Interpretive reading of in vitro antibiotic susceptibility

tests (the antibiogramme)

Patrice Courvalin
Centre National de RCfCrence des Antibiotiques, Unit6 des Agents AntibactCriens, CNRS EP 50058,
Institut Pasteur, 25 rue du Docteur Roux, 75724 Paris Cedex 15, France

INTRODUCTION production of a single enzyme. It also applies to

combined phenotypes resulting fiom the coexistence in
In recent years, there has been important progress in the same host of several mechanisms conferring
our knowledge of bacterial resistance to antibiotics. resistance to antibiotics belonging to the same group.
The availability of an increasing number of antibiotics However, the study of strains harboring two cross-
has allowed more precise individualization of resistance resistance determinants indicated that in general the
phenotypes, and enzyme inhibitors have provided clues genes contribute, in an additive fashion, to the degree
concerning certain mechanisms of resistance. Study of of antibiotic resistance, resulting in unambiguous and
the biochemical mechanisms of resistance of strains easily detectable phenotypes.
representative of the various phenotypes has led to
elucidation of cross-resistance, and examination of large
numbers of clinical isolates has provided information RATIONALE
on co-resistance to antibiotics. Detailed analysis of the
bacteriostatic and bactericidal activity of antibiotics, Antibiotics are not lonely individuals but members of tight
alone or in combination, has indicated the limits of in families
vitro antimicrobial susceptibility tests in the detection The vast majority of antibiotics used in human therapy
of resistance resulting in clinical failure. The goal of can be assigned to a few families, such as p-lactams,
combined molecular and therapeutic interpretation of aminoglycosides, quinolones, etc., whlch, in turn, can
susceptibility tests is to provide an improved logical be divided into classes (for example, p-lactams include
basis for decision-making in antibiotic therapy by benzylpenicillin, the methicillin class, amino-carboxy-,
taking into account the recent progress in the acylureido-, and amidinopenicillins, carbapenems,
understanding of bacterial resistance. cephalosporins and monobactams). Since group assign-
ment is based on structural criteria, antibiotics belong-
PRINCIPLE ing to the same family or the same class are closely
related. This implies that they have similar modes of
The molecular analysis and therapeutic interpretation, action and spectra of activity, and that they are likely to
designated ‘interpretive reading’, of the in vitro anti- be affected, albeit in certain cases to various degrees, by
biotic susceptibility tests (the antibiogramme) consists the same resistance mechanisms (cross-resistance).
of three steps: (1) characterization of the resistance Certain families such as macrolide, lincosamide and
phenotype with a judicious assortment of antibiotics streptogramin (MLS) antibiotics, although chemically
belonging to the same family or the same class; (2) distinct, have the same intracellular target and can
deduction from the observed phenotype of the therefore be considered, in view of interpretation of in
corresponding biochemical mechanism of resistance; vitro tests, as a single functional group.
and (3) inference fiom the deduced mechanism of the As a consequence, the results of in vitro anti-
predicted resistance phenotype. biotic susceptibility tests (the antibiogramme) must be
As can be seen from the examples provided in analyzed simultaneously for antibiotics belonging to
Table 1, this approach is most useful for detection the same family or class and not individually for every
and characterization of low-level resistance due to antibiotic considered separately. This notion has clear

526
Courvalin: Interpretive reading of the in vitro antibiotic susceptibility tests (the antibiograrnrne) S27

Table 1 Examples of interpretive reading of antibiogrammes

Host Observed phenotypeJ Inferred mechanismh Predicted phenotype‘

Gram-positive coccid KmRTmSGmSAk’ APH(3‘) KmRTm‘GGm5Ak’~K

KmRTmRGmsAk’ ANT(4‘) KmRTmRGm‘Ak”R
GmR APH(Z”)-AAC(6’) Resistant to all aniinoglycosides

Streptococcus EmK rRNA rnethylase Resistant to all macrolides

Enterobacteriaceae Amino, carboxypenicillinsR TEM1-2, SHVl Acylureidopenicillins’/R, mecillinamR,

Amino, carboxypenicillinsR, Extended-spectrum Resistant to all p-lactams except
synergy clavulanic acid P-lactamase cephamycins, carbapenems, moxalactam
+ cefotaxime or ceftazidime
’Ak, amikacin; Gm, gentamicin; Em, erythromycin; Km, kanamycin; Tm, tobramycin; R , resistant; S, susceptible.
‘AAC, aminoglycoside acetyltransferase; ANT, aminoglycoside nucleotidyltransferase; APH, aminoglycoside phosphotransferase.
‘I, intermediate.
dGentamicin and sisomicin are equivalent except against Enferococcusfaecium because of the chromosomally encoded AAC(6’),which confers
resistance to the latter but not to the former antibiotic.

implications for the selection of antibiotics that should population of susceptible bacteria into two clinical
be used for in vitro susceptibility testing. categories. These limitations or pitfalls of in vitro tests
can often be overcome by complementary analysis (e.g.
Limits of antimicrobial susceptibility tests use of a chromogenic cephalosporin for the detection
Partly for technical reasons, in particular limits of of p-lactamases) or, as indicated in Tables 1 and 2
optical scanning devices resulting in poor quantification and discussed below, by testing antibiotics not used
of bacterial populations lower than lo6 to lo7 CFU/ in clinical practice or by interpreting the results
mL, routine susceptibility tests explore only the obtained with other members of the same group of
bacteriostatic activity of antibiotics. Under these con- antibiotics.
ditions, decreases in activity of certain antibiotics
against certain strains cannot be detected. This lack of Co-resistance
apparent resistance is independent of the technique Although they result from distinct biochemical
used, since it is intrinsic to the biochemical mechanism mechanisms, certain antibiotic resistance traits often
of resistance itself. These limitations are observed co-reside within the same clinical isolate. Apart from
generally with antibiotics that are poor substrates for strain epidemics, this arises following multiple anti-
inactivating enzymes. For example, although bacterio- biotic pressure, spreads because of favorable epidemio-
static activity of amikacin and netilmicin against Gram- logic conditions and persists due to coordinated
positive cocci harboring a 3’-phosphotransferase or a expression of antibiotic resistance genes that are
2”-phosphotransferase-6’-acetyltransferase is apparently frequently clustered along the genome. Advantage can
unaffected, the bactericidal activity of the two be taken of this physical link between different
antibiotics is abolished and they no longer synergize p- determinants for detection of antibiotic resistance. For
lactams. They are inactive from a therapeutic point of example, in certain hospitals more than 99% of
view and the strains should thus be reported as methicillin-resistant Stuphylocncctrs uureus bacteria are
intermediate or resistant to both drugs (Table 1). also resistant to gentamicin. Because of phenotypic
Similarly, staphylococci resistant to high levels of heterogeneity, the presence of the former type of
lincomycin through production of a lincosamide resistance is difficult to ascertain, whereas the latter, due
nucleotidyltransferase apparently remain susceptible to to synthesis of the 2”-aminoglycoside phosphotrans-
clindamycin. However, the minimum bactericidal ferase-6’-aminoglycoside acetyltransferase bifunctional
concentrations (MBCs) of clindamycin against these enzyme, is readily detectable. As the degree of cor-
isolates are greatly increased and the strains should be relation between the two characters is much higher
considered as resistant. Reducing the lower breakpoint than the efficacy of techniques to detect methicillin
that discriminates the susceptible from the intermediate resistance, gentamicin is routinely used in those
and resistant populations of bacteria is therefore of particular ecosystems for detection of resistance to
no help and would only result in distributing the methicillin in strains of S. uureus.
S28 C l i n i c a l M i c r o b i o l o g y a n d I n f e c t i o n , V o l u m e 2 S u p p l e m e n t 1, D e c e m b e r 1 9 9 6

Table 2 Antibiotics and combinations that detect best certain resistance mechanisms
Antibiotic Phenotwe Mechanism' Host

Penicillin+pH indicator Penicillin resistance Penicillinase Gram-positive cocci.

Haemophilus, Neisseriu
Oxacillin P-Lactam resistance Additional PBP Staphylococcus
Oxacillin Penicillin resistance PBP alteration Streptococcus pneumoniae

Aminopenicillin, Penicillin resistance Penicillinase Enterobacteria

aminopenicillin + penicillinase inhibitor
Cefotaxime or cefiazidime P-Lactam (except Extended spectrum p-lactamaseb
+penicillinase inhibitor' cephamycins and (plasmid-mediated)
carbapenems) resistance
Cefoxitin+cephalosporind
Clavulanic acid+cephalosporind Cephalosporin resistance Cephalosporinase
Imipenem+cephalosporind (chromosome-mediated)
Cephamycins, moxalactam+cephalosporind Antibiotic resistance Porin alteration

Cefoxitin Antibiotic resistance Porin alteration Escherichia coli,

Klebsiella
Kanamycin Amikacin resistance APH(3'), ANT(4') Gram-positive cocci
Gentamicin Aminoglycoside resistance APH(2")-AAC(6')
2'-N-Ethylnetilmicin+6'-N-ethylnetilmicine Aminoglycoside resistance AAC(2'),AAC(6') Gram-negative bacteria
Apramycin' Aminoglycoside resistance AAC(3)-IV
Fortimicin' Aminoglycoside resistance AAC(3)-1
Amikacin+tobramycin Aminoglycoside resistance APH(3')-VI Acinetobacter
Netilmicin+tobramycin Aminoglycoside resistance AAC(3) Pseseudomonas

Erythromycin + lincomycin Inducible MLS resistance Ribosomal methylation Gram-positive cocci

Nalidixic acid Quinolone resistance DNA gyrase or porin alteration Gram-negative bacteria

Tinidazole lmidazole resistance Reductase Anaerobes

"C, aminoglycoside acetyltransferase; ANT, aminoglycoside nucleotidyltransferase;APH, aminoglycoside phosphotransferase;PBP,

penicillin binding protein.
bExcept Proteus penneri and I? vulgaris, which resist through production of a cephalosporinase susceptible to penicillinase inhibitors.
'In case of synergism.
case of indifference
'Not used in therapy.

Clinical categorization is already interpretation same species. This approach should thus rely on MIC
As mentioned above, limits of in vitro susceptibility determination, or MIC equivalent in the case of
tests reside in the fact that the increase in MIC of an automated devices. In addition, since there is no
antibiotic due to a resistance mechanism is often low international agreement on breakpoints for inter-
and insuflicient to cross the lower barrier of clinical pretation of in vitro antibiotic susceptibility tests (the
categorization. This confirms that there is an important antibiogramme), a system that intends to be universal
loss of information in expressing results obtained in a cannot be based on subjective and local criteria, as
multiple-class system (MICs are usually determined by resistance mechanisms and their bacterial hosts are
serial twofold dilution of the antibiotic) with three similar worldwide. It is noteworthy that in clinical
(susceptible, intermediate, and resistant) or even two practice there are only two categories, success or failure,
(susceptible and resistant) classes. Clearly, 0.006 and the intermediate category being confined to antibiotics
1 pg/mL of penicdhn are distinct values even though, with dosages that can be significantly increased (e.g. p-
in both cases, the strains could be considered as lactams but not aminoglycosides).
susceptible. A major goal of interpretive reading of in
vitro tests is better detection of antibiotic resistance Facilitation of resistance
mechanisms by comparison of the phenotype of the The attention of the physician should be drawn to
clinical isolate with that of a susceptible strain of the situations where emergence of resistance can be
Courvalin: Interpretive reading of the in vitro antibiotic susceptibility tests (the antibiogramme) S29

anticipated. Mutations (e.g. gyrA) leading to high-level Table 3 Antibiotics used in therapy that should not be
resistance to nalidixic acid in enterobacteria also confer tested
resistance to all other quinolones. However, the degree Antibiotic Microorganism
of cross-resistance depends upon the intrinsic activity
Penicillins (except benzylpenicillin) Staphylococcus
of the drug: fluoroquinolones with only slightly elevated
Cloxacillin, dicloxacillin,
MICs remain clinically active. Nevertheless, these flucloxacillin, nafcillin
strains already possess a resistance mechanism and are Cephalosporins
more likely to become resistant to ciprofloxacin under Clindamycin
monotherapy with t h s antibiotic following a second
Amikacin Gram-positive cocci
mutational event. Similarly, staphylococci that are
Netilmicin
inducibly resistant to MLS are good candidates for
constitutive generalized cross-resistance to these Penicillins susceptible to penicillinases Haemophilus', Nersreria
antibiotics if treated with 16-membered macrolides or (except benzylpenicillin)
with lincosamides.
Topical drugs All bacteria
Prodrugs
~

'Except against strains that are intermediate or resistant to

Carte or menu? penicillins and do not produce a penicillinase.
Under no circumstances can, or even should, all
commercially available antibiotics be tested. Therefore,
what are the best criteria for selection of antibiotics that microbiologists, who conduct in vitro antibiotic
have to be tested in vitro? For proper detection and susceptibility tests (the antibiogramme) only occasion-
characterization of resistance phenotypes, the approach ally. They avoid redundancy and contribute to the
proposed here implies the testing of a minimum of standardization of in vitro tests, a prerequisite for
antibiotics belonging to the same family in the case of comparison of results, epidemiologic studies and
various resistance mechanisms (e.g. aminoglycosides or efficient quality control. The rigidity in the antibiotic
MLS antibiotics), or of a representative of each class in assortment should not preclude the possibility of adding
the case of large f a d e s (e.g. p-lactams). As already antibiotics that are necessary under certain circum-
mentioned, there is often cross-resistance, but to stances (due to the patient, the bacterium or the
various degrees, among antibiotics that are closely environment), for evaluation of new drugs or for
related in structure. Antibiotics most affected by a specific epidemiologic studies.
resistance mechanism, and thus more apt to detect
resistance, have a lower intrinsic activity compared to The antibiotic family as a functional unit
other members of the family. These drugs should The relationship between bacteria and antibiotics is
therefore be tested, as well as combinations of substrate best defined at the species level. This is due to the fact
plus enzyme inhibitor or substrate plus inducer of that chromosomal intrinsic resistance, which results
enzyme synthesis (Table 2 ) . Similarly, antibiotics and from lack of penetration, active efflux, lack of target for
combinations not used (or no longer used) in clinical the antibiotic or production of a detoxift-ing enzyme,
practice should also be included (Table 2). By contrast, is species rather than genus, family or Gram-stain
certain antibiotics largely prescribed in human therapy specific. In cases of acquired resistance this is due to a
should not be tested in vitro, since they consistently limited host range of the plasmids and to barriers in
provide misleading results as to their efficacy both in heterologous gene expression. O n the other hand, and
vitro and in vivo (Table 3 ) . Adequate use of these drugs as already discussed, antibiotics belonging to the same
should be based on the testing of other members of the group display similar spectra of activity and are subject
family (Table 1). to bacterial cross-resistance. For certain combinations
These various technical or biological constraints of bacterial species and antibiotic class, it appears
favor fixed menus that can vary depending upon the justified (and even advisable) to report a result for an
Gram stain, the species, or the site of isolation of the entire family of antibiotics even though a single
bacterial pathogen. As far as possible, they should member has been tested in vitro (Table 1).
include the antibiotics that are currently most
prescribed. Exerting a selective pressure, these drugs are Correct bacterial identification is required
likely to detect early emergence of resistance. Provided Because of the multiplicity of acquired and intrinsic
that they are composed by experts, these menus are resistances that are often combined, analysis, in terms
likely to be more judicious than those established by of biochemical mechanisms, of resistance phenotypes is
S30 C l i n i c a l M i c r o b i o l o g y a n d I n f e c t i o n , V o l u m e 2 S u p p l e m e n t 1, D e c e m b e r 1996

almost impossible if the identity of the bacterial host is codbcting results that have to be verified. If confirmed,
not known. In addition, the clinical relevance of a given the discrepancy may be due to a ‘new’ mechanism that
resistance mechanism also depends upon the host. For does not phenotypically mimic an ‘older’ one.
instance, acquisition of MLS resistance by members of Detection of new mechanisms of resistance leads to
the famdy Enterobacteriaceae is far less important as reconsideration of breakpoints for clinical categoriz-
compared to Gram-positive cocci. Finally, levels of ation (e.g. reduction of the low breakpoint of penicillin
phenotypic resistance achieved by the same determin- for pneumococci) and thus contributes to improve-
ant in different host cells can vary dramatically, being ments in in vitro methods. Interpretation of resistance
easily detectable in one species and barely detectable in phenotypes is based on the comparison of clinical
another (for example, 6’-aminoglycoside acetyltrans- isolates with ‘prototype’ susceptible bacteria belonging
ferase determines in Pseudomonus a broad-resistance to the same species. It is therefore easier to perform
spectrum, whereas in Escherichia coli it confers resistance with bacteria possessing a single resistance mechanism
to only a few antibiotics). Fortunately, there are per class of antibiotic. However, coexistence of various
efficient techniques for rapid bacterial identification at mechanisms conferring resistance to the same group of
the species level, which is required for interpretive drugs is common in human pathogens. As already
reading of the in vitro antibiotic susceptibilitytests (the discussed, this does not constitute a problem, since, in
antibiogramme). Conversely, species identification also the majority of the cases studied so far, these
leads to identification of intrinsic resistance mech- mechanisms act synergistically and confer high-level
anisms, e.g. 6’-aminoglycoside acetyltransferase in resistance that is readdy detectable. A remaining
Enterococcus faecium and Serratia, 2’-amino-glycoside limitation is intrinsic resistance of new opportunistic
acetyltransferase in Providencia, and 3’-aminoglycoside pathogens that has not yet been totally explored.
phosphotransferase or active efflux in I? aeruginosa.
Biochemical and clinical resistances are not synonymous
Biochemical resistance is secondary to mutations or
acquisition of exogenous DNA and refers to the
Not all medical microbiologists are experts in antibiotics parental strain considered as susceptible. It does not
Since interpretive reading of in vitro antibiotic always correlate with clinical resistance. For example,
susceptibility tests (the antibiogramme) requires an Escherichia coli produces a chromosomal cephalo-
immense knowledge of antibiotics (chemical structure sporinase and, although amino-, carboxy-, and acyl-
and mode of action, mechanisms of resistance and ureidopenicillins and cephalosporins are less active
corresponding phenotypes, intrinsic resistance, fluid against this species, these antibiotics remain active in
and tissue pharmacokinetics, metabolism, MIC distri- therapy. Conversely, staphylococci, Haemophilus, and
bution of bacterial populations, correlation between gonococci that produce a penicillinase should be
in vitro results and therapeutic outcome, etc.), this considered as clinically resistant, irrespective of their
approach remains the domain of a few experts. As a MIC.
matter of fact, because of the amount of information
required, interpretive reading is best achieved by Permeability mutants and active efflux
computerized expert systems. Several of these systems Antibiotic resistance through production of a
have been developed in recent years in France and are detoxifying enzyme entails resistance to antibiotics of
used for routine work or for teaching purposes. the same class. Usually, the MICs of antibiotics that are
Fortunately, the rapid spread of automated sus- substrates are high whereas those of antibiotics that are
ceptibility devices and of computers in clinical not modified remain low, resulting in an enzymatic ‘in
laboratories wdl, no doubt, favor the popularization of vivo substrate profile’ that is easily recognizable. By
artificial intelligence applied to the evaluation of contrast, mutations in porins or in lipopolysaccharide
antibiotic activity, which should result in improved that alter permeability of the cells and active efflux of
quality and safety of in vitro tests and better education the drugs can confer low-level cross-resistance to
of medical microbiologists. structurally unrelated antibiotics such as p-lactams,
chloramphenicol, quinolones, tetracycline and tri-
Not all antibiotic resistance mechanisms are known methoprim. This pathway to resistance is therefore
Obviously one can only routinely analyze resistance difficult to detect in vitro, and also by molecular
mechanisms that have been previously reported. techniques that are not suited for detection of point
Interpretive reading, by utilizing bacterial identification mutations. In a few instances, certain antibiotics that are
and antibiotic susceptibility,is the most efficient way to more affected can help in detecting this type of
detect unknown mechanisms. It draws attention to resistance (Table 2 ) .
Courvaiin: interpretive reading of the i n vitro antibiotic susceptibility tests (the antibiogramme) S31

CONSEQUENCES Table 4 Antibiotics useful for bacterial identification.

Antihiotic Microorganism
Individualization of bacterial species
Amino-, carboxypenicillins Citrobacter diverstrs, Klebsiella
For reasons already considered, interpretation of in
Aminopenicillins, Citrobacterfreundii, Etzterobacter,
vitro antibiotic susceptibility tests (the antibiogramme)
first-generation Movanella, Proteus v d p r i s ,
cannot be dissociated from bacterial identification. For cephalosporins Providencia, Serratia
example, the modal MIC of ampicillin for susceptible
Cefoxitin C. freundii, Clostridium dijficile,
pneumococci is 0.01 pg/mL and these bacteria are Enterobacter
considered as resistant when the MIC is 20.1 p g / d , Cefalotin, Cefotaxime Bacillus
whereas MICs of the same antibiotic for susceptible E.
Clavulanic acid Campylobacter
coli strains are 2 to 4 p g / d . Conversely, groups of
Sulbactam Acinetobacter, Pseudomonas cepacia
bacteria need to be individualized for optimum in vitro
Imipeneni Xantkomonas maltoplzilia
study (i.e. staphylococci, streptococci, enterobacteria,
Pseudomonas, Haemophihr, Gram-positive anaerobes Aminoglycosides I? cepacia, l? maltuphilia, streptococci,
and Gram-negative anaerobes). For each group of anaerobes
microorganisms, different antibiotics have to be tested Tetracycline Proteus mirabilis
and various breakpoints should be used. This necessity
is recognized, at least in part, by the various national Lincomycin Eikenella corrodens, Ew terococcusfaecalis,
Haemophilus, Staphylococcus xylosus,
committees for antimicrobial susceptibility testing.
Listeria, Neisseria, Gram-negative
However, this subdivision represents an additional bacteria
degree of complexity for analysis of the results that,
again, can be best handled by expert systems. Bacitracin Streptococcus pyqenrs
Colistin Sensitivity: l? aeruxinosa
Aid in bacterial identification Resistance: Proteus, Providencia,
If proper identification is required for analysis of Serratia, I? cepacia, Gram-positive
bacteria
susceptibility tests, conversely, intrinsic resistance can
be extremely useful in correcting or confirming Trimethoprim Acinetobacter, Brucella spp.,
bacterial identification. This is the reason why certain Campylobacter, Neisseria, Nocardia,
antibiotics that often do not constitute the best choice Pseudomonas spp.
in therapy should be tested in vitro against particular Nitrofurans Proteus, Providencia, Acinetobacter,
species or genera (Table 4). micrococci
Novobiocin Staphylococcus saprophytieus
Improved and permanent quality control Fosfomycin Acinetobacter, S. saproplryticus
Together with reports of more logical results, a major Vancomycin Erysipelothrix rhusopathiae, Lactobacillus,
outcome of interpretive reading of in vitro antibiotic Leuconostoc, Nocardia, Pediococcus,
susceptibility tests (the antibiogramme) is the establish- Gram-negative bacteria
ment of continuous quality control. This is due to (1) Metronidazole Cardnerella vaginalis, Propionibacterrum,
aerobes
simultaneous analysis of identification and antibiotic
susceptibility profile of the clinical isolate that ensures Optochin” Pneumococci
coherence between the two types of results obtained Differentiation Micrococcns,
O / 129d
independently with different techniques, and (2) Staphylococcus
critical interpretation of the resistance phenotype Differentiation enterobacteria,
observed. This quality control is complementary to the Pasteurella, Vibrio
technical one with recommended test strains. “Not used in therapy

Impossible phenotypes
The so-called impossible phenotypes, although almost not correspond to a new resistance mechanism, reflect
anything is possible in biology, correspond to resistance erroneous identification or susceptibility testing, or a
phenotypes that have not yet been observed in nature mixed culture.
(Table 5 ) . They are the result of intrinsic resistance or
susceptibility but also of cross-resistance and co- Abnormal and isolated resistance or susceptibility
resistance to antibiotics. Their discovery implies that at In this case, a single resistance or susceptibdity character
least one of each identification or drug susceptibility does not fit with the other results. This anomaly is most
should be rechecked. Impossible phenotypes, if they do often associated with a technical defect. For example,
S32 C l i n i c a l M i c r o b i o l o g y a n d I n f e c t i o n , V o l u m e 2 S u p p l e m e n t 1 , D e c e m b e r 1996

Table 5 Examples of impossible antibiotic resistance phenotypes

Resistance phenotype Microorganism

GentamicinR, other aminoglycosidesS Gram-positive cocci

MinocyclineR, tetracyclineS

Methicillin or oxacillinR,cephalosporins or carhapenem’ Staphylococcus aureus

Methicillin or oxacillinR,penicillins

PenicillinR Group A, C, G streptococci

TeicoplaninR,vancomycin’ Enterococci

Amino-, carboxypenicillinss Citrobacter diuersus

Aminopenicillinss, first-generation cephalosporinss, Citrobacterjeundii, Enterobacter cloacae, Morganella

aminopenicillin+clavulanic acidS (except Proteus penneri and I? vulgaris) morpnii, Proteus vulgaris, Providencia, Serratia, Yersinia

Cefamandole or cefuroximeS l? uul,,ris, Serratia

Cefoxitin’ Citrobacterjeundii, Enterobacter

Third-generation cephalosporinsR, amino-, carboxypenicillins? Enterobacteriaceae

and/or first-generation cephalosporins’

Colistin’ Proteus, Prouidencia, Serratia

microorganisms susceptible to amoxicillin and resistant Table 6 Examples of rare antibiotic resistance phenotypes
to the combination amoxicillin plus clavulanic acid or Phenotype Microorganism
enterobacteria resistant to imipenem attest to antibiotic
instability. Isolated resistance to lincosamides Staphylococcus aureus
Isolated resistance to tobramycin
Streptogramin resistance
Teicoplanin resistance
High incidence of rare phenotypes
Rare phenotypes (Table 6) represent resistance Resistance to cefotaxime-ceftriaxone Streptococcuspneumoniae
mechanisms that either have been recently detected or
that have not been epidemiologically successful (e.g. Penicillin resistance through Enterococci
4‘-aminoglycoside nucleotidyltransferase in entero- penicillinase productiona
cocci, lincosamide nucleotidyltransferase in S. aureur). Isolated resistance to gentamicin
Isolated resistance to tobramycin
Apart from strain epidemics, for which they constitute
Glycopeptide resistance”
an excellent marker, a sudden change in their incidence
is suspect. For example, frequent isolation of strains of Carbapenem resistance Acinetobacter, enterobacteria,
Staphylococcus resistant to lincomycin suggests that the Barfemidesjagitis
Gram-positive cocci examined may well be, in fact,
enterococci, whereas numerous Enterococcus faecium Resistance to third-generation Escherichia coli,
strains resistant to low levels of vancomycin evoke cephalosporins Proteus mirabilis,
Enterococcus gallinarum or E. casselij-lavus-jlavescens. A Aztreonam resistance Providencia stuartii,
Amikacin resistance Salmonella
high number of isolates of various species resistant to
trimethoprim should lead us to verifjl the thymidine Amino-, carboxypenicillin Klebsiella
content of the culture medium. Of course, the notion susceptibility
of rarity is relative and varies with the ecosystem
considered. Enterococci resistant to penicdhn through Amikacin resistance+tobramycin Pseudomonns aeruginosa
production of a penicillinase are encountered with susceptibility
increasing frequency in North and South America,
Penicillin resistance through Neisseria meningitidis
whereas they have not yet been reported in Europe.
penicillinase production
Also, one does not expect a similar prevalence of
resistance determinants in patients from intensive care 5-Nitromidazole resistance
units and in members of rural communities.
‘Outside North America.
Courvalin: Interpretive reading of the in vitro antibiotic susceptibility tests (the antibiogramme) s33

Epidemiology of resistance characters Table 7 Detection of resistance genes in the absence of

The reasoning in terms of biochemistry substitutes the growth
study of incidence of resistance mechanisms for that of Slow growers
resistance to isolated drugs. This approach is much Myio bacterilrni spp.
more subtle and informative. It explains, for example,
co-selection of resistances due to cross-resistance and Fastidious bacteria
Anaerobes, B. !yydo$ri, Campylohartrr spp., capnophilic Gram
showed how antibiotics restricted to veterinary use can
negatives (Hnen~op/iilus spp.. Candiobacterilrnr /ronrini.c, Artinobacter
select mechanisms found later in clinical isolates where attiiiomyietrmairlrita~is,Eihenella iorrodctir, Kin(qe1la kicr‘yae) , Bartoriella
they confer resistance to drugs used in human therapy qnirrtana, Clilamydiii spp., Helimbniter pylon, Lqiuirrlla pnei~mop/iila,
(e.g. gentamicin resistance by production of a 3- L q t o y i r a spp., Myroplarma spp., Neiwria spp., Rur/ioli,rea Irriirt~lai~,
aminoglycoside acetyltransferase selected by apramycin; Streptoioirlrs pnnrmoniar, I Jreaplarma rtrealyticunr
Table 2).
In vivo cultivable bacteria
RickEttsia, T r i p r e m a spp.
GENOTYPIC DETECTION OF RESISTANCE
‘Dangerous’ bacteria
As already discussed, there are certain limitations to Bntcel/a spp., E tirlaremis. M . tubercrrlosii.
interpretative reading of the in vitro antibiotic suscepti-
bility tests (the antibiogramme) which, in fact, con-
stitute indications to detection of resistance at the based on selective amplification by the polymerase
genotypic level. chain reaction of genes involved in vancomycin
resistance has been developed. This technique allows
Detection of resistance in the absence of growth not only rapid, sensitive and specific detection of
Because the detection level of molecular biology glycopeptide resistance in enterococci but also
techniques can be as low as a single gene copy, identification of enterococci at the species level, since,
genotypic detection of resistance is effective against as already mentioned, resistance can be species specific.
non-growing and even dead bacteria (Table 7). This
not only provides results that could not be obtained by
another approach (e.g. for in vivo cultivable bacteria) CONCLUSION: THE GOAL JUSTIFIES THE MEANS
but also improves the safety in the laboratory when one
is dealing with ‘dangerous’ bacteria. Another major Extensive study of the biochemistry of bacterial
advantage of these techniques is the rapidity since the resistance to antibiotics has had two major practical
results can be obtained in a few hours. applications: (1)the design of new, semisynthetic, drugs
that avoid resistance; and (2) a better Comprehension
Detection of inducible or low-level resistance and hence better detection of resistance.
Glycopeptide resistance in enterococci represents a In vitro antibiotic susceptibility tests (the antibio-
good example of the indications of detection of gramme), like other tests in biology, should provide
resistance at the genome level. There are three types of objective quantitative data, e.g. MICs. For various
resistance, VanA, VanB and VanC, each being difficult historical reasons, they also provide subjective inter-
to detect phenotypically. The VanA type, which is pretation of the data such as clinical categories.
high-level resistance to both vancomycin and Interpretive reading of antimicrobial susceptibility tests
teicoplanin, should be in principle easily detected is an attempt to reconcile these two notions by basing
phenotypically. However, resistance of this type is interpretation on the most recent knowledge in the
slowly inducible by glycopeptides, which explains why field of antibiotic study, in particular that of
VanA resistance cannot be detected by rapid automated mechanisms of resistance. The best that one can ask of
systems. In addition, this resistance is expressed at antibiotic susceptibility testing is detection of resistance,
various levels when transferred to other bacterial in particular of low-level resistance. This can be
genera. The latter observation may well account, at achieved by improved interpretation of the results of in
least in part, for the delay in detection of this type of vitro sensitivity tests or by the design of certain
resistance in other bacterial species. The VanB type of genotypic approaches. The goal of the proposed
resistance is even more difficult to detect, since not only approach is to provide the clinician with the necessary
is it inducible but it also confers various levels of results for judicious decision-making in antibiotic
resistance, which can be very low, to vancomycin only. therapy utilizing available information and to draw his
The VanC type is also a low level of resistance to or her attention to the combinations of bacteriutn and
vancomycin alone and is species specific. An approach antibiotic for which there is a therapeutic risk. This
s34 C l i n i c a l M i c r o b i o l o g y a n d I n f e c t i o n , V o l u m e 2 S u p p l e m e n t 1 , D e c e m b e r 1996

should result not only in better therapy but also in 3. Courvalin P, Flandrois JP, Goldstein F, Philippon A, Quentin
lower selection of bacterial resistance and, hence, C, Sirot J. L'Antibiogramrne automatis& Paris: MPC/Vigot,
longer half-life of antimicrobial agents. 1988.
4. Leclercq R, Dutka-Malen S, Brisson-Noel A, et al. Resistance
References of enterococci to aminoglycosides and glycopeptides. Clin
1. Acar JE Goldstein F W Disk susceptibility test. In Lorian V, Infect Dls 1992; 15: 495-501.
ed. Antibiotics in laboratory medicine. Philadelphia: Williams 5 . Shimizu K, Kumada T, Hsieh WC, et al. Comparison of
&. Wilhns, 1991: 17-52. aminoglycoside resistance patterns in Japan, Formosa, and
2. Courvalin P, Goldstein F, Philippon A, Sirot J. L'Antibio- Korea, Chile, and the United States. Antimicrob Agents
gramme. Paris: MPC-Videom, 1985. Chemother 1985; 28: 282-8.