Chapter 3
Pharmacology of Drug Resistance
G.L. Drusano
1 Introduction
Resistance of pathogens to antimicrobial agents is a major
problem for all fields of anti-infective therapy. Certainly, we
are in a crisis of resistance with respect to antibacterial
agents (1). Therapy of human immunodeficiency virus (HIV)
is often restricted by multidrug resistance (2). Multidrug-
resistant TB is exploding in several areas of the world. Most
of the discussion below will focus on bacteria, but the ideas
are applicable to any pathogen.
In order to start to address the problem, it is important to
start with a definition of antimicrobial resistance. Resistance
to antimicrobial agents may be defined in a number of ways.
Here, the focus will be on two differing definitions. Each
definition has important differences in the outcomes that
result. The first definition of resistance, the one with which
most clinicians are familiar, is the idea that a specific patho-
gen, when infecting a patient, will have a low probability of
responding to a “normal” drug regimen. This idea is intrinsi-
cally bound up with the idea of breakpoints. Another defini-
tion of resistance is one often used by microbiologists who
perform large surveillance studies. Here, “resistance” is
defined by a strain’s acquisition of DNA allowing a changed
minimum inhibitory concentration (MIC) through drug destruc-
tion or other mechanisms (e.g., β-lactamase-containing plas-
mids), an altered target site, pump overexpression, or any
other mechanism that increases its MIC over that of a wild-
type isolate, even if it is still treatable by a standard therapeu-
tic dosing regimen. This definition has been used for
epidemiological purposes. However, it is also critical to the
idea of suppression of emergence of resistance through dose,
schedule, and duration choice. We shall examine each of
these definitions in turn.
G.L. Drusano (*)
Ordway Research Institute, Albany, NY, USA
gdrusano@ordwayresearch.org
D.L. Mayers (ed.), Antimicrobial Drug Resistance,
DOI 10.1007/978-1-59745-180-2_3, © Humana Press, a part of Springer Science+Business Media, LLC 2009
2 Resistance Defined by Breakpoint: Good
Clinical Response for “Sensitive”
Breakpoints dividing measured MICs into different catego-
ries (sensitive, intermediate, resistant, nonsusceptible) can
be determined in a logical manner. The first question that
must be asked and answered is what result is desired by
labeling a pathogen as being sensitive (or resistant) to a spe-
cific drug (and, by implication, drug dose and schedule)?
Normally, when an organism is labeled as “sensitive” to a
drug, we mean that a clinician treating a patient infected with
this pathogen will have an “acceptably high” probability of
the patient responding to therapy with a specific drug dose
and schedule.
In order to rationally choose such a breakpoint for deter-
mination of susceptibility/resistance, there are a number of
pieces of information required:
1. What goal of therapy is desired
2. What is the protein binding of the drug in man and, if the
goal of therapy is set pre-clinically, what is the protein
binding of the drug in the animal species used for target-
setting
3. What is the distribution of drug exposure in a population
of patients after being given a specific dose
4. What is the distribution of MIC values (or EC50 values, if
dealing with viruses) for the pathogen(s) for which the
drug is to be used
2.1 Goal of Therapy
Generally, exposure–response relationships in clinical patients
are not available a priori. Consequently, such relationships
must be developed in preclinical models, such as animal model
systems or in vitro systems, such as a hollow fiber infection
model. Such models allow linkage of differing amounts of drug
exposure to differing amounts of microbiological activity.
33
34 G.L. Drusano

Craig and colleagues have popularized the mouse-thigh and

mouse-lung infection models (3, 4), in which the pathogen of
interest is injected into the mouse thigh or inhaled into the
lungs, a time period elapses (usually 2 h) to allow the infection
to progress, and this is followed by treatments that differ by
dose and schedule. After a period of time (usually 24 h), the
mouse is sacrificed, the posterior thigh is dissected off or the
lungs removed, homogenized, and plated after serial dilution to
identify the bacterial burden at the time of sacrifice. Since dif-
ferent doses and schedules are administered, this allows link-
age of the true dynamically linked variable (peak/MIC ratio,
area under the concentration–time curve/MIC ratio (AUC/MIC
ratio) or time > MIC) which serves as the independent variable,
to the effect achieved, which serves as the dependent variable.
This can be done with microbiological effect, as in Fig. 1a
or with a dichotomous end point, such as survivorship, as
shown in Fig. 1b. Resistance suppression can also serve as a
dependent variable, but will be discussed separately.
In Fig. 1a, the dashed horizontal line is the organism con-
centration present at the time of therapy initiation. Having
the number of organisms at that concentration 24 h later is a
definition of stasis for a drug regimen (i.e., no net growth),
although it should be recognized that the number of organ-
Fig. 1 Demonstration of the pharmacodynamically linked variable for a
isms in between the 0 h and 24 h time points can take many microbiological endpoint (top panel), as demonstrated by Craig (after
trajectories to wind up at the same place. One can also see Reference 4). Demonstration of the pharmacodynamically linked variable
that it is straightforward to estimate 1, 2, or 3 log10 (CFU/g) for a survivorship endpoint. Three isogenic strains with 3 different MIC
decline from stasis and see at what level the measure of drug values were studied: parent strain, fluoroquinolone MIC = 1 mg/L;
Daughter Mutant #1, fluoroquinolone MIC = 4 mg/L; Daughter Mutant
exposure needs to be to mediate that particular degree of #2, fluoroquinolone MIC = 8 mg/L. Regimen/Strain pairs examined were:
effect. Finally, it is important to recognize that, in this ( ) Parent Strain: Rx: 80 mg/kg/day; (Ñ) 4X Daughter Mutant, Rx: 80 mg/
instance, where a cephalosporin-type β-lactam antibiotic is kg/day; (Δ) 8X Daughter Mutant, Rx: 80 mg/kg/day; (Ä) Parent strain, Rx
being tested, the time that drug concentrations exceed the 20 mg/kg/day, producing the same Peak/MIC ratio and AUC/MIC ratio as
the 4X Daughter Mutant treated with 80 mg/kg/day. After Ref. (6)
MIC (time > MIC) is the measure of drug exposure most
closely linked to microbiological effect.
In Fig. 1b, neutropenic rats are challenged intraperitoneally ratio as well as AUC/MIC ratio. Clearly, the survivorship curves
with 109 CFU of Pseudomonas aeruginosa (6). Previous work are nearly identical. So, even though the doses were different
had demonstrated that the AUC/MIC ratio or peak/MIC ratio and the MICs were different, the same outcome results when
was most closely linked to outcome (6, 7). Here, three isogenic the peak/MIC and AUC/MIC ratios are the same.
mutants were created that had three differing MIC values to From Fig. 1a, if one is contemplating the administration
the fluoroquinolone antibiotic lomefloxacin being evaluated of ceftazidime for pneumonia, it will require that drug con-
(1, 4, and 8 mg/L). Control cohorts are not shown for clarity. centrations remain above the MIC for about 30–35% of the
Three cohorts with differing MIC values are treated with dosing interval to achieve a static effect. For one, two and
80 mg/kg/day and survivorship was observed. As can be seen, three log kills of microorganisms, it is clear that about 40%,
the cohort infected with the MIC = 1 mg/L had a 65% survi- 50%, and 60–70% of the dosing interval needs to be covered
vorship, while the cohort infected with the isogenic isolate to achieve the desired microbiological endpoint. Consequently,
with an MIC of 4 mg/L had a 15% survivorship and the cohort the target will differ, depending on the drug class as well as
with the MIC of 8 mg/L had a survivorship of zero. Clearly, the ultimate effect one wishes to achieve.
MIC does make a difference. However, there was a fourth
cohort, where the infecting organism with an MIC of 1 mg/L
was treated with a dose of 20 mg/kg/day. Dose also matters, as 2.2 Protein Binding
the survivorship is 10% here versus 65% when the dose was
80 mg/kg/day. It is also important to recognize that the 4 mg/L While this topic is sometimes controversial, it should be stated
cohort treated with 80 mg/kg/day and the cohort infected with that, in the vast majority of instances, protein binding mat-
the strain having an MIC of 1 mg/L had the same peak/MIC ters, in that free drug is microbiologically active. There are