-

Cytostatics in the aquatic environment:
analysis, occurrence, and possibilities for removal

Von der Fakultt fr Mathematik, Informatik und Naturwissenschaften der RWTH Aachen University
zur Erlangung des akademischen Grades einer Doktorin der Naturwissenschaften genehmigte Dissertation
vorgelegt von
MSc
Lubomira Kovalova
aus Presov, Slowakei
Berichter: Prof. Dr.-Ing. Juliane Hollender

Univ.-Prof. Dr.rer.nat. Andreas Schffer
Tag der mndlichen Prfung: 21. August 2009
Diese Dissertation ist auf den Internetseiten der Hochschulbibliothek online verfgbar.
The presented doctoral study was conducted under the supervision of Prof. Dr. Juliane Hollender and Univ.-Prof. Dr.rer.nat.
Wolfgang Dott at the Institute of Hygiene and Environmental Medicine at the University Hospital Aachen, Germany and at the
Swiss Federal Institute of Aquatic Science and Technology (EAWAG), Dbendorf, Switzerland.
The study was part of the AQUAbase project, a training site on Organic Micropollutants in Aquatic Environment -
Interdisciplinary Concepts for Assessment and Removal (2004 2007), which was a Marie Curie fellowship for early stage
research training hosted by RWTH Aachen University, Germany. It was supported by the European Communitys Sixth
Framework Programme under contract number MEST-CT-2004-505169 (http://www.aquabase.rwth-aachen.de ).
The study of sorption of micropollutants to activated carbon described in Chapter 5 was part of an AiF project Nano-filtration
kombiniert mit Adsorption an Pulverkohle zur Entfernung von organischen Spurenstoffen aus Klranlagen-Ablaufwasser (2006
- 2008), supported by the German Federal Ministry of Economics and Technology through the German Federation of Industrial
Research Associations "Otto von Guericke under project number 14773 N1/2.
2
Contents
ABSTRACT 7
Chapter 1
GENERAL INTRODUCTION 9
1.1 Cytostatics: general characterization, classification,
and target selection 12
1.2 Target cytostatics: 5-fluorouracil, cytarabine
and gemcitabine 15
1.1.1 Indications and pharmacokinetics 15
1.1.2 Physical-chemical properties and impact on environmental
behavior and analytical procedure design 17
1.1.3 Current state of knowledge on environmental
occurrence, effects and fate 18
References 20
Chapter 2
OBJECTIVES 23
Chapter 3
ANALYTICAL METHOD DEVELOPMENT 27
3.1 Experimental 30
3.1.1 Chemicals and material 30
3.1.2 Sample collection 30
3.1.3 Hydrophilic interaction chromatography 30
3.1.4 Solid phase extraction 31
3.1.5 Tandem mass spectrometry 31
3.1.6 High-resolution mass spectrometry 32
3.1.7 Quantification, identification and quality control 32
3.1.8 Safety considerations 33
3.2 Results and discussion 33
3.2.1 Hydrophilic interaction chromatography. 33
3.2.2 Solid-phase extraction 35
3.2.3 Mass spectrometry 37
3.2.4 Application 39
References 40
3
Chapter 4
OCCURRENCE AND STABILITY IN HOSPITAL WASTEWATER 43
4.1 Experimental 46
4.1.1 Sampling of wastewater in a Swiss cantonal hospital 46
4.1.2 Description of municipal WWTP and sampling
of its wastewater 47
4.1.3 Analytes and analytical methods 47
4.1.4 Stability tests in different wastewater matrices 48
4.2.1 Occurrence in hospital wastewater and correlation
with pharmacokinetic data 49
4.2.2 Correlation between consumption data and emissions
in hospital wastewater 50
4.2.3 Stability studies 53
4.2.4 Environmental risk assessment 55
References 57
Chapter 5
REMOVAL BY ACTIVATED CARBON 59
5.1 Experimental 62
5.1.1 Adsorbents 62
5.1.2 Adsorbates 62
5.1.3 Solvents 62
5.1.4 Adsorption kinetics and isotherm set-up 63
5.1.5 Adsorbate removal prediction 63
5.2.1 Adsorption characteristics on two activated carbons
in wastewater 64
5.2.2 Influence of various parameters on adsorption to
activated lignite HOK Super 67
5.2.2.1 Influence of the solution pH 67
5.2.2.2 Influence of ionic strength 69
5.2.2.3 Influence of temperature 69
5.2.2.4 Influence of competing compounds 69
5.2.3 Prediction of adsorbate removal from the wastewater 70
References 71
Chapter 6
CONCLUSIONS 73
4
5
To my little niece Anna, who treasures her

first book of experiments and already
knows well that the beautiful colors of the
rainbow have something to do with
science. Although she still needs to work
for a while on figuring out what exactly...
6
7
Abstract
Many pharmaceuticals have attracted the attention of environmental scientists and
are of concern due to their observed occurrence in surface water, ground water,
drinking water, sediment, and soil. Further investigations have considered their
removal efficiencies in wastewater treatment processes and their ecotoxicological
effects. Cytostatics are a class of pharmaceuticals used in the treatment of cancer
and have the potential to negatively impact the environment. It is hypothesized that
due to their mode of action, practically all eukaryotic organisms are vulnerable to
damage, with teratogenicity being the greatest concern at low ng/L levels. This
thesis will discuss the contributions to this field in the form of 1) development of a
method for chemical analysis of cytostatics in wastewater, 2) occurrence studies,
and 3) potential for removal with advanced treatment processes.
To expand the knowledge on the environmental occurrence of cytostatics and to
compliment a range of cytostatics for which occurrence data are available, a solid
phase extraction and HPLCMS/MS method was established for highly consumed
cytostatics 5-fluorouracil, cytarabine, and gemcitabine and the human metabolites
uracil 1--d-arabinofuranoside and 2,2-difluorodeoxyuridine. Hydrophilic
interaction chromatography (HILIC), a chromatography for polar analytes in
aqueous samples, was used to achieve a good retention of the polar analytes. Along
with the method development, retention mechanisms on the HILIC stationary
phase were studied. Both partitioning and adsorption play a role in the retention on
the tested sulfoalkylbetaine modified silica HILIC column material. The
contribution of these two processes changes over the 1.640% range of water in the
mobile phase. Although the specific break point is difficult to determine,
adsorption becomes more significant as the fraction of water in the mobile phase
decreases below approximately 16%.
Wastewater from a Swiss hospital was monitored for 5-fluorouracil, gemcitabine
and 2,2-difluorodeoxyuridine (metabolite of gemcitabine). The limits of
quantification were 5.0, 0.9, and 9.0 ng/L and the maximum concentrations
detected were 27, 38, and 840 ng/L, respectively. Emission levels within one day
and over several days were found to correlate with the pharmacokinetic excretion
pattern and the consumed amounts in the hospital during these days. On average,
1.1%, 1.4% and 3.7% of the total excreted amounts of the cytostatics
5-fluorouracil, gemcitabine and 2',2'-difluorodeoxyuridine were found in the
hospital wastewater, respectively. Due to low recoveries, stability studies were
performed and showed that the half-lives of the analytes in the raw wastewater are
very short: 1.0, 0.7 and 2.5 hours, respectively.
Environmental impact of 5-fluorouracil, gemcitabine and 2,2-difluorodeoxy-
uridine seem to be of minor importance, as the environmental risk assessment
quotients PEC/PNEC are protective (in orders of magnitude of 10-3 - 10-6).
Nevertheless, cytostatics are a special group of pharmaceuticals with carcinogenic
potency and until data with specific tests and end-points designed especially for
this class of pharmaceuticals are available, the precautionary principle should be
applied.
To study removal possibilities, sorption to powdered activated carbon was
investigated. Adsorption is an effective process for the removal of hydrophobic
substances from water. Employing it as a wastewater treatment process, polar
compounds may also be simultaneously removed to a certain extent. Laboratory
scale batch experiments were performed to evaluate the adsorption kinetics and
isotherm data of 5-fluorouracil and cytarabine on two powdered activated carbons,
Norit SAE Super and Activated Lignite HOK Super. The results of the adsorption
characterization were compared with those of two hydrophobic substances,
8
bisphenol A and 17--ethinylestradiol, produced with the identical experimental
conditions and matrix. A dose of 30 mg.L-1 of Norit SAE Super that resulted in
removal of the two hydrophobic control compounds below 1% was found to
remove 30 and 65% of 5-fluorouracil and cytarabine, respectively. Influence of the
solution pH, ionic strength, temperature, and presence of matrix compounds
competing for the sorption sites was studied on one of the tested powdered
activated carbons, showing that the extent of the competing compounds influence
is the greatest. Applying a pseudo single-solute isotherm and a concept directly
linking carbon dose with relative removal of the micropollutant, capacity of the
powdered activated carbon was successfully predicted for different initial
micropollutant concentrations.
9
Chapter 1
General Introduction
It is better to know some of the questions than all of the answers.

(James Thurber)
10
General Introduction Chapter 1 11
1 General Introduction
It has been proven and become common knowledge among environmental scientists, as
well as among a part of the general public, that endocrine disruptor compounds that
make their way into the aquatic environment are causing shifts in population of fish due
to the feminization of males (Sumpter, 2008), and that antibiotics in the environment are
causing an increase in bacterial resistance (Zhang, 2009). But until now, no one has
confidently stated whether or not cytostatics have an impact on the environment; though
the potential is there. Cytostatics are designed to damage DNA, inhibit DNA synthesis,
and interrupt cell replication. They act unselectively on all growing cells and
furthermore possess a carcinogenic potency which implies that no threshold values for
lowest effect concentrations can be estimated. It is hypothesized that due to their mode
of action, practically all eukaryotic organisms are vulnerable to damage, with
teratogenicity being the greatest concern at low ng/L levels (Johnson, 2008). On the
other hand, consumption of cytostatics is low compared to consumption of other
pharmaceutical classes. As an example, the annual consumptions of analgesics
acetylsalicylic acid, paracetamol, and metamizol reached in 1999 in Land Brandenburg,
Germany were around 24 000, 7 000, and 5 000 kg/year, respectively, for antibiotics
roxythromycin, ampicilin, and ciprofloxacin the annual consumption was around 900,
500, and 400 kg/year, respectively, while for cytostatics 5-fluorouracil,
cyclophosphamide, and ifosfamide consumption was only 22, 8, and 8 kg/year,
respectively (LUA BB, 2002). The wastewater concentrations are expected to be the in
ng/L range, and surface water concentrations presumably under 1 ng/L (Kmmerer,
2001). This was proven for the two, most studied cytostatics in the environment:
ifosfamide and cyclophosphamide. Wastewater concentrations up to 146 ng/L were
measured in Germany (Steger-Hartmann, 1996) and up to 11 ng/L in Switzerland,
where the surface water concentrations were as low as 50 to 170 pg/L (Buerge, 2006).
The Swiss study concludes that concentrations were orders of magnitude lower than the
concentrations at which acute ecotoxicological effects have been reported (mg/L) but
due to lack of knowledge on chronic toxicity and occurrence and effects of metabolites,
it was not feasible to perform a final risk assessment.
The cytostatics consumption data from University Hospital Aachen, Germany, were
kindly provided over last six years for the purpose of our study by the hospital
pharmacy, enabling us to point out the hot targets and to follow the trends in the
prescriptions. It is not known if hospitals act as point sources of cytostatic input into the
environment. To consider this possibility, the following chapter describes, along with
consumption and pharmacokinetics, the manner of administration and the trends in
in-patient or out-patient treatment of oncologic patients. Attention is furthermore paid to
identify the urinary excreted human metabolites, as they might also be environmentally
relevant.
To understand the research needs and to recognize the gaps in the current knowledge,
the first chapter briefly summarizes the available information relevant to the topic
Cytostatics in the environment. Chapter 2, which follows this introduction, will later
list the objectives of this study the questions that have risen while reviewing the field,
and to which this thesis will provide the answers.
12 Chapter 1 General Introduction
1.1 Cytostatics: general characterization, classification,

and target selection
Cytostatic drugs are used in chemotherapy of oncological patients, who suffer from
cancer - a disease of uncontrolled multiplication of the bodys own cells and spread of
abnormal forms within the body. Chemotherapy, together with surgical excision and
irradiation, is one of the three main approaches to treat established cancer. Cytostatics
prevent the growth and proliferation of cells although they are not specific to cancer
cells. According to their mechanism of action, they can be divided into eight groups:
DNA synthesis inhibitors, angiogenesis inhibitors, DNA intercalators/cross-linkers,
DNA-RNA transcription regulators, enzyme inhibitors, gene regulators, microtubule
inhibitors, and other anti-tumor agents. The Anatomical Therapeutic Classification
System (ATC), recognized and used by the World Health Organization, groups the
cytostatics under Class L - Antineoplastic and Immunomodulating Agents according to
their chemical structures and therapeutic properties (Table 1.1).
Chemotherapy doses are usually computed on the basis of body surface area and are
expressed in mg/m2. The majority of cytostatic administrations are intravenous and are
executed at a clinic or hospital. Other ways of administration include intramuscular,
intraosseous, intralesional, topical, or oral. The trend in cytostatic administration is
shifting towards increasing oral administration executed at home or external ambulatory
infusion pumps (e.g. for 7 days) installed at the health-care facility but taken home for
the time of treatment. Approximately 75% of oncology patients are out-patients,
receiving the treatment at the oncology wards and leaving for home after the infusion or
injection is administered (Johnson, 2008; Mahnik, 2007). Those patients may excrete
part of the cytostatics in the hospital, as the treatment takes up to a couple of hours and
the pharmacokinetics of some cytostatics is relatively fast. European hospitals use
around 50 different active substances to treat oncology patients (Johnson, 2008). Table 1.1
shows the variety of cytostatics used at University hospital Aachen, Germany in 2008.
3
[kg / year]
0
e
e
in
ab
e
il
ne
ne
id
in
id
id
at
ac
at
m
zi
bi
ab
m
os
am
ex
ur
pl
ba
xi
ta
ha
tr
ar
op
bo
ro
sf
itu
ci
ar
ho
sp
yt
uo
I fo
em
Et
ar
ac
R
C
et
o
C
Fl
D
G
ph
M
lo
yc
2003 - 2008
C
Figure 1.1 Ten cytostatics with the highest consumption at University hospital Aachen, Germany (1 510 beds) in
2003-2008.
Table 1.1 Anatomical Therapeutic Chemical Classification System (ATC) of cytostatics. Example of an annual
consumption and expected emission (based on the consumption, urinary excretion rates, and assumption of 75%
out-patients) is given for a case of University hospital Aachen, Germany (1 510 beds) in 2008.
Hospitals annual consumption: Expected emission:
100 - 2 500 g/year 12 - 150 g/year
1 - 100 g/year 0.25 - 12 g/year
0.01 - 1 g/year 0.0025 0.25 g/year
no consumption ? urinary excretion data not available
Anatomical Therapeutic Chemical Classification System

Class L
Antineoplastic and Immunomodulating Agents
L01 ANTINEOPLASTIC AGENTS / L01CA03 Vindesine L01XE04 Sunitinib

L01A ALKYLATING AGENTS / L01CA04 Vinorelbine L01XE05 Sorafenib
L01AA Nitrogen mustard analogues L01CB Podophyllotoxin derivatives L01XE06 Dasatinib
/ L01AA01 Cyclophosphamide / L01CB01 Etoposide L01XE07 Lapatinib
L01AA02 Chlorambucil /? L01CB02 Teniposide L01XX Other antineoplastic agents
/ L01AA03 Melphalan L01CC Colchicine derivatives L01XX01 Amsacrine
L01AA05 Chlormethine L01CC01 Demecolcine /? L01XX02 Asparaginase
/ L01AA06 Ifosfamide L01CD Taxanes L01XX03 Altretamine
L01AA07 Trofosfamide / L01CD01 Paclitaxel L01XX05 Hydroxycarbamide
L01AA08 Prednimustine / L01CD02 Docetaxel L01XX07 Lonidamine
/? L01AA09 Bendamustine L01CX Other plant alkaloids L01XX08 Pentostatin
L01AB Alkyl sulfonates /? L01CX01 Trabectedin L01XX09 Miltefosine
/ L01AB01 Busulfan L01D CYTOTOXIC ANTIBIOTICS AND L01XX10 Masoprocol
/? L01AB02 Treosulfan RELATED SUBSTANCES L01XX11 Estramustine
L01AB03 Mannosulfan L01DA Actinomycines L01XX14 Tretinoin
L01AC Ethylene imines / L01DA01 Dactinomycin L01XX16 Mitoguazone
/ L01AC01 Thiotepa L01DB Anthracyclines and related / L01XX17 Topotecan
L01AC02 Triaziquone substances L01XX18 Tiazofurine
L01AC03 Carboquone / L01DB01 Doxorubicin / L01XX19 Irinotecan
L01AD Nitrosoureas / L01DB02 Daunorubicin L01XX22 Alitretinoin
/ L01AD01 Carmustine / L01DB03 Epirubicin L01XX23 Mitotane
L01AD02 Lomustine L01DB04 Aclarubicin L01XX22 Alitretinoin
L01AD03 Semustine L01DB05 Zorubicin L01XX23 Mitotane
L01AD04 Streptozocin / L01DB06 Idarubicin /? L01XX24 Pegaspargase
L01AD05 Fotemustine / L01DB07 Mitoxantrone L01XX25 Bexarotene
L01AD06 Nimustine L01DB08 Pirarubicin L01XX27 Arsenic trioxide
L01AD07 Ranimustine L01DB09 Valrubicin L01XX28 Imatinib
L01AG Epoxides L01DC Other cytotoxic antibiotics L01XX29 Denileukin diftitox
L01AG01 Etoglucid / L01DC01 Bleomycin L01XX31 Gefitinib
L01AX Other alkylating agents L01DC02 Plicamycin /? L01XX32 Bortezomib
L01AX01 Mitobronitol / L01DC03 Mitomycin L01XX33 Celecoxib
L01AX02 Pipobroman L01DC04 Ixabepilon L01XX35 Anagrelid
L01AX03 Temozolomide L01X OTHER ANTINEOPLASTIC L01XY Combinations of
/ L01AX04 Dacarbazine AGENTS antineoplastic agents
L01B ANTIMETABOLITES L01XA Platinum compounds
L01BA Folic acid analogues / L01XA01 Cisplatin L02 ENDOCRINE THERAPY
/ L01BA01 Methotrexate / L01XA02 Carboplatin L02A HORMONES AND RELATED
L01BA03 Raltitrexed / L01XA03 Oxaliplatin AGENTS
/ L01BA04 Pemetrexed L01XB Methylhydrazines L02AA Estrogens
L01BB Purine analogues L01XB01 Procarbazine L02AB Progestogens
L01BB02 Mercaptopurine L01XC Monoclonal antibodies L02AE Gonadotropin releasing
L01BB03 Tioguanine L01XC01 Edrecolomab hormone analogues
/? L01BB04 Cladribine /? L01XC02 Rituximab L02AX Other hormones
/ L01BB05 Fludarabine /? L01XC03 Trastuzumab L02B HORMONE ANTAGONISTS AND
L01BB06 Clofarabine /? L01XC04 Alemtuzumab RELATED AGENTS
L01BB07 Nelarabine /? L01XC05 Gemtuzumab L02BA Anti-estrogens
L01BC Pyrimidine analogues /? L01XC06 Cetuximab L02BB Anti-androgens
/ L01BC01 Cytarabine /? L01XC07 Bevacizumab L02BG Enzyme inhibitors
/ L01BC02 Fluorouracil L01XC08 Panitumumab
L01BC03 Tegafur L01XD Agents used in L03 IMMUNOSTIMULANTS
L01BC04 Carmofur photodynamic therapy L03AA Colony stimulating factors
/ L01BC05 Gemcitabine L01XD01 Porfimer sodium L03AB Interferons
L01BC06 Capecitabine L01XD02 Verteporfin L03AC Interleukins
L01BC52 Fluorouracil, comb. L01XD03 Methyl aminolevulinate
L01BC53 Tegafur, comb. L01XD04 Aminolevulinic acid L04 IMMUNOSUPPRESSANTS
L01C PLANT ALKALOIDS AND L01XD05 Temoporfin L04AA Selective immunosuppr.
OTHER NATURAL PRODUCTS L01XE Proteinkinase Inhibitors L04AB Tumor necrosis factor
L01CA Vinca alkaloids and analogues L01XE01 Imatinib alpha (TNF-) inhibitors
/ L01CA01 Vinblastine L01XE02 Gefitinib L04AC Interleukin inhibitors
/ L01CA02 Vincristine L01XE03 Erlotinib L04AD Calcineurin inhibitors
Consumption and expected emission ranges for the used drugs are included in the same
table. The ten cytostatics with the highest consumption at University hospital Aachen in
2003 - 2008 are shown in Figure 1.1. Significant consumption of 5-fluorouracil makes it
a suitable candidate for this study. Ifosfamide, gemcitabine, cyclophosphamide and
cytarabine are the next most used cytostatics; with annual consumption rates slightly
bellow 1 kg/year. Their consumption reaches only one half or one third of the 5-
fluorouracil consumption.
Urinary excretion rates of the unchanged drug

[% of administered drug]
< 5% 5 15% 15 25% 25 45% 45 75% > 75%
Busulfan Cytarabine Bleomycin Cisplatin Carboplatin Methotrexate
Carmustine Dactinomycin Cyclophosphamide Etoposide Oxaliplatin Pemetrexed
Docetaxel Daunorubicin Dacarbazine Ifosfamide
Epirubicin Doxorubicin Irinotecan Vincristine
Fludarabine Fluorouracil Mitomycin
Paclitaxel Gemcitabine Topotecan
Thiotepa Idarubicin Vinorelbine
Melphalan
Table 1.2 Mitoxantrone
Urinary excretion rates of some Vinblastine
cytostatics (SCM, 2007; Dorr, 1994) Vindesine
It is assumed that the hospital wastewater concentration of the cytostatics depends on

the amount of each drug excreted by patients via urine, although few cytostatics are also
excreted by faeces (e.g. doxorubicin, irinotecan, methotrexate, vincristin, and
docetaxel). Urinary excretion rates of the broad range of cytostatics widely vary, as can
be seen from Table 1.2. Beside the variation between the cytostatics, the excretion rates
for a single active substance shows strong inter-patient variation as it is dependent on
dose, manner of administration, age, and renal function of the patient. For the
University hospital Aachen, the list of the cytostatics with the highest consumption (Fig.
1.1) overlaps greatly with the list of compounds with the highest expected emission by
the hospital (Tab. 1.3). The expected emission was calculated by assuming 25% in-
patient treatment and solely urinary excretion:
Consumption [g/year] Excretion [%] In-patients [%]
Emission [g/L] =
Water Consumption [L/year]
The target cytostatics of this study were selected from the list of cytostatics with the
highest emission rates at the University hospital Aachen, Germany (2003-2004), with
the criterion that for some of them environmental occurrence data existed, while for
some not. They are three antimetabolites, all of them pyrimidine analogues:
5-fluorouracil; cytarabine; and gemcitabine. At the beginning of this work in 2004, there
have been a previously reported study on 5-fluorouracil in hospital wastewater (Mahnik,
2004), while no occurrence data existed at that time on gemcitabine and cytarabine.
Expected emission by
Consumption the hospital
ATC Drug
[g/year]
[g/year] [g/L]
Table 1.3
Cytostatic drugs with the highest L01BA01 Methotrexate 678 136 9.29
expected annual emission in 2008 L01AA06 Ifosfamide 908 91 6.22
by a University hospital Aachen L01BC02 Fluorouracil 2 212 55 3.79
(1 510 beds, 14 600 m3/year L01XA02 Carboplatin 214 37 2.56
L01AA01 Cyclophosphamide 736 37 2.52
water consumption) and their L01BC05 Gemcitabine 934 23 1.60
expected emission. Calculations L01CB01 Etoposide 245 18 1.26
are based on the consumption and L01BC01 Cytarabine 614 15 1.05
urinary excretion rates known from L01BA04 Pemetrexed 63 13 0.86
pharmacokinetics, assumption of L01AX04 Dacarbazine 178 9 0.61
75% out-patient treatment, and no L01XA01 Cisplatin 61 6 0.42
degradation of the drug in the L01XX19 Irinotecan 73 4 0.25
L01XA03 Oxaliplatin 24 3 0.20
wastewater.
1.2 Target cytostatics: 5-fluorouracil, cytarabine and

gemcitabine
1.2.1 Indications and Pharmacokinetics
Metabolic pathways of the three target antimetabolites are schematically represented in

figures 1.2, 1.3, and 1.4. 5-Fluorouracil, cytarabine and gemcitabine are pro-drugs that
require for their anticancer activity intracellular activation to diphosphate or
triphosphate form. They act by preventing the building blocks of DNA from becoming
incorporated during the S phase of the cell cycle, causing DNA replication inhibition.
Gemcitabine is a ribonucleotide reductase inhibitor, cytarabine a DNA polymerase
inhibitor, and 5-fluorouracil a thymidylate synthase inhibitor. The majority of the
administered doses do not follow this anabolic pathway but are rather catabolised into
one or more inactive metabolites that are subsequently urinary excreted.
5-Fluorouracil is applied to patients with carcinoma of the colon, rectum, breast,

stomach and pancreas (intravenously), as well as active solar keratoses, and superficial
basal cell carcinoma (topically). Capecitabine, tegafur, and carmofur are oral 5-
fluorouracil drugs. Elimination of 5-fluorouracil in the human body takes place in the
liver and other tissues. When administered intravenously (i.v.), elimination half-life in
plasma varies from 5 to 20 min. A total of 60-90% of the administered dose is excreted
in urine within 24 h (6 h for unchanged 5-fluorouracil), primarily as -fluoro--alanine
(approximately 80%) and as unchanged 5-fluorouracil (approximately 10% of the
administered dose) (Heggie, 1987). However, the excretion of 5-fluorouracil can vary
from patient to patient, with extreme ranges of 2% or 89% (SCM, 2007; Diasio, 1989;
Dorr, 1994). The urinary excretion rate of 5-fluorouracil after administration of
capecitabine, the oral form of 5-fluorouracil, is only 0.5% (Judson, 1999).
ELIMINATION ACTIVATION IN TARGET CELLS
dihydro-5-fluorouracil 5-fluorouracil 5-fluorouridine
-fluoro-ureidopropionic acid
5-fluorouridine-
-fluoro--alanine 5-monophosphate
urinary
excretion
5-fluorouridine- 5-fluorouridine-
5-diphosphate 5-triphosphate
5-fluoro-2-deoxy-
uridine-5-
monophosphate
Figure 1.2 Metabolic pathway of 5-fluorouracil in humans (Diasio, 1989), and plot of urinary excretion of ten
patients as reported by Heggie, 1987: total excreted radioactivity; -fluoro--alanine; 5-fluorouracil; and
-fluoro-ureidopropionic acid.
uracil 1--D- - cytarabine cytarabine 5-monophosphate

arabinofuranoside
cytarabine 5-diphosphate
urinary
excretion cytarabine 5-triphosphate
Figure 1.3 Metabolic pathway of cytarabine in humans (Hamada, 2002).
Cytarabine is applied to patients with hematologic malignancies, most commonly acute

myelogenous leukemia and lymphocytic leukemia in adults. It can be administered
intravenously (either as a bolus push or continuous infusion), intramuscularly or
subcutaneously. Cytarabine is not orally effective because of extensive and rapid
deamination within the gut lumen (Dorr, 1994). It is rapidly catabolized to uracil 1--D-
arabinofuranoside, which is subsequently eliminated in the urine. Within 24 hours, 80%
of the administered dose is urinary excreted, of which approximately 90% is uracil 1--
D-arabinofuranoside. The elimination of cytarabine is biphasic, with initial plasma half-
life 7 to 20 minutes and terminal half-life of 2 to 3 hours (Hamada, 2002).
Similar to cytarabine, gemcitabine is more potent and the most novel of the three
antimetabolites, originally synthesized as an antiviral agent. It is used for treating lung
cancer, pancreatic cancer, bladder cancer, and breast cancer. Gemcitabine can be
administered intravenously or orally. The elimination is rapid, with a half-life of
approximately 8 min. The gemcitabine metabolite 2,2-difluorodeoxyuridine (dFdU)
has a long terminal phase, with biphasic elimination kinetics with half-lives between 2.5
and >24 h (Abbruzzese, 1991). On average, 5% of the total drug is recovered as
gemcitabine in urine, and it all appears within the first 6 h after treatment, and 60% of
the administered gemcitabine is excreted as the metabolite 2,2-difluorodeoxyuridine
by the urinary tract within 24 h (Table 1.4). The table further illustrates the excretion
variability of individual patients that is also typical for other target drugs and shows the
calculated metabolite-to-drug ratios discussed later in Chapter 4 (Occurrence and
Stability in Hospital Wastewater).
2,2-difluoro- gemcitabine gemcitabine monophosphate

deoxyuridine
urinary
excretion
gemcitabine diphosphate
gemcitabine triphosphate
Figure 1.4 Metabolic pathway of gemcitabine in humans (SCM, 2007) and plot of urinary excretion of eight patients
with continuous infusion of 1000 mg/m2 gemcitabine for 30 min (Lin, 2004):
dFdU - 2,2-difluorodeoxyuridine; dFdC gemcitabine.
Table 1.4 Urinary excretion of gemcitabine and its metabolite 2.2-difluorodeoxyuridine by eight patients
(Abbruzzese, 1991).
percentage of the total dose in urine dFdU/GemC

[%] ratio
GemC dFdU TOTAL drug

0-6 h 0-6 h 6-12 h 12-18 h 18-24 h Total dFdU 0-24 h 0-6 h 0-24 h
dose [mg/m ]
2
790 1 12 10 7 0 29 30 12 29
5 55 - 6 9 70 75 11 14
10 74 2 5 5 86 96 7 9
1 000 5 30 2 5 5 42 47 6 8
3 18 3 9 2 32 35 6 11
4 66 2 5 1 74 78 17 19
7 44 26 0 9 79 86 6 11
5 34 27 6 4 71 76 7 14
AVERAGE 5 42 10 5 4 60 65 9 14
MEDIAN 5 39 3 6 5 71 76 7 13
MINIMUM 1 12 2 0 0 29 30 6 8
MAXIMUM 10 74 27 9 9 86 96 17 29
1.2.2 Physical-chemical properties and impact on environmental behavior and

analytical procedure design
The physical-chemical properties determine the behavior and fate of the target drugs in
the environment. Further, they have a major impact on the design of the analytical
procedure. The relevant physical-chemical properties are listed in Fig. 1.5. The selected
cytostatic drugs are highly polar, with the logarithm of the octanol/water partition
coefficient for the un-ionized solute (log P) between 0.89 and 2.09. Further, they are
non-volatile and either acidic or basic compounds, but uncharged at pH around 6 (with
exception of zwitterionic FBAL that was not studied further in this work).
The high polarity and low volatility predetermines that the compounds are likely to stay
in the water phase and their sorption or partitioning to solids or air is not to be expected
in high fractions. Occurrence in wastewater is expected, as well as low or no sorption to
particles or sludge. 5-Fluorouracil was reported to be photo-labile (Alsarra, 2004).
For the solid phase extraction procedure design, the properties imply that the retention
of all target compounds together with a single procedure at pH 6 is possible. This may
be achieved using polymeric materials developed specially for retention of polar
compounds (polymers with a hydrophilic monomer or chemically modified polymers)
as has been previously reported for 5-Fu and CytR (Kiffmeyer, 1998; Mahnik, 2004;
Tauxe-Wuersch, 2006), or carbon-based sorbents. Alternatively, two different
procedures with different sorbent materials and sample pHs could be exploited: an anion
exchange sorbent for acids at high pH; and a cation exchange sorbent for bases at low
pH. The pKa values of the target analytes are not compatible with the pKa values of
common weak ion exchange sorbents and only strong ion exchange materials can be
considered.
High polarity is also the major issue in the selection of a chromatographic method that
would be suitable for the target cytostatics and metabolites. A classical approach of
applying reversed phase chromatography on C18 materials is not conceptually suitable
for highly polar analytes and another approach has to be selected.
CYTOSTATIC DRUGS
5-Fu CytR GemC
5-fluorouracil cytarabine gemcitabine

CAS number 51-21-8 147-94-4 95058-81-4
formula C4H3FN2O2 C9H13N3O5 C9H11F2N3O4
MW [g/mol] 130.02 243.09 263.07
Hpc [atm m3/mol] [1] 1.6610-10 1.5710-19 1.7010-17
solubility [g/L] 12.5[3] 200[4] 51.4 [1]
Log P[2] -0.89 -2.09 [5] -1.22
pKa [2] 8.00 acidic 4.2 basic [5] 3.6 basic
URINARY EXCRETED HUMAN METABOLITES
FBAL* araU dFdU
-fluoro--alanine uracil 1--D-arabinofuranoside 2,2-difluorodeoxyuridine

CAS number 3821-81-6 3083-77-0 114248-23-6
formula C3H6FNO2 C9H12N2O6 C9H10F2N2O5
MW [g/mol] 107.04 244.07 264.06
Hpc [atm m3/mol] [1] 1.4910-10 2.8210-19 3.0610-17
solubility [g/L] 1000 61.2 15.6
Log P[2] < -2 -1.98 -1.04
pKa[2] 1.5 acidic, 9.3 basic 9.6 acidic 9.6 acidic
Figure 1.5 Physicalchemical properties of the three selected cytostatics and their major urinary excreted human
metabolites. (*) -Fluoro--alanine was not included in the analytical method described in the Chapter 3. Data
sources: estimated using EPI Suite v 3.20 software from US EPA [1]; estimated using ADME Boxes v 2.5 software
from Pharma Algorithms [2]; and experimental values from (Osol, 1975, p.1079)[3]; (Osol, 1980, p. 1088)[4]; and
(Romanova, 1996)[5].
1.2.3 Current state of knowledge on environmental occurrence, effects and fate
The currently available data on the occurrence of the target cytostatics are limited to one
of the three drugs, 5-fluorouracil, and are listed in Table 1.4. The levels in wastewater
from an oncology clinic, connected exclusively to 3 toilets and 3 showers used by 18
patients reached up to 122 g/L (Mahnik, 2004). In two other European studies,
5-fluorouracil was not detected in hospital or municipal wastewater above the limits of
quantification of the analytical method used (Tauxe-Wuersch, 2006; Yu, 2006). No data
on the occurrence of cytarabine or gemcitabine in the aquatic environment, or on the
occurrence of any of the above described urinary excreted human metabolites, are
available, to our best knowledge.
Tests of the ecotoxicity and genotoxicity for the target cytostatics and their metabolites
were performed in parallel to this study at the Institute of Hygiene and Environmental
Medicine of RWTH Aachen by Radka Zounkova (Table 1.5).
Table 1.4 Available occurrence data for the target cytostatics.

Pharmaceuti Measured
Analytical LOD/LOQ
cal Sample concentrations Reference
method [ng/L]
[ng/L]
Composite samples of wastewater SPE,

5-fluorouracil from oncology clinic of Vienna capillary 1000 / 5000 20 000 - 122 000 (Mahnik, 2004)
University Hospital, Austria electrophoresis
Composite samples of:

- wastewater from University
Hospital of Lausanne, Switzerland SPE, (Tauxe-Wuersch,
5-fluorouracil 15 / 50 Not detected
- WWPT influent GC-MS 2006)
- WWTP effluent
- wastewater from residential area
Grab samples of:

SPE,
5-fluorouracil - WWPT influent n.a. Not detected (Yu, 2006)
GC-MS
- WWTP effluent
The lowest observed effect concentrations (LOEC) of the ecotoxicity tests range from
micrograms per liter to milligrams per liter and are in general lower then the minimal
genotoxic concentrations (MGC), which are exclusively in milligrams per liter for all
compounds. MCG of 5-fluorouracil in the data set of Table 1.5 was not determined
because cytotoxicity of this compound was higher than its genotoxicity (growth
inhibition >50%). 5-Fluorouracil is showing the lowest EC50 concentrations in all
performed tests, except for the algae test, in which gemcitabine is slightly more
sensitive. Parent cytostatic compounds show higher toxicity than their metabolites.
No observed effect concentrations (NOEC) used later for environmental risk assessment
(Chapter 4.2.4) were chosen from Daphnia magna reproduction test data in Table 1.5,
as it was the most sensitive tested system. If NOEC value was higher than the highest
tested concentration, this concentration was taken as the worst case NOEC.
Table 1.5 Ecotoxicity tests (Pseudomonas putida - bacteria - endpoint growth, Desmodesmus subspicatus - algae -
endpoint growth rate, Daphnia magna - fresh water crustacean - acute test 48 h - endpoint immobilization,
reproduction test 21 days - endpoint reproduction), and genotoxicity tests (umu test minimal genotoxic
concentrations (MGC) with and without metabolic activation (MA)). (Zounkova, 2010, submitted to Chemosphere).
ECOTOXICITY TESTS [mg/L]
5-Fu CytR GemC FBAL AraU dFdU

NOEC 0.02 <10 <50 50 > 500 > 500
Pseudomonas putida LOEC 0.03 10 50 100 > 500 > 500
EC50 0.044 17 100 140 > 500 > 500
NOEC 10 20 <10 <25 > 200 20
D. subspicatus LOEC 40 40 10 25 > 200 200
EC50 48 53 45 80 > 200 > 200
Daphnia magna NOEC 1.0 50 5.0 > 100 > 100 60
acute LOEC 5.0 100 50 > 100 > 100 100
EC50 15 200 110 > 100 > 100 > 100
Daphnia magna NOEC 0.01 1.0 > 1.0 > 10 > 10 > 1.0
reproduction LOEC 0.05 3.7 > 1.0 > 10 > 10 > 1.0
EC50 0.1 10 > 1.0 > 10 > 10 > 1.0
GENOTOXICITY TESTS [mg/L]
5-Fu CytR GemC FBAL AraU dFdU

umu test MGC without MA - 167 167 667 > 667 > 667
MGC with MA - 333 42 > 667 > 667 > 667
Biodegradability data available from the literature show different degradability of the
three cytostatics obtained at different conditions, although data are in general not very
consistent (Table 1.6). While the results of Mahnik et al (2007) and Kiffmeyer et al
(1998) show complete degradation for 5-fluorouracil within the tested conditions,
results of Yu et al (2006) show only partial degradation. At high concentrations in
experiments performed by Kummerer et al (1997), no degradation was reported at all
within 28 days.
Table 1.6 Biodegradability of the target cytostatics
5-Fluorouracil Cytarabine Gemcitabine
concentrations 9 mg/L 4.5 mg/L 7 mg/L

time 28 days (40 days) 28 days (40 days) 28 days (40 days)
Closed bottle test 40% biodegraded 42% biodegraded

fate not biodegraded
(Kummerer, 1997) (85%) after lag-phase of 3 weeks (43%) after lag-phase of 3 weeks
analysis O2 with oxygen electrode O2 with oxygen electrode O2 with oxygen electrode
matrix mineral medium in deionised water mineral medium in deionised water mineral medium in deionised water
concentrations 854 mg/L (175 mg/L) 228 mg/L 1660 mg/L

time 28 days 28 days 4 days (28 days)
Zahn-Wellens test fate 2% biodegraded (17%) >90% biodegraded 45% biodegraded (50%)
(Kummerer, 1997) analysis DOC DOC DOC
mineral medium in deionised water
matrix mineral medium in deionised water mineral medium in deionised water
(effluent from communal hospital)
concentrations 5 mg/L 12.5 mg/L

time 2 days 10 days
fate 92% biodegraded 70% biodegraded
OECD confirmatory test analysis HPLC-DAD HPLC-DAD
(Kiffmeyer, 1998)
matrix synthetic sewage synthetic sewage
the higher the initial concentration
remark
of 5-Fu the slower the degradation
concentrations 1 g/L
time 50 days
Aerobic batch
fate 50% biodegraded
biodegradation
(Yu, 2006) analysis GC-MS
1:1000 diluted
matrix
waste activated sludge
concentrations 5 g/L
time 24 h
Elimination by activated
fate >95% biodegraded
sludge
(Mahnik, 2007) analysis LSC ([2-14C] 5-Fu )
municipal ww from local sewer,
matrix
8-18 g/L suspended solids
1.2.4 References
LUA BB, 2002. kotoxikologische Bewertung von Humanarzneimitteln in

aquatischen kosystemen. Landesumweltamt Brandenburg,
SCM, 2007. Swiss Compendium of Medicines. SA, D., Basel, Switzerland.
Abbruzzese, J. L., Grunewald, R., Weeks, E. A., Gravel, D., Adams, T., Nowak,
B., Mineishi, S., Tarassoff, P., Satterlee, W., Raber, M. N. and Plunkett,
W., 1991. A phase I clinical, plasma, and cellular pharmacology study of
gemcitabine. J Clin Oncol 9 (3), 491-498.
Alsarra, I. A. and Alarifi, M. N., 2004. Validated liquid chromatographic
determination of 5-fluorouracil in human plasma. J Chromatogr B Analyt
Technol Biomed Life Sci 804 (2), 435-439.
Buerge, I. J., Buser, H. R., Poiger, T. and Muller, M. D., 2006. Occurrence and fate
of the cytostatic drugs cyclophosphamide and ifosfamide in wastewater
and surface waters. Environmental Science & Technology 40 (23), 7242-
7250.
Diasio, R. B. and Harris, B. E., 1989. Clinical pharmacology of 5-fluorouracil. Clin
Pharmacokinet 16 (4), 215-237.
Dorr, R. T. and Von Hoff, D. D., 1994. Cancer Chemotherapy Handbook.
Appleton & Lange,
Hamada, A., Kawaguchi, T. and Nakano, M., 2002. Clinical pharmacokinetics of
cytarabine formulations. Clinical Pharmacokinetics 41 (10), 705-718.
Heggie, G. D., Sommadossi, J. P., Cross, D. S., Huster, W. J. and Diasio, R. B.,
1987. Clinical Pharmacokinetics of 5-Fluorouracil and Its Metabolites in
Plasma, Urine, and Bile. Cancer Research 47 (8), 2203-2206.
Johnson, A. C., Jurgens, M. D., Williams, R. J., Kummerer, K., Kortenkamp, A.
and Sumpter, J. P., 2008. Do cytotoxic chemotherapy drugs discharged
into rivers pose a risk to the environment and human health? An overview
and UK case study. Journal of Hydrology 348 (1-2), 167-175.
Judson, I. R., Beale, P. J., Trigo, J. M., Aherne, W., Crompton, T., Jones, D., Bush,
E. and Reigner, B., 1999. A human capecitabine excretion balance and
pharmacokinetic study after administration of a single oral dose of 14C-
labelled drug. Invest New Drugs 17 (1), 49-56.
Kiffmeyer, T., Gotze, H. J., Jursch, M. and Luders, U., 1998. Trace enrichment,
chromatographic separation and biodegradation of cytostatic compounds in
surface water. Fresenius Journal of Analytical Chemistry 361 (2), 185-
191.
Kmmerer, K., 2001. Pharmaceuticals in the Environment; Sources, Fate, Effects
and Risks. Springer,
Kummerer, K. and Al-Ahmad, A., 1997. Biodegradability of the anti-tumour
agents 5-fluorouracil, cytarabine, and gemcitabine: Impact of the chemical
structure and synergistic toxicity with hospital effluent. Acta
Hydrochimica Et Hydrobiologica 25 (4), 166-172.
Lin, N. M., Zeng, S., Ma, S. L., Fan, Y., Zhong, H. J. and Fang, L., 2004.
Determination of gemcitabine and its metabolite in human plasma using
high-pressure liquid chromatography coupled with a diode array detector.
Acta Pharmacol Sin 25 (12), 1584-1589.
Mahnik, S. N., Lenz, K., Weissenbacher, N., Mader, R. M. and Fuerhacker, M.,
2007. Fate of 5-fluorouracil, doxorubicin, epirubicin, and daunorubicin in
hospital wastewater and their elimination by activated sludge and treatment
in a membrane-bio-reactor system. Chemosphere 66, 30-37.
Mahnik, S. N., Rizovski, B., Fuerhacker, M. and Mader, R. M., 2004.
Determination of 5-fluorouracil in hospital effluents. Anal Bioanal Chem
380 (1), 31-35.
Osol, A., 1980, p. 1088. (Editor), Remington's Pharmaceutical Sciences. Osol, A.,
1088, Mack Publ., Easton, PA, 16th ed.
Osol, A. and Hoover, J. E., 1975, p.1079. (Editors), Remington's Pharmaceutical
Sciences. Osol, A., 1079, Mack Publ., Easton, PA, 15th ed.
Romanova, D. and Novotny, L., 1996. Chromatographic properties of cytosine,
cytidine and their synthetic analogues. Journal of Chromatography B-
Biomedical Applications 675 (1), 9-15.
Steger-Hartmann, T., Kummerer, K. and Schecker, J., 1996. Trace analysis of the
antineoplastics ifosfamide and cyclophosphamide in sewage water by
twostep solid-phase extraction and gas chromatography-mass
spectrometry. Journal of Chromatography A 726 (1-2), 179-184.
Sumpter, J. P. and Johnson, A. C., 2008. 10th Anniversary Perspective: Reflections
on endocrine disruption in the aquatic environment: from known knowns
to unknown unknowns (and many things in between). Journal of
Environmental Monitoring 10 (12), 1476-1485.
Tauxe-Wuersch, A., de Alencastro, L. F., Grandjean, D. and Tarradellas, J., 2006.
Trace determination of tamoxifen and 5-fluorouracil in hospital and urban
wastewaters. International Journal of Environmental Analytical Chemistry
86 (7), 473-485.
Yu, J. T., Bouwer, E. J. and Coelhan, M., 2006. Occurrence and biodegradability
studies of selected pharmaceuticals and personal care products in sewage
effluent. Agricultural Water Management 86 (1-2), 72-80.
Zhang, X. X., Zhang, T. and Fang, H., 2009. Antibiotic resistance genes in water
environment. Applied Microbiology and Biotechnology 82 (3), 397-414.
23
Chapter 2
Objectives
You can know the name of a bird in all the

languages of the world, but when you're finished,
you'll know absolutely nothing whatever about the
bird... So let's look at the bird and see what it's
doing - that's what counts. I learned very early the
difference between knowing the name of something
and knowing something.
(Richard Feynman)
24
Objectives Chapter 2 25
2 Objectives
This work aimed at investigating cytostatics in aquatic environment, understanding how
their consumption and environmental concentrations correlate, and how can they be
eliminated. The main objectives addressed in this work are:
A: to develop a robust and sensitive analytical method for the analysis of the
selected cytostatics and their metabolites in wastewater;
B: to apply the analytical method for determination of cytostatic concentrations in

hospital and municipal wastewater;
C: to asses the environmental risk caused by the selected cytostatics (and

metabolites);
D: to evaluate the sorption behavior of cytostatics to powdered activated carbon as

a surrogate for the removal of polar micropollutants and as a potential
wastewater treatment process.
26
27
Chapter 3
Analytical Method Development
To have a great idea, have a lot of them!

(Thomas Alva Edison)
28
Analytical Method Development Chapter 3 29
3 Analytical Method Development

An analytical method for the target pharmaceuticals in wastewater is a tool
necessary for any studies requiring the concentration measurements (e.g. stability
tests or sorption isotherms) as well as for revealing the environmental occurrence
data or mass flow analysis. The method development for the target compounds
faces three main challenges: 1) the analytes are very polar chemicals for which
analysis is generally regarded as difficult; 2) the estimated environmental
concentrations are very low; and 3) the heavy matrix of the raw wastewater
interferes with the analysis of the compounds of interest.
HPLC is nowadays a mature analytical technique, nevertheless its development has
generally mainly focused on reverse phase (RP) chromatography of non-polar
analytes, effectively retained on non-polar stationary phases such as C18. HPLC
retention of highly polar molecules, like the analytes described here, on reverse
phase materials is by definition not effective. If the octanol/water partition
coefficient of the molecule (P) is smaller than 1 (or log P < 0), there is not much
driving force for partitioning of a polar molecule from the polar aqueous mobile
phase to the non-polar stationary phase. As a consequence, polar molecules pass
through the non-polar stationary phase with little or no retention, unless they are
retained by a secondary interaction, e.g. aromatic interaction. Chromatography on
reversed phase materials modified for retention of polar compounds was previously
employed for analysis of antimetabolites in environmental (Kiffmeyer, 1998),
surface wipe (Floridia, 1999), as well as biological samples (Fahmy, 2004).
Alternative and conceptually more suitable liquid chromatographic techniques are
hydrophilic interaction chromatography (HILIC), employed previously for 5-Fu in
plasma and tissue (Pisano, 2005), porous graphitic carbon chromatography, as
employed for 5-Fu and CytR in plasma (Hsieh, 2007), ion pairing chromatography,
as used for CytR in plasma (Hsieh, 2007), micellar electrokinetic capillary
chromatography, as used for CytR and araU in human serum (Houze, 2001) or
normal phase chromatography, as used for GemC and dFdU in plasma and urine
(Freeman, 1995). In our method development we considered hydrophilic
interaction chromatography to be the most suitable approach. Apart from being the
method of choice for HPLC separation of polar compounds, the high amount of
organic modifier in HILIC has a positive influence on the sensitivity of mass
spectrometric detection (Hemstrom, 2006). Hydrophilic interaction chromatography
was first time referred to under the acronym HILIC by Andrew J. Alpert in 1990.
The analytes interact with polar stationary phase material that is more polar than
10-40% aqueous mobile phase. The more polar an analyte the better its interaction
with the stationary phase. Although the mechanism of retention has not been
completely elucidated, it was proposed to be partitioning between the dynamic
mobile phase and a slow-moving layer of water with which the polar stationary
phase is hydrated (Alpert, 1990). This mechanism was however contradicted
(Hemstrom, 2006) and is still being discussed. This thesis also contributes to the
better understanding of the HILIC retention process.
30 Chapter 3 Analytical Method Development
3.1 Experimental
3.1.1 Chemicals and material
Reference compounds 5-fluorouracil, cytarabine, uracil 1--D-arabinofuranoside

and uridine were purchased from Sigma-Aldrich (Seelze, Germany), and cytidine
from Fluka (Deisenhofen, Germany). Gemcitabine hydrochloride and 2,2-
difluorodeoxyuridine were kindly provided by Lilly Research Laboratories
(Indianapolis, IN, USA). Isotopically labeled internal standards 13C15N2 uridine and
15
N2 5-fluorouracil were supplied by Dr Ehrenstofer GmbH (Augsburg, Germany),
13
C3 cytarabine and 13C15N2 5-deoxy-5-fluorocytidine by Toronto Research
Chemicals Inc. (North York, ON, Canada). SPE cartridges Isolute ENV+ (90 m,
1000 mg, 6mL) were supplied by Separtis (Grenzach-Wyhlen, Germany) and
Bakerbond Speedisk H2O-philic SA-DVB disc cartridges (15 m, 200mg, 6 ml) by
Mallinckrodt Baker (Griesheim, Germany). Other tested extraction materials were
Oasis HLB and Oasis MAX (Waters, Eschborn, Germany), Accubond Evidex
(Agilent, Waldbronn, Germany), Empore MPC-SD, Bond Elut Certify I , SPEC
MP1 and SPEC MP3 (Varian, Darmstadt, Germany), Strata X, Strata SAX and
Strata XC (Phenomenex, Aschaffenburg, Germany), Bakerbond Carbon,
Bakerbond Speedisk Narc-2 and Bakerbond Speedisk H2O-philic SC-DVB
(Mallinckrodt Baker, Griesheim, Germany), and ZIC-HILIC SPE (SeQuant, Marl,
Germany).
3.1.2 Sample collection
Hospital wastewater from a Swiss cantonal hospital (415 occupied beds) was
collected automatically and flow-proportionally between April 30 and May 8,
2007. Composite samples of approximately 18-hours, were each prepared and
analyzed in duplicate. The samples were filtered through a 0.7 m GF/F glass-fiber
filter (Whatman, Dassel, Germany) and further through a 0.2 m cellulose acetate
membrane filter (Sartorius, Goettingen, Germany). Subsequently, the samples were
stored in glass bottles at -20C until analysis. For the method development, grab
samples of hospital wastewater from a University Hospital, Germany (1 510 beds)
were used.
3.1.3 Hydrophilic interaction chromatography
The HPLC system HP1100 (Agilent, Waldbronn, Germany) consisted of a binary

pump, an auto-sampler and a column thermostat. The separations were performed
on a 150 2.1 mm ZIC-HILIC PEEK column (3.5 m, 100 ) from SeQuant
(Marl, Germany) equipped with a ZIC-HILIC precolumn (20 2.1 mm, 5 m).
Eluent A was of 30 mM ammonium acetate and acetonitrile (2/3, v/v) and eluent B
was acetonitrile. The pH of ammonium acetate was not adjusted. The column
thermostat was set to 35C, flow rate to 0.2 mL/min and injection volume to 2-6
L. An Agilent HP1100 UV detector was used for HILIC method development and
retention mechanism studies. Standards of 10 mg/L in acetonitrile were monitored
at wavelengths of 267 nm (5-Fu) and 281 nm (CytR, GemC, araU, dFdU). Toluene
was used as a void volume marker for calculation of retention factors used in
Figure 3.1.
3.1.4 Solid Phase Extraction
3.1.4.1 First screening approach

During the SPE method development, different SPE materials were tested by a
simple procedure in which 10 mL of spiked tap water (500 g/L) were percolated
through a conditioned cartridge and collected. Tap water was before loading
adjusted to pH 6.3 for the materials Isolute ENV+, Oasis HLB and Strata X, to pH
2.0 for the materials Bond Elut Certify I, Accubond Evidex, SPEC MP1, SPEC
MP3, Empore MPC-SD, Speedisk Narc-2, Speedisk H2O-philic SC-DVB and
Strata XC (retention by cation exchange for basic analytes), and to pH 11.0 for the
materials Oasis MAX, Speedisk H2O-philic SA-DVB and Strata SAX (retention by
anion exchange for acidic analytes). Additionally, ZIC-HILIC SPE material was
tested as above but spiking 5 mL of acetonitrile/water mixture (9:1) with 100 g/L
of the analytes. Efficiency of eluting solvent for Bakerbond Carbon was
determined by spiking 2 mL of eluting solvent with analytes (25 g/L) and passing
it through a conditioned cartridge.
3.1.4.2 Isolute ENV+ method
Analytes were extracted from 50 mL of wastewater using Isolute ENV+ 1000 mg
cartridges preconditioned with 15 mL of methanol and 15 mL of 10 mM
ammonium acetate buffer pH 6.0. After adjusting the sample with acetic acid to pH
6.0 and spiking 50 ng of labeled internal standards, the samples were transferred
into 60 mL SPE reservoirs fixed on top of each preconditioned cartridge. The
wastewater, filtered as described above, was percolated through the cartridges by
gravity at a flow below 0.3 mL/min. After extraction, the cartridges were washed
with 3 mL of water and dried until the Isolute ENV+ material presented a uniform
light orange color. Analytes were eluted with methanol, and 4.5 mL of the eluate
was evaporated slowly to dryness under a gentle nitrogen stream. Samples were re-
dissolved in 200 L of HPLC mobile phase at initial composition.
3.1.4.3 Optional SPE for difficult matrices
Both materials, Isolute ENV+ and Speedisk H2O-philic SA-DVB, were separately
preconditioned using 15 mL of methanol and 15 mL of tap water adjusted to pH
11.2. A Speedisk H2O-philic SA-DVB disc cartridge was connected beneath each
Isolute ENV+ cartridge. Samples of 50 mL of filtered wastewater were adjusted to
pH 11.2 with 25% ammonium hydroxide, spiked with 50 ng of labeled internal
standards and loaded through the reservoirs connected to the Isolute ENV+
cartridges. After slow loading (0.3 ml/min) and washing with 3 mL of water
adjusted to pH 11.2, the two adjacent cartridges were disconnected and dried. The
Speedisk H2O-philic SA-DVB disc cartridges were eluted with 3 mL of
methanol:formic acid (98:2). Isolute ENV+ cartridges were eluted separately, as
described above. The eluates were carefully evaporated in a nitrogen stream, each
re-dissolved in 100 L of mobile phase and separately analyzed by different
HPLC-MS/MS methods.
3.1.5 Tandem mass spectrometry.
A triple quadrupole mass spectrometer API 4000 (Applied Biosystems, Foster City,
CA) with a Turbo IonSpray source was used for MS/MS detection. The detection
was performed at source temperature 390C in multiple reaction monitoring
(MRM) mode with negative or positive electrospray ionization. The analysis was
separated into two HPLC-MS/MS methods. The HPLC gradient for analytes in
negative ion mode (5-Fu, dFdU, araU) and respective internal standards (Table 3.1)
was as follows: 0 min (88% B), 10 min (88% B), 19 min (35% B), 19.8 min (35%
B), 26 min (88% B) and finished at 37 min. The flux from the column was directed
by a 6-port Valco valve to the detector (2.6 23 min) or discharged to waste.
Sample injection volume was set to 5 L. The ion spray voltage and declustering
potential of the MS/MS were set to -3.7 kV and -40 V respectively. For the
analysis of GemC and CytR as well as the two internal standards (Table 3.1),
positive ion mode was used and following chromatographic gradient was applied: 0
min (77% B), 2 min (77% B), 10 min (26% B), 12 min (26% B), 20 min (77% B)
and finished at 30 min. The flux from the column at 4.2 30 min was directed to
the MS/MS. The sample injection volume was set to 2 L, the ion spray voltage
and declustering potential were 4 kV and 31 V, respectively. Table 3.1 lists further
the measured transitions, optimum values of collision energy and collision cell exit
potential, as well as time segments of the analysis.
3.1.6 High-resolution mass spectrometry.
High-resolution MS was performed using the LTQ-Orbitrap mass analyzer

(Thermo, Waltham, MA, USA). The HPLC system consisted of a Flux Instruments
Rheos 2200 pump (Flux Instruments AG, Basel, Switzerland) and a PAL
autosampler (CTC Analytics AG, Zwingen, Switzerland). Chromatographic
analysis was performed as described above. Ionization was done in the negative
mode of electrospray with source voltage of 4.5 V and capillary voltage of 25 V.
Masses of 5-Fu and the identified interfering peak were subjected to chemical
formula calculator (Thermo Electron, San Jose, CA).
3.1.7 Quantification,identification and quality control.
Analyst 1.4.1 software was used for API 4000 instrument control and data
processing. Quantification was based on 1/x weighted linear regression calibration
curves of peak area ratio (analyte product ion 1 / internal standard product ion 1)
versus concentration. Comparison of retention times of internal standard and
analyte in the same chromatogram was used for identity confirmation.
Furthermore, two product ions of each analyte and internal standard were measured
and their ratio was calculated. This was not feasible for 5-Fu in SPE preconcentrated
wastewater samples due to interference in product ion 2 (see section 3.2.3).
Calibration standards were measured at the beginning and at the end of each
sequence, and one spiked real sample or calibration standard were measured
Table 3.1. HPLC-MS/MS parameters for the target substances and their internal standards. Product ion 1 was used
for quantification. Instrumental limits of detection were estimated from signal to noise ratio (S/N = 3) of a low
concentration calibration standard in mobile phase. tR - retention time, CE - collision energy, CXP - collision cell exit
potential, n.d. not determined.
time molecular product ion 1 product area ratio of

tR
compound segment ion instrumental ion 2 two product
[min] [m/z] CE CXP
[min] [m/z] LOD [pg] [m/z] ions
5-Fu 5.91 0.22 129.0 59.0 -34 -4 0.48 42.0 0.09

dFdU 0 - 12.5 6.14 0.17 262.8 220.0 -16 -10 0.06 111.0 8.4
METHOD 1
15N 5-Fu 5.91 0.15 131.0 43.1 -30 -1 n.d. 59.0 29.6
ESI-
araU 21.18 0.03 242.8 200.0 -14 -10 0.39 110.3 1.9
13C15N urd
12.5 - 22.5
2 20.44 0.09 246.0 110.9 -22 -6 n.d. 201.0 1.2
GemC 6.76 0.13 264.0 112.0 38 8 0.03 95.0 3.2

4.2 - 9
METHOD 2
13C15N dFC 5.33 0.10 249.1 132.7 59 8 n.d. 115.1 7.8

2
ESI+
CytR 14.72 0.14 244.0 112.0 23 8 0.09 95.0 5.0

13C CytR
9 - 30
3 14.72 0.11 247.1 115.0 23 5 n.d. 97.7 5.8
repeatedly throughout the sequence to check for signal stability. Blanks of mobile
phase with and without internal standards were used to check for possible carry-
over. Instrumental limits of detection were estimated from signal to noise ratios
(S/N = 3) of low concentration calibration standards in mobile phase. Method
limits of quantification were estimated from the signal to noise ratio (S/N = 10) of
a real sample. Run-to-run variations were assessed from 5 injections, day-to-day
variations over 3 days. Relative SPE recoveries were determined by spiking the
analytes and internal standards in wastewater prior to SPE. Absolute recoveries
were determined by spiking the analytes prior to SPE and internal standards
afterwards, prior to analysis. For the Isolute ENV+ method, the recovery
experiments were carried out in triplicates. The absolute recovery for the
combination of Speedisk H2O-philic SA-DVB and Isolute ENV+ were determined
for two replicates, relative recovery for 6 replicates.
3.1.8 Safety considerations.
The handling of hospital wastewater and cytostatic drugs requires safety measures
for the protection of personnel and the workplace from contamination from
pathogens and carcinogenic substances.
3.2 Results and Discussion
3.2.1 Hydrophilic interaction chromatography.
Retention of one of the target analytes (5-Fu) by hydrophilic interaction on amino

and silica columns has been previously reported (Olsen, 2001; Pisano, 2005).
Nevertheless, 5-Fu has also been used as a void volume marker on an Atlantis
HILIC Silica column (Grumbach, 2004).Baseline separation is required for
simultaneous determination when using UV detection. This is feasible for 5-Fu,
CytR, GemC, dFdU and araU on the tested ZIC-HILIC column. At isocratic 88%
acetonitrile all peaks can be separated with retention factors k of 0.5 7.6 and
resolution greater than 2.4. The elution order follows the Log P pattern, with the
most polar CytR being retained longest (Figure 3.1), and is the inverse to the
retention order observed on most RP columns (data not shown).
In Fig. 3.1, the logarithms of retention factors of the target compounds, log k, are
plotted against the volume fractions of water in the mobile phase, , in linear as
well as logarithmic scales. A linear relationship of log k = f()would provide an
evidence for partitioning, while linearity in the log k = f(log ) plot would fit the
adsorption model (Hemstrom, 2006). The results show linearity of the loglog plot
with the slope of the function changing at approximately 14% water (Fig. 3.1a).
The coefficients of determination, r2, of the linear regression at both regions (40
14% water and 141.6% water in the mobile phase) are between 0.9916 and
0.9996. The observed changes in the retention characteristics at around 14% water
apply to all five investigated compounds, albeit the change of the slope tends to be
less pronounced as the polarity of the compounds increase. The ratios of the slope
of the linear regression below 14% water and the slope of the regression above
14% water in the mobile phase are 1.2, 1.4, 1.8, 1.8, and 2.5 for cytarabine, uracil
1--D-arabinofuranoside, gemcitabine, 5-fluorouracil, and 2,2-difluorodeoxyuridine,
respectively. The linlog plots at 1.640% water are non-linear (Fig. 3.1b).
(a)
1.0
14 %
0.5
log k'
0.0
-0.5
-1.5 -1.0 -0.5

log (Volume fraction of water)
(b)
1.0
0.5
log k'
Figure 3.1 0.0

Logarithmic (a) and linear (b) plots
of log.retention factor (log k) vs.
volume fraction of water in ACN /
-0.5
30 mM ammonium acetate, as
mobile phase.
dFdU, 5-Fu, GemC, araU, 0.0 0.1 0.2 0.3 0.4
Volume fraction of water
CytR
Considering splitting the plot into two regions at around 18% water in the mobile
phase, it can be observed that the less polar compounds, for which the change in
the slope of loglog plot was the most evident, show linearity at the region of
18-40% water in the mobile phase. This indicates that there is a change in the
retention mechanism as the relative contributions of partitioning and adsorption are
changing over the range 1.640% water, although it is difficult to assign a specific
break point.
Decreasing the fraction of water below 10% increases the retention times
dramatically. Though HILIC is defined in 1040% aqueous mobile phase (Alpert,
1990), below 10% water the chromatographic interaction in the column is already
out of the scope of that definition. Nevertheless, applying 100% of acetonitrile did
not decrease retention. On the contrary, it resulted in retention times larger than the
tested 420 min for all five analytes. Elution occurred when water was again
introduced into the mobile phase. This indicates either that the previously bonded
water molecules are strongly attached to the stationary phase and partitioning into
the water layer can take place or that in the absence of water there is an interaction
with the sulfobetaine stationary phase itself. Employed stationary phase of the
ZIC-HILIC column is based on silica modified with a covalently bound
zwitterionic polymeric material with a ratio of strong cation to strong anion
exchange groups one to one, resulting in zero surface charge. The zwitterionic
3-sulfopropyldimethylalkylammonio inner salt has a good water-binding ability
(Hemstrom, 2006) that can be exploited in HILIC to create an interactive layer of
stagnant aqueous phase into which polar solutes might partition. A partitioning
mechanism in which analytes do not interact directly with the stationary phase but
partition into the water layer attached to it was proposed for HILIC in 1990
(Alpert, 1990). Recently, the evidence was shown that surface adsorption may also
play a role in retention (Hemstrom, 2006). Our results, presented above, for very
polar compounds of interest and ZIC-HILIC column confirm this evidence.
It is to be noted, that in this study we do not consider weak electrostatic

interactions that can take place on a zwitterionic column, although we are aware
that it might be an oversimplification. With permanently charged stationary phases,
the electrostatic interactions depend only on the ionization state of the analytes and
can take place when those are ionized. This is not very likely for the target analytes
at the working conditions (with exception of 5-Fu, of which less then 10% is
expected to be negatively charged with the remaining 90 + % uncharged) but
cannot be excluded. The pKa values listed in Figure 1.5 and used throughout the
study, refer to aqueous solutions and are, with exception of CytR, software
predicted. Furthermore exact pH values for the acetonitrile/water mixtures of the
mobile phases are not known, and thus neither the exact ionization state of the
molecules.
The effect of the water fraction on the retention was stronger than the influence of
pH, salt concentration or used organic solvent, acetonitrile versus methanol (data
not shown). At 90 or 95% acetonitrile, which were the initial conditions of the two
developed HPLCMS/MS methods, this effect was so pronounced that it resulted
in retention time fluctuations due to occasional changes in solvent ratio delivered
by pumps, operating at the margin of their range (at flow 0.1 mL/min).
Reproducible retention times were achieved with the same pumps by premixing the
eluents and increasing the eluent flow to 0.2 mL/min, which only slightly increased
the theoretical plate height.
Aqueous samples are not suitable for the injection as water is the strongest solvent
in HILIC. We found that the analysis of such samples is possible, when only a
small volume (2 L) is injected. Up to 20 L of calibration standard solution in the
initial mobile phase can be injected without deterioration of the peak shapes. For
the application to wastewater samples we consider an injection volume of 2 to 6 L
of SPE extract in the initial mobile phase composition as optimal.
3.2.2 Solid-phase extraction
The SPE screening test for retention with 10 mL of tap water spiked with analytes,
collected after percolation through a cartridge and analyzed, showed satisfactory
results of less than 10% of the spiked amount in the percolate for Isolute ENV+,
Speedisk H2O-philic SA-DVB, Speedisk H2O-philic SC-DVB, Strata XC and
Speedisk Narc-2.
The hydroxylated highly-crosslinked polystyrene-divinylbenzene copolymer of

Isolute ENV+ showed better retention, especially for 5-Fu and dFdU, than
hydrophilic macroporous poly(N-vinylpyrrolidone-divinylbenzene) copolymer of
Oasis HLB, as also reported for 5-Fu elsewhere (Tauxe-Wuersch, 2006). Both SPE
materials, Oasis HLB and Isolute ENV+, are commercially available polymers
designed to retain a broad range of molecules, including polar ones. Some studies
have compared the two sorbents with similar or slightly worse results for Isolute
ENV+ as summarized in a comprehensive overview (Fontanals, 2005). In many
cases, co-polymeric sorbents with a polar monomer are considered to be more
efficient at extracting polar compounds (e.g. phenols) than the hydrophobic
sorbents that are chemically modified with a polar moiety. It has to be noticed that
many compounds commonly addressed as polar or very polar, for example phenols
(log P above 1), are considerably less polar than the studied cytostatics (log P from
-0.8 to -2.1). Superior performance of Isolute ENV+ in retention of studied
cytostatics suggests that the retention mechanism characterized by interactions
between delocalized electrons of the aromatic ring of the sorbent and the system
of the analytes is favored in the case of very polar aromatic compounds when
compared to other noncovalent interactions like van der Waals forces or hydrogen
bonds.
Before testing retention on the carbon-based sorbent Bakerbond Carbon, a test for
evaluation of eluting solvents was performed. It has been reported that elution from
carbonaceous sorbents is difficult due to the strong - interactions as retention has
been shown to be strong for a wide range of compounds (Hennion, 1999). We
tested methanol, acetonitrile, methylene chloride, tetrahydrofuran, and their
mixtures and observed that analytes undesirably partitioned from the eluting
solvent into the carbon material. The most effective was the mixture methylene
chloride:methanol (8:2) with 2% of formic acid. Still, only two analytes were
recovered above 20%: dFdU (122%) and araU (63%). No efficient solvent for
eluting all analytes was found and therefore the material was not considered
further.
Aside from Isolute ENV+, the tested SPE materials capable to retain the selected
antimetabolites are the ion exchangers. Using the Isolute ENV+ has the advantage
of retaining all target analytes with one cartridge. The drawback is a low degree of
clean-up owing to the ability of the Isolute ENV+ to retain not only a broad range
of chemicals of interest, but also matrix compounds.
Relative recoveries of target analytes in treated wastewater and hospital wastewater

with Isolute ENV+ ranged between 79% 117% and 83% 125%, respectively.
Absolute recoveries ranged from 87% to 132% in treated wastewater and from
72% to 127% in hospital wastewater. Recoveries were verified for spiked
concentrations of 50 ng/L and 2 g/L, with no considerable difference. Evaluation
of recoveries for 50 ng/L of CytR and araU was not possible due to interferences
(see section 3.2.3).
Although the recoveries of the Isolute ENV+ method are satisfying, we have
occasionally observed a high ion suppression during MS/MS for all negatively
ionizable analytes eluting before 10 minutes. Figure 3a shows the extracted ion
chromatogram (XIC) of 5-Fu product ion 1 in the Isolute ENV+ preconcentrated
hospital wastewater sample spiked with cytostatics and injected to the HPLC-
MS/MS system. The same sample after a ten-fold dilution was again injected,
producing a signal of much higher quality and intensity (Figure 3.2b).
For a higher degree of clean-up we tested combinations of Isolute ENV+ with

Speedisk H2O-philic SA-DVB or with Speedisk Narc-2 at pH 11.2 and 1.3,
respectively. The Isolute ENV+ on top of the Speedisk H2O-philic SA-DVB proved
to be the best option. At pH 11.2 the basic compounds were retained uncharged on
Isolute ENV+ and deprotonated acidic compounds were held by an anion- exchanger
on Speedisk H2O-philic SA-DVB. Although the absolute recoveries of 5-Fu, GemC
and dFdU (40 79%) in hospital wastewater at spiked concentrations of 600 ng/L
and 1.8 ug/L are lower compared to the recoveries of the Isolute ENV+ method, the
matrix causes less ion suppression and the HPLC column is not overloaded. The
deficiency of this method is a partial retention of dFdU on the Isolute ENV+. The
internal standard 15N2 5-fluorouracil does not account for this loss, as it passes
through the Isolute ENV+ at pH 11.2 unretained. Concentrations of dFdU were
therefore corrected for the value of relative recovery.
-MRM
2e+3 129.0/58.9
(a)
(a)
1e+3
Intensity [cps]
0 (b)
-MRM
2e+3 129.0/58.9
(b)
Figure 3.2 1e+3

MRM chromatograms of 5-Fu in:
a) spiked hospital wastewater
preconcentrated by Isolute ENV+: 0
ion suppression due to insufficient
sample clean-up; b) same sample 4 5 6 7 8
as above, ten times diluted Time [min]
3.2.3 Mass spectrometry
In environmental samples of municipal or hospital wastewater we often observed

an interfering fragment ion of 129.0/42.0 co-eluting, within only a 0.2 minute
difference, with the most intense 5-Fu transition (5-Fu product ion 2). The presence
of the interfering substance is evident by comparing the retention times of the 5-Fu
and the labeled 15N2 5-Fu in the extracted ion chromatogram of the same
measurement (Figure 3.3). Comparing with the average retention time of the
standards for calibration or spiked samples measured within the same sequence is
not feasible, as the retention time of 5-Fu has a deviation of up to 0.22 min. The
presence of the isotopically labeled internal standards specific to the target analyte
is in this case crucial not only for quantification but also for confirming the identity
of the measured signal. High resolution MS analysis (LTQ Orbitrap) with mass
resolution of 100 000 and a mass accuracy within 5 ppm revealed, that while the 5-
Fu produces a molecular ion of 129.0109 (m/m = 2.3 ppm) in negative mode, the
mass of the interfering molecular ion was 155 ppm higher (Figure 4e, f). Elemental
composition of the interfering substance (elements in use: 1H, 12C, 14N, 16O, 32S)
was proposed to be C4H5O3N2 (m/m = 11.1 ppm). No matrix interference was
observed for the 5-Fu transition 129.0/59.0 (product ion 1) and the transition could
be reliably used for identification and quantification of 5-Fu. The intensity of the
129.0/59.0 signal is however one order of magnitude lower compared to
129.0/42.0, which increases considerably the lowest detectable concentration.
Antimetabolites are intentionally designed to closely resemble pyrimidine, purine

or folic acid so that they can be involved in cell division machinery and can later
cause its blockage. If the analogues differ only very slightly, their chromatographic
separation is difficult. Co-eluting peaks in the extracted ion chromatogram of both
product ions of CytR, 244.0/112.0 and 244.0/95.0, are present in all preconcentrated
wastewater samples, including WWTP effluent. The interfering substance is a
naturally occurring nucleoside, cytidine, which is an epimer of CytR. The two
compounds differ on the sugar moiety at the 2 position, where 2-hydroxyl group
of cytidine is equatorial, while in the case of CytR the group is axial and so in a
position trans to the 3-hydroxyl of the sugar, arabinose. The exact masses and
fragmentation pattern of the two isomers are identical and a small difference in
MS/MS on triple quadrupole high resolution MS

spiked hospital wastewater hospital wastewater hospital wastewater
-MRM -MRM m/z ( 3ppm)
131.0/43.1 8e+5
8e+5 131.0/43.1 131.0050
1e+6 2e+6
(a) (d)
0 0 0
-MRM
-MRM -MRM 3e+4 m/z ( 3ppm) C4H3FN2O2
7e+4
Intensity [cps]
129.0/59.0
129.0/59.0 1e+3 129.0/59.0
1e+3 129.0109
(b) (e)
0 0 0
-MRM
-MRM -MRM m/z ( 3ppm) (?) C4H5N2O3
1e+6 8e+5
129.0/42.0
129.0/42.0 3e+5
3e+5 129.0/42.0 129.0309
(c) (f)
0 0 0
4 5 6 7 8 4 5 6 7 8 3 4 5 6
Time [min] Time [min]
Figure 3.3 5-fluorouracil and an interfering unknown substance in MS/MS extracted ion chromatograms on
triple quadrupole (a-c) and MS extracted ion chromatograms on high resolution LTQ Orbitrap (d-f) of hospital
wastewater samples and hospital wastewater spiked with 0.6 g/L of 5-Fu. (a,d): internal standard 15N2 5-Fu,
(b): 5-Fu product ion 1, (c): 5-Fu product ion 2, (e): 5-Fu, (f): interfering substance
retention times on the ZIC-HILIC column (0.09 min) is not sufficient for
evaluation of the CytR signal. This is especially the case when the concentrations
of cytidine in wastewater samples are 2-3 orders of magnitude larger than the
expected environmental concentrations of CytR. The employed ZIC-HILIC column
does not enable baseline separation of the two epimers. Nevertheless, there have
been reported methods where RP chromatographic separation of CytR and cytidine
is possible employing 100% aqueous mobile phase (Romanova, 1996; Wermeling,
1989) , and other methods of CytR separation from an unidentified endogenous
6e+4 -MRM
242.8/200.0
(a)
4e+4 (a)
2e+4
Intensity [cps]
0 (b)
1.5e+4
-MRM
246.0/110.9
1.0e+4
(b)
Figure 3.4
a) Interfering isomer uridine (tR
20.52) in XIC of 242.0/200.0 of 5.0e+3
preconcentrated hospital waste-
water sample spiked with 2 g/L
of araU (tR 21.21); b) Internal 0.0
standard 13C15N2 uridine (tR 20.54) 16 18 20 22 24
in the same sample Time [min]
compound, that we assume could be cytidine (Hsieh, 2007; Hsieh, 2007). The
consequences of the presence of naturally occurring epimer of araU, uridine, are
not as severe as in the case of CytR described above. The retention times differ by
0.7 min resulting in a separation factor around 1.2 for spiked 2g/L araU in SPE
pre-concentrated hospital wastewater (Figure 3.4).
Isobaric interferences of CytR, araU and 5-Fu are not of large concern in spiked
samples with the amounts of cytostatics in microgram per liter range or higher,
where no preconcentration by SPE is necessary (e.g. lab-scale experiments of
removal from wastewater or ecotoxicity tests).
3.2.4 Application
The developed method employing clean-up and preconcentration on combination

of the Isolute ENV+ cartridges with Speedisk H2O-philic SA-DVB disc cartridges,
followed by separation on ZIC-HILIC column and detection by triple quadrupole
MS with an electrospray ionization was successfully applied to hospital wastewater
samples of the Cantonal Hospital Winterthur for monitoring concentrations of 5-
Fu, GemC and dFdU. The antimetabolite drug 5-Fu was detected in 76% of 17
investigated samples, while for GemC and its metabolite dFdU the number of
samples with concentrations above the limit of detection accounted for 65 % and
88 %, respectively. Figure 3.5 shows a typical extracted ion chromatograms of the
transitions used for quantification. Both cytostatics, 5-Fu and GemC, were detected
only in samples of the days at which they were administered at the oncology clinic
of the hospital. Measured environmental concentrations as well as the method
validation parameters for the three analytes were in the low nanogram per liter
range (Table 3.2).
5-Fu GemC dFdU

-MRM: 129.0/59.0 +MRM: 264.0/112.0 -MRM: 262.8/220.0
2e+3 7e+3
7e+3
3e+3
(a)
0
Intensity [cps]
1e+4
1e+4
1e+3 2e+4
(b)
0 0
2e+3 4e+3
4e+3
1e+3
(c)
0
4 5 6 7 8 6 7 8 9 4 5 6 7 8
Time [min]
Figure 3.5 MRM chromatograms of two cytostatics (5-Fu, GemC) and the human metabolite of GemC (dFdU) in: a)
calibration standard in the mobile phase (7.5 g/L 5-Fu, GemC and dFdU); b) spiked hospital wastewater
preconcentrated by Speedisk H2O-philic SA-DVB on top of Isolute ENV+ by a factor of 500 (spiked: 15 ng/L 5-Fu
and GemC, 60 ng/L dFdU); c) hospital wastewater preconcentrated as above (calculated concentrations of 5-Fu,
GemC and dFdU: 20.0 ng/L, 1.0 ng/L and 68.3 ng/L, respectively).
Table 3.2. Method application to hospital wastewater samples. SPE preconcentration by factor 500 with Speedisk
H2O-philic SA-DVB on top of Isolute ENV+, method limits of quantification (LOQ) were estimated from signal to
noise ratio (S/N = 10) in real sample, precision based on 5 injections per three days. ND - no peak detected, <LOQ -
concentration between detection and quantification limit.
SPE recoveries precision hospital ww sampling campaign

quantification
600 ng/L in hospital ww 60 ng/L in hospital ww April 30 to May 8, 2007 (n= 17)
absolute relative LOQ linear range run-to-run day-to-day ND / >LOQ maximum median
[%] [%] [ng/L] [pg] [ng/L] rsd [%] rsd [%] [samples] [ng/L] [ng/L]
5-Fu 46 7 95 2 5.0 12.5 4 - 1200 1.7 5.6 4/6 26.6 <LOQ

GemC 79 3 118 18 0.9 0.9 0.5 - 1000 14.3 22.0 6/6 37.8 <LOQ
dFdU 40 5 54 13 9.0 22.0 4 - 1200 11.4 19.7 2 / 12 838 68.3
3.2.5 Refernces
Alpert, A. J., 1990. Hydrophilic-Interaction Chromatography for the Separation of

Peptides, Nucleic-Acids and Other Polar Compounds. Journal of
Chromatography 499, 177-196.
Fahmy, O. T., Korany, M. A. and Maher, H. M., 2004. High performance liquid
chromatographic determination of some co-administered anticancer drugs
in pharmaceutical preparations and in spiked human plasma. J Pharm
Biomed Anal 34 (5), 1099-1107.
Floridia, L., Pietropaolo, A. M., Tavazzani, M., Rubino, F. M. and Colombi, A.,
1999. Measurement of surface contamination from nucleoside analogue
antineoplastic drugs by high-performance liquid chromatography in
occupational hygiene studies of oncologic hospital departments. Journal of
Chromatography B 724 (2), 325-334.
Fontanals, N., Marc, R. M. and Borrull, F., 2005. New hydrophilic materials for
solid-phase extraction. TrAC Trends in Analytical Chemistry 24 (5), 394-
406.
Freeman, K. B., Anliker, S., Hamilton, M., Osborne, D., Dhahir, P. H., Nelson, R.
and Allerheiligen, S. R. B., 1995. Validated Assays for the Determination
of Gemcitabine in Human Plasma and Urine Using High-Performance
Liquid-Chromatography with Ultraviolet Detection. Journal of
Chromatography B-Biomedical Applications 665 (1), 171-181.
Grumbach, E. S., Wagrowski-Diehl, D. M., Mazzeo, J. R., Alden, B. and Iraneta,
P. C., 2004. Hydrophilic interaction chromatography using silica columns
for the retention of polar analytes and enhanced ESI-MS sensitivity. Lc Gc
North America 22 (10), 1010-+.
Hemstrom, P. and Irgum, K., 2006. Hydrophilic interaction chromatography. J Sep
Sci 29 (12), 1784-1821.
Hennion, M.-C., 1999. Solid-phase extraction: method development, sorbents, and
coupling with liquid chromatography. Journal of Chromatography A 856
(1-2), 3-54.
Houze, P., Deschamps, F., Dombret, H., Bousquet, B. and Gourmel, B., 2001.
Micellar electrokinetic capillary chromatography quantification of cytosine
arabinoside and its metabolite, uracil arabinoside, in human serum. Journal
of Chromatography B: Biomedical Sciences and Applications 754 (1),
185-192.
Hsieh, Y. and Duncan, C. J., 2007. An ion-pairing liquid chromatography/tandem

mass spectrometric method for the determination of cytarabine in mouse
plasma. Rapid Commun Mass Spectrom 21 (4), 573-578.
Hsieh, Y., Duncan, C. J. and Brisson, J. M., 2007. Porous graphitic carbon
chromatography/tandem mass spectrometric determination of cytarabine in
mouse plasma. Rapid Commun Mass Spectrom 21 (5), 629-634.
Kiffmeyer, T., Gotze, H. J., Jursch, M. and Luders, U., 1998. Trace enrichment,
chromatographic separation and biodegradation of cytostatic compounds in
surface water. Fresenius Journal of Analytical Chemistry 361 (2), 185-
191.
Olsen, B. A., 2001. Hydrophilic interaction chromatography using amino and silica
columns for the determination of polar pharmaceuticals and impurities.
Journal of Chromatography A 913 (1-2), 113-122.
Pisano, R., Breda, M., Grassi, S. and James, C. A., 2005. Hydrophilic interaction
liquid chromatography-APCI-mass spectrometry determination of 5-
fluorouracil in plasma and tissues. Journal of Pharmaceutical and
Biomedical Analysis 38 (4), 738-745.
Romanova, D. and Novotny, L., 1996. Chromatographic properties of cytosine,
cytidine and their synthetic analogues. Journal of Chromatography B-
Biomedical Applications 675 (1), 9-15.
86 (7), 473-485.
Wermeling, J. R., Pruemer, J. M., Hassan, F. M., Warner, A. and Pesce, A. J.,
1989. Liquid-Chromatographic Monitoring of Cytosine-Arabinoside and
Its Metabolite, Uracil Arabinoside, in Serum. Clinical Chemistry 35 (6),
1011-1015.
42
43
Chapter 4
Occurrence and Stability in Hospital

Wastewater
The most exciting phrase to hear in science, the

one that heralds new discoveries, is not 'Eureka!'
('I found it! ') but 'That's funny ...'
(Isaac Asimov)
44
Occurrence and Stability in Hospital Wastewater Chapter 4 45
4 Occurrence and Stability in Hospital

Wastewater
Hospitals are sources for pharmaceuticals and disinfectants, and can be seen, besides
households and industries, as hotspots for the discharge of these emerging contaminants
into the sewer network. If not degraded in the municipal wastewater treatment plants,
they reach surface waters, with potential impact on the ecosystem and human health
(Kmmerer, 2001). Within the top 100 active compounds list of the Swiss national sales
data of pharmaceuticals, 18 % of the medicaments total volume is being administered
in hospitals (Moser, 2007).
Not much knowledge about occurrence of cytostatics in general is up to now available.
Two cytostatics that are not in focus of this work, cyclophosphamide and ifosfamide,
were previously detected in wastewater and even surface waters (Buerge, 2006), while
5-fluorouracil was so far detected only in wastewater of an oncological clinic (Mahnik,
2004). To expand the knowledge about occurrence of cytostatics as well as range of
cytostatics for which occurrence data are available, this chapter describes a sampling
campaign of 5-fluorouracil, gemcitabine and 2,2-difluorodeoxyuridine in a Swiss
cantonal hospital.
46 Chapter 4 Occurrence and Stability in Hospital Wastewater
4.1 Experimental
4.1.1 Sampling of wastewater in a Swiss cantonal hospital
The cantonal hospital has 415 occupied beds and produces a volume of 135 000
m3/year or 1 m3/(bedday) wastewater. It contains 1% out of a total of 41 215 occupied
stationary beds within Switzerland. This number is based on statistics of major hospitals
in 2004 (KSS, 2004) and additional calculation of beds for smaller hospitals based on
the official hospital typology of the Swiss Federal Office for Statistics. The sampling
location included all medical departments of interest, in particular radiology and
oncology institutes, and excluded laboratory chemical water, kitchen and laundry
wastewater. In case of rain events, storm water was discharged from roofs and from half
of the hospital parking area (0.13 ha). Therefore, dry weather conditions were chosen
for the sampling period in order to avoid sample dilution. The campaign was divided in
three successive phases as illustrated in Figure 4.1: a one-week test phase (flow
measurements), a 12-days sampling phase (18h- composite day-samples from 5:00-
23:00) and a one-day sampling phase (2-4 hours composite sample over 24 hours).
Table 4.1 lists the samples in which the concentration of cytostatics was determined.
The hospital wastewater flow rate profile was measured continuously during the test
phase, using a flow meter (American Sigma 950) triggering the auto-sampler (American
Sigma 900max). During the sampling period the average flow was 15.8 m3/h (min 4.0
m3/h; max 205.7 m3/h). Since concentration variations are a priori unknown and
expected to be high, samples must be collected with a flow- or volume-proportional
mode to obtain representative composite samples. Due to the sewer depth of eight
meters, at least 90 seconds of dead time is required between samples for back flushing
the tube. To achieve sufficient accuracy for the volume of an individual sample, a
minimum of 75 ml must be pumped each time. The auto-sampler was set to pump 75 ml
400
380
Test phase Sampling phase
50
360
45
340
Hospital wastewater flow (m /h)
40
320
3
35
Hospital wastewater flow (m /h)
300
30
3
280
25
260
20
240 05.05
15
220 10
Rain event
200 5
180 0
06
09
12
15
18
21
00
00
03
06
09
12
15
18
21
00
03
06
09
12
15
18
21
00
03
160
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
0
0
Mon 07.05.2007 Tue 08.05.2007 Wed 09.05.2007
140
120
100
80
60
40
20
0
17.04 18.04 19.04 20.04 21.04 22.04 23.04 24.04 25.04 26.04 27.04 28.04 29.04 30.04 01.05 02.05 03.05 04.05 05.05 06.05 07.05 08.05 09.05 10.05 11.05
Monday Monday Monday
Figure 4.1 Hospital wastewater flow rate at the sampling point during the test phase (April 18 April 23, 2007)
and the sampling phase (April 29 May 10, 2007).
each time after 2 m3 of wastewater had passed the sampling station. This resulted in an
average sampling interval of eight minutes. The flow in the sewer during the nights of
the test phase was substantially higher than the flow during the day. This was caused by
back flushing the hospitals reverse osmosis installations. As the concentrations of
cytostatics were presumed to be non-detectable in diluted night or even 24-hour
composite samples, day and night samples were separated and only 18-hours day
samples were analyzed. For the constitution of the mass flow profile over one day, nine
2h- to 4h- composite samples were analyzed.
As hospital wastewater is highly contaminated and for certain pharmaceuticals and

pathogens up to about 100 times more polluted than municipal WWTP influents,
considerations of full personal and instrumental safety measures are recommended for
the work with hospital effluents (Oppliger, 2005). All instruments in contact with raw
hospital wastewater were autoclaved or chemically disinfected when thermal
sterilization was not applicable.
Sampling scheme
18h-composite samples 2h- to 4h- composite

5:00 23:00 samples
Table 4.1
Monday, April 30, 2007 6:55 9:35
Sampling scheme of the sampling
Monday, May 7, 2007

Tuesday, May 1, 2007 9:35 11:45
campaign at a Swiss cantonal Wednesday, May 2, 2007 12:05 14:25
hospital: 18-hours (5:00-23:00) Thursday, May 3, 2007 14:25 17:05
flow-proportional composite Friday, May 4, 2007 17:05 20:35
samples were collected during one Saturday, May 5, 2007 20:35 00:05
week, and 2-4-hours flow- Sunday, May 6, 2007 0:05 04:25
proportional composite samples Monday, May 7, 2007 4:25 07:55
were collected over one day. Tuesday, May 8, 2007 7:55 09:55
4.1.2 Description of municipal WWTP and sampling
The hospital wastewater is discharged to a municipal wastewater treatment plant

(WWTP) which treats the wastewater of 115 000 inhabitants. The influent flow rate
reaches 49 000 m3/d at dry weather, what is by a factor of 130 higher then the flow rate
of the hospital. The hydraulic retention time in the WWTP amounts to 16 h, while the
travel time of the wastewater from the hospital to the WWTP is only 1 h. On two
successive Mondays, two 24h- composite flow proportional wastewater samples were
taken after the primary clarifier (WWTP influent) and after sand filtration (WWTP
effluent), both starting at 7:00 am.
4.1.3 Analytes and analytical methods
For the quantification of 5-fluorouracil, gemcitabine and 2,2-difluorodeoxyuridine in

the Swiss cantonal hospital wastewater, an analytical method described in Chapter 3,
employing SPE as described in Section 3.1.4.3 (Optional SPE for difficult matrix) was
employed and a more sensitive mass spectrometer API 4 000 from Applied Biosystems
was used for the detection.
Experiments of 5-fluorouracil, gemcitabine, uracil 1--D-arabinofuranoside and 2,2-

difluorodeoxyuridine stability in wastewater from University Hospital Aachen,
Germany were quantified by the analytical method described in Chapter 3, employing
SPE as described in Section 3.1.4.2 (Isolute ENV+ method). An Applied Biosystems

mass spectrometer, API 3 000, was used for detection. Limits of quantification of both
analytical methods employed in this chapter are listed in Table 4.2. The differences in
LOQs are partially due to different sensitivity of the detectors, and partially due to
different preconcentration factors of the SPE methods.
Table 4.2 Limits of quantification (LOQ) of the employed analytical methods

LOQ in hospital wastewater [ng/L]
API 4 000 API 3 000
Occurrence in hospital wastewater study Stability in hospital wastewater study
5-fluorouracil 5 150
uracil 1--D-arabinofuranoside - 250
gemcitabine 0.9 10
2,2-difluorodeoxyuridine 9 100
4.1.4 Stability in different wastewater matrices
Stability tests of the target compounds were performed in the following matrices and
conditions:
1. fresh grab-samples of wastewater from the University Hospital Aachen,
Germany (1 510 beds), 20C, 100 rpm (rounds per minute), 2 g/L
2. filtered (GF/F glass-fiber filters) fresh grab-samples of wastewater from the
same University Hospital, 4C, 2 g/L
3. filtered (GF/F glass-fiber filters) fresh grab-samples of wastewater from the
same University Hospital, -20C, 2 g/L
4. filtered (GF/F glass-fiber filters) municipal wastewater treatment plant effluent,
used in Chapter 5 for sorption experiment studies, 20C, 100 rpm, 200 g/L
The unfiltered wastewater was divided into 12 Erlenmeyer flasks (50 mL in each flask)
and spiked with 2 g/L of cytostatics (5-fluorouracil and gemcitabine). The contents of
the flasks were shaken in darkness in isothermal incubator (Multitron 2, Infors HT) at
100 rpm and 20C. Samples were taken as follows: t0 (duplicates), t30min, t1h (duplicates),
t2h, t4h (duplicates), t8h, t18h (duplicates) and t30h. After spiking with internal standards,
the exact time was recorded. The same procedure was applied for the metabolites (uracil
1--D-arabinofuranoside and 2,2-difluorodeoxyuridine), which were spiked separately
in another twelve Erlenmeyer flasks at 2 g/L level.
Stability tests at -20C and 4C for 2 days were performed with filtered hospital
wastewater. The aliquots of 50 mL were spiked with 2 g/L of cytostatics or
metabolites (as named above) and frozen or refrigerated in 250 mL glass bottles for two
days. After adjustment to room temperature, they were spiked with internal standards,
preconcentrated and analyzed.
Stability tests at 20C with filtered municipal wastewater treatment plant effluent, after
the sand filtration, were spiked with 200 g/L of 5-fluorouracil or gemcitabine and
agitated in isothermal incubator for 16 hours. Then they were spiked with internal
standards and analyzed without preconcentration.
4.2.1 Occurrence in Hospital Wastewater and Correlation with Pharmacokinetic Data
The mass flow profiles of cytostatics show high daily variations (Figure 4.2). Well
detectable levels of all three compounds were measured between 11:00-20:30. In the
night fractions, no parent cytostatic compound was detected above LOQ, while the
human metabolite dFdU was observed also during the night. The measured mass flow
profile is in agreement with the pharmacokinetic knowledge (Table 1.4). The oncology
treatments are done during the day, and the major part of the parent compounds were
expected to be emitted also during the day, after 1 h for 5-fluorouracil and within 6 h for
gemcitabine. The profile of gemcitabine presents the excretion pattern of one hospitalized
patient the only patient treated with this substance (2300 mg) that day staying
overnight (Table 4.3). Assuming a morning treatment, he excretes the gemcitabine and
its metabolite dFdU in a first (09:35-11:45) and second period (14:25-17:05), which is
physiologically relevant. The metabolite is excreted in higher amounts:
dFdU-to-gemcitabine ratios are 10 resp. 3 in the two periods, which are comparable to
the expected ratio of 6-17 in the excreted urine within the first six hours after infusion
(Table 1.4). Longer elimination half lives for the metabolite dFdU explain the presence
of the metabolized form and absence of the parent compound, gemcitabine in the night
samples.
Figure 4.2 Mass flow profile of 5-fluorouracil, gemcitabine and 2,2-difluorodeoxyuridine over one day (Monday,
May 7, 2009). Points are the calculated values based on the measured concentrations within the composite
sampling time interval. Minimum and maximum of SPE duplicates are described with the error bars.
Table 4.3. Exact consumption data from the oncology clinic of the investigated Swiss cantonal hospital over the
sampling period. The number of treated patients is in parenthesis. 5-fluorouracil data do not include the out-patient
treatment by pump infusion for 22 hours to 7 days.
Daily consumption of active substances [mg/day]

5-Fluorouracil Gemcitabine
Ambulant Hospitalized Ambulant Ambulant Hospitalized
patients patients patients patients patients
(i.v.) (i.v.) (Capecitabine tabl.) (i.v.) (i.v.)
Mon. 30.04.2007 2'500 (3 pat.) 0 0 5'800 (3 pat.) 2'300 (1 pat.)
Tue. 01.05.2007 0 0 0 0 0
Wed. 02.05.2007 3'600 (5 pat.) 43 (1 pat.) 2'600 (1 pat.) 3'800 (2 pat.) 0
Thu. 03.05.2007 1'300 (2 pat.) 43 (1 pat.) 0 0 0
Fri. 04.05.2007 0 43 (1 pat.) 2'500 (1 pat.) 1'800 (1 pat.) 0
Sat. 05.05.2007 0 43 (1 pat.) 0 0 0
Sun. 06.05.2007 0 43 (1 pat.) 0 0 0
Mon. 07.05.2007 950 (2 pat.) 43 (1 pat.) 0 0 2'300 (1 pat.)
Tue. 08.05.2007 0 3'793 (1 pat.) 0 3'300 (2 pat.) 0
Total on 9 days (mg) 8'350 (12 pat.) 4'051 (7 pat.) 5'100 (2 pat.) 14'700 (8 pat.) 4'600 (2 pat.)
Cytostatics total consumption on 9 days = 36.8 g (31 pat.)
4.2.2 Correlation between Consumption Data and Emissions in Hospital Wastewater
The exact consumption amounts for the year 2006 (Table 4.4) and the exact
consumption amounts during the sampling period and numbers of treated patients
during that time (Table 4.3) were made available by oncology department of the
investigated hospital. On average, 3-4 patients per day are treated in oncology with the
investigated drugs. During the sampling period, 4 g/day of those cytostatics have been
consumed. Compared to the average consumption in the investigated hospital in 2006,
4.5 times less cytostatics are used.
Calculated
Consumption Excretion rate [%] excreted amounts
[g/year] in urine and faeces
Urine Faeces [g/year]
5-fluorouracil 1 623.8 10 0
186.8
5-fluorouracil (from capecitabine) 4 518.0 0.54 0
methotrexate 92.4 81 15 88.4
cyclophosphamide 415.3 20 0 83.1
carboplatin 99.8 70 0 69.9
gemcitabine 498.4 5 0 24.9
etoposide 49.0 50 0 24.5
Table 4.4 irinotecan 27.6 22 33 15.2
Cytostatics excreted in the highest ifosfamide 41.0 34 0 13.9
amount in the investigated cantonal cisplatin 31.5 40 0 12.6
hospital in 2006 doxorubicin 22.0 10 45 12.1
Measured emissions and the comparison to their consumption in oncology clinic during
the sampling period are shown in Figure 4.3 and Table 4.5. The highest 5-fluorouracil
loads of 7.5 mg/d (Wednesday) corresponds also to the highest daily consumption
(Figure 4.3). The mass flow profile of 5-fluorouracil over the whole week correlates
linearly with the intravenous consumption pattern (r2 = 0.975, last point rejected, Figure
4.4). The consumption of capecitabine the oral form of 5-fluorouracil, only
administered to out-patients is not detected in the hospital sewer, as it was probably
5-Fluorouracil
14 6000
Consumtion 5-Fu [mg/day]

12
5-Fu mass flow [mg/day]

5000
10
4000
8
3000
6
2000
4
2 1000
0 0
Mo Tue Wed Thu Fri Sat Sun Mon Tue
5-Fu i.v. 5-Fu oral 5-Fu measured - mass flow [mg/day]
Gemcitabine
9000
9,7
Consumtion GemC i.v. [mg/day]

2.5 8000
GemC mass flow [mg/day]
7000
2.0
6000
1.5 5000
4000
1.0 3000
2000
0.5
1000
0.0 0
GemC i.v. GemC measured - mass flow [mg/day]
2',2'-Difluorodeoxyuridine
(human metabolite of Gemcitabine)
300.0 9000
2',2'-Difluorodeoxyuridine mass
8000
250.0
Consumtion of GemC
7000
200.0
i.v.[mg/day]
6000
flow [mg/day]
5000
150.0
4000
100.0 3000
2000
50.0
1000
0.0 0
GemC i.v. 2',2'-Difluorodeoxyuridine measured - mass flow [mg/day]
Figure 4.3. Mass flow profiles in hospital wastewater over 9 days (horizontal lines) compared to total predicted
excreted amounts (bars) during the sampling period. Error bars of the measurement account for the analytical error
of the duplicate measurements. Prediction of excreted amounts of 5-fluorouracile formed from oral capecitabine is
represented by yellow bars, intravenous (i.v.) 5-fluorouracil is represented by orange bars.
Table 4.5. Calculated average emission of drugs in cantonal hospital during sampling period and comparison with
measured emission. For 5-fluorouracil only i.v. applications was considered. The consumption of dFdU was set
equal to the consumption of gemcitabine considering the molecular weight difference*.Predicted loads were
calculated based on the consumption and excretion, not including the consumption by out-patients. Overall
uncertainties account for analytical, sampling and flow-measurement uncertainty and were calculated for each
day: minimum-maximum are given. Average, minimum and maximum recoveries were calculated based on
measured load relative to the predicted load for each day.
Measured
Average Duration Overall Average
Predicted average
consumption Excreted of full uncertainty recovery in
load load
in sampling amount excretion of measured hosp. sewer
[g/day] (min-max)
period [g/day] [h] load [%] (min-max) [%]
[g/day]
10% 0.14 0.0021 1.1

5-fluorouracil 1.38 6 74-112%
(2-98%) (0.03-1.23) (0-0.008) (0-3)
5% 0.11 0.0013 1.4

gemcitabine 2.14 6 86-123%
(1-10%) (0.02-0.21) (0-0.01) (0.1-6)
10% 1.29 0.058 3.7

2,2-difluorodeoxyuridine 2.15* >24 20-123%
(29-86%) (0.62-1.85) (0.007-0.215) (0.7-11)
excreted entirely in patients households. Moreover, the urinary excretion rate of

5-fluorouracil after administration of capecitabine is about 20 times smaller then for the
intravenous application. During the last sampling day no 5-fluorouracil was detected,
even though, there should have been an input. However, this input is predicted from the
treatment of one single hospitalized patient (Tab. 4.3), who might not have excreted into
the toilet of the hospital for personal reasons.
For gemcitabine and its metabolite, a weaker correlation exists between measured loads
and the consumption (Fig. 4.3). Measurable concentrations were found when a
treatment took place, and whenever gemcitabine is present, also dFdU was found. The
highest load of 9.7 mg/d for gemcitabine (last sampling day) was found when also a
relatively high daily consumption occurred. Such high variations of mass loads of
cytostatics in hospital wastewater were also observed in an Austrian study (Mahnik,
2007) and are caused by the low number of patients receiving chemotherapy. In average
over the 9 days sampling period, the amount of 5-fluorouracil detected corresponds to
1.1% of the excreted quantity; for gemcitabine 1.4% of excretion, and for its metabolite
dFdU 3.7% of excretion (Tab. 4.5).
Figure 4.4
Linear correlation between the
intravenous consumption of 5-Fu
at the hospital and the measured
emission of this compound in
hospital wastewater. The point of
the last Tuesday was rejected
(x,y)=(0 mg/day, 3793 mg/day).
There might be several reasons for the recoveries below 5% of the predicted load:
- excretion rates vary from patient to patient (Table 4.5)
- the oncology department provides a high amount of ambulatory treatments: for
5-fluorouracil, 63% were out-patients and 67% of the total quantity is
administered to out-patients; for gemcitabine, the percentages are slightly
higher (80% resp. 76%). Therefore, only about 30% of the administered dose is
fully excreted on-site into the hospital sewer, remaining 70% is excreted in the
hospital only partially.
- relatively high sampling uncertainty occurred due to the sampling procedure
and low number of treated patients in the hospital, as described in detail in
(Weissbrodt, 2009). Because only a few patients are contributing to the daily
load, the chance that at average sampling frequency of 8 minutes, some toilet
pulses were missed is high and an uncertainty of up to 120-130% was
calculated. This high sampling uncertainly is causing the large overall
uncertainties listed in Tab. 4.5.
- not much is known about stability of the target cytostatics in wastewater and
published data on biodegradability of 5-fluorouracil differ (Tab. 1.6)
A comparison for 5-fluorouracil can be done to the study performed by (Mahnik, 2004)
Vienna University Hospital. They sampled wastewater directly from the oncology in-
patient ward fed exclusively by 3 toilets and 3 showers (used by 18 oncology patients).
The concentrations of 5-fluorouracil were as high as 124 g/L and loads recovered were
in line with calculated loads when considering an excretion rate of 2%. In our study,
when using a 2% excretion rate (instead of the average rate of 10% used for our
calculation, Tab. 4.5), 5.5% of the expected excreted amount of 5-fluorouracil is
recovered in the hospital sewer.
Apart from hospital wastewater, we analyzed also influent and effluent of the receiving
municipal wastewater treatment plant. Target cytostatics were neither present above
limit of quantification in the 24-hours composite samples of the municipal wastewater
treatment plant influent nor effluent. This is for 5-fluorouracil in accordance with the
previously published study (Tauxe-Wuersch, 2006), Table 1.4.
4.2.3 Stability Studies
To obtain information on stability of the target compounds, and confirm or disprove one
of the above stated hypothesis for explanation of losses in material balances of
cytostatics in hospital wastewater, tests were performed in different matrices and at
different conditions as described in section 4.1.4.
Filtered hospital wastewater samples stored at -20C are stable and contain after 2 days
89, 98, 93 and 95% of spiked amount (2 g/L) of 5-Fu, araU, GemC and dFdU
respectively (dashed blue lines in Fig. 4.5). No decrease of spiked concentration within
10% was also observed for the above analytes when the filtered samples were stored for
2 days at 4C. The same results were obtained for 200 g/L spiked 5-Fu and CytR at
20C for 24 hours in filtered matrix used for the sorption experiments described in
Chapter 5.
The four investigated analytes, 5-Fu, araU, GemC and dFdU, showed to be unstable in
raw hospital wastewater at the temperature relevant for wastewater in a sewer system
(Figure 4.5). As the figure shows, in case of 5-Fu and GemC, we can speak of first order
degradation reaction. The plot of araU shows slight increase in concentration within the
1st order kinetics k t1/2 steady state

r2
dc/dt = -k * c [h-1] [h] [h]
5-Fu 5-fluorouracil 0.68 0.9913 1.0 5

araU uracil 1--D-arabinofuranoside 0.12 0.9943 5.6 28
GemC gemcitabine 1.05 0.9737 0.7 3
dFdU 2,2-difluorodeoxyuridine 0.28 0.9618 2.5 13
Figure 4.5 Stability tests of four target compounds in raw hospital wastewater at +20C (symbols), filtered hospital
wastewater at -20C (dashed line), and first order kinetics fits (solid line). Limits of quantification are 8, 13, 1 and
5% of spiked concentration 2 g/L of 5-Fu, araU, GemC and dFdU, respectively.
first 5 hours, which could be caused by presence of CytR in the hospital wastewater
metabolizing into araU. This assumption was not confirmed. In case of dFdU, the first six
points fit the curve of the first order kinetics, but the concentration stays on a threshold value
and does not decrease to zero. The decrease in concentration is faster for cytostatics than
for metabolites. The half-times are around 3-5 times lower for cytostatic then for
metabolites, as well as the time when steady state is achieved (five half-times).
The human metabolite of gemcitabine, dFdU, was found to be formed from gemcitabine
in the raw hospital wastewater and its concentration was further decreasing after
showing maximum at approximately 1.5 hours after the start of the experiment.
Assuming that dFdU is not the only transformation product of GemC:
GemC k
k met 2
dFdU k
met 1

dFdU
and setting the differential equations:

dc(GemC) (1)
= k GemC * c(GemC) = (k met1 + k met 2 ) * c(GemC)
dt
dc(dFdU ) (2)
= k met1 * c(GemC) k dFdU * c(dFdU )
dt
the data of dFdU formation and eventual decrease (Fig. 4.5c) were fitted using
Modelmaker Version 4 software and the rate constant of dFdU formation, kmet1 = 0.35 h-1,
was calculated. The rate constant of gemcitabine is equal to the sum of rate constants of
dFdU and the unknown metabolite(s) (kmet2 = 0.70 h-1). The processes involved have
biological transformation character, and it can be theorized that the same processes are
also likely for 5-fluorouracil. Decrease of the concentration due to sorption to
suspended solids is due to low Kow of the target compounds not likely.
The literature reported stability of gemcitabine and dFdU in urine under refrigerated
conditions was at least 1 month and at room temperature was at least 1 week (Freeman,
1995). In the study at Vienna university hospital, 5-fluorouracil was reported to be
stable in wastewater at -20C for 28 days and for at least 7 days at 4C and room
temperature (Mahnik, 2004). If the reported stability tests were performed in filtered
wastewater, they are in accordance with our results. In another publication of the same
research group (Lenz, 2007) it is reported, that even though 5-Fu was detected in the
oncology ward wastewater in concentrations up to 120 g/L, no 5-Fu above the
detection limits was detected in the in-flow of the MBR-system, witch was fed by a
1 000-L tank with retention of approximately 24-hours. This report is fully in agreement
with our results. On the other hand, there is a contradiction between our results and the
results published in another study (Kummerer, 1997), where 5-fluorouracil was not
bio-degraded at all and gemcitabine was degraded to 50% in ready biodegradability, as
well as inherent biodegradability tests with concentrations in mg/L range (Tab. 1.6).
The biodegradation rate in the study was measured by non-specific parameters.
Our results described in this section indicate that the main reason for the recoveries in
hospital wastewater below 5% of the predicted load of cytostatics in the wastewater
revealed in sections 4.2.1-4.2.1 lays in fast biodegradability of the target compounds in
unfiltered raw wastewater. Collecting 18-hours composite sample from 5:00-23:00
implies that the sample is unfiltered for approximately 26 hours from the beginning of
the sampling until next morning when it is transported to the laboratory. Within that
time, as the results depicted in Figure 4.5 show, are most of the target compounds
degraded.
4.2.4 Environmental Risk Assessment
Predicted environmental concentration (PEC) in surface water from the measured

concentration in the hospital wastewater was in this study estimated for two cases: (1)
Case A assumes that the concentrations measured at hospital sewer are the only source
of target compounds into surface waters; (2) Case B assumes that the concentrations
measured at hospital sewer are 20% of target compound input into surface waters.
Further, as depicted in Figure 4.6, it accounts for:

- 100 times dilution of the hospital wastewater before it enters municipal
wastewater treatment plant. In the case of the Swiss cantonal hospital and the
receiving municipal WWTP the dilution factor is 130 (average flows are 15.8
m3/h and 49 000 m3/d respectively).
- no elimination in the municipal WWTP. Although this is a cautious worst
case scenario as the target compounds are biodegradable and unstable in raw
wastewater (Table 1.6, Figure 4.5).
- 10 times dilution of the WWTP effluent by discharge to the receiving surface
water. This is the case when a treatment plant discharges into a small river.
Figure 4.6
Schematic drawing of factors and
flows considered in calculation of
the predicted environmental
concentration (PEC) in surface
water from the measured
environmental concentration (MEC)
in a Swiss hospital wastewater.
The exposure data of the target cytostatics in the aquatic compartment reported here
were related to the toxicity data of the most sensitive test organism from the test
systems listed in the table 1.5. The predicted no effect concentration (PNEC) in surface
waters was obtained by applying a conservative assessment factor of 100 to no observed
effect concentration (NOEC). The choice of an assessment factor influences greatly the
final results of the risk assessment. Factors of 10-1000 are usually employed, depending
on the degree of uncertainty in the extrapolation from the test data on a limited number
of species to the actual environment. The Guideline on the Environmental Risk
Assessment of Medicinal Products for Human Use published by EMEA (European
Medicines Agency) in 2006, recommends in the Tier A when using long-term toxicity
tests to employ an assessment factor of 10, accounting for inter-species variations of
differences in sensitivity, intra-species variability, and laboratory data to field
extrapolation.
The risk quotients PEC/PNEC calculated in Table 4.6 are protective and environmental
impact of 5-fluorouracil, gemcitabine and 2,2-difluoro-deoxyuridine seems to be of
minor importance. Nevertheless, cytostatics are special group of pharmaceuticals with
carcinogenic potency which implies that in general no threshold values for lowest effect
concentrations can be estimated. Until data with specific tests and end-points designed
especially for this class of pharmaceuticals are available, precautionary principle should
be applied.
Table 4.6 Environmental risk assessment

PNEC
PEC (surface water) [ng/L] PEC / PNEC
NOEC [ng/L]
[ng/L] assessment factor: 100
Case A Case B Case A Case B
5-fluorouracil 1 .104 1 .102 0.03 0.14 3 .10-4 1 .10-3

gemcitabine >1 .106 >1 .104 0.04 0.19 <4 .10-6 <2 .10-5
2,2-difluorodeoxyuridine >1 .106 >1 .104 0.88 4.20 <8 .10-5 <4 .10-4
4.2.5 References
KSS, 2004. Kennzahlen der Schweizer Spitler. Bundesamt fr Gesundheit BAG.

Buerge, I. J., Buser, H. R., Poiger, T. and Muller, M. D., 2006. Occurrence and fate
of the cytostatic drugs cyclophosphamide and ifosfamide in wastewater
and surface waters. Environmental Science & Technology 40 (23), 7242-
7250.
Kmmerer, K., 2001. Drugs in the environment: emission of drugs, diagnostic aids
and disinfectants into wastewater by hospitals in relation to other sources-
a review. Chemosphere 45, 957-969.
Kummerer, K. and Al-Ahmad, A., 1997. Biodegradability of the anti-tumour
agents 5-fluorouracil, cytarabine, and gemcitabine: Impact of the chemical
structure and synergistic toxicity with hospital effluent. Acta
Hydrochimica Et Hydrobiologica 25 (4), 166-172.
Lenz, K., Mahnik, S. N., Weissenbacher, N., Mader, R. M., Krenn, P., Hann, S.,
Koellensperger, G., Uhl, M., Knasmuller, S., Ferk, F., Bursch, W. and
Fuerhacker, M., 2007. Monitoring, removal and risk assessment of
cytostatic drugs in hospital wastewater. Water Science and Technology 56
(12), 141-149.
Mahnik, S. N., Lenz, K., Weissenbacher, N., Mader, R. M. and Fuerhacker, M.,
2007. Fate of 5-fluorouracil, doxorubicin, epirubicin, and daunorubicin in
hospital wastewater and their elimination by activated sludge and treatment
in a membrane-bio-reactor system. Chemosphere 66, 30-37.
Mahnik, S. N., Rizovski, B., Fuerhacker, M. and Mader, R. M., 2004.
Determination of 5-fluorouracil in hospital effluents. Anal Bioanal Chem
380 (1), 31-35.
Moser, R., McArdell, C. S. and Weissbrodt, D., 2007. Micropollutants from urban
drainage: Pretreatment of hospital wastewater. GWAGas, Wasser,
Abwasser (11), 869875.
Oppliger, A., Hilfiker, S. and Duc, T. V., 2005. Influence of seasons and sampling
strategy on assessment of bioaerosols in sewage treatment plants in
Switzerland. Annals of Occupational Hygiene 49 (5), 393-400.
86 (7), 473-485.
Weissbrodt, D., Kovalova, L., Pazhepurackel, V., Ort, C., Moser, R., Hollender, J.,
Siegrist, H. and McArdell, C. S., 2009. Mass flows of X-ray contrast media
and cytostatics in hospital wastewater. Environmental Science &
Technology 43 (13), 4810-4817.
58
59
Chapter 5
Removal by Activated Carbon
All science is either physics or stamp collecting.

(Ernest Rutherford)
60
Removal by Activated Carbon Chapter 5 61
5 Removal by Activated Carbon

The development and application of more efficient and cost-effective advanced
wastewater treatment steps is presently an intensively discussed issue (Jones, 2007;
Joss, 2008). Post-treatment based on processes such as adsorption to activated
carbon, membrane processes or ozonation might soon become common also in
wastewater, not only drinking water treatment. Adsorption is widely used for water
purification and has been studied for decades. Most of the research in the field is
focused on apolar (hydrophobic) or moderately polar adsorbates, as the process is
the most effective for those compounds. Generally, compounds with a low
octanol/water partitioning coefficient adsorb rather poorly to activated carbon from
the water phase. Despite low adsorption capabilities, the potential of polar
micropollutant adsorption can be exploited in wastewater treatment processes, at
least to a certain extent, when the process is applied to non-polar compounds.
This chapter aims to study the removal of polar compounds, such as the target
cytostatics, from wastewater by sorption to powdered activated carbon. To
investigate the sorption process, laboratory scale batch experiments were
conducted to evaluate the adsorption kinetics and isotherm data of 5-fluorouracil
and cytarabine on two powdered activated carbons (Norit SAE Super and
Activated Lignite HOK Super). The influence of the solution chemistry (pH and
ionic strength) and other parameters (temperature and competing background
organic compounds) on adsorption was also analyzed. Further, the adsorption
capacity of carbon at different initial micropollutant concentrations was estimated
by a modeling approach that can be used in wastewater treatment to prediction the
minimum powdered activated carbon doses required for elimination of polar
adsorbates to a certain concentration c1 (e.g. calculation of the carbon dose for
removal of the polar micropollutant down to 30 ng.L-1), or a certain c1/c1,0 ratio
(e.g. calculation of the carbon dose for 95% removal of the polar micropollutant).
Interactions between the adsorbent and adsorbate are primarily controlled by non-
specific dispersive interactions (e.g. van der Waals interactions), although
electrostatic interactions between ionic adsorbates and charged adsorbent surface
are also present. Besides the ionization state of the adsorbate, surface charge of the
adsorbent also plays a role in sorption processes (Newcombe, 2006). The surface of
activated carbon is always charged due to surface functional groups. The point of
zero surface charge (pHPZC) determines the pH at which the net surface charge of
the carbon equals zero, since at this pH there are as many positively as negatively
charged functional groups. Below the pHPZC the carbon surface is mostly positively
charged. Despite pHPZC being crucial characteristic of each adsorbent, the values
for commercially available powdered activated carbons are not available. For the
two studied adsorbents, Norit SAE Super and Activated Lignite HOK Super, the
pHPZC values were determined experimentally in this study.
62 Chapter 5 Removal by Activated Carbon
5.1 Experimental
5.1.1 Adsorbents
Two commercially available powdered activated carbons were tested: Norit SAE
Super (Norit, Amersfoort, The Netherlands) and Activated Lignite HOK Super
(RWE Power - Rheinbraun Brennstoff, Frechen, Germany). The adsorbents were
dried in an oven for 3 hours at 150C (ASTM, 2003) prior to storage in glass
bottles with Teflon-coated screw caps. The pH of the point of zero surface charge
(pHPZC) was established using the mass titration method (Li, 2002). Table 5.1
summarizes the properties of the studied adsorbents.
Table 5.1. Characterization of the adsorbents. Values of pHPZC were determined in this study, remaining data
provided by suppliers.
Surface area Particle size d50 Raw Activation Price

pHPZC Supplier
[m2/g] [m] material method [/t]
SAE Super 9.8 1300 15 peat/wood steam Norit 1200

HOK 10.0 300 24 lignite steam RWE 350
5.1.2 Adsorbates
Reference compounds 5-fluorouracil and cytarabine were purchased from Sigma-

Aldrich (Seelze, Germany). The 200 mg.L-1 stock solutions were prepared in
Millipore water and stored in the dark at 4C for up to 3 weeks. Analysis of the
target adsorbates by HPLC-MS/MS was performed on an API 3000 (Applied
Biosystems, Foster City, CA) as described in the Chapter 3. The method involved a
2 L injection of filtered aqueous sample diluted to 50% with acetonitrile. Labeled
internal standards, [15N2] 5-fluorouracil and [13C3] cytarabine, were used for quality
assurance of the analysis. Limits of quantification without SPE were 0.5 and 1
g.L-1 for 5-fluorouracil and cytarabine, respectively. SPE preconcentrated samples
could be quantified for concentrations above 15 and 3 ng.L-1 for 5-fluorouracil and
cytarabine, respectively.
5.1.3 Solvents
All solvents for the adsorption experiments were handled in organic-solvent-free

and surfactant-free glass or polyethylene. Wastewater treatment plant (WWTP)
effluent used for the experiments originated from the municipal wastewater
treatment plant in Aachen Soers, Germany. Approximately 100 L of treatment
plant effluent were collected, thoroughly mixed and stored in 2L bottles at -20C
until they were thawed at room temperature and filtered with GF/F filter
(Whatman, Dassel, Germany) prior to each experiment. The pH of this WWTP
effluent was 7.8 and the total organic carbon concentration was 5 mg.L-1.
Concentrated WWTP effluent was obtained by a four-fold concentration of WWTP

effluent by nanofiltration membrane with a molecular weight cut-off of 200
Daltons. The concentrated effluent originated from the same wastewater treatment
plant as the WWTP effluent described above but was collected at a different time
point. Total organic carbon concentration in the concentrated effluent was 20 mg.L-1.
Single solute isotherms were performed in 5 or 100 mM phosphate buffers with

pHs of 6.2, 7.8 or 9.6. The 5 mM phosphate buffer pH 7.8 was carefully selected,
so that it had the same ionic strength and pH as the above described WWTP
effluent after carbon additions.
5.1.4 Adsorption kinetics and isotherm set-up
Kinetics experiments for each adsorbate were performed separately in 1L

Erlenmeyer flasks containing a given amount of carbon. The higher and lower
doses of Activated Lignite HOK Super were 500 and 50 mg.L-1 for both
adsorbates. The higher dose of Norit SAE Super was 220 mg.L-1 for 5-fluorouracil
and 100 mg.L-1 for cytarabine. The lower dose of Norit SAE Super was 40 mg.L-1
for 5-fluorouracil and 20 mg.L-1 for cytarabine. Subsequently, 400 mL of WWTP
effluent with 200 or 0.8 g.L-1 of a single adsorbate was transferred into the flasks.
Two flasks for each kinetics curve construction were used to obtain measurements
in duplicate. The contents of the flasks were shaken in isothermal incubators
(Multitron 2, Infors HT) at 100 rpm (rounds per minute) and 20C. At the time
point before the addition of wastewater to the activated carbon and in the course of
up to 22 hours, samples of 1 mL were taken by syringe and were immediately
filtered by a syringe filter (IC Acrodisc, 45 m, Pall). The sum of all of the
withdrawn samples decreased the total volume by no more than 2.5%. From the
filtrate, 0.5 mL was transferred into HPLC vials, diluted with acetonitrile and
spiked with an internal standard to 100 g.L-1. The liquid phase concentration of
the adsorbate was measured and the solid phase concentration of the powdered
activated carbon was calculated by a mass balance.
Adsorption isotherm experiments for each adsorbate were performed separately in

250 mL Erlenmeyer flasks containing different amounts of carbon. Each amount of
carbon was weighed twice to obtain the measurements in duplicate; two blanks
with no added carbon were included with each isotherm experiment. Subsequently,
100 mL of WWTP effluent or buffer with 200 or 0.8 g.L-1 of a single adsorbate
was transferred into the flasks. The content of the flasks was shaken in isothermal
incubators at 100 rpm and 4 or 20C for 16 hours. Samples were filtered and
analyzed as described above.
5.1.5 Adsorbate removal prediction
Prediction of the adsorbate removal from the same wastewater matrix at different
initial adsorbate concentrations was performed using a pseudo-single solute
isotherm, derived from a simplification of ideal adsorbed solution theory (IAST),
q1 = K1 (n1 / n2 )c2(1, 0 n1 ) / n1 (c1, 0 c1 )( n1 1) / n1 c11 / n1 (1)
and the direct link between relative removal and the carbon dose
cc =
n2 (11 / n1 )
c2 , 0 (c1, 0 / c1 1)1 / n1 (2)
n1K1
both as described in Qi et al., 2007. Subscripts 1 and 2 refer to the micropollutant

and equivalent background compound (EBC), respectively. K is the Freundlich
isotherm coefficient [(g.mg-1)(g.L-1)-(1/n)], 1/n is the Freundlich isotherm
exponent [dimensionless], c is the aqueous-phase concentration [g.L-1], q is the
solid-phase concentration [g.mg-1], and initial concentrations are represented by

the subscript 0.
Experimental data derived from WWTP effluent on Norit SAE Super were used for
calibration of the model and were fitted by a nonlinear regression routine in RGui,
version 2.6.2, to Eq. (3),
q1 = A(c1, 0 c1 ) ( n1 1) / n1 c11 / n1 (3)
which is the pseudo single-solute isotherm (Eq. (1)) with the term
K1 (n1 / n2 )c2(1,0 n ) / n substituted by A. The two EBC parameters, c2,0 and 1/n2, in the term A
1 1
are correlated and their individual values can not be resolved independently (Qi,
2007). The parameters A and n1 determined by Eq. (3), were used in Eq. (4),
derived from the Eq. (2),
1 c1, 0 (4)
cc = ( 1) 1 / n1
A c1
to calculate the carbon doses for any initial adsorbate concentration c1,0 and a set of
c1 concentrations in the range (c1,010-4; c1,0). From the obtained data, adsorptive
capacities q1 were calculated,
q1 = (c1, 0 c1 ) / cc (5)
and plots of q1=f(c1), predicting the adsorbate removal at different initial

concentrations, were constructed.
5.2.1 Adsorption characteristics on two activated carbons in wastewater
To illustrate the extent of the difference in adsorption of the target hydrophilic

micropollutants, 5-fluorouracil and cytarabine, and two other selected hydrophobic
compounds, bisphenol A and 17--ethinylestradiol, Figure 6.1 compares adsorption
of the four compounds to Norit SAE Super, under the same experimental set-up
and conditions and from the same WWTP effluent with pH 7.8 (Lehnberg, 2009).
Fig 5.1 illustrates the difference in polarities of the 4 adsorbates. Assuming that the
log Dow of 5-fluorouracil is at the working pH, due to its ionization, lower than log
Dow of cytarabine, it can be observed that the sorption behavior corresponds to
what is expected from octanol/water partition coefficients. The apolar
17--ethinylestradiol adsorbs most readily, followed by bisphenol-A, cytarabine
and 5-fluorouracil. To remove 90% of the adsorbates from WWTP effluent, 7, 15,
65 and 160 mg.L-1 of Norit SAE Super is required, respectively. Simplifying the
comparison, to remove polar compounds to the same extent requires an
approximately one order of magnitude higher carbon doses in the wastewater
treatment process. When 30 mg.L-1 of Norit SAE Super is applied to the WWTP
effluent, which is a carbon dose sufficient to remove nearly all of the two apolar
adsorbates, the remaining percentage of cytarabine and 5-fluorouracil are about 35
and 70%, respectively.
Figure 5.1
Polar versus apolar adsorbate. 1.0
Comparison of fraction adsorbate
remaining at equilibrium as a
function of carbon dose for polar
5-fluorouracil, and cytarabine versus
apolar 17--ethinylestradiol, and
c/c0
bisphenol-A from WWTP effluent 0.5
on Norit SAE Super.
Experimental conditions: WWTP
effluent spiked with 200 g.L-1
(open or light solid symbols) or 0.8
g.L-1 (dark solid symbols) of a
single adsorbate, pH 7.8, 20C, 0.0
100 rpm, 16 h contact time.
5-fluorouracil, cytarabine,
0 50 100 150
17--ethinylestradiol,
bisphenol-A -1
carbon dose [mg.L ]
The single solute isotherms of the same four adsorbates in 5 mM phosphate buffer
fit well to a Freundlich isotherm. The Freundlich parameters (Tab. 5.2) of the
isotherms obtained on the second tested carbon, Activated Lignite HOK Super,
reveal that the difference in the Freundlich coefficient (KF) representing sorption
capacity of Activated Lignite HOK Super, is one order of magnitude lower for the
polar adsorbates when compared to the apolar ones. The Freundlich exponent,
determining the slope of the linear curve, is higher for polar then for apolar
adsorbates, resulting in a steeper slope for polar adsorbates which indicates bigger
differences between higher and lower carbon concentrations.
Table 5.2. Freundlich single solute isotherm parameters of HOK in 5mM phosphate buffer pH 7.8, c0=200g.L-1,
t=20C, 16 h contact time, 100 rpm. Isotherms are fitted at range c/c0 at which pH was not influenced by the
adsorbent addition.
KF Range c/c0
q = KF.c1/n 1/n R2
[(g.mg-1)( g.L-1)-(1/n)] [%]
5-fluorouracil 0.55 0.43 0.996 7-76

cytarabine 0.67 0.43 0.987 5-72
17--ehynylestradiol 5.25 0.28 0.953 1-46
bisphenol A 5.24 0.20 0.921 4-48
Figure 5.1 further proves that the percentage of adsorbate removal is independent
of the initial adsorbate concentration, as shown for the two concentrations of
cytarabine and 5-fluorouracil, 200 and 0.8 g.L-1. This allowed for performing all
subsequent experiments at an initial concentration of 200 g.L-1, avoiding sample
preconcentration prior to quantitative analysis. For the tested micropollutants, the
proportionality between the powdered activated carbon (PAC) capacity and the
initial adsorbate concentration was at levels not influencing the adsorption of EBC
previously shown by Knappe et al., 1998 to be consistent with the IAST
(Sontheimer, 1985). The necessary assumptions are that the adsorbent surface
loading is dominated by the competing EBC (q2q1) and the Freundlich isotherm
exponents of both, micropollutant and EBC, are of the same range (0.1-1) (Knappe,
1998). The first assumption holds for the poorly adsorbing highly polar
micropollutants well. And as the typical Freundlich isotherm exponents are in the
range 0.2-1 (Qi, 2007), it can be assumed that the second assumption also holds.
1.0 A
0.5
0.0
1-c/c0
Figure 5.2 1.0 B 0 5 10 15 20
Adsorption kinetics of 5-fluoro-
uracil (circles), and cytarabine
(triangles) from WWTP effluent on
carbon dose [mg/L]
Norit SAE Super (A), and Activated 0.5
Lignite HOK Super (B) at two
carbon doses: higher dose (light
color), lower dose (dark color).
Experimental conditions: spiked 0.0
200 g.L-1 of a single adsorbate,
pH 7.8, 20C, 100 rpm. 0 5 10 15 20
contact time [h]
Kinetic experiments are regarded as a prerequisite for isotherm studies, providing

the information about the equilibrium times. Further, they are of much interest
when designing a wastewater treatment process. Figure 5.2 shows the sorption
kinetics of 200 g.L-1 of 5-fluorouracil and cytarabine in the WWTP effluent, when
two doses of Activated Lignite HOK Super and Norit SAE Super were applied.
The higher and lower doses were selected so that they remove approximately 95
and 40% of the adsorbate, respectively. There is a difference between the sorption
rates at different carbon doses (Fig. 5.2), with longer equilibration times needed for
lower carbon doses. This is consistent with the assumption that adsorption rate is
directly proportional to the concentration of vacant surface sites. The kinetics of
both target adsorbates is comparable and is similar on both studied adsorbents.
Contact times above 5 hours yield results close to equilibrium, for higher as well as
for lower carbon doses, and would be recommended contact time for the proposed
wastewater treatment applications. An equilibration time of 16 hours was applied
in the adsorption isotherm set-up. The kinetics of 0.8 g.L-1 5-fluorouracil and
cytarabine in WWTP effluent provided the same conclusions (data not shown).
Experimental adsorption isotherms of both polar adsorbates from WWTP effluent

to Activated Lignite HOK Super and Norit SAE Super are depicted in Figure 5.3.
Norit SAE Super provides higher adsorption capacities for both target adsorbates
when compared to Activated Lignite HOK Super, which is in accordance with the
different surface areas of the two adsorbents (Table 5.1). The powdered activated
carbon doses required to achieve 90% removal of the target polar adsorbates are 3
to 5-fold higher for Activated Lignite HOK Super than for Norit SAE Super, while
the surface area is 4.3-fold lower. The difference in the adsorption capacities
between cytarabine and 5-fluorouracil is more pronounced on Norit SAE Super
than on Activated Lignite HOK Super. Affinities of the two compounds are equal
on Norit SAE Super, which is not observed on Activated Lignite HOK Super. Non-
linearity in the plot at the region of high surface concentrations is apparent on both
tested adsorbents for cytarabine as well as 5-fluorouracil and its possible cause is
discussed further in section 5.2.2.4.
10
q [g.mg-1]
1
Figure 5.3
Adsorption isotherms of 5-fluoro-
uracil (circles), and cytarabine
(triangles) from WWTP effluent on
Norit SAE Super (open symbols),
and Activated Lignite HOK Super
(solid symbols). 0.1
Experimental conditions: spiked 1 10 100
200 g.L-1 of a single adsorbate,
pH 7.8, 20C, 16 h contact time,
100 rpm. c [g.L-1]
5.2.2 Influence of various parameters on adsorption to activated lignite HOK Super
5.2.2.1 Influence of the solution pH
The Solution pH, ionic strength, adsorbent pHPZC, and adsorbate pKa are the key
factors controlling the Coulombic adsorbent-adsorbate interactions (Newcombe,
2006). Figure 6A shows the adsorption isotherms of 5-fluorouracil and cytarabine
on Activated Lignite HOK Super from 0.1 M phosphate buffer at three different pH
values. In case of 5-fluorouracil, raising the pH above 6 increases the fraction of
negatively charged species (Figure 5.4) and it can be observed that the adsorption
capacity of Activated Lignite HOK Super at pH 9.6 is considerably lower than at
lower pHs. At the same time, increasing the pH also influences the ionization state
of the carbon, increasing the number of negatively charged functional groups and
thus the possibility of repulsion. However, the surface charge stays though on
average positive until the solution pH reaches the pHPZC of the carbon, in the case
of Activated Lignite HOK Super pH 10.0 (Table 5.1). Varying pH from 6.2 to 9.6
has no influence on the adsorption characteristics of cytarabine, as the adsorbate is
always in its neutral form in this range.
H
+
cytarabine
cation
o pKa
PA
C+
neutral o
fluorouracil pHpzc
species PA
C-
pKao
Figure 5.4
anion
Changes in ionization state of

5-fluorouracil, cytarabine and -
Activated Lignite HOK Super, one
of the tested powdered activated
carbons (PAC), with changing pH of
the solution. The arrows indicate 0 2 4 6 8 10 12
the pH at which the adsorption
experiments were performed. pH
The addition of different carbon doses into unbuffered solution results in a gradual
change in the pH of the system. Performing the adsorption isotherm experiment on
Activated Lignite HOK Super in Millipore water (data not shown) resulted in a
curve overlapping the plot of pH 6.2 in Figure 5.5A when the pH was in the range
of 5-8 (low carbon doses) and the plot of pH 9.6 when the pH was in the range of
9.5-10.5 (high carbon doses). The part of the curve in between those two ranges
had a steep slope and represented carbon doses 150-300 mg.L-1. This implies the
fact that isotherms in unbuffered systems without pH control might lead to
substantial errors, as the possible change of solution pH after addition of increasing
doses of carbon influences both the ionization state of adsorbate and the net surface
charge of the carbon.
A pH ionic strength B
pH 6.2: Freundlich 1/n=0.39, K=0.44 5 mM: Freundlich 1/n=0.43, K=0.55
pH 7.8: Freundlich 1/n=0.36, K=0.51
10 10 100 mM: Freundlich 1/n=0.36, K=0.51
q [g.mg ]
q [g.mg ]
-1
-1
5-Fu 5-Fu
1 1
pH 6.2: Freundlich 1/n=0.36, K=0.75 5 mM: Freundlich 1/n=0.43, K=0.67
10 carbon dose [mg/L] 10 100 mM: Freundlich 1/n=0.48, K=0.43
carbon dose [mg/L]
CytR CytR
1 1
30 50 100 200 30 50 100 200
-1
c [g.L ] -1
c [g.L ]
C temperature matrix D
4C: Freundlich 1/n=0.15, K=1.87 5 mM phosphate buffer
10 WWTP effluent
10 20C: Freundlich 1/n=0.36, K=0.51
concentrated WWTP effluent
1
q [g.mg ]
q [g.mg ]
-1
-1
5-Fu 0.1 5-Fu

1
4C: Freundlich 1/n=0.13, K=2.73
10
10 20C: Freundlich 1/n=0.48, K=0.43
carbon dose [mg/L]
CytR 0.1 CytR

1
20 50 100 200 1 10 100
-1
c [g.L ]
-1 c [g.L ]
Table 5.5. Effects of pH, ionic strength, temperature and matrix compounds on adsorption isotherms of
5-fluorouracil (circles), and cytarabine (triangles) on Activated Lignite HOK Super. Freundlich isotherm coefficients
are listed for the single solute isotherms in a phosphate buffer (q = K.c1/n, K is listed in (g.mg-1)( g.L-1)-(1/n)).
Experimental conditions: (A)-(D): spiked 200 g.L-1 of a single adsorbate; 100 rpm; contact time 16 h.
(A): 100 mM phosphate buffers; pH 6.2, 7.8 or 9.6; 20C. (B): 5 or 100 mM phosphate buffers; pH 7.8; 20C. (C):
100 mM phosphate buffer; pH 7.8; 4 or 20C. (D): 100 mM phosphate buffer, WWTP effluent or concentrated
WWTP effluent; pH 7.8; 20C.
5.2.2.2 Influence of ionic strength
To determine the effect of the solution ionic strength on the adsorption process, the
adsorption isotherms of 5-fluorouracil and cytarabine were derived from two
buffers of different molarity at pH 7.8 (Figure 5.5B). Comparison of the
corresponding plots shows that the increase in the ionic strength decreases the
sorption capacity of 5-fluorouracil, while for cytarabine is the decrease apparent
only at low surface concentrations. This is due to the different ionization state of
the adsorbates and is in agreement with the theory presented in (Moreno-Castilla,
2004). At solution pH 7.8, the Activated Lignite HOK Super surface charge is
positive on average, 5-fluorouracil is partially negatively charged (less than 50%)
and cytarabine is in its neutral form. Screening of the attractive electrostatic
interactions between Activated Lignite HOK Super and 5-fluorouracil as well as
when surface concentrations of cytarabine are low, is more pronounced at higher
ionic strength conditions, resulting in lower sorption capacity.
For uncharged hydrophobic compounds it was previously demonstrated that an

increase of ionic strength can lead to a considerable increase in adsorption capacity
(Toledo, 2005). This is partially explained due to the salting-out effect, decreasing
the solubility of the hydrophobic solute and enhancing its adsorption. For the
uncharged polar compounds, as the results of cytarabine show, is this is due to their
good solubility not the case.
5.2.2.3 Influence of temperature
A study was also conducted on the temperature effects on the adsorption capacity
of Activated Lignite HOK Super. The adsorption isotherms of 5-fluorouracil and
cytarabine were obtained at the same experimental conditions, but different
temperatures, at 4C and 20C. Figure 5.5C depicts the results showing that lower
temperature provides more favorable conditions for the adsorption of both solutes.
As adsorption is an exothermic process, decrease in temperature increases the
uptake, which is also expected from the adsorption theory (Moreno-Castilla,
2004). It should be noted that the extent of the influence is not negligible, but
comparable with the extent of the influence of the solution pH and ionic strength.
5.2.2.4 Influence of competing compounds
Presence of competing matrix compounds greatly influences the adsorption of

both polar adsorbates, as is noted on different scale of the y-axes when comparing
plots in Figure 5.5A,B,C,D. The single solute isotherms of 5-flourouracil and
cytarabine obtained in 5 mM phosphate buffer pH 7.8 fit the Freundlich isotherm
model. Table 6.2 lists the Freundlich parameters of the target adsorbates. Isotherms
in multi-compound mixture of WWTP effluent show lower adsorption capacity and
non-linearity (Figure 5.5D). The effect is even more pronounced in the
concentrated WWTP effluent. Generally poorly adsorbing polar compounds
experience direct competition with natural organic matter. The competition is less
pronounced at high carbon doses than at low carbon doses, when the number of
sorption sites is very limited. The change in the extent of the influence is reflected
in the non-linearity of the isotherms obtained in the WWTP effluent.
5.2.3 Prediction of adsorbate removal from the wastewater
While experimentally determined single solute isotherms of 5-fluorouracil and

cytarabine fit the Freundlich isotherm model well, isotherms in the WWTP effluent
have a sharp drop in q to the right of the maximum and fit well a pseudo-single
solute isotherm in log-log q=f(c) plot (Qi, 2007).
As shown and discussed in section 5.2.1., PAC capacity and the initial adsorbate
concentration are in case of the target adsorbates in the WWTP effluent
proportional even if the adsorption isotherms are not linear. This can be used for
prediction of adsorbate removal from the same matrix (wastewater) at different
initial adsorbate concentrations if the data for one initial micropollutant
concentration are available. Figure 6.6 shows the pseudo-single solute isotherm fit
(Qi, 2007) of the experimental data of cytarabine with an initial concentration of
200 g.L-1 in WWTP effluent on Norit SAE Super. No single solute isotherm data
are required for the employed approach. The two unknown parameters from
Eq. (3), A and n1, were determined to be 0.04860.0016 L.mg-1 and 2.0090.089,
respectively.
Further, Fig.5.6 shows the predicted isotherms of cytarabine in WWTP effluent for
initial concentrations 33, 6.2 and 0.8 g.L-1 as well as experimental data for 0.8
g.L-1. Good agreement between experimental and predicted data confirms the
suitability of Eq. (1) and (2) for such a prediction and proves that the concept of Qi
et al., 2007 is suitable for highly polar adsorbates. This simple and powerful
approach might be used in wastewater treatment for prediction of minimal PAC
doses required for elimination of polar adsorbates to a certain concentration c1, e.g.
30 ng.L-1 as depicted in Fig. 5.7, or a certain c1/c1,0 ratio.
10
1
q [g.mg-1]
Figure 5.6
Adsorption isotherm predictions 0.1 c0=200 g.L
-1
for cytarabine on Norit SAE Super c0=33 g.L

-1
in WWTP effluent from municipal

wastewater treatment plant Aachen 0.01 c0=6.2 g.L
-1
Soers, Germany. Experimental -1

data points for cytarabine c0=0.8 g.L
concentrations 200 and 0.8 g.L-1
are shown as grey and black
0.001
triangles, respectively. Solid and
dashed lines represent model fit 0.001 0.01 0.1 1 10 100
and model predictions (Qi, 2007),
respectively.
c [g.L-1]
5.2.4 Conclusions
The presented study focused on the adsorption behavior of highly polar

micropollutants with a log Dow below -0.8 and the potential of powdered activated
carbon for their removal from wastewater. The findings can be summarized as
follows:
(1) Influence of various parameters on adsorption
- Influence of the solution pH on the carbon adsorption capacity was observed to be
relevant only for charged adsorbate and was comparable with the magnitude of the
PAC dose prediction [mg.L-1]

1500
1000
500
Figure 5.7
Prediction of the minimal Norit
SAE Super doses required to
remove cytarabine at different 0
initial concentrations from WWTP
effluent (TOC 5 mg.L-1, pH 7.8, 0 50 100 150 200
20C) to reach final concentration
30 ng.L-1. Initial cytarabine concentration [g.L-1]
influence of ionic strength and temperature. Competing background organic matrix

decreases adsorption capacities to much greater extent than the three above
mentioned factors.
- Performing isotherm experiments with different carbon doses in unbuffered
systems (Millipore water) might lead to pH shifting and consequently substantial
errors. This is especially the case when the pKa or pHPZC of the adsorbates are
close to or within the working pH range.
- Comparing the two adsorbent materials with similar pHPZC, the performance was
corresponding with their surface areas.
(2) Potential for wastewater treatment application
- Doses of carbon sufficient to almost completely remove apolar micropollutants
(bisphenol A and 17 ehinylestradiol) from WWTP effluent are able to remove
approximately 50% of highly polar micropollutants (5-fluorouracil and cytarabine).
- Contact times above 5 hours yield results very close to equilibrium and would be
advisable for maximum efficiency at a wastewater treatment process.
(3) Prediction of the adsorption capacity at different initial micropollutant
concentrations
- Prediction using a pseudo single-solute isotherm for micropollutants in a
wastewater matrix was feasible in this study, as the percentage removal of target
micropollutants was experimentally confirmed to be independent of their initial
concentration. The two main assumptions for IAST simplification, q2q1 and n1n2,
were found valid for the polar adsorbates of interest.
- The pseudo single-solute isotherm equation and the explicit relation between the
carbon dose and equilibrium removal proved to be reliable and simple tool for
prediction of the adsorption capacity at different initial concentrations of the target
polar micropollutants.
5.2.5 References
ASTM (2003). ASTM Standard Practice for Determination of Adsorptive Capacity

of Activated Carbon by Aqueous Phase Isotherm Technique, D 3860 98,
Reapproved 2003.
Jones, O. A. H., Green, P. G., Voulvoulis, N. and Lester, J. N., 2007. Questioning
the excessive use of advanced treatment to remove organic micropollutants
from wastewater. Environmental Science & Technology 41 (14), 5085-

5089.
Joss, A., Siegrist, H. and Ternes, T. A., 2008. Are we about to upgrade wastewater
treatment for removing organic micropollutants? Water Science and
Technology 57 (2), 251-255.
Knappe, D. R. U., Matsui, Y., Snoeyink, V. L., Roche, P., Prados, M. J. and
Bourbigot, M. M., 1998. Predicting the capacity of powdered activated
carbon for trace organic compounds in natural waters. Environmental
Science & Technology 32 (11), 1694-1698.
Lehnberg, K., Kovalova, L., Kazner, C., Wintgens, T., Schettgen, T., Melin, T.,
Hollender, J. and Dott, W., 2009. Atmospheric and Biological
Environmental Monitoring: Removal of Selected Organic Micropollutants
from WWTP Effluent with Powdered Activated Carbon and Retention by
Nanofiltration: Kim, Y. J., Platt, U., Gu, M. B. and Iwahashi, H., Springer,
Heidelberg.
Li, L., Quinlivan, P. A. and Knappe, D. R. U., 2002. Effects of activated carbon
surface chemistry and pore structure on the adsorption of organic
contaminants from aqueous solution. Carbon 40 (12), 2085-2100.
Moreno-Castilla, C., 2004. Adsorption of organic molecules from aqueous
solutions on carbon materials. Carbon 42 (1), 83-94.
Newcombe, G. and Dixon, D., 2006. Interface science in drinking water treatment.
Theory and Applications. Chapter 9: Surface chemistry effects in activated
carbon adsorption of industrial pollutants (D.R.U. Knappe). 155-177,
Elsevier, London, United Kingdom.
Qi, S., Schideman, L., Marinas, B. J., Snoeyink, V. L. and Campos, C., 2007.
Simplification of the IAST for activated carbon adsorption of trace organic
compounds from natural water. Water Res 41 (2), 440-448.
Sontheimer, H., Frick, B. R., Fettig, J., Hrner, G., Hubele, C. and Zimmer, G.,
1985. Adsorption processes for water treatment (in German). 156-172,
DVGW-Forschungsstelle, Karlsruhe, Germany.
Toledo, I. B., Ferro-Garcia, M. A., Rivera-Utrilla, J., Moreno-Castilla, C. and
Fernandez, F. J. V., 2005. Bisphenol A removal from water by activated
carbon. Effects of carbon characteristics and solution chemistry.
Environmental Science & Technology 39 (16), 6246-6250.
73
Chapter 6
Conclusions
I am among those who think that science has great

beauty. A scientist in his laboratory is not only a
technician: he is also a child placed before natural
phenomena which impress him like a fairy tale.
(Marie Curie)
74
Conclusions Chapter 6 75
6 Conclusions
The main objectives of this work lied in: (A) developing a robust and sensitive
analytical method for the analysis of the selected cytostatics and their metabolites in
wastewater; (B) applying the analytical method for determination of cytostatic
concentrations in hospital and municipal wastewater; (C) assesing the environmental
risk caused by the selected cytostatics (and metabolites); (D) evaluating the sorption
behavior of cytostatics to powdered activated carbon as a surrogate for the removal of
polar micropollutants and as a potential wastewater treatment process. This concluding
chapter summarizes to which extent the objectives were achieved, points out the
novelties and highlights of this work, and gives a brief outlook for future work.
A robust and sensitive HPLC-MS/MS method with SPE pre-concentration was

successfully developed and validated for analysis of the selected, highly polar
cytostatics and their metabolites in wastewater. The limits of quantification (LOQ)
were in low ng/L range, which is an improvement when compared to other recently
published methods for the similar matrices. Concerning the retention mechanism studies
investigated in parallel with the method development, this is one of the first studies that
confirms the hypothesis that the hydrophilic interaction chromatography (HILIC)
retention mechanism is not based solely on partitioning, but also on surface adsorption.
The method was applied for the determination of two cytostatics with high consumption
and one known human metabolite in the wastewater effluent of a Swiss hospital, along
with the influent and effluent of the corresponding municipal wastewater treatment
plant (WWTP). None of the measured analytes were present above the detection limits
in any of the municipal wastewater samples, while the concentrations in hospital
wastewater ranged between <5 and 27 or <0.9 and 38 ng/L for 5-fluorouracil and
gemcitabine, respectively. The gemcitabine metabolite 2,2-difluorodeoxyuridine was
detected in the hospital wastewater in concentrations up to 838 ng/L. To the best of our
knowledge, gemcitabine and its urinary excreted human metabolite 2,2-
difluorodeoxyuridine had never been previously monitored in wastewater or any other
compartment of the aquatic environment. Further, this is the first study to show that
they are present in the wastewater and that 2,2-difluorodeoxyuridine is a metabolite of
gemcitabine formed not only in human body, but also in wastewater. Apart from
measuring concentrations in the hospital effluent, the consumption data of 5-
fluorouracil and gemcitabine at the hospital were collected during every sampling day
(number of patients and doses applied). Good correlation of 5-fluorouracil consumed
and measured amounts (r2 = 0.975) was observed. Further, gemcitabine-to-metabolite
76 Chapter 6 Conclusions
ratios were calculated from concentrations in hospital wastewater samples at the time
periods when only one patient in the hospital was treated. The calculated ratios agree
with values in urine reported in medical literature dealing with pharmacokinetics of
gemcitabine. The target compounds were shown to be readily biotransformed in raw
hospital wastewater and degraded within 5 to 30 hours, which is not the case in other
matrices.
Following the hospital wastewater screening, environmental concentrations could be

estimated and compared with the available ecotoxicity and genotoxicity data. The
conservative risk assessment shows that the environmental risk caused by the target
compounds is not relevant (PEC/PNEC 10-3 - 10-6). Compounds are not stable in the
wastewater and will be partially degraded already before reaching the wastewater
treatment plant and fully degraded by the time they leave the treatment plant and reach
the environment. This however does not mean that the risk assessment of other, more
persistent cytostatics (e.g. cyclophosphamid) would provide the same results and that
our results are valid for cytostatics as a pharmaceutical class. The future research should
focus on defining the appropriate toxicity tests for cytostatics that would consider their
mode of action and account for the effects they may cause in the environment due to
their pharmacological design. Consequentially, a risk assessment with PNEC of such an
origin can be reliably performed.
Sorption to powdered activated carbon was chosen for investigation as one of the
possible solutions for removal of cytostatics from wastewater, during the wastewater
post-treatment process. Batch-scale experiments elucidated sorption of 5-fluorouracil
and cytarabine to Norit SAE Super. A 30 mg/L dose of the carbon, sufficient to remove
completely apolar micropollutants (bisphenol A and 17--ehinyl-estradiol) from a
municipal WWTP effluent, removes 25 - 65 % of the two cytostatics from the same
matrix. Prediction of minimal dose of powdered activated carbon, required to eliminate
the target compounds to certain concentration or ratio, is feasible using the pseudo single-
solute isotherm model.
Last but certainly not least I would like to express my big gratitude
to
all involved professors, project collaborators, technicians, students, and

colleagues from IHU, AQUAbase, EAWAG, and RECETO
for their valuable contribution.

ubomra Kovaov
luba.kovalova@gmail.com
EDUCATION
2009 PhD. Environmental Chemistry
RWTH Aachen University, Germany
2004 MSc. Analytical Chemistry

Slovak University of Technology in Bratislava, Slovakia
2002 BSc. Chemical Technology

Slovak University of Technology in Bratislava, Slovakia
RESEARCH INTERESTS
Pharmaceuticals and their metabolites in aquatic environment - occurrence, fate, transport, as well as
wastewater treatment / post-treatment processes for micro-pollutant removal (mass-flow analysis, hospital
wastewater decentralized treatment, adsorption to powdered activated carbon, ozonation).
Retention mechanisms and analytical method development for determination of organic micro-pollutants in
aqueous matrices (MS-MS, HPLC, RP, HILIC, SPE).
AWARDS AND FELLOWSHIPS

1st Price at Young Scientists Competition of XENOVAC 2009 (International Conference on Xenobiotics in the
Urban Water Cycle, Cyprus). Analysis of Cytostatics and their Human Metabolites in Wastewater by HPLC-
MS/MS: Retention Mechanisms, Interferences and Method Applications.
Marie Curie fellowship for early stage research training supported by the European Communitys Sixth
Framework Programme under contract number MEST-CT-2004-505169. AQUAbase: Organic Micropollutants
in Aquatic Environment - Interdisciplinary Concepts for Assessment and Removal, (2004 2007).
PUBLICATIONS
Kovalova L., McArdell C.S., Hollender J. (2009). Challenge of high polarity and low concentrations in analysis
of cytostatics and metabolites in wastewater by hydrophilic interaction chromatography/tandem mass
spectrometry. J. Chromatography A, 1216, 1100-1108.
Weissbrodt D., Kovalova L., Pazhepurackel V., Ort C., Moser R., Hollender J., Siegrist H., McArdell C.S.
(2009). Mass flows of X-rax contrast media and cytostatics in hospital wastewater. Environ. Sci.
Technol. 43, 4810-4817.
Lehnberg, K., Kovalova, L., Kazner, C., Wintgens, T., Schettgen, T., Melin, T., Hollender, J. and Dott, W.
(2009). Atmospheric and Biological Environmental Monitoring: Removal of Selected Organic
Micropollutants from WWTP Effluent with Powdered Activated Carbon and Retention by
Nanofiltration (book chapter). Kim, Y. J., Platt, U., Gu, M. B. and Iwahashi, H., Springer, Heidelberg.
Kazner, C., Lehnberg, K., Kovalova, L., Wintgens, T., Melin, T., Hollender, J., Dott, W. (2008). Removal of
endocrine disruptors and cytostatics from effluent by nanofiltration in combination with adsorption on
powdered activated carbon. Water Science and Technology. 58, 1699-1706.

-

Uploaded by

Copyright:

Available Formats

-

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

-

Uploaded by

Copyright:

Available Formats

Cytostatics in the aquatic environment:

analysis, occurrence, and possibilities for removal

Berichter: Prof. Dr.-Ing. Juliane Hollender

Tag der mndlichen Prfung: 21. August 2009

To my little niece Anna, who treasures her

It is better to know some of the questions than all of the answers.

1.1 Cytostatics: general characterization, classification,

Anatomical Therapeutic Chemical Classification System

L01 ANTINEOPLASTIC AGENTS / L01CA03 Vindesine L01XE04 Sunitinib

Urinary excretion rates of the unchanged drug

It is assumed that the hospital wastewater concentration of the cytostatics depends on

1.2 Target cytostatics: 5-fluorouracil, cytarabine and

1.2.1 Indications and Pharmacokinetics

Metabolic pathways of the three target antimetabolites are schematically represented in

5-Fluorouracil is applied to patients with carcinoma of the colon, rectum, breast,

ELIMINATION ACTIVATION IN TARGET CELLS

dihydro-5-fluorouracil 5-fluorouracil 5-fluorouridine

uracil 1--D- - cytarabine cytarabine 5-monophosphate

Figure 1.3 Metabolic pathway of cytarabine in humans (Hamada, 2002).

Cytarabine is applied to patients with hematologic malignancies, most commonly acute

ELIMINATION ACTIVATION IN TARGET CELLS

2,2-difluoro- gemcitabine gemcitabine monophosphate

percentage of the total dose in urine dFdU/GemC

GemC dFdU TOTAL drug

1.2.2 Physical-chemical properties and impact on environmental behavior and

5-Fu CytR GemC

5-fluorouracil cytarabine gemcitabine

URINARY EXCRETED HUMAN METABOLITES

FBAL* araU dFdU

-fluoro--alanine uracil 1--D-arabinofuranoside 2,2-difluorodeoxyuridine

1.2.3 Current state of knowledge on environmental occurrence, effects and fate

Table 1.4 Available occurrence data for the target cytostatics.

Composite samples of wastewater SPE,

Composite samples of:

Grab samples of:

5-Fu CytR GemC FBAL AraU dFdU

GENOTOXICITY TESTS [mg/L]

5-Fu CytR GemC FBAL AraU dFdU

Table 1.6 Biodegradability of the target cytostatics

5-Fluorouracil Cytarabine Gemcitabine

concentrations 9 mg/L 4.5 mg/L 7 mg/L

Closed bottle test 40% biodegraded 42% biodegraded

concentrations 854 mg/L (175 mg/L) 228 mg/L 1660 mg/L

concentrations 5 mg/L 12.5 mg/L

LUA BB, 2002. kotoxikologische Bewertung von Humanarzneimitteln in

You can know the name of a bird in all the

B: to apply the analytical method for determination of cytostatic concentrations in

C: to asses the environmental risk caused by the selected cytostatics (and

D: to evaluate the sorption behavior of cytostatics to powdered activated carbon as

Analytical Method Development

To have a great idea, have a lot of them!

3 Analytical Method Development

3.1.1 Chemicals and material

Reference compounds 5-fluorouracil, cytarabine, uracil 1--D-arabinofuranoside

3.1.2 Sample collection

3.1.3 Hydrophilic interaction chromatography

The HPLC system HP1100 (Agilent, Waldbronn, Germany) consisted of a binary

3.1.4 Solid Phase Extraction

3.1.4.1 First screening approach