent
Statement: Statement: Not applicable.
Not applicable.
J. Pers. Med. 2021, 11, 258 10 of 14

Data Availability Statement: Data supporting the finding are presented in the text and Supplemen-
tary Material.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Bauchet, L.; Ostrom, Q.T. Epidemiology and Molecular Epidemiology. Neurosurg. Clin. N. Am. 2019, 30, 1–16. [CrossRef]
2. Negendank, W.G.; Sauter, R.; Brown, T.R.; Evelhoch, J.L.; Falini, A.; Gotsis, E.D.; Heerschap, A.; Kamada, K.; Lee, B.C.; Mengeot,
M.M.; et al. Proton magnetic resonance spectroscopy in patients with glial tumors: A multicenter study. J. Neurosurg. 1996, 84,
449–458. [CrossRef]
3. Morgan, L.L. The epidemiology of glioma in adults: A “state of the science” review. Neuro Oncol. 2015, 17, 623–624. [CrossRef]
4. Kowalczyk, T.; Ciborowski, M.; Kisluk, J.; Kretowski, A.; Barbas, C. Mass spectrometry based proteomics and metabolomics in
personalized oncology. Biochim. Biophys. Acta Mol. Basis Dis. 2020, 1866, 165690. [CrossRef]
5. Dolecek, T.A.; Propp, J.M.; Stroup, N.E.; Kruchko, C. CBTRUS statistical report: Primary brain and central nervous system tumors
diagnosed in the United States in 2005–2009. Neuro Oncol. 2012, 14 (Suppl. 5), v1–v49. [CrossRef]
6. Ostrom, Q.T.; Bauchet, L.; Davis, F.G.; Deltour, I.; Fisher, J.L.; Langer, C.E.; Pekmezci, M.; Schwartzbaum, J.A.; Turner, M.C.; Walsh,
K.M.; et al. The epidemiology of glioma in adults: A “state of the science” review. Neuro Oncol. 2014, 16, 896–913. [CrossRef]
7. Lim, M.; Xia, Y.; Bettegowda, C.; Weller, M. Current state of immunotherapy for glioblastoma. Nat. Rev. Clin. Oncol. 2018, 15,
422–442. [CrossRef]
8. Touat, M.; Idbaih, A.; Sanson, M.; Ligon, K.L. Glioblastoma targeted therapy: Updated approaches from recent biological insights.
Ann. Oncol. 2017, 28, 1457–1472. [CrossRef] [PubMed]
9. Malkki, H. Trial Watch: Glioblastoma vaccine therapy disappointment in Phase III trial. Nat. Rev. Neurol. 2016, 12, 190. [CrossRef]
[PubMed]
10. Delgado-Lopez, P.D.; Corrales-Garcia, E.M. Survival in glioblastoma: A review on the impact of treatment modalities. Clin. Transl.
Oncol. 2016, 18, 1062–1071. [CrossRef] [PubMed]
11. Clish, C.B. Metabolomics: An emerging but powerful tool for precision medicine. Cold Spring Harb. Mol. Case Stud. 2015,
1, a000588. [CrossRef]
12. Lopez de Maturana, E.; Alonso, L.; Alarcon, P.; Martin-Antoniano, I.A.; Pineda, S.; Piorno, L.; Calle, M.L.; Malats, N. Challenges
in the Integration of Omics and Non-Omics Data. Genes 2019, 10, 238. [CrossRef] [PubMed]
13. Crisan, A.; Munzner, T.; Gardy, J.L.; Wren, J. Adjutant: An R-based tool to support topic discovery for systematic and literature
reviews. Bioinformatics 2019, 35, 1070–1072. [CrossRef] [PubMed]
14. R Core Team. A Language and Environment for Statistical Computing; R Foundation: Vienna, Austria, 2020.
15. Connelly, J.M.; Malkin, M.G. Environmental risk factors for brain tumors. Curr. Neurol. Neurosci. Rep. 2007, 7, 208–214. [CrossRef]
[PubMed]
16. Wingren, C.; James, P.; Borrebaeck, C.A.K. Strategy for surveying the proteome using affinity proteomics and mass spectrometry.
Proteomics 2009, 9, 1511–1517. [CrossRef]
17. Yamanaka, R.; Hayano, A.; Kanayama, T. Radiation-induced gliomas: A comprehensive review and meta-analysis. Neurosurg.
Rev. 2018, 41, 719–731. [CrossRef]
18. Philips, A.; Henshaw, D.L.; Lamburn, G.; O’Carroll, M.J. Brain Tumours: Rise in Glioblastoma Multiforme Incidence in England
1995-2015 Suggests an Adverse Environmental or Lifestyle Factor. J. Environ. Public Health 2018, 2018, 7910754. [CrossRef]
19. Dobes, M.; Khurana, V.G.; Shadbolt, B.; Jain, S.; Smith, S.F.; Smee, R.; Dexter, M.; Cook, R. Increasing incidence of glioblastoma
multiforme and meningioma, and decreasing incidence of Schwannoma (2000–2008): Findings of a multicenter Australian study.
Surg. Neurol. Int. 2011, 2, 176. [CrossRef]
20. Yang, M.; Guo, W.; Yang, C.; Tang, J.; Huang, Q.; Feng, S.; Jiang, A.; Xu, X.; Jiang, G. Mobile phone use and glioma risk: A
systematic review and meta-analysis. PLoS ONE 2017, 12, e0175136. [CrossRef]
21. Karipidis, K.; Elwood, M.; Benke, G.; Sanagou, M.; Tjong, L.; Croft, R.J. Mobile phone use and incidence of brain tumour
histological types, grading or anatomical location: A population-based ecological study. BMJ Open 2018, 8, e024489. [CrossRef]
22. Repacholi, M.H.; Lerchl, A.; Roosli, M.; Sienkiewicz, Z.; Auvinen, A.; Breckenkamp, J.; d’Inzeo, G.; Elliott, P.; Frei, P.; Heinrich,
S.; et al. Systematic review of wireless phone use and brain cancer and other head tumors. Bioelectromagnetics 2012, 33, 187–206.
[CrossRef]
23. Benke, G.; Turner, M.C.; Fleming, S.; Figuerola, J.; Kincl, L.; Richardson, L.; Blettner, M.; Hours, M.; Krewski, D.; McLean, D.; et al.
Occupational solvent exposure and risk of glioma in the INTEROCC study. Br. J. Cancer 2017, 117, 1246–1254. [CrossRef]
24. Parent, M.E.; Turner, M.C.; Lavoue, J.; Richard, H.; Figuerola, J.; Kincl, L.; Richardson, L.; Benke, G.; Blettner, M.; Fleming, S.; et al.
Lifetime occupational exposure to metals and welding fumes, and risk of glioma: A 7-country population-based case-control
study. Environ. Health 2017, 16, 90. [CrossRef] [PubMed]
25. Rice, T.; Lachance, D.H.; Molinaro, A.M.; Eckel-Passow, J.E.; Walsh, K.M.; Barnholtz-Sloan, J.; Ostrom, Q.T.; Francis, S.S.; Wiemels,
J.; Jenkins, R.B.; et al. Understanding inherited genetic risk of adult glioma—A review. Neurooncol. Pract. 2016, 3, 10–16.
[CrossRef] [PubMed]
26. Yuile, P.; Dent, O.; Cook, R.; Biggs, M.; Little, N. Survival of glioblastoma patients related to presenting symptoms, brain site and
treatment variables. J. Clin. Neurosci. 2006, 13, 747–751. [CrossRef]
J. Pers. Med. 2021, 11, 258 11 of 14

27. Vecht, C.J.; Kerkhof, M.; Duran-Pena, A. Seizure prognosis in brain tumors: New insights and evidence-based management.
Oncologist 2014, 19, 751–759. [CrossRef] [PubMed]
28. Wirsching, H.G.; Galanis, E.; Weller, M. Glioblastoma. Handb. Clin. Neurol. 2016, 134, 381–397. [CrossRef] [PubMed]
29. Yan, J.L.; Li, C.; Boonzaier, N.R.; Fountain, D.M.; Larkin, T.J.; Matys, T.; van der Hoorn, A.; Price, S.J. Multimodal MRI
characteristics of the glioblastoma infiltration beyond contrast enhancement. Ther. Adv. Neurol. Disord. 2019, 12. [CrossRef]
[PubMed]
30. Peeken, J.C.; Goldberg, T.; Pyka, T.; Bernhofer, M.; Wiestler, B.; Kessel, K.A.; Tafti, P.D.; Nusslin, F.; Braun, A.E.; Zimmer, C.; et al.
Combining multimodal imaging and treatment features improves machine learning-based prognostic assessment in patients
with glioblastoma multiforme. Cancer Med. 2019, 8, 128–136. [CrossRef]
31. Law, M.; Yang, S.; Wang, H.; Babb, J.S.; Johnson, G.; Cha, S.; Knopp, E.A.; Zagzag, D. Glioma grading: Sensitivity, specificity, and
predictive values of perfusion MR imaging and proton MR spectroscopic imaging compared with conventional MR imaging.
AJNR Am. J. Neuroradiol. 2003, 24, 1989–1998.
32. Price, S.J.; Young, A.M.; Scotton, W.J.; Ching, J.; Mohsen, L.A.; Boonzaier, N.R.; Lupson, V.C.; Griffiths, J.R.; McLean, M.A.; Larkin,
T.J. Multimodal MRI can identify perfusion and metabolic changes in the invasive margin of glioblastomas. J. Magn. Reson.
Imaging 2016, 43, 487–494. [CrossRef]
33. Barajas, R.F., Jr.; Hodgson, J.G.; Chang, J.S.; Vandenberg, S.R.; Yeh, R.F.; Parsa, A.T.; McDermott, M.W.; Berger, M.S.; Dillon, W.P.;
Cha, S. Glioblastoma multiforme regional genetic and cellular expression patterns: Influence on anatomic and physiologic MR
imaging. Radiology 2010, 254, 564–576. [CrossRef]
34. Lonjon, M.; Mondot, L.; Lonjon, N.; Chanalet, S. Clinical factors in glioblastoma and neuroradiology. Neurochirurgie 2010, 56,
449–454. [CrossRef]
35. Louis, D.N.; Perry, A.; Reifenberger, G.; von Deimling, A.; Figarella-Branger, D.; Cavenee, W.K.; Ohgaki, H.; Wiestler, O.D.;
Kleihues, P.; Ellison, D.W. The 2016 World Health Organization Classification of Tumors of the Central Nervous System: A
summary. Acta Neuropathol. 2016, 131, 803–820. [CrossRef]
36. Louis, O.H.; Wiestler, O.D.; Cavenee, W.K. The 2016 World Health Organization Classification of Tumors of the Central Nervous System,
4th ed.; IARC Publication: Geneva, Switzerland, 2016; Volume 1.
37. Figarella-Branger, D.; Bouvier, C.; Moroch, J.; Michalak, S.; Burel-Vandenbos, F. Morphological classification of glioblastomas.
Neurochirurgie 2010, 56, 459–463. [CrossRef]
38. Banan, R.; Hartmann, C. The new WHO 2016 classification of brain tumors-what neurosurgeons need to know. Acta Neurochir.
2017, 159, 403–418. [CrossRef]
39. Zhao, Y.H.; Wang, Z.F.; Cao, C.J.; Weng, H.; Xu, C.S.; Li, K.; Li, J.L.; Lan, J.; Zeng, X.T.; Li, Z.Q. The Clinical Significance of
O(6)-Methylguanine-DNA Methyltransferase Promoter Methylation Status in Adult Patients With Glioblastoma: A Meta-analysis.
Front. Neurol. 2018, 9, 127. [CrossRef]
40. Zhang, Y.; Dube, C.; Gibert, M., Jr.; Cruickshanks, N.; Wang, B.; Coughlan, M.; Yang, Y.; Setiady, I.; Deveau, C.; Saoud, K.; et al.
The p53 Pathway in Glioblastoma. Cancers 2018, 10, 297. [CrossRef] [PubMed]
41. Eskilsson, E.; Rosland, G.V.; Solecki, G.; Wang, Q.; Harter, P.N.; Graziani, G.; Verhaak, R.G.W.; Winkler, F.; Bjerkvig, R.; Miletic, H.
EGFR heterogeneity and implications for therapeutic intervention in glioblastoma. Neuro Oncol. 2018, 20, 743–752. [CrossRef]
[PubMed]
42. Binabaj, M.M.; Bahrami, A.; ShahidSales, S.; Joodi, M.; Joudi Mashhad, M.; Hassanian, S.M.; Anvari, K.; Avan, A. The prognostic
value of MGMT promoter methylation in glioblastoma: A meta-analysis of clinical trials. J. Cell Physiol. 2018, 233, 378–386.
[CrossRef] [PubMed]
43. Benitez, J.A.; Ma, J.; D’Antonio, M.; Boyer, A.; Camargo, M.F.; Zanca, C.; Kelly, S.; Khodadadi-Jamayran, A.; Jameson, N.M.;
Andersen, M.; et al. PTEN regulates glioblastoma oncogenesis through chromatin-associated complexes of DAXX and histone
H3.3. Nat. Commun. 2017, 8, 15223. [CrossRef] [PubMed]
44. Lacroix, M.; Abi-Said, D.; Fourney, D.R.; Gokaslan, Z.L.; Shi, W.; DeMonte, F.; Lang, F.F.; McCutcheon, I.E.; Hassenbusch, S.J.;
Holland, E.; et al. A multivariate analysis of 416 patients with glioblastoma multiforme: Prognosis, extent of resection, and
survival. J. Neurosurg. 2001, 95, 190–198. [CrossRef] [PubMed]
45. Brown, T.J.; Brennan, M.C.; Li, M.; Church, E.W.; Brandmeir, N.J.; Rakszawski, K.L.; Patel, A.S.; Rizk, E.B.; Suki, D.; Sawaya, R.;
et al. Association of the Extent of Resection With Survival in Glioblastoma: A Systematic Review and Meta-analysis. JAMA Oncol.
2016, 2, 1460–1469. [CrossRef] [PubMed]
46. Chambless, L.B.; Kistka, H.M.; Parker, S.L.; Hassam-Malani, L.; McGirt, M.J.; Thompson, R.C. The relative value of postoperative
versus preoperative Karnofsky Performance Scale scores as a predictor of survival after surgical resection of glioblastoma
multiforme. J. Neurooncol. 2015, 121, 359–364. [CrossRef] [PubMed]
47. Marcus, H.J.; Vakharia, V.N.; Ourselin, S.; Duncan, J.; Tisdall, M.; Aquilina, K. Robot-assisted stereotactic brain biopsy: Systematic
review and bibliometric analysis. Childs Nerv. Syst. 2018, 34, 1299–1309. [CrossRef]
48. McGirt, M.J.; Villavicencio, A.T.; Bulsara, K.R.; Friedman, A.H. MRI-guided stereotactic biopsy in the diagnosis of glioma:
Comparison of biopsy and surgical resection specimen. Surg. Neurol. 2003, 59, 279–283. [CrossRef]
49. Eseonu, C.I.; Rincon-Torroella, J.; ReFaey, K.; Lee, Y.M.; Nangiana, J.; Vivas-Buitrago, T.; Quinones-Hinojosa, A. Awake
Craniotomy vs Craniotomy Under General Anesthesia for Perirolandic Gliomas: Evaluating Perioperative Complications and
Extent of Resection. Neurosurgery 2017, 81, 481–489. [CrossRef]
J. Pers. Med. 2021, 11, 258 12 of 14

50. Foster, C.H.; Morone, P.J.; Cohen-Gadol, A. Awake craniotomy in glioma surgery: Is it necessary? J. Neurosurg. Sci. 2019, 63,
162–178. [CrossRef]
51. Obermueller, T.; Schaeffner, M.; Shiban, E.; Droese, D.; Negwer, C.; Meyer, B.; Ringel, F.; Krieg, S.M. Intraoperative neuromonitor-
ing for function-guided resection differs for supratentorial motor eloquent gliomas and metastases. BMC Neurol. 2015, 15, 211.
[CrossRef]
52. Senders, J.T.; Muskens, I.S.; Schnoor, R.; Karhade, A.V.; Cote, D.J.; Smith, T.R.; Broekman, M.L. Agents for fluorescence-guided
glioma surgery: A systematic review of preclinical and clinical results. Acta Neurochir. 2017, 159, 151–167. [CrossRef]
53. Cho, S.S.; Salinas, R.; Lee, J.Y.K. Indocyanine-Green for Fluorescence-Guided Surgery of Brain Tumors: Evidence, Techniques,
and Practical Experience. Front. Surg. 2019, 6, 11. [CrossRef] [PubMed]
54. Stummer, W.; Pichlmeier, U.; Meinel, T.; Wiestler, O.D.; Zanella, F.; Reulen, H.-J. Fluorescence-guided surgery with 5-
aminolevulinic acid for resection of malignant glioma: A randomised controlled multicentre phase III trial. Lancet Oncol. 2006, 7,
392–401. [CrossRef]
55. Norred, S.E.; Johnson, J.A. Magnetic resonance-guided laser induced thermal therapy for glioblastoma multiforme: A review.
BioMed Res. Int. 2014, 2014, 761312. [CrossRef] [PubMed]
56. Kamath, A.A.; Friedman, D.D.; Akbari, S.H.A.; Kim, A.H.; Tao, Y.; Luo, J.; Leuthardt, E.C. Glioblastoma Treated With Magnetic
Resonance Imaging-Guided Laser Interstitial Thermal Therapy: Safety, Efficacy, and Outcomes. Neurosurgery 2019, 84, 836–843.
[CrossRef] [PubMed]
57. Carpentier, A.; Chauvet, D.; Reina, V.; Beccaria, K.; Leclerq, D.; McNichols, R.J.; Gowda, A.; Cornu, P.; Delattre, J.Y. MR-guided
laser-induced thermal therapy (LITT) for recurrent glioblastomas. Lasers Surg. Med. 2012, 44, 361–368. [CrossRef] [PubMed]
58. Pirro, V.; Alfaro, C.M.; Jarmusch, A.K.; Hattab, E.M.; Cohen-Gadol, A.A.; Cooks, R.G. Intraoperative assessment of tumor margins
during glioma resection by desorption electrospray ionization-mass spectrometry. Proc. Natl. Acad. Sci. USA 2017, 114, 6700–6705.
[CrossRef]
59. Santagata, S.; Eberlin, L.S.; Norton, I.; Calligaris, D.; Feldman, D.R.; Ide, J.L.; Liu, X.; Wiley, J.S.; Vestal, M.L.; Ramkissoon,
S.H.; et al. Intraoperative mass spectrometry mapping of an onco-metabolite to guide brain tumor surgery. Proc. Natl. Acad. Sci.
USA 2014, 111, 11121–11126. [CrossRef]
60. Stupp, R.; Mason, W.P.; van den Bent, M.J.; Weller, M.; Fisher, B.; Taphoorn, M.J.; Belanger, K.; Brandes, A.A.; Marosi, C.; Bogdahn,
U.; et al. Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma. N. Engl. J. Med. 2005, 352, 987–996.
[CrossRef]
61. Hegi, M.E.; Diserens, A.C.; Gorlia, T.; Hamou, M.F.; de Tribolet, N.; Weller, M.; Kros, J.M.; Hainfellner, J.A.; Mason, W.; Mariani,
L.; et al. MGMT gene silencing and benefit from temozolomide in glioblastoma. N. Engl. J. Med. 2005, 352, 997–1003. [CrossRef]
[PubMed]
62. Perry, J.R.; Laperriere, N.; O’Callaghan, C.J.; Brandes, A.A.; Menten, J.; Phillips, C.; Fay, M.; Nishikawa, R.; Cairncross, J.G.; Roa,
W.; et al. Short-Course Radiation plus Temozolomide in Elderly Patients with Glioblastoma. N. Engl. J. Med. 2017, 376, 1027–1037.
[CrossRef]
63. Keime-Guibert, F.; Chinot, O.; Taillandier, L.; Cartalat-Carel, S.; Frenay, M.; Kantor, G.; Guillamo, J.S.; Jadaud, E.; Colin, P.;
Bondiau, P.Y.; et al. Radiotherapy for glioblastoma in the elderly. N. Engl. J. Med. 2007, 356, 1527–1535. [CrossRef] [PubMed]
64. Mor, V.; Laliberte, L.; Morris, J.N.; Wiemann, M. The Karnofsky Performance Status Scale. An examination of its reliability and
validity in a research setting. Cancer 1984, 53, 2002–2007. [CrossRef]
65. Malmstrom, A.; Gronberg, B.H.; Marosi, C.; Stupp, R.; Frappaz, D.; Schultz, H.; Abacioglu, U.; Tavelin, B.; Lhermitte, B.; Hegi,
M.E.; et al. Temozolomide versus standard 6-week radiotherapy versus hypofractionated radiotherapy in patients older than 60
years with glioblastoma: The Nordic randomised, phase 3 trial. Lancet Oncol. 2012, 13, 916–926. [CrossRef]
66. Chukwueke, U.N.; Wen, P.Y. Use of the Response Assessment in Neuro-Oncology (RANO) criteria in clinical trials and clinical
practice. CNS Oncol. 2019, 8, CNS28. [CrossRef]
67. Weller, M.; Cloughesy, T.; Perry, J.R.; Wick, W. Standards of care for treatment of recurrent glioblastoma–are we there yet? Neuro
Oncol. 2013, 15, 4–27. [CrossRef] [PubMed]
68. Wann, A.; Tully, P.A.; Barnes, E.H.; Lwin, Z.; Jeffree, R.; Drummond, K.J.; Gan, H.; Khasraw, M. Outcomes after second surgery
for recurrent glioblastoma: A retrospective case-control study. J. Neurooncol. 2018, 137, 409–415. [CrossRef]
69. Xing, W.K.; Shao, C.; Qi, Z.Y.; Yang, C.; Wang, Z. The role of Gliadel wafers in the treatment of newly diagnosed GBM: A
meta-analysis. Drug Des. Dev. Ther. 2015, 9, 3341–3348. [CrossRef]
70. Sage, W.; Guilfoyle, M.; Luney, C.; Young, A.; Sinha, R.; Sgubin, D.; McAbee, J.H.; Ma, R.; Jefferies, S.; Jena, R.; et al. Local
alkylating chemotherapy applied immediately after 5-ALA guided resection of glioblastoma does not provide additional benefit.
J. Neurooncol. 2018, 136, 273–280. [CrossRef]
71. Grangeon, L.; Ferracci, F.X.; Fetter, D.; Maltete, D.; Langlois, O.; Gilard, V. How safe are carmustine wafers? Rev. Neurol. 2018, 174,
346–351. [CrossRef]
72. Taal, W.; Oosterkamp, H.M.; Walenkamp, A.M.E.; Dubbink, H.J.; Beerepoot, L.V.; Hanse, M.C.J.; Buter, J.; Honkoop, A.H.;
Boerman, D.; de Vos, F.Y.F.; et al. Single-agent bevacizumab or lomustine versus a combination of bevacizumab plus lomustine
in patients with recurrent glioblastoma (BELOB trial): A randomised controlled phase 2 trial. Lancet Oncol. 2014, 15, 943–953.
[CrossRef]
J. Pers. Med. 2021, 11, 258 13 of 14

73. Stupp, R.; Wong, E.T.; Kanner, A.A.; Steinberg, D.; Engelhard, H.; Heidecke, V.; Kirson, E.D.; Taillibert, S.; Liebermann, F.; Dbaly,
V.; et al. NovoTTF-100A versus physician’s choice chemotherapy in recurrent glioblastoma: A randomised phase III trial of a
novel treatment modality. Eur. J. Cancer 2012, 48, 2192–2202. [CrossRef] [PubMed]
74. Lombardi, G.; De Salvo, G.L.; Brandes, A.A.; Eoli, M.; Ruda, R.; Faedi, M.; Lolli, I.; Pace, A.; Daniele, B.; Pasqualetti, F.; et al.
Regorafenib compared with lomustine in patients with relapsed glioblastoma (REGOMA): A multicentre, open-label, randomised,
controlled, phase 2 trial. Lancet Oncol. 2019, 20, 110–119. [CrossRef]
75. Szopa, W.; Burley, T.A.; Kramer-Marek, G.; Kaspera, W. Diagnostic and Therapeutic Biomarkers in Glioblastoma: Current Status
and Future Perspectives. BioMed Res. Int. 2017, 2017, 8013575. [CrossRef]
76. Sasmita, A.O.; Wong, Y.P.; Ling, A.P.K. Biomarkers and therapeutic advances in glioblastoma multiforme. Asia Pac. J. Clin. Oncol.
2018, 14, 40–51. [CrossRef] [PubMed]
77. Chaddad, A.; Daniel, P.; Desrosiers, C.; Toews, M.; Abdulkarim, B. Novel Radiomic Features Based on Joint Intensity Matrices for
Predicting Glioblastoma Patient Survival Time. IEEE J. Biomed. Health Inform. 2019, 23, 795–804. [CrossRef] [PubMed]
78. Chaddad, A.; Daniel, P.; Sabri, S.; Desrosiers, C.; Abdulkarim, B. Integration of Radiomic and Multi-omic Analyses Predicts
Survival of Newly Diagnosed IDH1 Wild-Type Glioblastoma. Cancers 2019, 11, 1148. [CrossRef]
79. Hood, L.; Rowen, L. The Human Genome Project: Big science transforms biology and medicine. Genome Med. 2013, 5, 79.
[CrossRef] [PubMed]
80. Zuo, S.; Zhang, X.; Wang, L. A RNA sequencing-based six-gene signature for survival prediction in patients with glioblastoma.
Sci. Rep. 2019, 9, 2615. [CrossRef]
81. Lin, W.; Huang, Z.; Xu, Y.; Chen, X.; Chen, T.; Ye, Y.; Ding, J.; Chen, Z.; Chen, L.; Qiu, X.; et al. A three-lncRNA signature predicts
clinical outcomes in low-grade glioma patients after radiotherapy. Aging 2020, 12, 9188–9204. [CrossRef] [PubMed]
82. Stackhouse, C.T.; Gillespie, G.Y.; Willey, C.D. Exploring the Roles of lncRNAs in GBM Pathophysiology and Their Therapeutic
Potential. Cells 2020, 9, 2369. [CrossRef]
83. Marziali, G.; Signore, M.; Buccarelli, M.; Grande, S.; Palma, A.; Biffoni, M.; Rosi, A.; D’Alessandris, Q.G.; Martini, M.; Larocca,
L.M.; et al. Metabolic/Proteomic Signature Defines Two Glioblastoma Subtypes With Different Clinical Outcome. Sci. Rep. 2016,
6, 21557. [CrossRef]
84. Zhai, X.H.; Xiao, J.; Yu, J.K.; Sun, H.; Zheng, S. Novel sphingomyelin biomarkers for brain glioma and associated regulation
research on the PI3K/Akt signaling pathway. Oncol. Lett. 2019, 18, 6207–6213. [CrossRef] [PubMed]
85. Heiland, D.H.; Haaker, G.; Watzlawick, R.; Delev, D.; Masalha, W.; Franco, P.; Machein, M.; Staszewski, O.; Oelhke, O.; Nicolay,
N.H.; et al. One decade of glioblastoma multiforme surgery in 342 elderly patients: What have we learned? J. Neurooncol. 2018,
140, 385–391. [CrossRef] [PubMed]
86. Guo, D.; Bell, E.H.; Chakravarti, A. Lipid metabolism emerges as a promising target for malignant glioma therapy. CNS Oncol.
2013, 2, 289–299. [CrossRef] [PubMed]
87. Moren, L.; Bergenheim, A.T.; Ghasimi, S.; Brannstrom, T.; Johansson, M.; Antti, H. Metabolomic Screening of Tumor Tissue and
Serum in Glioma Patients Reveals Diagnostic and Prognostic Information. Metabolites 2015, 5, 502–520. [CrossRef]
88. Petralia, F.; Tignor, N.; Reva, B.; Koptyra, M.; Chowdhury, S.; Rykunov, D.; Krek, A.; Ma, W.; Zhu, Y.; Ji, J.; et al. Integrated
Proteogenomic Characterization across Major Histological Types of Pediatric Brain Cancer. Cell 2020. [CrossRef]
89. Batchelor, T.T.; Mulholland, P.; Neyns, B.; Nabors, L.B.; Campone, M.; Wick, A.; Mason, W.; Mikkelsen, T.; Phuphanich, S.; Ashby,
L.S.; et al. Phase III randomized trial comparing the efficacy of cediranib as monotherapy, and in combination with lomustine,
versus lomustine alone in patients with recurrent glioblastoma. J. Clin. Oncol. 2013, 31, 3212–3218. [CrossRef]
90. Weller, M.; Butowski, N.; Tran, D.D.; Recht, L.D.; Lim, M.; Hirte, H.; Ashby, L.; Mechtler, L.; Goldlust, S.A.; Iwamoto, F.;
et al. Rindopepimut with temozolomide for patients with newly diagnosed, EGFRvIII-expressing glioblastoma (ACT IV): A
randomised, double-blind, international phase 3 trial. Lancet Oncol. 2017, 18, 1373–1385. [CrossRef]
91. Jain, K.K. A Critical Overview of Targeted Therapies for Glioblastoma. Front. Oncol. 2018, 8, 419. [CrossRef]
92. Ene, C.I.; Holland, E.C. Personalized medicine for gliomas. Surg. Neurol. Int. 2015, 6, S89–S95. [CrossRef]
93. Harder, B.G.; Blomquist, M.R.; Wang, J.; Kim, A.J.; Woodworth, G.F.; Winkles, J.A.; Loftus, J.C.; Tran, N.L. Developments in
Blood-Brain Barrier Penetrance and Drug Repurposing for Improved Treatment of Glioblastoma. Front. Oncol. 2018, 8, 462.
[CrossRef] [PubMed]
94. Jahangiri, A.; Chin, A.T.; Flanigan, P.M.; Chen, R.; Bankiewicz, K.; Aghi, M.K. Convection-enhanced delivery in glioblastoma: A
review of preclinical and clinical studies. J. Neurosurg. 2017, 126, 191–200. [CrossRef] [PubMed]
95. Parodi, A.; Rudzińska, M.; Deviatkin, A.; Soond, S.; Baldin, A.; Zamyatnin, A. Established and Emerging Strategies for Drug
Delivery Across the Blood-Brain Barrier in Brain Cancer. Pharmaceutics 2019, 11, 245. [CrossRef]
96. Franklin, C.; Livingstone, E.; Roesch, A.; Schilling, B.; Schadendorf, D. Immunotherapy in melanoma: Recent advances and future
directions. Eur. J. Surg. Oncol. 2017, 43, 604–611. [CrossRef] [PubMed]
97. Kong, Z.; Wang, Y.; Ma, W. Vaccination in the immunotherapy of glioblastoma. Hum. Vaccines Immunother. 2018, 14, 255–268.
[CrossRef]
98. Kong, D.S.; Nam, D.H.; Kang, S.H.; Lee, J.W.; Chang, J.H.; Kim, J.H.; Lim, Y.J.; Koh, Y.C.; Chung, Y.G.; Kim, J.M.; et al. Phase
III randomized trial of autologous cytokine-induced killer cell immunotherapy for newly diagnosed glioblastoma in Korea.
Oncotarget 2017, 8, 7003–7013. [CrossRef]
J. Pers. Med. 2021, 11, 258 14 of 14

99. Swartz, A.M.; Batich, K.A.; Fecci, P.E.; Sampson, J.H. Peptide vaccines for the treatment of glioblastoma. J. Neurooncol. 2015, 123,
433–440. [CrossRef]
100. Eagles, M.E.; Nassiri, F.; Badhiwala, J.H.; Suppiah, S.; Almenawer, S.A.; Zadeh, G.; Aldape, K.D. Dendritic cell vaccines for
high-grade gliomas. Ther. Clin. Risk Manag. 2018, 14, 1299–1313. [CrossRef]
101. Mitchell, D.A.; Batich, K.A.; Gunn, M.D.; Huang, M.N.; Sanchez-Perez, L.; Nair, S.K.; Congdon, K.L.; Reap, E.A.; Archer, G.E.;
Desjardins, A.; et al. Tetanus toxoid and CCL3 improve dendritic cell vaccines in mice and glioblastoma patients. Nature 2015,
519, 366–369. [CrossRef]
102. Landry, A.P.; Balas, M.; Alli, S.; Spears, J.; Zador, Z. Distinct regional ontogeny and activation of tumor associated macrophages in
human glioblastoma. Sci. Rep. 2020, 10, 19542. [CrossRef]
103. Gregoire, H.; Roncali, L.; Rousseau, A.; Cherel, M.; Delneste, Y.; Jeannin, P.; Hindre, F.; Garcion, E. Targeting Tumor Associated
Macrophages to Overcome Conventional Treatment Resistance in Glioblastoma. Front. Pharmacol. 2020, 11, 368. [CrossRef]
104. Takenaka, M.C.; Gabriely, G.; Rothhammer, V.; Mascanfroni, I.D.; Wheeler, M.A.; Chao, C.C.; Gutierrez-Vazquez, C.; Kenison,
J.; Tjon, E.C.; Barroso, A.; et al. Control of tumor-associated macrophages and T cells in glioblastoma via AHR and CD39. Nat.
Neurosci. 2019, 22, 729–740. [CrossRef] [PubMed]
105. Yee, P.P.; Wei, Y.; Kim, S.Y.; Lu, T.; Chih, S.Y.; Lawson, C.; Tang, M.; Liu, Z.; Anderson, B.; Thamburaj, K.; et al. Neutrophil-induced
ferroptosis promotes tumor necrosis in glioblastoma progression. Nat. Commun. 2020, 11, 5424. [CrossRef]
106. Wang, T.; Cao, L.; Dong, X.; Wu, F.; De, W.; Huang, L.; Wan, Q. LINC01116 promotes tumor proliferation and neutrophil
recruitment via DDX5-mediated regulation of IL-1beta in glioma cell. Cell Death Dis. 2020, 11, 302. [CrossRef] [PubMed]
107. Stupp, R.; Taillibert, S.; Kanner, A.; Read, W.; Steinberg, D.; Lhermitte, B.; Toms, S.; Idbaih, A.; Ahluwalia, M.S.; Fink, K.; et al.
Effect of Tumor-Treating Fields Plus Maintenance Temozolomide vs. Maintenance Temozolomide Alone on Survival in Patients
With Glioblastoma: A Randomized Clinical Trial. JAMA 2017, 318, 2306–2316. [CrossRef]
108. Cammarata, F.P.; Torrisi, F.; Forte, G.I.; Minafra, L.; Bravata, V.; Pisciotta, P.; Savoca, G.; Calvaruso, M.; Petringa, G.; Cirrone,
G.A.P.; et al. Proton Therapy and Src Family Kinase Inhibitor Combined Treatments on U87 Human Glioblastoma Multiforme
Cell Line. Int. J. Mol. Sci. 2019, 20, 4745. [CrossRef]
109. Torrisi, F.; Minafra, L.; Cammarata, F.P.; Savoca, G.; Calvaruso, M.; Vicario, N.; Maccari, L.; Peres, E.A.; Ozcelik, H.; Bernaudin,
M.; et al. SRC Tyrosine Kinase Inhibitor and X-rays Combined Effect on Glioblastoma Cell Lines. Int. J. Mol. Sci. 2020, 21, 3917.
[CrossRef]
110. Torrisi, F.; Vicario, N.; Spitale, F.M.; Cammarata, F.P.; Minafra, L.; Salvatorelli, L.; Russo, G.; Cuttone, G.; Valable, S.; Gulino,
R.; et al. The Role of Hypoxia and SRC Tyrosine Kinase in Glioblastoma Invasiveness and Radioresistance. Cancers 2020, 12, 2860.
[CrossRef] [PubMed]
CHAPITRE III : TRAVAIL EXPERIMENTAL

1. Définition du problème

Le travail réalisé dans les deux premiers chapitres de cette thèse met en exergue plusieurs
problématiques. Dans le Chapitre I, nous avons montré par une perspective historique, comment la
neurochirurgie, qui est une discipline ancienne, a su s’approprier les grandes avancées

technologiques de chaque époque pour progresser. Ensuite, nous avons rappelé le rôle émergent

de la médecine de précision dans de nombreux domaines médicaux en insistant sur la neuro-

oncologie qui est aujourd’hui à l’orée d’une révolution grâce à l’interdisciplinarité et les outils du

big data qui permettent d’intégrer et exploiter les données des sciences omiques. L’exploitation de

ces informations va permettre de servir la neuro-oncologie et d’en faire une médecine de précision

centrée sur le patient. Dans de nombreux essais cliniques récents, les thérapies innovantes sont
fréquemment adaptées à chaque patient, se basant sur les caractéristiques intrinsèques de la tumeur

ce qui est l’essence même de la médecine personnalisée.

En parallèle de ce bond technologique, nous avons développé dans le Chapitre II un résumé

des connaissances actuelles du glioblastome. Cette pathologie est extrêmement complexe, infiltre la

substance blanche et s’accompagne ainsi d’une symptomatologie insidieuse. Historiquement, les

avancées sur cette pathologie se sont faites par des analyses de la sémiologie, de la radiologie et de

l’histologie rendues possibles par les progrès des techniques d’imagerie en coupes et de la

microscopie optique puis électronique. Depuis quelques années, la biologie moléculaire a modifié
les outils diagnostiques et bousculé les classifications établies. Néanmoins, le diagnostic, et la

thérapeutique du glioblastome passent par un geste chirurgical, invasif par définition et à risque de
morbidité. Sur le plan thérapeutique, la chirurgie doit s’accompagner d’une chimiothérapie et d’une
radiothérapie. Malgré ces traitements lourds, la médiane de survie est inférieure à 2 ans. Il y a donc

une nécessité d’une nouvelle approche pour améliorer le diagnostic de cette maladie et son
pronostic.
Le concept de biopsie liquide a émergé depuis quelques années. Son concept est de pouvoir

établir un diagnostic de manière non invasive pour le patient en évitant cette étape de chirurgie.
Les sciences omiques, et notamment la métabolomique, pourraient permettre de proposer ces
nouveaux outils diagnostiques. La petite taille des métabolites leur permet de franchir la barrière

hématoencéphalique et ainsi d’exprimer dans le plasma des patients, un profil métabolomique des

tumeurs intracérébrales telles que le glioblastome. L’objectif de notre travail expérimental était ainsi
de définir le profil métabolomique du glioblastome dans le plasma en le corrélant au tissu tumoral.

58
 2 Mise en place du design expérimental

2.1 Sélection des patients contrôles

Notre étude reposant sur une étude métabolomique ciblée et non ciblée, la sélection d’un
groupe contrôle est une étape prépondérante dans la mise en place d’un tel type d’étude. Pour le

choix du tissu contrôle, une analyse de la littérature retrouvait comme principaux groupes contrôles
du cortex cérébral issu de cortectomies ou les méningiomes. Nous avons fait des études
préliminaires avec ces deux tissus et retenu le cortex cérébral. La principale raison est que même si

plus rare, le tissu est macroscopiquement similaire et le méningiome correspond à une lésion
extraparenchymateuse ce qui est moins proche du gliome, lésion par définition

intraparenchymateuse. Concernant le plasma, nous avons opté pour du plasma de sujets dits sains

issu du commerce. En effet, il est facilement accessible, relativement peu onéreux et le caractère

hautement symptomatique du glioblastome fait qu'il peut être supposé aisément les patients
comme indemnes d’un gliome.

2.2 Type d’étude métabolomique

Les études métabolomiques peuvent être ciblées ou non ciblées. La différence est le postulat des

métabolites d’intérêt. Ainsi, une étude métabolomique ciblée permettra de chercher au sein de

classes de métabolites (lipides par exemple), ceux qui diffèrent entre le groupe d’intérêt et le groupe
contrôle. Une étude non ciblée explorera le maximum de métabolites d’intérêt sans préjugés de

classes ou de tailles des métabolites. Ces études ont des avantages et inconvénients qui leur sont

propres. Une étude ciblée sera plus précise dans le screening, mais pourra écarter à tort des

métabolites hautement discriminants si le postulat de base est faux. A l’inverse une étude non ciblée
donnera une analyse moins fine des métabolites, mais en exclura moins. Nous avons opté pour

deux types d’analyses ciblées et non ciblées pour profiter de leurs avantages respectifs. Il s’agit

également d’une validité externe de notre étude de vérifier si les métabolites d’intérêt en analyse
non ciblée sont discriminants dans l‘analyse ciblée.

Enfin, notre schéma d’étude s’appuie sur l’imagerie métabolomique à l’aide de la technologie

FTICR (spectrométrie de masse à résonance cyclotronique ionique). La technique FTICR est une
technique à très haute résolution qui permet de déterminer les masses moléculaires avec une grande

précision. De nombreuses applications de la FTICR-MS utilisent cette précision des masses pour

aider à déterminer la composition des molécules sur la base de la masse exacte. Il s’agit d’une
technologie de pointe, encore peu répandue dans la recherche médicale, mais qui dispose d’un

59
 potentiel significatif. L’intérêt est de permettre une analyse de coupes histologiques avec leurs

métabolites d’intérêt. De plus, après traitement des données, il est possible d’obtenir une analyse
spatiale des métabolites avec leur répartition au sein de la coupe étudiée. Ceci est particulièrement

intéressant pour l’analyse de phénomènes dynamiques, dans notre cas, la différence entre le
cerveau sain, le centre tumoral et la zone péri lésionnelle.

3 Application : Article III

Ce chapitre est présenté sous forme d’article accepté dans le journal Cancers.

Gilard V, Ferey J, Marguet F, Fontanilles M, Ducatez F, Pilon C, Lesueur C, Pereira T, Basset C, Schmitz-
Afonso Iet al (2021) Integrative Metabolomics Reveals Deep Tissue and Systemic Metabolic Remodeling
in Glioblastoma. Cancers. 2021:13, 20:5157. https://www.mdpi.com/2072-6694/13/20/5157.

60
 cancers
Article
Integrative Metabolomics Reveals Deep Tissue and Systemic
Metabolic Remodeling in Glioblastoma
Vianney Gilard 1 , Justine Ferey 2 , Florent Marguet 3 , Maxime Fontanilles 4,5 , Franklin Ducatez 2,6 ,
Carine Pilon 2 , Céline Lesueur 2 , Tony Pereira 4 , Carole Basset 3 , Isabelle Schmitz-Afonso 5 , Frédéric Di Fioré 7,8 ,
Annie Laquerrière 3 , Carlos Afonso 5 , Stéphane Derrey 9 , Stéphane Marret 6 , Soumeya Bekri 2, *
and Abdellah Tebani 2

1 Department of Neurosurgery, UNIROUEN, CHU Rouen, INSERM U1245, Normandie University,

76000 Rouen, France; Vianney.gilard@chu-rouen.fr
2 Department of Metabolic Biochemistry, UNIROUEN, CHU Rouen, INSERM U1245, Normandie University,
76000 Rouen, France; justine.ferey@inrae.fr (J.F.); franklin.ducatez@gmail.com (F.D.);
carine.pilon@chu-rouen.fr (C.P.); celine.lesueur@chu-rouen.fr (C.L.); abdellah.tebani@chu-rouen.fr (A.T.)
3 Department of Pathology, UNIROUEN, CHU Rouen, INSERM U1245, Normandie University,
76000 Rouen, France; florent.marguet@chu-rouen.fr (F.M.); carole.basset@chu-rouen.fr (C.B.);
annie.laquerriere@chu-rouen.fr (A.L.)
4 Institut de Biologie Clinique, CHU Rouen, 76000 Rouen, France; maxime.fontanilles@chb.unicancer.fr (M.F.);
tony.pereira@chu-rouen.fr (T.P.)
5 INSA Rouen, CNRS IRCOF, 1 Rue TesnieÌre, COBRA UMR 6014 Et FR 3038 University Rouen,
Normandie University, CEDEX, 76821 Mont-Saint-Aignan, France;
isabelle.schmitz-afonso@univ-rouen.fr (I.S.-A.); carlos.afonso@univ-rouen.fr (C.A.)



6 Intensive Care and Neuropediatrics, Department of Neonatal Pediatrics, INSERM U1245, CHU Rouen,


UNIROUEN, Normandie University, 76000 Rouen, France; stephane.marret@chu-rouen.fr
7 Normandy Centre for Genomic and Personalized Medicine, IRON Group, INSERM U1245, UNIROUEN,
Citation: Gilard, V.; Ferey, J.;
Marguet, F.; Fontanilles, M.; Ducatez, Normandie University, 76000 Rouen, France; Frederic.difiore@chu-rouen.fr
8 Department of Medical Oncology, Cancer Centre Henri Becquerel, Rue d’Amiens, 76000 Rouen, France
F.; Pilon, C.; Lesueur, C.; Pereira, T.;
9 Department of Neurosurgery, CHU Rouen, INSERM U1073, UNIROUEN, Normandie University,
Basset, C.; Schmitz-Afonso, I.; et al.
76000 Rouen, France; stephane.derrey@chu-rouen.fr
Integrative Metabolomics Reveals
* Correspondence: soumeya.bekri@chu-rouen.fr
Deep Tissue and Systemic Metabolic
Remodeling in Glioblastoma. Cancers
Simple Summary: This study aims to explore metabolic remodeling in plasma and tissue samples
2021, 13, 5157. https://doi.org/
in patients with glioblastoma through an integrated targeted and untargeted metabolomics-based
10.3390/cancers13205157
strategy. We report phospholipids, sphingomyelins, acylcarnitines, and acylglycerols as key impaired
Academic Editor: Peter Hau metabolic classes in glioblastoma.

Received: 8 September 2021 Abstract: (1) Background: Glioblastoma is the most common malignant brain tumor in adults. Its
Accepted: 3 October 2021 etiology remains unknown in most cases. Glioblastoma pathogenesis consists of a progressive infiltra-
Published: 14 October 2021 tion of the white matter by tumoral cells leading to progressive neurological deficit, epilepsy, and/or
intracranial hypertension. The mean survival is between 15 to 17 months. Given this aggressive
Publisher’s Note: MDPI stays neutral prognosis, there is an urgent need for a better understanding of the underlying mechanisms of
with regard to jurisdictional claims in glioblastoma to unveil new diagnostic strategies and therapeutic targets through a deeper under-
published maps and institutional affil- standing of its biology. (2) Methods: To systematically address this issue, we performed targeted
iations.
and untargeted metabolomics-based investigations on both tissue and plasma samples from pa-
tients with glioblastoma. (3) Results: This study revealed 176 differentially expressed lipids and
metabolites, 148 in plasma and 28 in tissue samples. Main biochemical classes include phospholipids,
acylcarnitines, sphingomyelins, and triacylglycerols. Functional analyses revealed deep metabolic
Copyright: © 2021 by the authors. remodeling in glioblastoma lipids and energy substrates, which unveils the major role of lipids
Licensee MDPI, Basel, Switzerland.
in tumor progression by modulating its own environment. (4) Conclusions: Overall, our study
This article is an open access article
demonstrates in situ and systemic metabolic rewiring in glioblastoma that could shed light on its
distributed under the terms and
underlying biological plasticity and progression to inform diagnosis and/or therapeutic strategies.
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).

Cancers 2021, 13, 5157. https://doi.org/10.3390/cancers13205157 https://www.mdpi.com/journal/cancers

Cancers 2021, 13, 5157 2 of 22

Keywords: glioblastoma; high-grade glioma; brain tumor; neuro-oncology; metabolomics; omics;

mass spectrometry; machine learning

1. Introduction
Glioblastoma (GBM) is the most common malignant brain tumor in adults [1]. Its
incidence, 2 to 5 per 100,000 [1,2], is still increasing [3,4], and its etiology remains un-
known in most cases. Glioblastoma pathogenesis consists of a progressive infiltration
of the white matter by tumoral cells leading to progressive neurological deficit, epilepsy,
and/or intracranial hypertension [5,6]. Multimodal magnetic resonance imaging (MRI) is
the first-line investigation to diagnose glioblastoma and guides treatment strategies [7].
The current standard of care in the management of glioblastoma is based on surgical re-
section or biopsy followed by adjuvant chemotherapy and radiotherapy [8–10]. Despite
many therapeutic trials [11–13], the prognosis of glioblastoma is poor, with a mean survival
between 15 to 17 months [14]. Given this poor prognosis, there is an urgent need for a better
understanding of the underlying mechanisms of glioblastoma to unveil new diagnosis
strategies and therapeutic targets through a deeper understanding of its biology. During
the last few years, Precision Medicine (PM) has been a new paradigm that has opened a
new era in personalized diagnosis and therapeutic approaches in oncology [15–18]. It is
a tailored approach in disease management driven by the patient’s attributes to deliver
personalized healthcare. The concept of PM is penetrating steadily into the clinical setting
with increasing evidence in different areas, including oncology [15], autoimmunity [19], or
inborn errors of metabolism [20]. The PM surge is shaped by omics-based technologies
and the data-driven medicine revolution [21–23]. Omics are technologies that allow for
a multilayer molecular deciphering of the cell, tissue, and organism in a holistic fashion.
These technologies have been explored in various avenues of medicine and health sciences,
such as screening, diagnosis, and prognosis [24]. Given their capacity to interrogate biology
at an unprecedented depth and breath, omics are also reshaping the biomarker discovery
landscape [25]. Metabolomics, one of these technologies [26], directly emulates the bio-
chemical activity of a biological system with high time sensitivity and spatial resolution.
Metabolomics is the closest biological information layer to the phenotype, given its position
in the downstream information flow. Thus, interrogating the metabolome, which is a
set of metabolites in a given biological system, is an appealing opportunity to parse the
biological process. Two main technologies are used in metabolomics: mass spectrometry
(MS) and nucleic magnetic resonance (NMR) [27,28]. The latter is nondestructive and
robust; however, the former exhibits higher sensitivity and broader metabolome coverage.
High-resolution mass spectrometry (HRMS) coupled to other analytical modalities, such
as liquid chromatography (LC-MS) or MALDI-MS, offers a great opportunity to achieve
higher metabolic coverage, thus, biological scope. Furthermore, ultra-high resolution mass
spectrometry instruments based on Fourier-transform ion cyclotron resonance (FTICR)
exhibits unique resolution, sensitivity, and specificity [29–31]. Thus, this technology offers
unprecedented exhaustive metabolome coverage and non-ambiguous molecular formula
assignments [31]. Thus, disease-related metabolic phenotype could be described by retriev-
ing metabolomic profiles through biospecimens, such as blood, urine, tissue samples, or
cerebrospinal fluid [32–34]. Previous human-based studies have described glioblastoma-
related metabolites [35–43], and to the best of our knowledge, no study has reported results
based on both plasma and tissue analysis from the same glioblastoma patients. The aim
of this study is to explore the glioblastoma biological landscape through an integrated
metabolomics analysis of plasma and tissue samples of glioblastoma patients. This strategy
presents a great opportunity to explore both the in-situ and systemic metabolic phenotypes
and unveil predictive metabolomic patterns that could help early non-invasive glioblas-
toma diagnosis at presentation or recurrence. It might also provide new insights regarding
glioblastoma biology and ultimately guide targeted treatment strategies.
Cancers 2021, 13, 5157 3 of 22

2. Materials and Methods

2.1. Patients and Blood Samples
From January 2020 to November 2020, twenty-nine patients with histologically doc-
umented glioblastoma were selected for the study. All patients were referred to our
neurosurgery department, and resection or biopsies of the tumor were performed in all
cases. Plasma and tissue samples were collected at surgery before initiating any treatment.
Medical charts were reviewed, and the following data were collected: gender, age at diagno-
sis, past medical history, revealing symptoms, radiological characteristics, type of surgery
performed, and histological and biological characteristics of the tumor. All samples used in
this study belong to a declared collection located in the Pathology Department (Prof. Annie
Laquerrière) of Rouen University Hospital in accordance with the relevant guidelines and
regulations and with the permission of the local authorities. All procedures performed in
studies involving human participants were in accordance with the ethical standards of the
institutional and/or national research committee and with the 1964 Helsinki Declaration
and its later amendments or comparable ethical standards.

2.2. Targeted Metabolomics

2.2.1. Reagents
All reagents, internal and calibration standards, quality controls, test mix, and a
patented 96-well filter plate required for the AbsoluteIDQ® p180 analysis are included in
the kit or provided by Biocrates Life Science AG (Innsbruck, Austria).

2.2.2. Sample Preparation

Sample preparation was carried out according to the manufacturer’s protocol [44,45].
Briefly, 10 µL of plasma were transferred to the upper 96-well plate and dried under
a nitrogen stream. Thereafter, 50 µL of a 5% PITC solution was added to derivatize
amino acids and biogenic amines. After incubation, the spots were dried again before
the metabolites were extracted using 5 mM ammonium acetate in methanol (300 µL) into
the lower 96-well plate for analysis after further dilution using the MS running solvent
A. Quantification was carried out using internal standards and a calibration curve [44,45].
The full list of 188 measured metabolites is presented in Table S1.

2.2.3. Liquid Chromatography and Tandem Mass Spectrometry

The AbsoluteIDQ® p180 kit is a fully automated assay based on phenyl isothiocyanate
(PITC) derivatization of the target analytes in bodily fluids using internal standards for
quantitation. Amino acids and biogenic amines are determined in LC-MS mode, acyl-
carnitines, phospholipids, sphingomyelins, and the sum of hexoses are analyzed in flow
injection analysis (FIA). The analyses were performed following the instructions of the kit
manufacturer using the liquid chromatography instrument prominence Shimadzu UFLC
System (Shimadzu, Prominence, Kyoto, Japan) coupled to the 4000 Qtrap mass spectrome-
ter (Sciex, Framingham, MA, USA) with an electrospray ion source. A system suitability
test was conducted before each batch of the samples (analysis of a standard mixture)
to warm up the LC-MS/MS system and check the inter-day performance of the system.
Data acquisition and processing were performed using the Analyst 1.5 software (Sciex,
Framingham, MA, USA). Twenty microliters of the sample extract were used in the flow
injection analysis (FIA) in positive mode to capture acylcarnitines, glycerophospholipids,
and sphingolipids, while hexoses were monitored in a subsequent negative mode run. All
FIA injections were carried out using the Biocrates MS Running Solvent. All metabolites
were identified and quantified using isotopically labeled internal standards and multiple
reaction monitoring (MRM) as optimized and provided by Biocrates Life Sciences AG
(Innsbruck, Austria).
Cancers 2021, 13, 5157 4 of 22

2.3. Untargeted Metabolic Phenotyping

2.3.1. Reagents
Water, acetonitrile, isopropanol, and formic acid were optima LCMS grade and pur-
chased from Fisher Chemicals. Ammonium formate was LCMS grade (Sigma–Aldrich,
Saint-Quentin Fallavier, France). Leucine Enkephalin (Sigma–Aldrich, Saint-Quentin
Fallavier, France) at a concentration of 2 mg/mL (in isopropanol/water, 50/50) was used
as a reference for mass measurements. Poly-DL-alanine was prepared in 50:50 (v/v) wa-
ter/acetonitrile at 10 mg/L and used for ion mobility cell calibration.

2.3.2. Sample Preparation

Human plasmas were prepared using a liquid biphasic approach. Lipids were ex-
tracted from 100 µL of plasma by adding 280 µL chloroform, 140 µL methanol, and 84 µL
water. The mixtures were vortex-mixed, cooled in ice for 15 minutes, and centrifuged
10 min at 12,000 rpm. The bottom phase was removed, transferred to glass UPLC vials,
and N2 evaporated. The resulting phase was re-suspended in 200 µL ACN/H2 O/IPrOH
(1:1:1). A pooled quality control (QC) was prepared by mixing 5 µL of each plasma sample.
Four QC dilutions at 1:2, 1:4, 1:8, and 1:16 were prepared from the QC sample. Samples
were stored at −20◦ C before use.

2.3.3. Chromatographic Conditions

The liquid chromatography separation was performed with a UHPLC system (Dionex
Ultimate 3000 UPLC+, Thermo Scientific, San Jose, CA, USA). Separation was carried out at
50 ◦ C using a 1.0 × 100 mm Acquity UPLC HSS T3 column (Waters), with a particle size of
1.8 µm, equipped with a 0.2 µm prefilter. Mobile phase A consisted of ACN/H2 O (40/60)
containing 10 mM ammonium formate; mobile phase B consisted of IPrOH/ACN (90/10).
Both solvents contained 0.1% formic acid. The flow rate was maintained at 100 µL min−1 .
A mobile phase gradient separation was performed over 20 min according to the following
steps: 10% B at 0 min, 56% B at 2 min, 75% B at 10 min, 99% B at 12–15 min, 10% B at
16–20 min. Sample analysis order was randomized to avoid the potential for confounding
critical variables with analytical run order effects.

2.3.4. Ion Mobility and Mass Spectrometry

The U-HPLC system was coupled to a hybrid quadrupole orthogonal time-of-flight
(TOF) mass spectrometer (SYNAPT G2 HDMS, Waters MS Technologies, Manchester, UK).
The mass spectrometer was operated in both positive and negative electrospray ionization
mode. A mass range of m/z 50−1200 was used. The source and IMS parameters are summa-
rized in Table S2. Leucine enkephalin was used as the lock mass [M + H]+ at m/z 556.2771
and [M-H]− at m/z 554.26202. Sodium formate solution was used for external instrument
calibration. The Synapt G2 HDMS was equipped with a traveling wave “Triwave™” ge-
ometry, including an ion mobility cell (IMS T-wave). The TOF analyzer was operated in
the V resolution mode with an average mass resolution of m/∆m 20,000 (full-width at half-
maximum definition). Data acquisition of an ion mobility experiment consisted of 200 bins.
The cross-collision section (CCS) values, obtained in nitrogen, were experimentally de-
termined using singly charged Poly-DL-alanine oligomers as the TWIM calibrant species
for ESI+. The CCS values were derived according to previously reported procedures [46]
with the following general equation ln(CCS’) = xln(dt’) + ln(A). Calibration values and
curves are shown in the Supplementary Materials (Table S3). The ion mobility resolution
was ~40 Ω/∆Ω (fwhm). The N2 CCS values reported were determined at the apex of the
ion-mobility peak.

2.3.5. Raw Data Processing

All LC-IM–MS raw data files data processing, peak detection, and peak matching
across samples using retention time (tR ) correction and chromatographic alignment along
with drift time were performed using Progenesis QI (Waters MS Technologies, Manchester,
Cancers 2021, 13, 5157 5 of 22

UK) to yield a data matrix containing retention times, accurate masses, drift time, and
peak intensities. Experimental CCS were determined, as stated above, to include CCS
instead of drift time values in the data matrix. CCS errors between experimental and
theoretical values (%) were then calculated from the LipidsCCS Web Server (Zhu Lab) [47].
The preprocessing step resulted in an X-matrix where tR, CCS, and m/z values were
concatenated into “tR _m/z_CCS” features (in columns) present in each sample (in rows)
with corresponding peak areas.

2.3.6. Quality Control

Aliquoted 10 µL extracts were mixed to generate pooled quality control samples (QCs).
The QCs and solvent blank samples (mobile phase) were injected sequentially in-between
the plasma samples. In addition, a dilution series of QC samples (1:2, 1:4, 1:8, and 1:16
original concentration) were used to assess the quality of the extracted features. In this
study, we used a filter strategy in which the features’ intensity must be correlated to the
matrix concentrations in a series of diluted QC samples in order to be included in further
analysis. Feature groups with a correlation coefficient of less than 0.70 were removed
from the dataset. Furthermore, datasets were refined by the removal of feature groups
that did not meet the threshold of peak area measurement precision prior to data analysis.
This approach used RSD values derived from repeated measurements of a pooled QC
sample. The threshold was set to RSD < 25% to enhance the biological interpretation of
metabolomics data.

2.4. Tissue Metabolic Phenotyping

2.4.1. Reagents
MALDI matrices, including 2.5-dihydroxybenzoic acid (DHB) and 9-aminoacridine
hydrochloride monohydrate (9-AA), were purchased from Sigma-Aldrich (St. Louis, MO,
USA). Sodium trifluoroacetate (NaTFA, Sigma–Aldrich) at 0.1 mg mL−1 (ACN/H2 O 50:50
[v/v]) was used for external calibration before each analysis.

2.4.2. Sample Preparation

Fresh frozen 10-µm sections of human brain tissue were mounted on conductive
ITO-coated glass slides 75 × 25 mm (Bruker, Bremen, Germany) and stored at −80 ◦ C until
analysis. Matrix solutions were applied with an automatic micro sprayer HTX TM-Sprayer
(HTX Imaging, Chapel Hill, NC, USA) as previously described [31]. In the positive-ion
mode, the DHB matrix (30 mg mL−1 in MeOH/H2 O 50:50 [v/v]) was deposited with the
following parameters: nozzle temperature 80 ◦ C, nozzle velocity 1200 mm min−1 , N2
pressure 10 psi, N2 flow rate 2 L min−1 , number of passes 12, flow rate 100 µL min−1 ,
and track spacing 3 mm. In negative-ion mode, 9-AA matrix (10 mg mL−1 , EtOH/H2 O
70:30 [v/v]) was sprayed using the following parameters: nozzle temperature 90 ◦ C, nozzle
velocity 1200 mm min−1 , N2 pressure 10 psi, N2 flow rate 3 L min−1 , number of passes 2,
flow rate 120 µl min−1 , track spacing 3 mm, and drying time 30 s between passes. Each
slide was vacuum-dried before analysis.

2.4.3. Data Acquisition and Data Processing

Data were acquired on a FTICR instrument (SolariX XR, Bruker, Bremen, Germany)
equipped with a 12-Tesla superconducting magnet and a dynamically harmonized ICR
cell. This instrument was also equipped with both a laser desorption ionization source
(Smartbeam II, Nd:YAG × 3 laser at 355 nm, Bruker, Bremen, Germany) and an electro-
spray (ESI) source. Each MALDI spectrum for each position was the result of 1 scan and
500 consecutive laser shots. Spectra were acquired over a 100 µm raster. Before imaging
analyses, the instrument was externally calibrated in the required mode by NaTFA infusion
via an ESI source, then internally calibrated by assigning known metabolites from m/z
150–1000 via MALDI source. Thus, the instrument was auto calibrated during image
acquisition. In positive-ionization mode, calibration was performed by assigning C7 H6 O4
Cancers 2021, 13, 5157 6 of 22

(m/z 155.033885 [M+H]+ , matrix peak), C7 H15 NO3 (m/z 162.112470 [M+H]+ , carnitine),
C7 H6 O4 (m/z 177.015829, [M+Na]+ , matrix peak), C5 H14 NO4 P (m/z 184.073321, [M+H]+ ,
phosphocholine), C9 H17 NO4 (m/z 204.123034, [M+H]+ , acetyl carnitine), C14 H8 O6 (m/z
273.039364, [M+H]+ , matrix peak), C21 H12 O9 (m/z 409.055408, [M+H]+ , matrix peak),
C40 H80 NO8 P (m/z 734.569432, [M+H]+ , lipid), and C42 H82 NO8 P (m/z 798.54096, [M+K]+ ,
lipid). In the negative-ion mode, the assigned peaks were C13 H10 N2 (m/z 193.077122,
[M−H]− , matrix peak), C10 H15 N5 O10 P2 (m/z 426.022139, [M-H]− , adenosine diphos-
phate), C10 H16 N5 O13 P3 (m/z 505.988470, [M−H]− , adenosine triphosphate), C41 H80 NO8 P
(m/z 744.554879, [M−H]− ), and C47 H83 O13 P (m/z 885.549853, [M−H]− ). Data size was set
at 2 million points for a transient length of 0.87 s, and spectra were acquired with a 97%
data file reduction. A single MSI measurement was performed for each specimen. Images
were captured using FTMS control and FlexImaging (v 5.0, Bruker) software. Images were
processed with SCiLS Lab Pro software (Bruker Daltonics, Bremen, Germany). The total
ion current method was used for normalization, and m/z intervals were automatically set
at ±1 ppm. Images were viewed using both FlexImaging and SCiLS Lab software (Bruker
Daltonics, Bremen, Germany).

2.5. Metabolite Annotation and Identification

Preliminary assignments based on accurate mass measurements were performed
using the mass spectrometry databases, METLIN [48] and HMDB [49], using a threshold
of ±2 ppm. For some metabolites, the precise raw formula led to one hit. For CCS-based
annotations, the AllCCS database was used [50]. Only positive ion mode highlighted
discriminative ions. Other ions were identified via “on-tissue” tandem mass spectrometry
experiments using MALDI tandem MS/MS. Ions of interest were first isolated using a
window of ± 1 Da, then fragmented by collision-induced dissociation with energy levels
between 10 and 40 eV. For each MS/MS analysis, 50 scans were accumulated for better
sensitivity. Spectra were reprocessed using Data Analysis 4.4 software (Bruker Daltonics,
Bremen, Germany) and recalibrated with the single-point calibration option. The full list of
annotated features is presented in Tables S4 and S5.

2.6. Circulating Cell-Free DNA Acquisition

Circulating cell-free DNA was extracted from blood samples right before surgery
and as previously described [51]. Briefly, six milliliters of whole blood was collected in
tubes containing ethylenediaminetetraacetic acid (EDTA). Within two hours after blood
collection, the tubes were centrifuged; the plasma was then extracted and stored at −80 ◦ C
until use. cfDNA was extracted from 1 mL to 5 mL of plasma using the QIAamp Cir-
culating Nucleic Acid kit® (Qiagen, Hilden, Germany) according to the manufacturer’s
instructions. The sample was eluted in a final volume of 30 µL and stored at −20 ◦ C.
Double-stranded DNA quantification was performed by a fluorometric method using a
Quantit™ PicoGreen®dsDNA Assay kit (Invitrogen, Carlsbad, CA, USA) and a Twinkle
LB970 microplate fluorimeter (Berthold, Bad Wildbad, Germany). For each sample, cfDNA
quantification was performed in duplicate from 2 µL of eluate, and normalization was
performed using a standard calibration curve of known concentrations of standard dsDNA
(from 0 to 10 ng). Overall, cfDNA concentration reflects both cell-free genomic DNA as
well as circulating tumor DNA (ctDNA).

2.7. Data Analysis

Ion intensities or absolute concentrations were log-transformed and Pareto-scaled [52].
Missing values were imputed using the nearest neighbor averaging algorithm using the
impute.knn function in the impute R package. The full data matrices are presented in
Tables S6–S8. Univariate analyses were performed using t-tests or Mann–Whitney U de-
pending on the normal distribution of the data. The Limma package [53] was used for
differential analysis using sex, and age was taken into account by adding it as a covariate.
Spearman correlation analysis was performed using R software. Euclidean distance was
Cancers 2021, 13, 5157 7 of 22

used as a similarity measure in the clustering analysis. Principal Component Analysis was
used as a dimension reduction technique using log-transformed and Pareto-scaled datasets.
False discovery rates were corrected using the Benjamini–Hochberg–Yekutieli method [54],
and p < 0.05 was considered statistically significant. For network analysis, the first step
was to compute several glasso-based networks (GLN) [55]. Three kinds of GLNs were
calculated; control + disease samples, control samples only, and all the combinations of
samples, including control + “disease-minus-one-patient” to get patient-specific networks.
Networks were then constructed from each GLNs data matrix and pruned with each other
to get specific networks. The idea of network pruning is to remove edges in a general
network that are also found in a more specific network. So, we pruned the “disease +
control” network with the “control” network in order to keep only the edges that were
disease-specific. Thus, this step resulted in a “disease-specific” network. This step was
done using the CTD R package [56]. Using the same strategy, networks of controls +
“disease-minus-one” samples were pruned with the controls samples network to obtain
a “disease-specific-minus-one-patient” network. This network was then pruned with the
“disease-specific” network calculated above in order to extract “patient-specific” metabolic
signatures. A summary overview of the network strategy is presented in Figure S1. The
metabolites present in more than 50% of these “patient-specific” networks were selected
to build a Consensus-Network, enabling the visualization of key metabolic signatures for
the disease. To test the discriminatory power of this signature, Random Forest models
were tuned for every possible combination of metabolites from the Consensus-Network.
Random Forest models were built using the ranger package [57] and the caret package in
R [58]. The models were tuned over ~50 repeats to obtain robust classification probabilities.
Performances of the models were assessed with the MLeval package in R. The main metric
for predictive performance assessment was the area under the curve (AUC) for the resulting
receiver operating characteristic (ROC) curve. All analyses were done using R software [59].
All data generated or analyzed during this study are included in this published article and
its Supplementary Materials.

3. Results
3.1. Cohort Description
The mean glioblastoma showed a male predominance of 82%. The mean age at
diagnosis was 66 ± 11.4 years for females and 64 ± 11.1 years for males. Regarding
the performed surgery, it consisted of a biopsy in 56.7% and resection in the other cases.
The mean controls age was 39 ± 12.5 years for females and 37 ± 10.9 years for males. A
cohort overview is presented in Figure 1, and a summary is presented in Table 1. The
detailed clinical and biological data are listed in Table S9. Three different metabolomics
methods were used to explore metabolic profile differences between glioblastoma and
control samples. They included untargeted mass spectrometry imaging on the tissue
samples and both targeted and untargeted analysis on plasma samples. The full list of
analyzed metabolites is presented in Tables S4 and S5, and the full data matrices are
presented in Tables S6–S8.

Table 1. Main cohort summary.

Characteristic Control (Brain), n = 6 1 Control (Plasma), n = 60 1 Glioblastoma, n = 22 1 p-Value 2

Sex 0.023
Female 2 (33%) 30 (50%) 4 (18%)
Male 4 (67%) 30 (50%) 18 (82%)
Age (Years) 39 (25, 46) 34 (27, 42) 62 (54, 66) <0.001
ATRX_mutation
Absent 1 (6.2%)
Present 15 (94%)
Cancers 2021, 13, 5157 8 of 22

Table 1. Cont.

Characteristic Control (Brain), n = 6 1 Control (Plasma), n = 60 1 Glioblastoma, n = 22 1 p-Value 2

Unknown 6
IDH_mutation
Absent 15 (88%)
Present 2 (12%)
Unknown 5
1 n (%); Median (IQR), IQR: Interquantile range; 2 Fisher’s exact test; Kruskal–Wallis rank-sum test.

Figure 1. Cohort overview. (A) Cohort description and sex distribution related to each dataset.
(B) Age distribution by sex and dataset.

3.2. Unsupervized Exploratory Analysis

To analyze the data, an exploratory approach was first used. This unsupervised
analysis aimed to track samples’ clustering trends based on their underlying metabolic
profiles. We unveiled these grouping trends using Spearman correlations and hierarchical
clustering of the samples. The results showed a clear separation of two distinct groups
related mainly to control and GBM samples Figure 2A–C. This was observed in all three
datasets (MSI, untargeted, or targeted plasma metabolomics) as shown in the heatmaps in
Figure 2A–C. Higher-resolution heatmaps are presented in Figure S2, and full correlation
matrices are presented in Tables S10–S12. The same clustering trends were also observed
using the PCA scores plots, Figure 2D, that showed two clear groups related to GBM and
control samples. This separation was mainly observed on the PC1 dimension. The PCA
scores matrices are presented in Table S13.
Cancers 2021, 13, 5157 9 of 22

Figure 2. Unsupervised exploratory analysis overview. (A) Heatmap of the sample’s spearman correlation analysis based
on untargeted FT-ICR-MS data. (B) Heatmap of the sample’s spearman correlation analysis based on untargeted LC-IMS-
MS data. (C) Heatmap of the sample’s spearman correlation analysis based on targeted LC-MS/MS data. (D) Principal
Component Analysis scores plot based on untargeted FT-ICR-MS (Explained variance: PC1 = 65%, PC2 = 7%), untargeted LC-
IMS-MS (Explained variance: PC1 = 55%, PC2 = 15%) and targeted LC-MS/MS (Explained variance: PC1 = 32%, PC2 = 16%)
datasets, respectively. (E) Principal Component Analysis loadings plot based on untargeted FT-ICR-MS (Explained variance:
PC1 = 65%, PC2 = 7%), untargeted LC-IMS-MS (Explained variance: PC1 = 55%, PC2 = 15%) and targeted LC-MS/MS
(Explained variance: PC1 = 32%, PC2 = 16%) datasets, respectively.
Cancers 2021, 13, 5157 10 of 22

3.3. Differential Expression Analysis

To go further in the analysis, we performed a differential analysis between the
two groups GBM versus Control, using each of the three datasets. The analysis identified
176 annotated metabolites that were differentially expressed between GBM and control sam-
ples. The plasma-based comparison yielded 148 metabolites, including 75 phospholipids
(69 glycerophosphocholines, 4 phosphatidylcholines, and 2 phosphosphingolipids), 20 sph-
ingomyelins, 15 acylcarnitines, 14 amino acids, 12 triacylglycerids, 10 biogenic amines,
1 cholestan steroid, and 1 diacylglycerol. The tissue-based comparison unveiled 28 metabo-
lites, including 17 phospholipids (4 glycerophosphocholines, 3 phosphoinositols, 1 phos-
phatidylcholine, 2 phosphosphingolipids, 2 glycerophosphoglycerophaosphate, 2 glyc-
erophosphoglycerid, 1 glycosphyngolipid, 1 glycerosphingoserine, and 1 glycerophospho-
ethanolamine), 5 acylcarnitines, 1 diacylglycerol, 2 sphingomyelins, 2 triacylglycerols, and
1 steroid ester. The summary results are presented in Figure 3A. The list of metabolites
and their statistics-related metrics is presented in Table S14. Furthermore, we compared
the retrieved metabolites with previously reported human GBM metabolomics-based lit-
erature [38–41,60]. The summary results are presented in Figure 3B, and the full results
are presented in Table S15. The overlap with previously reported metabolites includes
4 biogenic amines (Putrescine, cis-4-Hydroxyproline, trans-4-Hydroxyproline, Spermine),
8 amino acids (Glutamine, Asparagine, Ornithine, Lysine, Tryptophan, Citrulline, Threo-
nine, Valine), 1 steroid ester (CE(22:6)) and Carnitine. The related boxplots and correlations
are presented in Figures S3 and S4, respectively. In contrast, the metabolites that were not
reported in the above-mentioned literature include mainly Acylcarnitines, Biogenic amines,
Cholesterol, Phosphatidylcholines, and Triacylglycerols. Based on adjusted p-values, the
top metabolites are presented in Figure 4, and their Spearman correlations are presented in
Figure S5.

3.4. Correlation Analysis

Given the opportunity to have in-tissue metabolomics with matched-plasma samples,
we compared the plasma-based and tissue-based metabolic profiles and found ten lipids
that overlap between the two lists. These include, phosphatidylcholines LPC(18:0), PC(36:5),
PC(40:6), PC(32:1), sphingomyelins SM(33:1), SM(36:2) and triacylglycerides TG(56:7),
TG(52:2), TG(52:4), TG(52:3). The boxplots of the top five lipids are presented in Figure 5.
They include PC(40:6), PC(32:1), SM(33:1), TG(52:2), and TG(52:3). We also evaluated
these lipid intercorrelations between tissue samples and their matched plasma samples.
The figure did not show a high correlation between plasma and brain levels. However,
the heatmap showed a high intra-matrix correlation. The top metabolites (abs(log fold
change > 1.5) and adjusted p-value < 0.01) are presented in Figure 5. Furthermore, we
explored the overall intra-GBM group variability using the coefficient of variation as a proxy
in both tissue and plasma samples. The results showed high inter-tissue variability (CV
median = 127 %) compared to plasma (CV median = 36%) (Figure S6A,B). This highlights
an inter-tumor metabolic heterogeneity. We also explored the variability of the above-
mentioned overlap lipids that clearly showed the same observation (Figure S6C). The full
metabolite correlation matrix is presented in Table S16.
Cancers 2021, 13, 5157 11 of 22

Figure 3. Differential analysis overview. (A) Differential analysis results between control and glioblastoma samples
highlighting metabolic classes. One hundred and seventy-six metabolites were identified based on an adjusted p-value of
0.05. One hundred and forty-eight plasma-based metabolites and 28 Tissue-based metabolites. (B) Upper: Ven diagram
representation of the overlap between the identified differentially expressed metabolites in Plasma versus Brain. Lower:
Ven diagram representation of the overlap between the identified differentially expressed metabolites in this study versus
reported human metabolomics-based GBM reported literature. (C) Heatmap showing the relative levels of statistically
differential metabolites in different patients’ tissues using FT-ICR-MS data. (D) Heatmap showing the relative levels of
statistically differential metabolites in different patients’ plasma using LC-IMS/MS data. (E) Heatmap showing the relative
levels of statistically differential metabolites in different patients’ plasma using LC-MS/MS data. Detailed results are
presented in Table S13.
Cancers 2021, 13, 5157 12 of 22

Figure 4. Boxplots of the top (based on adjusted p-values) twenty newly reported metabolites and related biological matrix.

Given the rising interest in using circulating free DNA (cfDNA) in oncology [51],
we also assessed the associations between the retrieved metabolic profile and cfDNA us-
ing Spearman correlations. The analysis yielded 13 positive correlations and 2 negative
correlations with a cut-off set at adjusted p-value < 0.05 and abs(Spearman rho) > 0.25.
The positively correlated metabolites included Ornithine (Spearman rho = 0.52), Lysine
(Spearman rho = 0.49), Phenylalanine (Spearman rho = 0.46), Propionylcarnitine (Spearman
rho = 0.45), Carnitine (Spearman rho = 0.45), Hexadecenoylcarnitine (Spearman rho = 0.42),
Tetradecadienylcarnitine (Spearman rho = 0.4), Hexenoylcarnitine (Spearman rho = 0.35),
Spermine (Spearman rho = 0.3), SM C18:1 (Spearman rho = 0.28), Butenylcarnitine (Spear-
man rho = 0.29), SM C16:1 (Spearman rho = 0.29), and Asymmetric dimethylarginine
(Spearman rho = 0.27). The negatively correlated metabolites included two biogenic amines
Serotonin (Spearman rho = −0.68) and Methionine sulfoxide (Spearman rho = −0.29). A
network visualization is presented in Figure 6, and full results are presented in Table S17.
The full heatmap correlation matrix is presented in Figure S7.
Cancers 2021, 13, 5157 13 of 22

Figure 5. Overlap analysis between tissue-based and plasma-based metabolomics results. (A) Brain tissue section distribu-
tion of selected overlap lipids. Ion images were generated using SCiLs Lab software. (B) Brain tissue sections boxplots of
the selected top five differentially expressed metabolites among the overlap lipids with related adjusted p-values. The y-axis
shows the log-scaled average intensity. (C) Plasma boxplots of the selected top five differentially expressed lipids among
the overlap metabolites with related adjusted p-values. The y-axis shows the log-scaled intensity. (D) Spearman correlation
analysis between the selected top five lipids that overlap between tissue-based and plasma-based metabolomics analysis.

3.5. Network and Predictive Machine Learning Analysis

Using the targeted plasma metabolomic dataset, different correlation networks were
built using control, disease, or both samples. Using these correlation networks, 29 patient-
specific metabolic signatures were extracted (Figures S8–S36, Tables S18–S20). These
patient-specific metabolic networks highlighted the metabolic individuality of each pa-
tient. Based-on this, we explored the similarity of these signatures between patients. The
results revealed 69 unique metabolites spanning phosphatidylcholines (57%), amino acids
(14%), acylcarnitines (12%), sphingomyelins (12%), and biogenic amines (6%). Eleven
metabolites were found in only one signature and included aspartate, proline, valine,
hexanoyl-carnitine, SM C16:0, lysoPC a C16:0, PC aa C42:1, PC aa C42:4, PC aa C42:5, PC
ae C32:2, PC ae C34:0, PC ae C38:6. Two patients with mutated IDH status, PGB_04 and
PGB_24, both males at 35 and 45 years old, respectively, yielded the following signatures.
Cancers 2021, 13, 5157 14 of 22

For PGB_04 the signature included 11 metabolites; 6 phosphatidylcholines, 3 acylcarnitines,

and 2 sphingomyelins (Figure S11) and for PGB_024 male at 35 years old, the signature
included 18 metabolites; 14 phosphatidylcholines, 1 acylcarnitine, 1 biogenic amine, and
2 sphingomyelins (Figure S31). Then, we identified the most redundant metabolites found
in more than 50% of the patients’ signatures which included eight metabolites, seven phos-
phatidylcholines (PC aa C28:1, PC aa C32:2, PC aa C34:1, PC aa C36:0, PC aa C36:5, PC
aa C36:6), and two sphingomyelins (SM C18:0, SM C24:1) (Figure 7A,B, Table S21). The
related boxplots are presented in Figure 7C. Based on this signature, we explored the
predictive performance of each lipid and all its combinations using predictive machine
learning. We built Random Forest predictive models based on each lipid alone as well all
possible combinations of the eight lipids. Area under curve and ROC curves were used
as performance metrics (Table S22). The eight univariate models and their combinations
are shown in Figure 7C. These models included PC aa C36:6 (AUC = 0.949: 0.891–1.000),
PC aa C34:1 (AUC = 0.907: 0.83–0.983), PC aa C36:5 (AUC = 0.864: 0.773–0.955), SM
C18:0 (AUC = 0.789: 0.68–0.897), PC aa C28:1 (AUC = 0.757: 0.643–0.871), PC aa C36:0
(AUC = 0.72: 0.601–0.839), PC aa C32:2 (AUC = 0.712: 0.591–0.832), SM C24:1 (AUC = 0.686:
0.563–0.809). All included lipids models showed an AUC = 0.936: 0.871–1.000. It is worth
mentioning that the most predictive model was the PC aa C36:6 (AUC = 0.949: 0.891–1.000).
The full list of investigated models is presented in Table S23.

Figure 6. Spearman correlation analysis between metabolic profiles and free circulating DNA. Cut-off was set at adjusted
p-value > 0.05 and abs(Spearman rho) > 0.25. Multiple test correction was applied for p-values using Benjamini and
Hochberg method is used.
Cancers 2021, 13, 5157 15 of 22

Figure 7. Network and machine learning analysis. (A) Negative correlation network visualization of the glioblastoma
consensus metabolic signature. (B) Positive correlation network visualization of the glioblastoma consensus metabolic
signature. Line width is proportional to the Spearman correlation. (C) Boxplots of the consensus plasma metabolic signature.
(D) ROC curves of the Random Forest predictive models, including seven phosphatidylcholines (PC aa C36:6, PC aa C32:2,
PC aa C36:0, PC aa C36:5, PC aa C34:1, and PC aa C28:1) and two sphingomyelins (SM C18:0 and SM C24:1). AUC and 95%
confidence intervals (CI) were calculated using balanced subsampling with 50 repeats.
Cancers 2021, 13, 5157 16 of 22

4. Discussion
Despite a sustained research effort in the fight against glioblastoma, the prognosis
of this condition remains devastating, and its metabolic signaling pathways are barely
understood. To tackle this challenge, omics-based technologies could shed light on this
aggressive disease. Metabolic phenotyping through metabolomics is one of the promising
tools to interrogate functional metabolic readouts.
In this study, we identified four metabolites’ classes as proxies of deep metabolic
remodeling in plasma and tissue samples from glioblastoma patients, which included
phospholipids, acylcarnitines, sphingomyelins, and Triacylglycerols. Some of the high-
lighted metabolites in our study have been previously reported. This is the case for a
number of amino acids, such as asparagine, glutamine, lysine, citrulline, or valine, for
example (Figure 3) [38,39,41]. Indeed, amino acid metabolism is an area of interest in the
race for new therapies for glioblastoma. The discovery of the Isocitrate Dehydrogenase
(IDH) metabolic pathway has revived this antimetabolic approach, which aims at reducing
the synthesis of these molecules, which play an essential role in tumor development and
progression [61]. Recently differences in the levels of certain amino acids in the blood of
patients with glioblastoma compared to healthy subjects have been demonstrated [62].
These data empower the central role of amino acids in the genesis of glioblastoma. In
parallel to amino acid metabolism, our study also highlighted the role of lipids in this
disease. Glycerophosphocholines (GPC) were one key class highlighted in this study. Most
prominent GPC were PC aa C34:1, PC aa C38:6, PC (14:2), PC aa C36:5, lysoPC a C18:0,
PC aa C28:1, PC aa C32:2, PC aa C36:0, and PC aa C36:6 (Figures 3A, 4, 5 and 7). GPC are
precursors lipids of acetylcholine [63]. Cholines are important membrane components
and known as markers of tumor progression [64]. A choline peak visualized by MRI
spectroscopy is a diagnostic criterion for glioblastoma that has been used in clinical practice
for several years [38]. Similarly, it has been shown that choline blood levels in glioblas-
toma patients are correlated with tumor progression [65]. More recently, following in-vivo
brain micro dialysis studies [66], it appears that choline levels in the peri-tumor tissue
also correlate with its degree of invasion. One of the pathophysiological hypotheses is
that glioblastoma modifies its environment to make it favorable to its development and
proliferation [67]. Furthermore, one of the key elements for glioblastoma progression is
energy substrate availability. The cerebral high lipid enrichment helps deliver the energy
necessary for tumor growth [68]. Unlike normal cells, which deliver energy from mito-
chondrial phosphorylation, tumoral glial cells get their energy from glycolysis and have
intense lipidogenesis. Taïb et al. [69] showed that oleic acid, a monounsaturated fatty acid,
increases triacylglycerol production in malignant glial cells but not in normal glial cells, ar-
guing for a metabolic reprogramming of the malignant glial cell environment to allow them
to spread [70]. Sphingolipids are also involved in this phenomenon and were highlighted
in our analysis through SM (0H) C22:2, SM (0H) C16:1, SM C18:0, and SM C18:1 and SM
C24:1 (Figures 3A, 4, 5 and 7). These lipids are abundant in the brain and are essential
membrane constituents highly expressed in glial cells. Alterations of the sphingolipid
pathway resulting in lower levels of ceramides which have pro-apoptotic properties, are
thought to play a role in malignant cell dissemination, and malignant glial cells have also
been reported to evade apoptosis by converting ceramides to sphingosine-1-phosphate,
thus preventing apoptosis [71]. Thus, the action of oncological treatments of glioblastoma,
which generate sphingomyelinases to lyse sphingomyelins into ceramides, escape this cell
death phenomenon [67]. The exact drivers of this process are not yet well understood.
Steroylcarnitine and L-palmitoylcarnitine, which belong to the Acylcarnitines (AC) class,
are also modulated during glioma genesis (Figures 3A and 5). AC primary function is to
allow fatty acids transport to feed mitochondrial β-oxidation and thus ATP production.
Based on a recent study on patient-derived xenograft models of GBM, it appears that AC
are abundantly found within and at the edge of these tumors, and this may provide a fertile
ground for glioblastoma spread [72]. These advances on how glioblastoma modulates its
biological environment to promote its spread, notably through lipid metabolism, opens,
Cancers 2021, 13, 5157 17 of 22

obviously, promising ways to new targeted therapeutic strategies and biomarker discovery.
Interestingly, this study showed that circulating free DNA level was positively correlated
with sphingomyelins (SM C16:1 and SM C18:1) and Acylcarnitines (Figure 6), which are
major substrates of tumor progression by providing the glioblastoma with the energy.
Lipid droplets (LDs) are organelles that are commonly found in fatty tissues storing lipids,
triglycerol, and cholesterol esters. Recently, the presence of LDs has been demonstrated in
glial and glioblastoma cell cultures in vivo [73]. These LDs are not only a simple reservoir
of fatty acids but have an impact on important cell processes such as cell cycles, cell migra-
tion of glial cells, and their resistance to apoptosis and conventional treatments [69]. LDs
seem to be promising biomarkers to probe tumor progression [74]. It has been established
for a long time [75] that GBM diverts brain lipid metabolism to its advantage to sustain the
energy stores needed for their own expansion. By focusing on these LDs, it has been shown
that they play a central role in this lipid metabolism reprogramming [73]. Furthermore,
it has been reported that autophagy-mediated hydrolysis of these LD maintains energy
homeostasis [76]. Thus, limiting the access of glioblastoma to this lipid stock represents a
potentially innovative treatment avenue. Despite intensive chemoradiotherapy, the median
survival of patients with GBM remains at around 15 months. In recent years, lipidomics has
opened the way to new perspectives [77]. The goal is to prevent glial cells from spreading
by modulating their environment. Sphingolipids are one of these targets by promoting
the formation of apoptosis-inducing ceramides. The action of temozolomide, the first-line
chemotherapy for glioblastoma, on the biological environment of glial cells is being actively
studied. Temozolomide modifies the extracellular vesicles both in size and content [78].
These extracellular vesicles release fatty acids and proteins that modulate the action of
tumor-associated macrophages [78]. Chemotherapy may thus have a modulating effect on
the environment of glial cells via these vesicles. The macrophages associated with tumors
are major components of tumor spread. Thus, targeting these macrophages could counter-
act the chemo- and radio resistance of glial cells. As lipids are ubiquitous in the brain, they
have been used as vectors for targeted therapies using CRISPR-Cas9 technology [79]. Based
on the seminal paper by Garofano et al. [80], a classification of glioblastomas based on cell
signaling pathways has revealed a therapeutic sensitivity in the mitochondrial glioblastoma
subgroup. These preliminary results, although encouraging, need to be validated and
replicated in humans. Some limitations have to be noted regarding our study. The sample
size was limited for the tissue-based metabolomics which is mainly related to practical
constraints and patients’ recruitment. Larger cohorts are needed to confirm the present
results, ideally, including other bulk and/or single-cell panomics to have a broader view of
the impaired biological pathways and their clinical effect.

5. Conclusions
In summary, our study demonstrated the potential of systems-based metabolomics
strategies to holistically interrogate biological plasticity in glioblastoma and parse the role
of lipids in tumor progression by the modulation of its own environment. The next step of
our work would be the study of therapeutic modulations of these metabolic signatures in
order to probe treatment effectiveness. Applying these plasma metabolic patterns could
also inform clinical practice to adapt treatment before MRI or clinical modification. A
better understanding of metabolic impairments underlying glioblastoma spread would
also drive the development of targeted therapies to prevent glioblastoma from promoting
its own extension to the surrounding tissues. Such systems-based strategies highlight
the importance of multi-omics and multimodal investigations to understand glial lesions
to pave the way to more personalized therapies and, ultimately, achieve the promise of
precision medicine.
Cancers 2021, 13, 5157 18 of 22

Supplementary Materials: The following are available online at https://www.mdpi.com/article/10

.3390/cancers13205157/s1, Figure S1: Network analysis overview. Networks are built using Partial
Correlation, Figure S2: Spearman correlations between samples, Figure S3: Boxplots of previously re-
ported metabolites, Figure S4: Heatmap of Spearman correlations of previously reported metabolites,
Figure S5: Heatmap of Spearman correlations of newly reported metabolites, Figure S6: Intra-group
(GBM) variation, Figure S7: Heatmap of Spearman correlations between all significant metabolites,
Figure S8: Network-based patient specific plasma metabolic signature (Patient PGB_01), Figure S9:
Network-based patient specific plasma metabolic signature (Patient PGB_02), Figure S10: Network-
based patient specific plasma metabolic signature (Patient PGB_03), Figure S11: Network-based
patient specific plasma metabolic signature (Patient PGB_04), Figure S12: Network-based patient spe-
cific plasma metabolic signature (Patient PGB_05), Figure S13: Network-based patient specific plasma
metabolic signature (Patient PGB_06), Figure S14: Network-based patient specific plasma metabolic
signature (Patient PGB_07), Figure S15: Network-based patient specific plasma metabolic signature
(Patient PGB_08), Figure S16: Network-based patient specific plasma metabolic signature (Patient
PGB_09), Figure S17: Network-based patient specific plasma metabolic signature (Patient PGB_10),
Figure S18: Network-based patient specific plasma metabolic signature (Patient PGB_11), Figure S19:
Network-based patient specific plasma metabolic signature (Patient PGB_12), Figure S20: Network-
based patient specific plasma metabolic signature (Patient PGB_13), Figure S21: Network-based
patient specific plasma metabolic signature (Patient PGB_14), Figure S22: Network-based patient
specific plasma metabolic signature (Patient PGB_15), Figure S23: Network-based patient specific
plasma metabolic signature (Patient PGB_16), Figure S24: Network-based patient specific plasma
metabolic signature (Patient PGB_17), Figure S25: Network-based patient specific plasma metabolic
signature (Patient PGB_18), Figure S26: Network-based patient specific plasma metabolic signature
(Patient PGB_19), Figure S27: Network-based patient specific plasma metabolic signature (Patient
PGB_20), Figure S28: Network-based patient specific plasma metabolic signature (Patient PGB_21),
Figure S29: Network-based patient specific plasma metabolic signature (Patient PGB_22), Figure S30:
Network-based patient specific plasma metabolic signature (Patient PGB_23), Figure S31: Network-
based patient specific plasma metabolic signature (Patient PGB_24), Figure S32: Network-based
patient specific plasma metabolic signature (Patient PGB_25), Figure S33: Network-based patient
specific plasma metabolic signature (Patient PGB_26), Figure S34: Network-based patient specific
plasma metabolic signature (Patient PGB_27), Figure S35: Network-based patient specific plasma
metabolic signature (Patient PGB_28), Figure S36: Network-based patient specific plasma metabolic
signature (Patient PGB_29), Table S1: Metabolites List used in Targeted Metabolomics, Table S2:
Sources and IMS parameters for Ion Mobility Mass Spectrometry analyses, Table S3: Cross Collision
Section Calculation Curve, Table S4: Variable metadata (Untargeted LC-IMS-MS Metabolomics),
Table S5: Variable metadata (Untargeted FT-ICR-MS Metabolomics), Table S6: Data matrix (Targeted
LC-MS/MS Metabolomics), Table S7: Data matrix (Untargeted LC-IMS-MS Metabolomics), Table S8:
Data matrix (Untargeted FT-ICR-MS Metabolomics), Table S9: Cohorte Overview, Table S10: Sample
Spearman correlation (Targeted LC-MS/MS Metabolomics), Table S11: Sample Spearman correla-
tion (Untargeted LC-IMS-MS Metabolomics, Table S12: Sample Spearman correlation (Untargeted
FT-ICR-MS Metabolomics), Table S13: Principal Component Analysis Scores, Table S14: Differential
Analysis Results, Table S15: Analysis of overlap with litterature, Table S16: Spearman Metabolite
Correlation Matrix, Table S17: Spearman Correlations with circulating free DNA, Table S18: Partial
Correlation Matrix including all samples, Table S19: Pruned netwrok using all samples network,
Table S20: Partial Correlation Matrix including only control samples, Table S21: Patient specfic
pruned networks, Table S22: Consensus network, Table S23: All Random Forest Models.
Author Contributions: Conceptualization: V.G., A.T., S.M., and S.B.; Data curation: F.M., M.F., F.D.,
A.L., C.A., S.B., V.G., S.D., I.S.-A., J.F., C.P., C.L, T.P., C.B., F.D.F., and F.D.; Formal analysis: A.T. and
F.D.; writing—original draft preparation: V.G., A.T., S.M., S.B.; Resources: S.B., C.A., and A.L.; Formal
analysis: I.S.-A., J.F., C.P., C.L., C.B., and F.D.; writing—original draft preparation. A.T. and S.B.;
writing-review and editing, all authors; visualization, A.T. and F.D.; supervision, S.B., S.M., and A.T.;
project administration, A.T., S.M., and S.B. All authors have read and agreed to the published version
of the manuscript.
Funding: This research received no external funding.
Cancers 2021, 13, 5157 19 of 22

Institutional Review Board Statement: All samples used in this study belong to a declared collection
located in the Pathology Department (Annie Laquerrière) of Rouen University Hospital in accordance
with the relevant guidelines and regulations and with permission of the local authorities.
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.
Data Availability Statement: Data supporting the finding are presented in the text and Supplemen-
tary Material.
Acknowledgments: At COBRA laboratory, for MALDI-FTICR imaging and LC-IM-MS experiments,
this work has been partially supported by the University of Rouen Normandy, INSA Rouen Normandy,
the Centre National de la Recherche Scientifique (CNRS), European Regional Development Fund (ERDF),
Labex SynOrg (ANR-11-LABX-0029), Carnot Institut I2C, the graduate school for research Xl-Chem
(ANR-18-EURE-0020 XL CHEM), and by the Region Normandie. In addition, this work is supported
by the European Union’s Horizon 2020 Research Infrastructures program (Grant Agreement 731077).
Access to a CNRS FTICR research infrastructure (FR3624) is gratefully acknowledged.
Conflicts of Interest: The authors have no conflict of interest to disclose.

References
1. Bauchet, L.; Ostrom, Q.T. Epidemiology and molecular epidemiology. Neurosurg. Clin. N. Am. 2019, 30, 1–16. [CrossRef]
2. Morgan, L.L. The epidemiology of glioma in adults: A “state of the science” review. Neuro-Oncology 2015, 17, 623–624. [CrossRef]
3. Khurana, V.; Jain, S.; Smee, R.; Cook, R.; Dobes, M.; Shadbolt, B.; Smith, S.; Dexter, M. Increasing incidence of glioblastoma
multiforme and meningioma, and decreasing incidence of Schwannoma (2000–2008): Findings of a multicenter Australian study.
Surg. Neurol. Int. 2011, 2, 176. [CrossRef]
4. Philips, A.; Henshaw, D.L.; Lamburn, G.; O’Carroll, M.J. Brain Tumours: Rise in Glioblastoma Multiforme Incidence in England
1995–2015 Suggests an Adverse Environmental or Lifestyle Factor. J. Environ. Public Health 2018, 2018, 7910754. [CrossRef]
5. Vecht, C.J.; Kerkhof, M.; Duran-Pena, A. Seizure Prognosis in Brain Tumors: New Insights and Evidence-Based Management.
Oncol. 2014, 19, 751–759. [CrossRef]
6. Yan, J.-L.; Li, C.; Boonzaier, N.R.; Fountain, D.M.; Larkin, T.J.; Matys, T.; Van Der Hoorn, A.; Price, S.J. Multimodal MRI
characteristics of the glioblastoma infiltration beyond contrast enhancement. Ther. Adv. Neurol. Disord. 2019, 12. [CrossRef]
7. Law, M.; Yang, S.; Wang, H.; Babb, J.S.; Johnson, G.; Cha, S.; Knopp, E.A.; Zagzag, D. Glioma Grading: Sensitivity, Specificity, and
Predictive Values of Perfusion MR Imaging and Proton MR Spectroscopic Imaging Compared with Conventional MR Imaging.
Am. J. Neuroradiol. 2003, 24, 1989–1998.
8. Stupp, R.; Mason, W.P.; Bent, M.V.D.; Weller, M.; Fisher, B.; Taphoorn, M.J.; Belanger, K.; Brandes, A.; Marosi, C.; Bogdahn,
U.; et al. Radiotherapy plus Concomitant and Adjuvant Temozolomide for Glioblastoma. N. Engl. J. Med. 2005, 352, 987–996.
[CrossRef]
9. Brown, T.J.; Brennan, M.C.; Li, M.; Church, E.W.; Brandmeir, N.J.; Rakszawski, K.L.; Patel, A.S.; Rizk, E.B.; Suki, D.; Sawaya,
R.; et al. Association of the extent of resection with survival in glioblastoma: A systematic review and meta-analysis. JAMA Oncol.
2016, 2, 1460–1469. [CrossRef]
10. Bloch, O.; Han, S.J.; Cha, S.; Sun, M.Z.; Aghi, M.K.; McDermott, M.W.; Berger, M.S.; Parsa, A.T. Impact of extent of re-section for
recurrent glioblastoma on overall survival: Clinical article. J. Neurosurg. 2012, 117, 1032–1038. [CrossRef]
11. Lim, M.; Xia, Y.; Bettegowda, C.; Weller, M. Current state of immunotherapy for glioblastoma. Nat. Rev. Clin. Oncol. 2018, 15,
422–442. [CrossRef]
12. Touat, M.; Idbaih, A.; Sanson, M.; Ligon, K.L. Glioblastoma targeted therapy: Updated approaches from recent biological insights.
Ann. Oncol. 2017, 28, 1457–1472. [CrossRef]
13. Malkki, H. Trial watch: Glioblastoma vaccine therapy disappointment in phase iii trial. Nat. Rev. Neurol. 2016, 12, 190. [CrossRef]
14. Delgado-López, P.D.; Corrales-García, E.M. Survival in glioblastoma: A review on the impact of treatment modalities. Clin. Transl.
Oncol. 2016, 18, 1062–1071. [CrossRef]
15. Ius, T.; Ciani, Y.; Ruaro, M.E.; Isola, M.; Sorrentino, M.; Bulfoni, M.; Candotti, V.; Correcig, C.; Bourkoula, E.; Manini, I.; et al. An
nf-kappab signature predicts low-grade glioma prognosis: A precision medicine approach based on patient-derived stem cells.
Neuro Oncol. 2018, 20, 776–787. [CrossRef]
16. Le Tourneau, C.; Borcoman, E.; Kamal, M. Molecular profiling in precision medicine oncology. Nat. Med. 2019, 25, 711–712.
[CrossRef]
17. Ogrinc, N.; Saudemont, P.; Takats, Z.; Salzet, M.; Fournier, I. Cancer surgery 2.0: Guidance by real-time molecular tech-nologies.
Trends Mol. Med. 2021. Available online: https://www.cell.com/trends/molecular-medicine/fulltext/S1471-4914(21)00096-4
(accessed on 8 September 2021). [CrossRef]
18. Birzu, C.; French, P.; Caccese, M.; Cerretti, G.; Idbaih, A.; Zagonel, V.; Lombardi, G. Recurrent glioblastoma: From molec-ular
landscape to new treatment perspectives. Cancers 2020, 13, 47. [CrossRef]
19. Aletaha, D. Precision medicine and management of rheumatoid arthritis. J. Autoimmun. 2020, 110, 102405. [CrossRef]
Cancers 2021, 13, 5157 20 of 22

20. Tebani, A.; Afonso, C.; Bekri, S. Advances in metabolome information retrieval: Turning chemistry into biology. Part II: Biological
information recovery. J. Inherit. Metab. Dis. 2017, 41, 393–406. [CrossRef]
21. Manzoni, C.; Kia, D.A.; Vandrovcova, J.; Hardy, J.; Wood, N.; Lewis, P.A.; Ferrari, R. Genome, transcriptome and proteome: The
rise of omics data and their integration in biomedical sciences. Brief. Bioinform. 2016, 19, 286–302. [CrossRef]
22. McColl, E.R.; Asthana, R.; Paine, M.F.; Piquette-Miller, M. The age of omics-driven precision medicine. Clin. Pharmacol. Ther. 2019,
106, 477–481. [CrossRef]
23. Lombardi, G.; De Salvo, G.L.; Brandes, A.A.; Eoli, M.; Rudà, R.; Faedi, M.; Lolli, I.; Pace, A.; Daniele, B.; Pasqualetti, F.; et al.
Regorafenib compared with lomustine in patients with relapsed glioblastoma (REGOMA): A multicentre, open-label, randomised,
controlled, phase 2 trial. Lancet Oncol. 2018, 20, 110–119. [CrossRef]
24. Arivaradarajan, P.; Misra, G. Omics Approaches, Technologies and Applications: Integrative Approaches for Understanding Omics Data,
1st ed.; Springer: Singapore, 2018; p. 1. Available online: https://www.springer.com/gp/book/9789811329241 (accessed on
8 September 2021).
25. Sarma, A.; Calfee, C.S.; Ware, L.B. Biomarkers and precision medicine: State of the art. Crit Care Clin. 2020, 36, 155–165. [CrossRef]
26. Clish, C.B. Metabolomics: An emerging but powerful tool for precision medicine. Mol. Case Stud. 2015, 1, a000588. [CrossRef]
[PubMed]
27. Tebani, A.; Afonso, C.; Bekri, S. Advances in metabolome information retrieval: Turning chemistry into biology. Part I: Analytical
chemistry of the metabolome. J. Inherit. Metab. Dis. 2018, 41, 393–406. [CrossRef] [PubMed]
28. Nicholson, J.; Lindon, J.; Holmes, E. ‘Metabonomics’: Understanding the metabolic responses of living systems to pathophysiolog-
ical stimuli via multivariate statistical analysis of biological NMR spectroscopic data. Xenobiotica 1999, 29, 1181–1189. [CrossRef]
[PubMed]
29. Shaw, J.B.; Lin, T.-Y.; Leach, F.E., III; Tolmachev, A.V.; Tolić, N.; Robinson, E.W.; Koppenaal, D.W.; Paša-Tolić, L. 21 tesla fourier
transform ion cyclotron resonance mass spectrometer greatly expands mass spectrometry toolbox. J. Am. Soc. Mass Spectrom.
2016, 27, 1929–1936. [CrossRef]
30. Kooijman, P.C.; Nagornov, K.O.; Kozhinov, A.N.; Kilgour, D.P.A.; Tsybin, Y.; Heeren, R.M.A.; Ellis, S.R. Increased throughput and
ultra-high mass resolution in DESI FT-ICR MS imaging through new-generation external data acquisition system and advanced
data processing approaches. Sci. Rep. 2019, 9, 1–11. [CrossRef]
31. Ferey, J.; Marguet, F.; Laquerrière, A.; Marret, S.; Schmitz-Afonso, I.; Bekri, S.; Afonso, C.; Tebani, A. A new optimization strategy
for MALDI FTICR MS tissue analysis for untargeted metabolomics using experimental design and data modeling. Anal. Bioanal.
Chem. 2019, 411, 3891–3903. [CrossRef]
32. Ciocan-Cartita, C.A.; Jurj, A.; Buse, M.; Gulei, D.; Braicu, C.; Raduly, L.; Cojocneanu, R.; Pruteanu, L.L.; Iuga, C.A.; Coza, O.; et al.
The Relevance of Mass Spectrometry Analysis for Personalized Medicine through Its Successful Application in Cancer “Omics”.
Int. J. Mol. Sci. 2019, 20, 2576. [CrossRef]
33. Duarte, T.T.; Spencer, C.T. Personalized Proteomics: The Future of Precision Medicine. Proteomes 2016, 4, 29. [CrossRef] [PubMed]
34. Rogachev, A.; Alemasov, N.; Ivanisenko, V.; Ivanisenko, N.; Gaisler, E.; Oleshko, O.; Cheresiz, S.; Mishinov, S.; Stupak, V.;
Pokrovsky, A. Correlation of Metabolic Profiles of Plasma and Cerebrospinal Fluid of High-Grade Glioma Patients. Metabolites
2021, 11, 133. [CrossRef] [PubMed]
35. Pandey, R.; Caflisch, L.; Lodi, A.; Brenner, A.J.; Tiziani, S. Metabolomic signature of brain cancer. Mol. Carcinog. 2017, 56,
2355–2371. [CrossRef]
36. Heiland, D.H.; Wörner, J.; Haaker, J.G.; Delev, D.; Pompe, N.; Mercas, B.; Franco, P.; Gäbelein, A.; Heynckes, S.; Pfeifer, D.; et al.
The integrative metabolomic-transcriptomic landscape of glioblastome multiforme. Oncotarget 2017, 8, 49178–49190. [CrossRef]
[PubMed]
37. Huang, J.; Weinstein, S.J.; Kitahara, C.M.; Karoly, E.D.; Sampson, J.N.; Albanes, D. A prospective study of serum metab-olites and
glioma risk. Oncotarget 2017, 8, 70366–70377. [CrossRef]
38. Hvinden, I.C.; Berg, H.E.; Sachse, D.; Skaga, E.; Skottvoll, F.S.; Lundanes, E.; Sandberg, C.J.; Vik-Mo, E.O.; Rise, F.; Wilson, S.R.
Nuclear Magnetic Resonance Spectroscopy to Identify Metabolite Biomarkers of Nonresponsiveness to Targeted Therapy in
Glioblastoma Tumor Stem Cells. J. Proteome Res. 2019, 18, 2012–2020. [CrossRef] [PubMed]
39. Kurake, N.; Ishikawa, K.; Tanaka, H.; Hashizume, H.; Nakamura, K.; Kajiyama, H.; Toyokuni, S.; Kikkawa, F.; Mizuno, M.;
Hori, M. Non-thermal plasma-activated medium modified metabolomic profiles in the glycolysis of U251SP glioblastoma. Arch.
Biochem. Biophys. 2018, 662, 83–92. [CrossRef] [PubMed]
40. Wang, L.-B.; Karpova, A.; Gritsenko, M.A.; Kyle, J.E.; Cao, S.; Li, Y.; Rykunov, D.; Colaprico, A.; Rothstein, J.H.; Hong, R.; et al.
Proteogenomic and metabolomic characterization of human glioblastoma. Cancer Cell 2021, 39, 509–528. [CrossRef] [PubMed]
41. Wibom, C.; Surowiec, I.; Mörén, L.; Bergström, P.; Johansson, M.; Antti, H.; Bergenheim, A.T. Metabolomic Patterns in Glioblas-
toma and Changes during Radiotherapy: A Clinical Microdialysis Study. J. Proteome Res. 2010, 9, 2909–2919. [CrossRef]
42. Yu, D.; Xuan, Q.; Zhang, C.; Hu, C.; Li, Y.; Zhao, X.; Liu, S.; Ren, F.; Zhang, Y.; Zhou, L.; et al. Metabolic Alterations Related to
Glioma Grading Based on Metabolomics and Lipidomics Analyses. Metabolites 2020, 10, 478. [CrossRef] [PubMed]
43. Jaroch, K.; Modrakowska, P.; Bojko, B. Glioblastoma Metabolomics—In Vitro Studies. Metabolites 2021, 11, 315. [CrossRef]
[PubMed]
Cancers 2021, 13, 5157 21 of 22

44. Ramsay, S.L.; Guggenbichler, W.; Weinberger, K.M.; Graber, A.; Stoeggl, W.M. Device for quantitative analysis of a drug or
metabolite profile. Google Patents. 2012. Available online: https://patents.google.com/patent/RU2008102835A/en (accessed on
8 September 2021).
45. Ramsay, S.L.; Stoeggl, W.M.; Weinberger, K.M.; Graber, A.; Guggenbichler, W. Apparatus and method for analyzing a metabolite
profile. Google Patents. 2012. Available online: https://patents.google.com/patent/US8265877B2/en (accessed on 8 Septem-
ber 2021).
46. Bush, M.F.; Campuzano, I.D.G.; Robinson, C.V. Ion Mobility Mass Spectrometry of Peptide Ions: Effects of Drift Gas and
Calibration Strategies. Anal. Chem. 2012, 84, 7124–7130. [CrossRef] [PubMed]
47. Zhou, Z.; Tu, J.; Xiong, X.; Shen, X.; Zhu, Z.-J. LipidCCS: Prediction of Collision Cross-Section Values for Lipids with High
Precision To Support Ion Mobility–Mass Spectrometry-Based Lipidomics. Anal. Chem. 2017, 89, 9559–9566. [CrossRef] [PubMed]
48. Smith, C.A.; O’Maille, G.; Want, E.J.; Qin, C.; Trauger, S.A.; Brandon, T.R.; Custodio, D.E.; Abagyan, R.; Siuzdak, G. Metlin: A
metabolite mass spectral database. Ther. Drug Monit. 2005, 27, 747. [CrossRef]
49. Wishart, D.S.; Jewison, T.; Guo, A.C.; Wilson, M.; Knox, C.; Liu, Y.; Djoumbou, Y.; Mandal, R.; Aziat, F.; Dong, E.; et al. Hmdb
3.0—The human metabolome database in 2013. Nucleic Acids Res. 2013, 41, D801. [CrossRef]
50. Zhou, Z.; Luo, M.; Chen, X.; Yin, Y.; Xiong, X.; Wang, R.; Zhu, Z.-J. Ion mobility collision cross-section atlas for known and
unknown metabolite annotation in untargeted metabolomics. Nat. Commun. 2020, 11, 1–13. [CrossRef]
51. Fontanilles, M.; Marguet, F.; Beaussire, L.; Magne, N.; Pépin, L.-F.; Alexandru, C.; Tennevet, I.; Hanzen, C.; Langlois, O.; Jardin,
F.; et al. Cell-free DNA and circulating tert promoter mutation for disease monitoring in newly-diagnosed glio-blastoma. Acta
Neuropathol. Commun. 2020, 8, 179. [CrossRef] [PubMed]
52. Van Den Berg, R.A.; Hoefsloot, H.C.; Westerhuis, J.A.; Smilde, A.K.; Van Der Werf, M.J. Centering, scaling, and trans-formations:
Improving the biological information content of metabolomics data. BMC Genomics 2006, 7, 142. [CrossRef]
53. Ritchie, M.; Phipson, B.; Wu, D.; Hu, Y.; Law, C.W.; Shi, W.; Smyth, G.K. limma powers differential expression analyses for
RNA-sequencing and microarray studies. Nucleic Acids Res. 2015, 43, e47. [CrossRef]
54. Benjamini, Y.; Hochberg, Y. Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. J. R.
Stat. Soc. Ser. B Stat. Methodol. 1995, 57, 289–300. [CrossRef]
55. Liang, F.; Song, Q.; Qiu, P. An Equivalent Measure of Partial Correlation Coefficients for High-Dimensional Gaussian Graphical
Models. J. Am. Stat. Assoc. 2015, 110, 1248–1265. [CrossRef]
56. Thistlethwaite, L.R.; Petrosyan, V.; Li, X.; Miller, M.J.; Elsea, S.H.; Milosavljevic, A. CTD: An information-theoretic algorithm
to interpret sets of metabolomic and transcriptomic perturbations in the context of graphical models. PLoS Comput. Biol. 2021,
17, e1008550. [CrossRef]
57. Wright, M.N.; Ziegler, A. ranger: A Fast Implementation of Random Forests for High Dimensional Data in C++ and R. J. Stat.
Softw. 2017, 77, 1–17. [CrossRef]
58. Kuhn, M. Caret: Classification and Regression Training. 2020. Available online: https://cran.r-project.org/web/packages/caret/
caret.pdf (accessed on 8 September 2021).
59. R Core Team. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing Website. 2020.
Available online: https://www.gbif.org/zh/tool/81287/r-a-language-and-environment-for-statistical-computing (accessed on
8 September 2021).
60. Kampa, J.M.; Kellner, U.; Marsching, C.; Ramallo Guevara, C.; Knappe, U.J.; Sahin, M.; Giampà, M.; Niehaus, K.; Bednarz, H.
Glioblastoma multiforme: Metabolic differences to peritumoral tissue and idh-mutated gliomas revealed by mass spec-trometry
imaging. Neuropathology 2020, 40, 546–558. [CrossRef] [PubMed]
61. Barritault, M.; Picart, T.; Poncet, D.; Fenouil, T.; D’Hombres, A.; Gabut, M.; Guyotat, J.; Jouanneau, E.; Ameli, R.; Joubert, B.; et al.
Avoiding New Biopsies by Identification of IDH1 and TERT Promoter Mutation in Nondiagnostic Biopsies From Glioma Patients.
Neurosurgery 2020, 87, E513–E519. [CrossRef]
62. Bobeff, E.J.; Szczesna, D.; Bieńkowski, M.; Janczar, K.; Chmielewska-Kassassir, M.; Wiśniewski, K.; Papierz, W.; Wozniak, L.A.;
Jaskólski, D.J. Plasma amino acids indicate glioblastoma with ATRX loss. Amino Acids 2021, 53, 119–132. [CrossRef] [PubMed]
63. Rand, J.B. Acetylcholine. WormBook 2007, 1–21. [CrossRef]
64. Sonkar, K.; Ayyappan, V.; Tressler, C.; Adelaja, O.; Cai, R.; Cheng, M.; Glunde, K. Focus on the glycerophosphocholine pathway
in choline phospholipid metabolism of cancer. NMR Biomed. 2018, 32, e4112. [CrossRef] [PubMed]
65. Shimizu, H.; Kumabe, T.; Shirane, R.; Yoshimoto, T. Correlation between Choline Level Measured by Proton MR Spectroscopy
and Ki-67 Labeling Index in Gliomas. Am. J. Neuroradiol. 2000, 21, 659–665.
66. Lima, E.; Otaduy, M.; Tsunemi, M.H.; Pincerato, R.; Cardoso, E.; Rosemberg, S.; de Aguiar, P.H.P.; Cerri, G.; Leite, C. The Effect of
Paramagnetic Contrast in Choline Peak in Patients with Glioblastoma Multiforme Might Not Be Significant. Am. J. Neuroradiol.
2012, 34, 80–84. [CrossRef]
67. Van Brocklyn, J.R. Sphingolipid signaling pathways as potential therapeutic targets in gliomas. Mini Rev. Med. Chem. 2007, 7,
984–990. [CrossRef]
68. Bruce, K.D.; Zsombok, A.; Eckel, R.H. Lipid processing in the brain: A key regulator of systemic metabolism. Front. Endocrinol.
2017, 8, 60. [CrossRef]
69. Taïb, B.; Aboussalah, A.M.; Moniruzzaman, M.; Chen, S.; Haughey, N.J.; Kim, S.F.; Ahima, R.S. Lipid accumulation and oxidation
in glioblastoma multiforme. Sci. Rep. 2019, 9, 1–14. [CrossRef]
Cancers 2021, 13, 5157 22 of 22

70. Zhou, L.; Wang, Z.; Hu, C.; Zhang, C.; Kovatcheva-Datchary, P.; Yu, D.; Liu, S.; Ren, F.; Wang, X.; Li, Y.; et al. Integrated
Metabolomics and Lipidomics Analyses Reveal Metabolic Reprogramming in Human Glioma with IDH1 Mutation. J. Proteome
Res. 2018, 18, 960–969. [CrossRef]
71. Hawkins, C.C.; Ali, T.; Ramanadham, S.; Hjelmeland, A.B. Sphingolipid Metabolism in Glioblastoma and Metastatic Brain
Tumors: A Review of Sphingomyelinases and Sphingosine-1-Phosphate. Biomolecules 2020, 10, 1357. [CrossRef] [PubMed]
72. Lin, H.; Patel, S.; Affleck, V.S.; Wilson, I.; Turnbull, D.; Joshi, A.R.; Maxwell, R.; Stoll, E.A. Fatty acid oxidation is required for the
respiration and proliferation of malignant glioma cells. Neuro-Oncology 2016, 19, 43–54. [CrossRef]
73. Yuan, Y.; Shah, N.; Almohaisin, M.I.; Saha, S.; Lu, F. Assessing fatty acid-induced lipotoxicity and its therapeutic potential in
glioblastoma using stimulated Raman microscopy. Sci. Rep. 2021, 11, 1–14. [CrossRef] [PubMed]
74. Geng, F.; Guo, D. Lipid droplets, potential biomarker and metabolic target in glioblastoma. Intern. Med. Rev. 2017, 3, 10–18103.
75. Guo, D.; Bell, E.H.; Chakravarti, A. Lipid metabolism emerges as a promising target for malignant glioma therapy. CNS Oncol.
2013, 2, 289–299. [CrossRef] [PubMed]
76. Wu, X.; Geng, F.; Cheng, X.; Guo, Q.; Zhong, Y.; Cloughesy, T.F.; Yong, W.H.; Chakravarti, A.; Guo, D. Lipid Droplets Maintain
Energy Homeostasis and Glioblastoma Growth via Autophagic Release of Stored Fatty Acids. iScience 2020, 23, 101569. [CrossRef]
77. Antal, O.; Péter, M.; Hackler, L.; Mán, I.; Szebeni, G.J.; Ayaydin, F.; Hideghéty, K.; Vigh, L.; Kitajka, K.; Balogh, G.; et al. Lipidomic
analysis reveals a radiosensitizing role of gamma-linolenic acid in glioma cells. Biochim. Biophys. Acta BBA Mol. Cell Biol. Lipids
2015, 1851, 1271–1282. [CrossRef] [PubMed]
78. Panzarini, E.; Tacconi, S.; Carata, E.; Mariano, S.; Tata, A.; Dini, L. Molecular Characterization of Temozolomide-Treated and Non
Temozolomide-Treated Glioblastoma Cells Released Extracellular Vesicles and Their Role in the Macrophage Response. Int. J.
Mol. Sci. 2020, 21, 8353. [CrossRef] [PubMed]
79. Rosenblum, D.; Gutkin, A.; Kedmi, R.; Ramishetti, S.; Veiga, N.; Jacobi, A.M.; Schubert, M.S.; Friedmann-Morvinski, D.; Cohen,
Z.R.; Behlke, M.A.; et al. CRISPR-Cas9 genome editing using targeted lipid nanoparticles for cancer therapy. Sci. Adv. 2020,
6, eabc9450. [CrossRef] [PubMed]
80. Garofano, L.; Migliozzi, S.; Oh, Y.T.; D’Angelo, F.; Najac, R.D.; Ko, A.; Frangaj, B.; Caruso, F.P.; Yu, K.; Yuan, J.; et al. Pathway-based
classification of glioblastoma uncovers a mitochondrial subtype with therapeutic vulnerabilities. Nat. Rev. Cancer 2021, 2, 141–156.
[CrossRef] [PubMed]
Supplementary material: Integrative Metabolomics Reveals
Deep Tissue and Systemic Metabolic Remodeling in
Glioblastoma
Vianney Gilard, Justine Ferey, Florent Marguet, Maxime Fontanilles, Franklin Ducatez, Carine Pilon, Céline
Lesueur, Tony Pereira, Carole Basset, Isabelle Schmitz‐Afonso, Frédéric Di Fioré, Annie Laquerrière, Carlos
Afonso, Stéphane Derrey, Stéphane Marret, Soumeya Bekri and Abdellah Tebani

Figure S1. Network analysis overview. Networks are built using Partial Correlation. The first network (“disease‐control”
network) is learned from disease profiles and control profiles. A second network composed of only control (“control‐only”
network). A third network is composed of control profiles and disease excluding one patient at a time (“Disease‐minus‐
one” network). This third network is built iteratively to cover all the patients. A final pruning stage subtracts edges from
the disease‐control network that are also found either in A) the control‐only network, outputting the disease‐specific
network B) “Disease‐minus‐one” specific network or C) Patient‐specific network.
A
FT−ICR−MSI (Brain)
1 Group
Group
G20
G01
0.8 Control
G21
G14
G15 Glioblastoma
G22
G13
G16 0.6
G17
G10
G06
G18
G05 0.4
G19
G09
G04
G08
G11 0.2
G12
G03
G02
G07
0
C06
C04
C01
C05
C02 −0.2
C03
G20

G01

G21

G14

G15

G22

G13

G16

G17

G10

G06

G18

G05

G19

G09

G04

G08

G11

G12

G03

G02

G07

C06

C04

C01

C05

C02

C03
−0.4

B
LC−IMS−MS (Plasma)
1 Group
Group
G28
G29 Control
C05
G17
G24
G19 0.8 Glioblastoma
G22
G04
G13
G12
G02
G15
G01
G05
G26
0.6
G20
G27
G16
G21
G08
G09
G07
G10
G18
0.4
G25
G11
G06
G23

C26
C28
C17
0.2
C18
C12
C01
C30
G03
C22
C03
C21
C08
0
C27
C07
C02
C09
C15
C06
C19
C20
G14
−0.2
C25
C29
C16
C10
C23
C04
C11
C13
C14
C24
G28
G29
C05
G17
G24
G19
G22
G04
G13
G12
G02
G15
G01
G05
G26
G20
G27
G16
G21
G08
G09
G07
G10
G18
G25
G11
G06
G23

C26
C28
C17
C18
C12
C01
C30
G03
C22
C03
C21
C08
C27
C07
C02
C09
C15
C06
C19
C20
G14
C25
C29
C16
C10
C23
C04
C11
C13
C14
C24

C
LC−MS/MS (Plasma)
1 Group
Group
C16
C36
C34
Control
C42
C31
C40
C41 0.99 Glioblastoma
C38
C35
C43
C19
C18
C44
C28
C60
C05
C59
C48
C49
0.98
C47
C55
C06
C09
C15
C52
C45
C54
C02
C37
C11
0.97
C51
C46
C32
C33
C56
C57
C50
C53
C58
C07
0.96
C13
C10
C14
C08
C12

G10
G26
G30
G24
0.95
G01
G21
G34
G16
G25
G31
G08
G11
G20
G29
G18
0.94
G35
G12
G27
C01
G09
G33
G13
G23
G03
G14
G05
G15
G02
G06
G07
G19
G04
G22
G28
G17
G32
C21
C23
C03
C24
C17
C25
C04
C20
C29
C26
C30
C39
C22
C27
C16
C36
C34
C42
C31
C40
C41
C38
C35
C43
C19
C18
C44
C28
C60
C05
C59
C48
C49
C47
C55
C06
C09
C15
C52
C45
C54
C02
C37
C11
C51
C46
C32
C33
C56
C57
C50
C53
C58
C07
C13
C10
C14
C08
C12

G10
G26
G30
G24
G01
G21
G34
G16
G25
G31
G08
G11
G20
G29
G18
G35
G12
G27
C01
G09
G33
G13
G23
G03
G14
G05
G15
G02
G06
G07
G19
G04
G22
G28
G17
G32
C21
C23
C03
C24
C17
C25
C04
C20
C29
C26
C30
C39
C22
C27

Figure S2. Spearman correlations between samples.

 Glioblastoma − Previously Reported Metabolites
Asparagine (Aminoacids) Carnitine (Acylcarnitines) cis−4−Hydroxyproline (Biogenic_amines) Citrulline (Aminoacids)
logFC = −1.28 p = 1.04e−09 logFC = −0.84 p = 1.13e−04 logFC = −0.8 p = 2.73e−04 logFC = 0.58 p = 9.24e−03

Plasma Plasma Plasma Plasma

5 1.2
4.0
4.0
4 0.9
3.5 3.5
3 0.6
3.0 3.0
2 0.3
2.5 2.5
1 0.0
Control Glioblastoma Control Glioblastoma Control Glioblastoma Control Glioblastoma

Glutamine (Aminoacids) Lysine (Aminoacids) Ornithine (Aminoacids) Putrescine (Biogenic_amines)

logFC = −1.3 p = 6.98e−10 logFC = −0.76 p = 5.36e−04 logFC = −0.84 p = 1.11e−04 logFC = −1.51 p = 5.22e−13

Plasma Plasma Plasma Plasma

7 0.20
6.0
5.0
0.15
6 5.5 4.5
0.10
5.0 4.0
5 0.05
4.5 3.5
Level (log scaled)

0.00
Control Glioblastoma Control Glioblastoma Control Glioblastoma Control Glioblastoma

Spermine (Biogenic_amines) Threonine (Aminoacids) trans−4−Hydroxyproline (Biogenic_amines) Tryptophan (Aminoacids)

logFC = 0.56 p = 1.23e−02 logFC = 0.53 p = 1.89e−02 logFC = 0.85 p = 9.65e−05 logFC = 0.66 p = 3.17e−03

Plasma Plasma Plasma Plasma

5.5 4.5
0.6
3
5.0 4.0
0.4
4.5 3.5
2
0.2 4.0 3.0

3.5 1
0.0 2.5
Control Glioblastoma Control Glioblastoma Control Glioblastoma Control Glioblastoma

Valine (Aminoacids)
logFC = −0.59 p = 9.15e−03

Plasma
6.5

6.0

5.5

5.0

4.5

4.0
Control Glioblastoma

Figure S3. Boxplots of previously reported metabolites.

 Previously Reported Metabolites

1
1 0.71 0.68 0.41 0.66 0.68 0.71 0.07 0.25 0.44 0.4 0.17 0.13 Valine (Plasma)

0.71 1 0.87 0.33 0.61 0.56 0.75 −0.02 0.27 0.23 0.35 0.17 0.37 Asparagine (Plasma)
0.8
0.68 0.87 1 0.17 0.72 0.43 0.63 0.11 0.36 0.31 0.26 0.27 0.26 Threonine (Plasma)

0.41 0.33 0.17 1 0.43 0.7 0.7 0.16 0.23 0.47 0.25 0.26 0.2 Ornithine (Plasma) 0.6
0.66 0.61 0.72 0.43 1 0.53 0.67 0.36 0.47 0.6 0.19 0.33 0.15 Glutamine (Plasma)

0.68 0.56 0.43 0.7 0.53 1 0.75 0.28 0.41 0.68 0.36 0.36 0.44 Carnitine (Plasma)
0.4

0.71 0.75 0.63 0.7 0.67 0.75 1 0.25 0.34 0.5 0.31 0.12 0.32 Lysine (Plasma)
0.2
0.07 −0.02 0.11 0.16 0.36 0.28 0.25 1 0.45 0.4 0.04 0.13 −0.12 cis−4−Hydroxyproline (Plasma)

0.25 0.27 0.36 0.23 0.47 0.41 0.34 0.45 1 0.58 −0.03 0.25 −0.07 trans−4−Hydroxyproline (Plasma) 0

0.44 0.23 0.31 0.47 0.6 0.68 0.5 0.4 0.58 1 0.18 0.51 0.27 Tryptophan (Plasma)

0.4 0.35 0.26 0.25 0.19 0.36 0.31 0.04 −0.03 0.18 1 0 0.27 Spermine (Plasma)

0.17 0.17 0.27 0.26 0.33 0.36 0.12 0.13 0.25 0.51 0 1 0.38 Citrulline (Plasma)

0.13 0.37 0.26 0.2 0.15 0.44 0.32 −0.12 −0.07 0.27 0.27 0.38 1 Putrescine (Plasma)
Valine (Plasma)

Asparagine (Plasma)

Threonine (Plasma)

Ornithine (Plasma)

Glutamine (Plasma)

Carnitine (Plasma)

Lysine (Plasma)

cis−4−Hydroxyproline (Plasma)

trans−4−Hydroxyproline (Plasma)

Tryptophan (Plasma)

Spermine (Plasma)

Citrulline (Plasma)

Putrescine (Plasma)

Figure S4. Heatmap of Spearman correlations of previously reported metabolites.

 Newly Reported Metabolites

1
1 −0.02 0.13 0.33 −0.11 0.05 0.31 0.46 0.11 0.09 0.16 0.05 −0.1 0.07 0.25 −0.09 −0.12 −0.02 0.29 0.27 PC aa C38:6 (Plasma)

−0.02 1 0.52 −0.13 −0.12 −0.03 0.25 0.12 0.29 0.3 0.25 0.06 0.13 −0.04 0.12 −0.02 0.19 0.22 0.08 −0.41 Cholesterol (Plasma)

0.13 0.52 1 −0.55 −0.53 −0.65 −0.24 −0.18 −0.14 −0.17 −0.21 −0.4 −0.49 −0.46 −0.25 −0.55 −0.4 −0.46 −0.56 −0.56 PC(14:2) (Tissue) 0.5

0.33 −0.13 −0.55 1 0.71 0.72 0.2 0.29 0.09 0.3 0.22 0.2 0.25 0.12 0.17 0.05 0.24 0.32 0.62 0.25 L−Palmitoylcarnitine (Tissue)

−0.11 −0.12 −0.53 0.71 1 0.78 0.1 0.16 −0.01 0.02 0 0.18 0.12 0.22 0.06 0.21 0.42 0.51 0.31 −0.01 Stearoylcarnitine (Tissue)
0
0.05 −0.03 −0.65 0.72 0.78 1 0.25 0.27 0.14 0.2 0.22 0.3 0.34 0.31 0.2 0.35 0.41 0.5 0.54 0.31 TG(52:2) (Tissue)

0.31 0.25 −0.24 0.2 0.1 0.25 1 0.83 0.61 0.4 0.52 0.6 0.63 0.73 0.82 0.6 0.62 0.78 0.36 0.28 TG(52:2) (Plasma)

−0.5
0.46 0.12 −0.18 0.29 0.16 0.27 0.83 1 0.53 0.45 0.51 0.58 0.54 0.66 0.8 0.32 0.23 0.47 0.12 0.14 TG(52:3) (Plasma)

0.11 0.29 −0.14 0.09 −0.01 0.14 0.61 0.53 1 0.83 0.93 0.89 0.79 0.77 0.88 0.65 0.49 0.44 0.43 0.53 TG(54:6) (Plasma)

0.09 0.3 −0.17 0.3 0.02 0.2 0.4 0.45 0.83 1 0.96 0.82 0.85 0.58 0.71 0.41 0.31 0.2 0.44 0.35 TG(54:4) (Plasma)

0.16 0.25 −0.21 0.22 0 0.22 0.52 0.51 0.93 0.96 1 0.92 0.86 0.73 0.82 0.58 0.42 0.32 0.48 0.5 TG(54:5) (Plasma)

0.05 0.06 −0.4 0.2 0.18 0.3 0.6 0.58 0.89 0.82 0.92 1 0.85 0.92 0.89 0.77 0.55 0.49 0.4 0.51 TG(52:5) (Plasma)

−0.1 0.13 −0.49 0.25 0.12 0.34 0.63 0.54 0.79 0.85 0.86 0.85 1 0.75 0.78 0.65 0.56 0.52 0.49 0.48 TG(54:3) (Plasma)

0.07 −0.04 −0.46 0.12 0.22 0.31 0.73 0.66 0.77 0.58 0.73 0.92 0.75 1 0.89 0.87 0.61 0.63 0.32 0.46 TG(50:3) (Plasma)

0.25 0.12 −0.25 0.17 0.06 0.2 0.82 0.8 0.88 0.71 0.82 0.89 0.78 0.89 1 0.64 0.48 0.54 0.3 0.45 TG(52:4) (Plasma)

−0.09 −0.02 −0.55 0.05 0.21 0.35 0.6 0.32 0.65 0.41 0.58 0.77 0.65 0.87 0.64 1 0.78 0.71 0.53 0.57 TG(48:2) (Plasma)

−0.12 0.19 −0.4 0.24 0.42 0.41 0.62 0.23 0.49 0.31 0.42 0.55 0.56 0.61 0.48 0.78 1 0.9 0.56 0.34 TG(48:1) (Plasma)

−0.02 0.22 −0.46 0.32 0.51 0.5 0.78 0.47 0.44 0.2 0.32 0.49 0.52 0.63 0.54 0.71 0.9 1 0.52 0.25 TG(50:2) (Plasma)

0.29 0.08 −0.56 0.62 0.31 0.54 0.36 0.12 0.43 0.44 0.48 0.4 0.49 0.32 0.3 0.53 0.56 0.52 1 0.65 PC aa C34:1 (Plasma)

0.27 −0.41 −0.56 0.25 −0.01 0.31 0.28 0.14 0.53 0.35 0.5 0.51 0.48 0.46 0.45 0.57 0.34 0.25 0.65 1 Putrescine (Plasma)
PC aa C38:6 (Plasma)

Cholesterol (Plasma)

PC(14:2) (Tissue)

L−Palmitoylcarnitine (Tissue)

Stearoylcarnitine (Tissue)

TG(52:2) (Tissue)

TG(52:2) (Plasma)

TG(52:3) (Plasma)

TG(54:6) (Plasma)

TG(54:4) (Plasma)

TG(54:5) (Plasma)

TG(52:5) (Plasma)

TG(54:3) (Plasma)

TG(50:3) (Plasma)

TG(52:4) (Plasma)

TG(48:2) (Plasma)

TG(48:1) (Plasma)

TG(50:2) (Plasma)

PC aa C34:1 (Plasma)

Putrescine (Plasma)

Figure S5. Heatmap of Spearman correlations of newly reported metabolites.

Figure S6. Intra‐group (GBM) variation. A) Histogram distribution of coefficient of variation in brain and plasma (Colored
according to matrice). B) Histogram distribution of coefficient of variation in brain and plasma (Colored according to
biochemical classes). C) Coefficient of variation of the 10 lipids correlated between plasma and brain samples.
 1 class
class
Ornithine
Hexadecenoylcarnitine Clinical
blood_circ_DNA
Lysine
total DMA Biogenic_amines
Asymmetric dimethylarginine
Aspartate
0.5 Phosphatidylcholine
Citrulline
Valine
Leucine
Tetradecadienylcarnitine
Aminoacids
Dodecenoylcarnitine
Dodecanoylcarnitine
Decenoylcarnitine
Sphingomyelins
Acetylcarnitine 0
Methionine sulfoxide
Decadienylcarnitine
Acylcarnitines
Acetylornithine
Proline
Hexenoylcarnitine
Valerylcarnitine
alpha−Aminoadipic acid
Phenylalanine
Butenylcarnitine
−0.5
Propionylcarnitine
Carnitine
ki67_percent
postoperative_karnofsky
trans−4−Hydroxyproline
PC aa C38:6
PC aa C40:6
PC aa C36:0
PC ae C40:6
PC aa C38:0
Glutamate
SM C26:1
PC ae C40:2
SM (OH) C16:1
SM C18:0
SM C24:1
cis−4−Hydroxyproline
Tryptophan
PC ae C34:3
PC ae C38:4
lysoPC a C18:2
PC ae C38:5
Octadecadienylcarnitine
Octadecenoylcarnitine
lysoPC a C20:4
lysoPC a C17:0
PC aa C40:3
PC aa C42:4
Serotonin
PC ae C38:0
PC ae C38:6
PC aa C42:5
lysoPC a C18:0
lysoPC a C16:0
PC aa C38:4
PC aa C38:5
PC aa C28:1
PC ae C38:1
egfr_percent
PC ae C42:1
PC aa C34:1
PC aa C36:1
PC ae C34:1
PC aa C32:1
PC aa C30:0
PC aa C34:4
lysoPC a C14:0
PC aa C36:6
PC aa C36:5
lysoPC a C16:1
PC aa C32:0
PC ae C36:0
PC ae C34:0
PC ae C36:1
PC aa C40:4
PC ae C38:3
lysoPC a C20:3
p53_percent
Threonine
Arginine
Asparagine
lysoPC a C28:0
SM (OH) C22:2
SM (OH) C14:1
tumor_mean_diameter_mm
PC ae C32:1
PC ae C32:2
PC ae C30:0
PC aa C32:2
PC aa C32:3
Putrescine
PC ae C40:3
PC ae C36:2
PC ae C42:3
PC aa C40:2
PC ae C44:5
PC ae C42:5
PC aa C42:0
PC aa C42:1
SM C18:1
SM C16:1
Glutamine
SM C16:0
Hexanoylcarnitine
PC ae C38:2
PC aa C36:2
PC aa C34:3
Spermine
Ornithine
Hexadecenoylcarnitine
blood_circ_DNA
Lysine
total DMA
Asymmetric dimethylarginine
Aspartate
Citrulline
Valine
Leucine
Tetradecadienylcarnitine
Dodecenoylcarnitine
Dodecanoylcarnitine
Decenoylcarnitine
Acetylcarnitine
Methionine sulfoxide
Decadienylcarnitine
Acetylornithine
Proline
Hexenoylcarnitine
Valerylcarnitine
alpha−Aminoadipic acid
Phenylalanine
Butenylcarnitine
Propionylcarnitine
Carnitine
ki67_percent
postoperative_karnofsky
trans−4−Hydroxyproline
PC aa C38:6
PC aa C40:6
PC aa C36:0
PC ae C40:6
PC aa C38:0
Glutamate
SM C26:1
PC ae C40:2
SM (OH) C16:1
SM C18:0
SM C24:1
cis−4−Hydroxyproline
Tryptophan
PC ae C34:3
PC ae C38:4
lysoPC a C18:2
PC ae C38:5
Octadecadienylcarnitine
Octadecenoylcarnitine
lysoPC a C20:4
lysoPC a C17:0
PC aa C40:3
PC aa C42:4
Serotonin
PC ae C38:0
PC ae C38:6
PC aa C42:5
lysoPC a C18:0
lysoPC a C16:0
PC aa C38:4
PC aa C38:5
PC aa C28:1
PC ae C38:1
egfr_percent
PC ae C42:1
PC aa C34:1
PC aa C36:1
PC ae C34:1
PC aa C32:1
PC aa C30:0
PC aa C34:4
lysoPC a C14:0
PC aa C36:6
PC aa C36:5
lysoPC a C16:1
PC aa C32:0
PC ae C36:0
PC ae C34:0
PC ae C36:1
PC aa C40:4
PC ae C38:3
lysoPC a C20:3
p53_percent
Threonine
Arginine
Asparagine
lysoPC a C28:0
SM (OH) C22:2
SM (OH) C14:1
tumor_mean_diameter_mm
PC ae C32:1
PC ae C32:2
PC ae C30:0
PC aa C32:2
PC aa C32:3
Putrescine
PC ae C40:3
PC ae C36:2
PC ae C42:3
PC aa C40:2
PC ae C44:5
PC ae C42:5
PC aa C42:0
PC aa C42:1
SM C18:1
SM C16:1
Glutamine
SM C16:0
Hexanoylcarnitine
PC ae C38:2
PC aa C36:2
PC aa C34:3
Spermine

Figure S7. Heatmap of Spearman correlations between all significant metabolites.

Figure S8. Network‐based patient specific plasma metabolic signature (Patient PGB_01). The signature includes 18
metabolites; 10 phosphatidylcholines, 5 sphingomyelin and 3 aminoacids.
Figure S9. Network‐based patient specific plasma metabolic signature (Patient PGB_02). The signature includes 19
metabolites; 11 phosphatidylcholines, 6 sphingomyelin, 1 aminoacid and 1 biogenic amine.
Figure S10. Network‐based patient specific plasma metabolic signature (Patient PGB_03). The signature includes 18
metabolites; 9 phosphatidylcholines, 3 sphingomyelin, 1 aminoacid and 3 biogenic amines.
Figure S11. Network‐based patient specific plasma metabolic signature (Patient PGB_04). The signature includes 11
metabolites; 6 phosphatidylcholines, 3 acylcarnitines and 2 sphingomyelins.
Figure S12. Network‐based patient specific plasma metabolic signature (Patient PGB_05). The signature includes 22
metabolites; 14 phosphatidylcholines, 6 sphingomyelin, 1 aminoacid and 1 acylcarnitine.
Figure S13. Network‐based patient specific plasma metabolic signature (Patient PGB_06). The signature includes 21
metabolites; 16 phosphatidylcholines, 4 sphingomyelins and 1 aminoacid.
Figure S14. Network‐based patient specific plasma metabolic signature (Patient PGB_07). The signature includes 24
metabolites; 16 phosphatidylcholines, 6 sphingomyelins and 2 biogenic amines.
Figure S15. Network‐based patient specific plasma metabolic signature (Patient PGB_08). The signature includes 19
metabolites; 10 phosphatidylcholines, 4 sphingomyelins, 2 acylcarnitines, 2 biogenic amines and 1 aminoacid.
Figure S16. Network‐based patient specific plasma metabolic signature (Patient PGB_09). The signature includes 19
metabolites; 5 phosphatidylcholines, 7 aminoacids, 3 acylcarnitines, 3 biogenic amines and 1 sphingomyelin.
Figure S17. Network‐based patient specific plasma metabolic signature (Patient PGB_10). The signature includes 17
metabolites; 11 phosphatidylcholines, 3 aminoacids, 1 biogenic amine and 2 sphingomyelins.
Figure S18. Network‐based patient specific plasma metabolic signature (Patient PGB_11). The signature includes 17
metabolites; 12 phosphatidylcholines, 1 aminoacid, and 4 sphingomyelins.
Figure S19. Network‐based patient specific plasma metabolic signature (Patient PGB_12). The signature includes 12
metabolites; 8 phosphatidylcholines, 2 aminoacids, 1 biogenic amine and 2 sphingomyelins.
Figure S20. Network‐based patient specific plasma metabolic signature (Patient PGB_13). The signature includes 19
metabolites; 11 phosphatidylcholines, 3 aminoacids, 3 acylcarnitines, 1 biogenic amine and 3 sphingomyelins.
Figure S21. Network‐based patient specific plasma metabolic signature (Patient PGB_14). The signature includes 11
metabolites; 5 phosphatidylcholines, 3 aminoacids, 2 acylcarnitines and 1 biogenic amine.
Figure S22. Network‐based patient specific plasma metabolic signature (Patient PGB_15). The signature includes 14
metabolites; 8 phosphatidylcholines, 2 aminoacids, 1 acylcarnitine, 1 biogenic amine and 2 sphingomyelins.
Figure S23. Network‐based patient specific plasma metabolic signature (Patient PGB_16). The signature includes 13
metabolites; 7 phosphatidylcholines, 2 aminoacids, 1 acylcarnitine and 3 sphingomyelins.
Figure S24. Network‐based patient specific plasma metabolic signature (Patient PGB_17). The signature includes 21
metabolites; 13 phosphatidylcholines, 2 aminoacids, 1 biogenic amine and 5 sphingomyelins.
Figure S25. Network‐based patient specific plasma metabolic signature (Patient PGB_18). The signature includes 19
metabolites; 11 phosphatidylcholines, 3 aminoacids, 2 biogenic amines and 3 sphingomyelins.
Figure S26. Network‐based patient specific plasma metabolic signature (Patient PGB_19). The signature includes 19
metabolites; 11 phosphatidylcholines, 1 aminoacid, 2 acylcarnitines, 3 biogenic amines and 2 sphingomyelins.
Figure S27. Network‐based patient specific plasma metabolic signature (Patient PGB_20). The signature includes 13
metabolites; 5 phosphatidylcholines, 3 aminoacids, 1 acylcarnitine, and 4 sphingomyelins.
Figure S28. Network‐based patient specific plasma metabolic signature (Patient PGB_21). The signature includes 23
metabolites; 14 phosphatidylcholines, 1 aminoacid, 1 acylcarnitine, 2 biogenic amines and 5 sphingomyelins.
Figure S29. Network‐based patient specific plasma metabolic signature (Patient PGB_22). The signature includes 22
metabolites; 9 phosphatidylcholines, 3 aminoacids, 1 biogenic amine and 3 sphingomyelins.
Figure S30. Network‐based patient specific plasma metabolic signature (Patient PGB_23). The signature includes 24
metabolites; 12 phosphatidylcholines, 2 aminoacids, 2 acylcarnitines, 4 biogenic amines and 4 sphingomyelins.
Figure S31. Network‐based patient specific plasma metabolic signature (Patient PGB_24). The signature includes 18
metabolites; 14 phosphatidylcholines, 1 acylcarnitine, 1 biogenic amine and 2 sphingomyelins.
Figure S32. Network‐based patient specific plasma metabolic signature (Patient PGB_25). The signature includes 14
metabolites; 6 phosphatidylcholines, 3 aminoacids, 1 acylcarnitine, 1 biogenic amine and 3 sphingomyelins.
Figure S33. Network‐based patient specific plasma metabolic signature (Patient PGB_26). The signature includes 27
metabolites; 16 phosphatidylcholine, 1 aminoacid, 2 acylcarnitines, 2 biogenic amines and 6 sphingomyelins.
Figure S34. Network‐based patient specific plasma metabolic signature (Patient PGB_27). The signature includes 19
metabolites; 9 phosphatidylcholines, 3 aminoacides, 3 acylcarnitine, 1 biogenic amine and 3 sphingomyelins.
Figure S35. Network‐based patient specific plasma metabolic signature (Patient PGB_28). The signature includes 28
metabolites; 17 phosphatidylcholines, 4 aminoacides, 1 acylcarnitine and 6 sphingomyelins.
Figure S36. Network‐based patient specific plasma metabolic signature (Patient PGB_29). The signature includes 25
metabolites; 15 phosphatidylcholines, 2 aminoacides, 1 biogenic amine and 7 sphingomyelins.
 CONCLUSIONS ET PERSPECTIVES

La neurochirurgie s’est enrichie de tout temps grâce aux progrès techniques de son

époque que ce soit par les avancées en radiologie, magnification optique ou de biologie. Un

domaine qui illustre au mieux cette interdisciplinarité est certainement la neuro-oncologie. Les

glioblastomes, qui sont les tumeurs cérébrales primitives les plus fréquentes s’accompagnent
toujours aujourd’hui d’un pronostic sombre, et ce malgré les travaux de recherche qui leur

sont consacrés. C’est pourquoi il y a une nécessité de repenser l’approche que l’on peut avoir

de cette pathologie en terme diagnostique et thérapeutique.

La médecine de précision propose une médecine centrée sur le patient avec des

approches personnalisées. Elle a contribué à des avancées dans de nombreux domaines de la

médecine. L’essor du Big Data permet d’intégrer aux données classiques médicales, les

différentes sciences Omiques. La métabolomique nous parait particulièrement intéressante,

car cette approche permet l'étude du produit final du métabolisme.

La compréhension des mécanismes métaboliques mis en jeu dans le cadre du Glioblastome

permettra à terme de proposer des stratégies diagnostiques et thérapeutiques. Le travail que

nous présentons dans cette thèse montre quels outils la biologie, et les sciences Omiques en

particulier, peuvent apporter à la neurochirurgie.

Les revues présentées sont une réflexion sur le changement de paradigme que connait
le neurochirurgien actuellement en s’appropriant une chirurgie de précision et en dressant un

état des connaissances de la lutte contre le Glioblastome. Notre travail expérimental, qui est

une étude multi-omique des patients présentant un Glioblastome a permis d’émettre des

hypothèses quant à la propagation des lésions gliales intracérébrales. L’un des déterminants
mis en évidence est la modulation par le Glioblastome de son environnement, notamment

énergétique, lui permettant de croître. Nous avons déterminé des perturbations dans le

métabolisme lipidique et relevé des métabolites d’intérêt dans le sérum et le tissu chez des

patients malades. Ces données pourraient permettre de proposer des marqueurs

diagnostiques après validation dans de plus grandes cohortes. La caractérisation plus précise

des lésions gliales est un prérequis indispensable pour proposer des biomarqueurs efficients

et ouvrir la voie à des thérapies ciblées, permettant ainsi de s’inscrire pleinement dans la

médecine de précision.

102
Résumé : La médecine de précision vise à placer le patient au centre du parcours de soins en intégrant
les données médicales et biologiques individuelles tout en tenant compte de la grande diversité entre
les individus. Ces nouvelles approches nous invitent à nous réinventer en tant que neurochirurgiens
et à remettre en question notre pratique et notre métier notamment en neuro-oncologie. Le
glioblastome est la tumeur cérébrale maligne la plus fréquente chez l'adulte et consiste en une
infiltration progressive de la substance blanche par des cellules tumorales. La survie moyenne est de
15 à 17 mois. Compte tenu de ce pronostic agressif, il est urgent de mieux comprendre les mécanismes
sous-jacents du glioblastome afin de dévoiler de nouvelles stratégies diagnostiques et des cibles
thérapeutiques grâce à une compréhension plus approfondie de sa biologie. Pour ces raisons, dans
ce travail, nous avons réalisé des études métabolomiques ciblées et non ciblées sur des échantillons
de tissus et de plasma provenant de patients atteints de glioblastome. Cette étude a révélé 176 lipides
et métabolites exprimés de manière différentielle, 148 dans le plasma et 28 dans les échantillons de
tissus. Les principales classes biochimiques comprennent les phospholipides, les acylcarnitines, les
sphingomyélines et les triacylglycérols. Les analyses fonctionnelles ont révélé un remodelage
métabolique profond des lipides et des substrats énergétiques des glioblastomes, ce qui dévoile le
rôle majeur des lipides dans la progression tumorale en modulant son propre environnement. Dans
l'ensemble, notre étude démontre un remodelage métabolique in situ et systémique dans le
glioblastome qui nous éclaire sur sa plasticité biologique sous-jacente et sa progression afin de faire
évoluer le diagnostic et/ou les stratégies thérapeutiques.

Mots clés : Glioblastome ; gliome de haut grade ; tumeur cérébrale ; neuro-oncologie ;

métabolomique ; omique ; spectrométrie de masse ; apprentissage automatique.

Abstract: Precision medicine aims to bring the patient at the center of the healthcare ecosystem by
integrating individual medical and biological data while taking into account the great diversity
between individuals. These new approaches invite us to reinvent ourselves as neurosurgeons and to
question our practice and profession especially in neuro-oncology. Glioblastoma is the most common
malignant brain tumor in adults and consists in a progressive infiltration of the white matter by
tumoral cells. The mean survival is between 15 to 17 months. Given this aggressive prognosis, there
is an urgent need for a better understanding of the underlying mechanisms of glioblastoma to unveil
new diagnostic strategies and therapeutic targets through deeper understanding of its biology. For
these reasons, in this work, we performed targeted and untargeted metabolomics-based
investigations on both tissue and plasma samples from patients with glioblastoma. This study
revealed 176 differentially expressed lipids and metabolites, 148 in plasma and 28 in tissue samples.
Main biochemical classes include phospholipids, acylcarnitines, sphingomyelins and triacylglycerols.
Functional analyses revealed deep metabolic remodeling in glioblastoma lipids and energy
substrates, which unveils the major role of lipids in tumor progression by modulating its own
environment. Overall, our study demonstrates in situ and systemic metabolic rewiring in
glioblastoma that could shed light on its underlying biological plasticity and progression to inform
diagnosis and/or therapeutic strategies.

Keywords: Glioblastoma; high grade glioma; brain tumor; neuro oncology; metabolomics; omics;
mass spectrometry; machine learning.

103
104