Disk Diffusion Versus Broth Dilution Techniques

Both methods of susceptibility testing require rapid growth of organisms and therefore may not be available for all organisms. Broth dilution data are particularly dependent on rapid growth, and for some organisms disk diffusion may be the only available means of obtaining data. The disk diffusion method (e.g., Kirby–Bauer) involves disks that contain a known amount of the drug of interest. The agar is streaked with a standardized inoculum of the isolated organism, and the disks are placed in standardized positions on the inoculated gel. Drug diffuses from the disk into the agar at a known rate (see Figure 6-4),⁴⁰ such that, at a standard time, the concentration in the agar correlates with the minimum inhibitory concentration (MIC) of the drug as would be determined by the broth dilution procedures (the most common method serving as a gold-standard to other methods). At the prescribed time (i.e., as specified by CLSI^35-37), a zone of no microbial growth (in mm) is measured around the disk. Because the concentration of the drug decreases with the distance (zone) diameter from the disk, the larger the zone, the lower the concentration of drug necessary to inhibit the growth of the organism and the more likely effective drug concentrations will be achieved at the site of the infection. A susceptible (“S”) designation is given if the zone is sufficiently large. Growth up to the designated zone indicates that the concentration of drug necessary to inhibit the organism is too high to achieve in the patient, leading to a resistant (“R”) designation. Intermediate (“I”) designation is provided for some drugs. Zone sizes necessary for an organism to be considered susceptible as opposed to resistant to a specific drug are variable and are very sensitive to disruptions in protocols, which underscores the importance of following standards. An advantage to the disk diffusion method is that multiple drugs might be simultaneously tested on one plate. This is in contrast to the more tedious and costly, yet more informative, broth dilution methods. Because disk diffusion results are reported as S, I, or R, it is described as semiquantitative.

In contrast to the disk diffusion method, the broth dilution method provides quantitative data regarding the amount of drug necessary to inhibit microbial growth (see Figure 6-5).⁴¹ For each drug of interest, tubes of liquid media are spiked with concentrations of the drug of interest, with the highest concentration generally being that just below the CLSI threshold of susceptibility (resistant MIC breakpoint). Subsequent test tubes containing serially diluted (by half) concentrations of the drug. As such, MICs are generally reported out as logarithmic fractions or multiples of 1 μg/mL (i.e., from lowest to highest 0.0312, 0.0615, 0.125, 0.25, 0.5, 1, 2, 4, 8, 16, 32, 64, 128, 256, 512; see Figure 6-5). Each drug to be tested must involve multiple test tubes or wells. The low and high range of concentrations tested for each drug will vary depending on concentrations achieved in tissues (including blood) when administered at recommended dosing regimens to the target species. For example, the ranges tested for ticarcillin would be expected to be much higher than the concentrations tested for enrofloxacin because the maximum concentration achieved in serum after administration of a recommended dose will be much higher for ticarcillin than for enrofloxacin (see Chapter 9). Occasionally, the MIC for some drugs deviates from the aforementioned tested concentrations; generally, these are drugs marketed as combinations (e.g., trimethoprim/sulfonamide combination). PD data generated for package inserts or scientific reports also may incorporate dilutions other than those delineated by CLSI. It is important to remember that CLSI guidelines are intended only to support clinical microbiological laboratories that provide direct support for patient care.

The tubes that contain broth (standardized type and amount) of the appropriate dilutions of the drug of interest must be inoculated with a standard number of the isolated bacterial organism during the logarithmic phase of growth. Microbial growth continues under standardized conditions for the standardized period (as set by CLSI^35-37). At the end of the incubation period, each tube is observed for evidence of growth. Evidence of growth is determined visually or using computer systems that allow miniaturized automation (see Figure 6-5 and Figure 6-6) The tube with the lowest concentration of drug that exhibits no detectable growth contains the MIC (in μg/mL), or the minimum amount of antimicrobial necessary to (in vitro) inhibit the growth of the organism cultured from the patient.^23,41 Because of the complexities of the procedures, laboratories that provide clinical C&S testing may find MIC results on the same isolate that vary, even if CLSI guidelines and interpretive standards are followed. Generally, variations within 1 broth dilution are not considered significant. Laboratories ensure that quality standards of testing are met by performing drug MIC determination for control isolates (i.e., obtained from American Testing Cell Culture: e.g, E. coli ATCC 25922, Staphylococcus intermedius ATCC 45222).⁴¹ Research publications that address bacterial PD likewise should demonstrate adherence to CLSI guidelines, including quality-control procedures.

Figure 6-6 A, A commercially available antibiogram card and an E-test (C-D) with interpretation. The commercially available antibiogram card is a miniaturized broth dilution procedure that generates minimum inhibitory concentration (MIC) using a microwell design. Generally, one card is made for gram-negative isolates and another for gram-positive isolates. The size of the card limits testing of the number of drugs and the range of concentrations, with concentrations approximating the susceptible and resistant breakpoints (indicated under each drug). Some cards test only the susceptible and resistant breakpoints. Growth is indicated by a color change (all wells had growth in row 1, indicating resistance to ampicillin, but no wells had growth in row 3, indicating susceptibility to enrofloxacin). The ranges tested (above the wells) and interpretations (to the right of each drug) are provided for four of the drugs tested on the card. None of the isolates tested intermediate. A limitation of the cards is ability to indicate how susceptible an isolate is. This limitation is largely overcome with the E-test system (C; strip is enrofloxacin). Each strip releases the drug into the medium at logarithmic rates. Growth in susceptible isolates follows a tear-shaped pattern, with the point of the tear indicating the MIC. Advantages of the E-test include a very broad range of test concentrations (over16,00-fold) indicated by outer bracket (see Table 6-3), exceeding both the susceptible and resistant breakpoints (indicated by inner bracket) by several magnitudes, thus allowing assessment of how susceptible or resistant the isolate might be to the drug of interest. The MIC of this isolate is 0.06 μg/mL; for comparison, the lowest concentration that would be tested on the antibiogram. The E-test suggests that this isolate is moderately susceptible to doxycycline. The differences in the MIC between the microwell dilution and the E-test may reflect subtle differences in methodology but also the lipophilic nature of doxycline (better penetrability), thus highlighting a caveat of susceptibility testing: model drugs do not always represent the drug of interest well. Another advantage of the E-test is the smaller increments of change, and thus greater precision provided compared with tube dilution procedures. A more precise dosing regimen can thus be designed. For example, with standard tube dilution, concentrations increased from 8 to 16, whereas with E-testing, concentrations increased from 8 to 12 to 16 μg/mL. Finally, the individual nature of the test strips allows a “pick and choose” approach to individualizing drug therapy. This also, however, is a disadvantage in that costs are higher when multiple drugs are tested.

Broth dilution reports provide both the MIC (a concentration, reported in μg/mL) and (assuming CLSI procedures are followed), the CLSI interpretation of S, I, or R for that MIC. The basis of the interpretation (S, I, or R) for broth dilution procedures is addressed later in this chapter. The MIC for selected drugs may be accompanied by a “≤” or “≥”. Using Figure 6-7 as an example, Pseudomonas have an MIC for amikacin of ≤4 μg/mL. The ≤ indicates the absence of growth in the lowest concentration of amikacin tested by this laboratory (8 μg/mL); this lowest concentration may be different among laboratories that use different systems. However, often the lowest dilution tested is at or just below the lower threshold of susceptibility (the susceptible MIC breakpoint; see later definition) set by CLSI. Testing at concentrations at or very close to the susceptible breakpoint of a drug is a major disadvantage of current susceptibility testing methods: either isolate may be very susceptible to amikacin such that their actual MIC may be several tube dilutions below the lowest concentration tested (see below). As such, the closest approximation to the actual MIC for either isolate will be the concentration below the lowest dilution tested by the laboratory (i.e., ≤4 μg/mL or <8 μg/ml both indicate the same result). The isolate will be accompanied by an “S” designation, indicating susceptibility because the MIC is at or below the susceptible breakpoint determined by CLSI (Table 6-2). At the other end of the testing range, an MIC accompanied by ≥ indicates that growth was present in the highest concentration tested by the laboratory. Generally, for most automated procedures, the highest concentration tested is 1 tube dilution below the upper threshold of susceptibility (the resistant MIC breakpoint) set by CLSI for each drug (Table 6-2). For example, for cephalothin (the model drug for cephalexin in this example), the upper threshold of susceptibility (the resistant MIC breakpoint) set by CLSI is 8 μg/mL. Thus for Pseudomonas, growth was present in the well containing 4 μg/mL, indicating that the actual MIC is equal to or higher than 8 μg/mL (≥16 μg/mL or >8 μg/mL both indicate the same result). However, again the testing range limitations of the current procedures emerges in that level of resistance of the isolate cannot be assessed. The isolate may be characterized by low-level resistance, although this is unlikely for P. aeruginosa and first-generation cephalosporins (indeed, testing of P. aerugniosa toward cephalothin is not appropriate). However, unless the range tested extends beyond the resistant breakpoint, all that is known is that the isolate is resistant, and the MIC will be accompanied by an R designation.

Figure 6-7 An example of a C&S report for broth dilution. The breakpoints have been added in parantheses this report; for ampicillin and penicillin, a second breakpoint in [brackets] is for the gram-negative organisms (see text and Table 6.2). The relative in vitro efficacy of antibiotics to which an organism is susceptible can be evaluated by comparing the minimum inhibitory concentration (MIC) of the organism. For Staphylococcus, the resistant MIC breakpoint to MIC ratio is 64/4, whereas that for gentamicin is 16/4 or two tube dilutions from the breakpoint. Although the isolate is considered susceptible to both drugs, amikacin presumably would be more effective (although neither generally should be used alone to treat Staphylococcus). Differences greater than one tube dilution should be considered significant. For the beta-lactams, effective treatment of Staphylococcus with cephalothin (the model drug for cephalexin) may be more easily achieved (ratio of 8/2) compared to amoxicillin clavulanic acid (ratio of 0.25/0.25). However, as time-dependent drugs, elimination half-life of both drugs would need to be considered. All values (concentrations) are in μg per mL. S = susceptible; R = resistant; I = intermediate.

Table 6-2 Interpretive Standards for Disk Diffusion Equivalent Minimum Inhibitory Concentration Breakpoints for Selected Antimicrobials

MIC, Minimum inhibitory concentration.

∗ Old breakpoints replaced by Clinical Laboratory Standards Institute (CLSI) for amoxicillin–clavulanic acid for all organisms were, for Staphylococcus ≤4/2 = S, and ≥ 8/4 = R and for non-staphylococci, ≤8/2,=S and ≥ 32/16 = R. The provision of a separate breakpoint of ≤8 μg/mL for UTI is new.

1 Clinical Laboratory Standards replaced by CLSI for cephalexin were ≤ 8 = S, and ≥ 32 = R. The new breakpoints were becoming official at the time of publication. Institute Interpretive standards that are based on animal pathogens are designated by an asterisk.

2 When testing Staphyloccocus organisms

3 When testing gram-negative enteric organisms

4 Ampicillin is used to test amoxicillin

5 When testing Pseudomonas organisms

6 Oxacillin is used to treat methicillin, cloxacillin

7 Cephalothin is used to test all first-generation cephalosporins. Does not represent cefazolin, which should be tested separately if a gram-negative organism.

8 Clindamycin is used to test lincomycin, which is less susceptible to Staphylococcus.

9 When testing Streptococcus (S. pneumoniae for levofloxacin)

10 When testing pathogens associated with food animal respiratory disease

11 Trimethoprim–sulfamethoxazole is used to test trimethoprim–sulfadiazine and ormetoprim–sulfadimethoxine ¹²For urinary tract infections

13 For soft tissue infections

14 Used to test chlortetracycline, oxytetracycline, minocycline, doxycycline

15 When testing Enterococcus organisms

16 A human criteria deemed relevant to dogs and cats. Note reduced oral bioavailability (mean of 40%) in dogs and negligible (0%-20%) in cats.

Among the pitfalls of C&S testing are the stepwise dilutions and the range of concentrations tested for each drug. The twofold dilutions at which MIC are tested affect the design of dosing regimens at the higher MIC. Precision in the design of dose would be facilitated if MICs could be determined between the tube dilutions. For example, the dose to target 64 μg/mL would be substantially cheaper and potentially safer than that necessary to target 128 μg/mL. The limited range of concentrations tested for each drug negatively affects the ability to identify the drug to which the isolate is most susceptible (see Figure 6-6).⁴¹ Ideally, concentrations tested by broth dilution procedures should span the range of drug concentrations that characterize the range of MICs established in a sample population of isolates of the organisms, with the highest concentration being at least one dilution above the highest drug concentration acheived in target biological fluids.⁴¹ However, automated systems test in a very narrow range. As previously noted, because the lowest concentrations are at or just below the lower limit of susceptibility, isolates that are very susceptible to the drug of interest cannot be indentified (see Table 6-2). Therefore standard antibiograms are more indicative of resistance rather than susceptibility.

A third testing system approved by the Food and Drug Administration (FDA) offers advantages to the standard commercial broth dilution card. The “E test” (Epsilon test) combines the simplicity of disk diffusion with the informative nature of broth dilution, but goes beyond standard broth dilution procedures. (see Figure 6-6). In general, MICs generated by the E-test correlate well with MICs generated from broth dilution procedures.⁴² A disadvantage of the E-test is that the length of the test strip limits the number of drugs that can be tested on a large plate (three strips for a large plate, one for a small plate), which contributes to the cost of the testing. However, because the drugs can be chosen for each patient, the method lends itself well to expanded susceptibility testing in the presence of a multidrug-resistant isolate. Although E tests are tedious and expensive, the wider range of concentrations tested (up to 1600-fold differences; Table 6-3) includes MICs well below the lower and higher thresholds of susceptibility, thus allowing identification of very susceptible isolates. Further, isolates with low-level resistance might be identified, potentially justifying the use of the drug, albeit at a higher dose or in combination with another drug.

Table 6-3 Examples of Minimum Inhibitory Concentration Ranges Covered by the E-test

Because of the inherent risks of inaccuracy associated with any C&S testing procedure, results yielded from procedures that are not based on CLSI standards should be interpreted with caution. Aspects subject to variability include pH; cation content and osmolality of the media; inoculum size; media volume; temperature and duration of incubation; humidity; and, for broth dilution, the method of observing growth.⁴¹ Accordingly, in practice, culture methods should be considered less than ideal unless CLSI protocols are followed. Further, preliminary data, quick “snap” tests, or other methods intended to generate rapid results must be interpreted cautiously; the role of the organisms in causing infection and the susceptibility of the organisms (unless identified as a multidrug-resistant microbe) may require full C&S testing. Whereas the FDA is responsible for approval of diagnostic tests for human medicine, such a pathway is not required for veterinary diagnostic tests.

Agar Gel Versus Broth Dilution Pharmacodynamics

A nonquantitative but helpful summary of PD data is an antibiogram that indicates the proportion of isolates that are susceptible (or resistant) to the drug of interest (Figure 6-8). Although it does not provide information regarding the level of susceptibility, it can provide direction regarding empirical drug selection by indicating the likelihood that an organism is susceptible to the drug of choice. The antibiogram might be generated for each practice on the basis of cumulative data summarized on an annual basis.

Figure 6-8 A cumulative antibiogram generated for the target species can be helpful in identifying drugs to which acquired resistance has emerged. The data will be specific to the facility (i.e., hospital). The number in each cell refers to the number of tested isolates designated as susceptible to the drug. When present, the number in parentheses in each cell refers to the number of isolates tested for that drug; otherwise, the number of isolates tested is indicated in the second left-hand column. Note that the data indicate that one species in a genera may not be well represented by another species in the same genera, particularly for Enterococcus and Staphylococcus genera.

Population statistics generated from MICs can provide even more useful information. They are particularly helpful if MIC data is not available for an isolate infecting a patient. Population MIC statistics can be generated from a sample population of the same organism; ideally, at least 100 isolates will be collected from different patients. Pertinent PD (MIC) statistics that describe the population distribution include the range (lowest and highest MICs recorded for any isolate representing the organism), mode (the most frequently reported MIC), median (the middle MIC, the 50^th percentile or MIC₅₀), and the MIC₉₀ (or the 90^th percentile MIC; the MIC at which 90% of the organisms are inhibited (see Figures 6-9 to 6-11). The two-fold dilution nature of MIC determination mandates that the geometric mean (converted to account for the noncontinous nature of MIC) be reported rather than arithmetic mean. If an MIC is not available for an organism infecting a patient, the MIC₉₀ (or even more ideally, the MIC₁₀₀) of a drug for an organism is the preferred surrogate indicator of “what is needed” by the author. For example, if S. pseudintermedius is a known or suspected cause of pyoderma in a patient and the drug to be chosen empirically is cephalexin, the MIC₉₀ of S. intermedius for cephelexin⁴³ can be used as an indicator of “what is needed”—that is, the PD target for therapy in the patient. PD information can be found on many package inserts scientific literature^43,44 or textbooks (veterinary for animal drugs, human if not), and other resources (see Table 6-4 and Chapter 7). However, the dynamic nature of microbes in response to the presence of antimicrobials may render some population data obsolete even within several years of collection. In addition to the species, a number of host factors are likely to affect the sample population statistics and its applicability to the patient. Among the more important factors is previous exposure to antimicrobials, which is likely to be associated with higher MICs compared with MICs of isolates collected from antimicrobial-naïve animals (i.e., not pathogens). Ideally, separate statistics might be promulgated for isolates collected from animals not previously exposed to antimicrobials.

Figure 6-9 Population pharmacodynamic data. Each sample collected from a different animal (same species) yields an isolate of the organism of interest. Ideally, at least 100 representative isolates will be tested. The minimum inhibitory concentration (MIC) of each isolate is determined, and all are plotted in a distribution curve. The range represents the lowest and highest MIC determined for the isolate; the mode would be the most frequent MIC reported and the median represents the middle MIC or the 50^th percentile (which, for normally distributed data, also represents the MIC) The MIC₉₀ is the 90^th percentile MIC. A representative package insert demonstrating the presentation of population pharmacodynamic data can be found in Figure 6-11.

Figure 6-10 Population distributions of canine and feline Escherichia coli pathogens’ minimum inhibitory concentration (MIC) based on E-testing for two time-dependent drugs (amoxicillin–clavulanic acid, and cefpodoxime, upper plots) and two concentration-dependent drugs (enrofloxacin, and gentamicin, lower plots). The susceptible (left) and resistant (right) breakpoints, as identifed by Committee on Laboratory Standards Institute, are indicated in brackets and lines except for amoxicillin. CLSI’s new breakpoints are indicated by arrows (≤8 μg/mL for the susceptible breakpoint for isolates collected from the urinary the resistant breakpoint for other isolates; indicated by arrows below x axis). The distribution is bimodal for all drugs except gentamicin, as is indicated by a second distribution of isolates with an MIC well above the resistant breakpoint. This second population of isolates will cause the MIC₉₀ (dashed arrow) to exceed the breakpoint. The range is represented by the lowest and highest MIC recorded (either may be limited by the range tested), the median is the 50^th percentile (or MIC₅₀) (solid arrow) the MIC₉₀ is the 90^th percentile, and the mode is the most common MIC reported for that isolate and drug. Because an E-test was used, the MIC tested are not limited to two-fold dilutions. Because these isolates are pathogens that have been cultured from dogs or cats with spontaneous disease, they may represent isolates already exposed to antimicrobials which may explain the bimodal distribution (i.e., these isolates may have undergone stepwise mutations). The population distribution of drug-naïve only isolates is likely to be somewhat lower.

Figure 6-11 Package insert information two fluoroquinolones, FQA (top) and FQB (bottom). Comparison of MIC₉₀ among isolates for FQA suggests that Pasteurella sp. should be more easily treated compared with Escherichia coli for both drugs. Integration of pharmacokinetic data (C_max for these concentration-dependent drugs) and pharmacodynamic data (MIC₉₀) can be used to identify which drug is best used to treat each microbe and which dose might be used to treat the microbe. For example, for FQA at the low dose of 2.5 mg/kg, when treating E. coli (and no patient-specific MIC is available), the C_max is 2 μg/mL and the MIC₉₀ is 0.06 μg/mL, resulting in a C_max/MIC₉₀ ratio of 25. For Proteus, the ratio is 2/0.125, or 16. For concentration-dependent drugs, the target ratio is ≥10, suggesting the low dose may be effective for both, but the large dose might be considered for Proteus if the patient is considered at risk. The number of isolates of P. aeruginosa is not sufficient to represent the population. If the process is repeated for FQB with a C_max of 1.8 μg/mL at the low dose, the MIC₉₀ for E. coli is 0.11 μg/mL, resulting in a ratio of 16. For Proteus, the MIC₉₀ is 1.8 μg/mL, resulting in a ratio of only 1. Although the dose might be sufficient for E. coli, the target ratio could not be reached even at the higher dose for Proteus. Note that the number of organisms on which the data are based for each organism often does not reach the ideal target of 100. The smaller the sample size, the more caution is indicated when extrapolating this data to the general population. (From Pfizer, Package Insert and Fort Dodge, Package Insert)

Comparing PD data of a drug reveals differential susceptibility among organisms toward each drug. For example, using the antimicrobial package insert, comparison of MIC₉₀ among different organisms reveals that P. aeruginosa tends to be susceptible (if at all) only at high concentrations compared with the more susceptible Pasteurella multocida (see Figure 6-11). The MIC₉₀ of P. aeruginosa more often than not approaches or surpasses the upper threshold of susceptibility for most drugs. Thus achieving effective antimicrobial concentrations is more likely to be difficult in the patient infected with P. aeruginosa compared with one infected with Pasteurella. A review of the antimicrobial package insert reveals other differential susceptibilities.

The distribution of the MICs of organisms for drugs can help identify emerging resistance. For example, the distribution of E. coli for several drugs (see Figure 6-10) is bimodal, representing two different populations. The majority of isolates in the first population are characterized by an MIC well below the susceptible threshold of susceptibility (i.e., susceptible MIC breakpoint). This data demonstrates that even isolates considered susceptible are characterized by MIC that are close to the susceptible breakpoint. Further, a substantial portion of the population is higher than the upper threshold of susceptibility—that is, the MIC₉₀ exceeds the resistant MIC breakpoint. It is very possible that the second population, characterized by higher MICs, probably represents isolates previously exposed to antimicrobials; as such, culture would be prudent for those animals previously exposed to antimicrobials. Finally, detecting increasing MICs determined from sequential cultures of the same organisms in a patient with recurrent infections might indicate emerging resistance, likewise, comparison of the MIC₉₀ of a sample population of an organism across time can reveal emerging resistance.

The Minimum Inhibitory Concentration: Determining Susceptibility Versus Resistance

Susceptibility data based on broth dilution procedures that are reported for a patient will include the MIC, as well as a susceptible, intermediate, or resistant (SIR) interpretation.

The clinical microbiology laboratory provides the interpretation on the basis of CLSI interpretive criteria. The criteria for broth dilution procedures are presented as thresholds or breakpoint MICs (MIC_BP) whose values will also be in terms of the concentrations tested for each drug (i.e., multiples or fractions of 1 μg/mL) (see Table 6-2). Two breakpoints are provided for each drug. An isolate inhibited at a concentration at or below the lower threshold or susceptible MIC breakpoint will be designated “S,” whereas an isolate that is able to grow after in vitro exposure to a drug concentration that equals the upper threshold or the resistant MIC breakpoint will be designated “R.” The susceptible breakpoint is at least one broth dilution below the resistant breakpoint for all drugs; for some drugs the susceptible breakpoint is 2 or more broth dilutions below the resistant breakpoint, allowing for an intermediate, or “I,” designation (see Figure 6-6). For example, for enrofloxacin, the susceptible and resistant MIC_BP are ≤0.5 and ≥4 μg/mL, respectively. Thus an isolate whose growth (under in vitro conditions specified by CLSI) is inhibited with as little as 0.5 μg/mL or less will be designated as “S.” On the other hand, if growth is present in the well that contains 2 μg/mL, then 4 μg/mL (the next broth dilution) or more will be necessary to inhibit the growth of the isolate, and the isolate will be designated as “R,” or resistant to enrofloxacin. An additional broth dilution occurs between 0.5 and 2 μg/mL. Isolates that are inhibited by enrofloxacin at 1 μg/mL will be designated as intermediate, or “I.” An isolate with an “I” designation has developed some level of resistance, and such isolates should be treated with that drug only cautiously, at higher doses, or in combination with a complementary antimicrobial drug. The more prudent approach would be to consider “I” isolates as “R” for that drug. Use might also be considered in circumstances in which the drug accumulates in active (i.e., unbound) form at the site of infection such that concentrations exceed that achieved in plasma. Examples might include urine (produced by the normally functioning kidney) or accumulation in phagocytic white blood cells (selected drugs). Note, however, that such concentrations may yet be insufficient.

CLSI determines the thresholds of susceptibilities—that is, the lower (susceptible) and upper (resistant) breakpoint MIC_BP for each drug after exhaustive evaluation of both PK data in the target species and PD data for the drug of interest toward the microbes of interest. For animal drugs approved in the last several decades (only since then have MIC become standard testing procedures), it is likely that some of the data were collected by the drug manufacturer during the approval process. However, both PK and PD data may be drawn from peer-reviewed literature or other sources, particularly for drugs not approved for use in the target species, or for drugs whose susceptibility thresholds are being re-evaluated by CLSI.

Three criteria must be met for CLSI to establish an MIC_BP for each drug. The primary and initial consideration is the population distributions, with a focus on both the statistics as well as the type (i.e., modal or bimodal; see Figure 6-10). Obvious patterns of low versus high MICs can be used to identify susceptible “cutoffs” or breakpoints. Statistics will be compared among different strains of the same species being tested. Note that some organisms may be much more susceptible to the drug of interest—that is, they have MIC₅₀ and MIC₉₀ that are much lower compared with other organisms (e.g., pasteurella and pseudomonas). However, CLSI generally provides only one set of criteria for all susceptible isolates.

The second consideration upon which criteria are based is, the clinical pharmacology of the drug, ideally in the target species. Among the more important PK parameters evaluated by CLSI are the peak and trough plasma drug concentrations (C_max and C_min), area under the curve (AUC) for a 24-hour dosing period, and the drug elimination half-life (see later discussion of PD indices) (Figure 6-12).^41,45-51 Volume of distribution; protein binding; and, when available, tissue (including urine) concentrations are also considered. Presumably, the MIC of an isolate considered susceptible should be below the peak plasma or tissue drug concentrations or C_max of a given drug when administered at a recommended dose. Indeed, selected resistant breakpoints correlate with C_max, as is demonstrated for amikacin when administered to dogs at 22 mg/kg. The C_max of 65 μg/mL (see Chapter 7) is similar to the resistant breakpoint for amikacin. However, for some breakpoints, the correlation does not exist, as is exemplified by amoxicillin/clavulanic. The C_max of the labeled dose of 13.5 mg/kg, administered orally, will generate a C_max of approximately 4 to 6 μg/mL of amoxicillin in dogs, yet the resistant breakpoint for nonstaphylococcal organisms has been ≥32 μg/mL, well beyond the concentrations that can be achieved in plasma at any reasonable dose. (In response to this disparity, CLSI has recently re-examined and readjusted the breakpoint for amoxicillin-clavulanic acid as is discussed below; see Table 6-2).Another limitation of setting breakpoints based on peak plasma drug concentrations is their lack of precision: breakpoints are limited to concentrations used for susceptibility testing and thus will be reported using twofold dilutions. As such, a resistant breakpoint concentration may be considerably higher or lower than the actual concentration achieved at the recommended dose. As such, the actual MIC reported for an infecting isolate should be compared to C_max reported in a sample population of the target species is an ideal default when selecting drugs (and the dosing regimen; see later discussion). The original veterinary fluoroquinolones were approved with “flexible labels” (multiple doses); for those drugs, the susceptible and resistant breakpoints reflect the C_max resulting from the lowest and highest labeled doses of the drug.

Figure 6-12 Pharmacodynamic indices (PDI) resulting from integration of pharmacokinetic (PK) data (from sample population of target or closely related surrogate species), represented here by C_max (μg/mL; target of 10 to 12) or area under the curve (AUC) (μg^∗hr/mL; target of 100-125), and microbial or pharmacodynamic (PD) data, represented here by minimum inhibitory concentration (MIC). Note that the PDI are based on (Mouton 2005 ⁴⁷ and Amsterdam 2005⁴¹). Note that activity should be based on free (unbound) drug. The pertinent PK parameters for this hypothetical drug and infecting microbe would be as follows: C_max of 75 μg/mL, half-life of approximately 2 hours and AUC of 159 μg^∗hr/mL. The PD parameter, or MIC of the infecting organism, is 2 μg/mL. The PDI for this drug and microbe combination would be as follows: C_max/MIC = 37.5 (surpasses target for a concentration-dependent drug); AUC/MIC = 80 (insufficient for a gram-negative organism if a fluorinated quinolone but potentially sufficient for a gram-positive organism); and T > MIC = 8 hours, which would allow a 12, 16, or 24- (32 hr is not a reasonable interval) hour dosing interval for a time-dependent drug if the target T > MIC were 75%, 50%, or 25%, respectively. The AUIC (area under the inhibitory curve) is the integrated area of the curve above the MIC. Although somewhat similar to AUC/MIC, its use among investigators has caused confusion, leading experts in the field to focus on AUC/MIC as the area-based PDI of choice.

The third criteria that must be met as CLSI determines thresholds of susceptibility is one of clinical relevance. The MIC_BP must be clinically relevant—that is, the microorganisms defined as susceptible should respond clinically to the drug, and in vitro data must correlate adequately with in vivo findings.^46,51 The MIC_BP values established by CLSI for each drug are generally not included in the susceptibility reports. However, having this information would be helpful because the MIC of the infecting organism might then be compared with the MIC_BP of the drug, allowing an assessment of “level” of susceptibility of the isolate to the drug of interest (discussed in more detail later). Breakpoints set by CLSI are based on isolates collected from across the country and accordingly are relevant for any clinical microbiology laboratory that uses CLSI protocols. As such, CLSI breakpoints generally can be obtained by any microbiology laboratory that uses CLSI criteria for its testing procedures (the preferred choice). Interpretive standards generally also are delineated on package inserts (including sources such as Physicians’ Desk Reference for human-marketed drugs or similar veterinary compendiums).

In order to simplify the use of interpretive criteria, when possible, CLSI breakpoints are generally inclusive to organisms whose spectrum is included in the drug, regardless of the site of infection. However, notable exceptions occur for both organisms and tissues. Organism exceptions generally are those either very susceptible to the drug or for organisms that easily develop in vivo resistance. For example, lower or more stringent penicillin breakpoints have been promulgated for Staphylococcus spp. because their beta-lactamases are particularly destructive toward penicillins compared to cephalosporins. Because destructive activity decreases the amount of drug at the site, a second set of lower breakpoints are provided. A higher breakpoint has been established for other susceptible isolates (e.g., gram-negative organisms; see Table 6-2). In contrast to Staphylococcus, P. aeruginosa is particularly susceptible to ticarcillin; accordingly, while another coliform is considered resistant if it is inhibited at 16 μg/mL P. aeruginosa is still considered susceptible even if, in vitro, it grows in the presence of 64 μg/mL (see Table 6-2). Another consideration regarding inclusivity is the generation of tissue specific breakpoints. Interpretive criteria are based on plasma drug concentrations. However, renally excreted drugs achieve much higher concentrations in urine compared to plasma. New interpretive criteria for ampicillin and amoxicillin–clavulanic acid includes a separate breakpoint that is tissue dependent: a higher breakpoint (≤8 μg/mL) has been set for E. coli-associated urinary tract infections, compared to much lower breakpoints for other tissues. However, caution is recommended when selecting a drug for treatment of a UTI when the “S” designation for a urinary isolate is based on breakpoints that differ from plasma (see Chapter 8) as it assumes, among other considerations, that infection occurs only in the bladder, that urine is concentrated by the patient, and the duration of exposure fulfills the needs of a time-dependent drug (see later discussion). Just as resistance can be detected in an infecting organism collected from a patient, across time, statistics may also indicate resistance (i.e., the MIC and MIC statistics [mode, median, MIC₅₀, and MIC₉₀]). Because CLSI reviews new data intermittently, their criteria for interpretive standards should result in new MIC_BP, and laboratories will implement these changes. The sequelae of increasing MICs in certain populations should result in an increasing number of isolates designated as “R” for drugs to which the organisms traditionally have been considered susceptible. An example is amoxicillin–clavulanic acid, whose susceptible breakpoint was recently decreased by CLSI (see Table 6-2).

Integration of Pharmacokinetics and Pharmacodynamics: How Much Is Needed Versus How Much is Achieved?

The information provided on the C&S report can be used effectively beyond the simple identification of “S” drugs. The MIC is an indicator of what is needed to target the organism and thus provides the PD information for the infecting isolate. Often, an organism will be designated as susceptible to several drugs. One advantage to the broth dilution method compared with the disk diffusion method of C&S is the ability of the

former to rank the drugs according to relative efficacy based on MIC. However, the relative efficacy of antimicrobials designated as “S” against a specific pathogen should not be determined by directly comparing MICs among different drugs, even drugs in the same class. The MIC varies among the drugs for a number of reasons beyond susceptibility. These include, but are not limited to, differences in molecular weight (one drug is simply heavier than another), the ability of the drug to penetrate the organism, the number of molecules necessary to “neutralize” the target, and differences in the mechanisms of action. Further, each drug achieves different concentrations in the patient as a result of differences in disposition. Thus antimicrobials differ in potency and MIC.⁴⁷ Rather than direct comparison to an isolate MIC for one drug to an MIC for another drug, one should also consider the concentration of drug that will be achieved in the patient when the drug is administered at the recommended dose (i.e., what is achieved; C_max). A less than ideal method of standardizing MIC is to compare how far the MIC of the drug of interest is from the resistant breakpoint (i.e., ratio of resistant MIC breakpoint of the drug to MIC of the organism). This is less ideal, as was previously discussed, because the resistant breakpoint does not always equate with the C_max.

The limitations in using MIC breakpoint as an indicator of what will be achieved in the patient reflect the serial concentration used for susceptibility testing. If a drug achieves a C_max of 24 μg/mL at the recommended dose, this concentration falls between 16 and 32 μg/mL serial dilutions. A susceptible breakpoint of 16 μg/mL might underestimate while a resistant breakpoint of 32 μg/mL would over overestimate what will be achieved at the recommended dose. Thus, a more relevant choice among the susceptible drugs might be based on comparing what is needed to what is achieved—in plasma (e.g., the C_max) when the drug is administered at the recommended dose. The ratio of C_max to MIC, however, requires that PK information be available for the drug of interest, ideally in the species of interest. This information often is available on package inserts for animal-approved drugs but must be collected from the literature for other drugs (see Chapter 7). The United States Pharmacopiea antimicrobial monographs published by the Journal of Veterinary Pharmacology and Therapeutics (2007) offers a compilation of PK information for a variety of antimicrobials and animal species. The integration of PD and PK is the first step in selecting a drug. However, this step is preliminary in that it does not take into account other factors that affect tissue concentrations, including host and microbial factors.

The relationship between the MIC of the infecting organism (what is needed) and the C_max also should be used as a basis for designing a dosing regimen.^48-50 The closer the MIC (or MIC₉₀) to the C_max (or the MIC_BP), the higher the dose and for time-dependent drugs, the shorter the interval that is indicated. For a more specific design of a dose, the MIC can serve as the targeted drug concentration (see Enhancing Antimicrobial Efficacy). If MIC data is not available from the isolate collected in the patient, population PD [MIC₉₀] data might be used as a surrogate PD indicator of how much is needed; (see Figure 6-11).

Caveats to Culture and Susceptibility Interpretation

Despite the usefulness of C&S testing, the information nonetheless reflects in vitro testing that must be applied to in vivo situations.^40,41 Results can be misleading or misinterpreted, despite ideal sampling and C&S techniques. Some of the pitfalls reflect the limitations presented by practicalities in testing (e.g., economics, technology), whereas others reflect limitations of applying in vitro data to an in vivo system. Examples include the following:

Bactericidal Versus Bacteriostatic Antimicrobials

The MIC is a drug concentration that inhibits but does not necessarily kill the target microbe. The MIC is a reasonable clinical outcome target because the success of antimicrobial therapy usually depends on host defenses that sequester and ultimately kill the microbial population after its inhibition by the drug. Antibacterials are frequently classified according to their ability to kill (bactericidal) rather than inhibit (bacteriostatic) microbial growth. Whereas bacteriostatic activity is indicated by the MIC, bactericidal activity of a drug is indicated by its minimum bactericidal concentration (MBC). However, this classification is based on in vitro methods. The MBC can be determined in several ways. For example, those tests tubes in which no visible sign of growth was observed following the broth dilution procedures can be reinoculated on nutrient-rich agar plates (see Figure 6-5). Those test tubes that yield no growth contained concentrations that killed, rather than inhibited, the microbe. Thus the test tube with the lowest concentration that yielded no growth contained the MBC of the drug. All antimicrobial drugs are characterized by an MBC; however, those drugs whose MBC approximates the MIC (e.g., within one broth dilution) might also be considered bactericidal. The MBC is most appropriately determined based on killing curves, which measure the number of surviving bacteria after exposure to fixed concentrations of drug; the concentrations are based on those achieved in serum at defined time intervals.⁴¹ For organisms noted as “S” to bacteridical drugs, achieving sufficient drug at the site to kill, rather than simply inhibit, the infecting pathogen is possible. For bacteriostatic drugs achieving the concentration necessary to kill the organisms without causing harm to the patient is much more difficult.^23,41 Exceptions might occur for drugs that are accumulated (in an unbound state) at the site of infection (e.g., urine or phagocytic white blood cells); in selected instances bactericidal concentrations of a bacteriostatic drug can be achieved.

Categorization of static versus cidal activity of a drug can be associated with its mechanism of action (Table 6-5). In general, drugs that target cell walls (beta-lactams, glycopeptides), cell membranes (polymixin B, colistin), or DNA (fluorinated quinolones) tend to act bactericidal in vitro. Ribosomal inhibitors that target more than one subunit (i.e., 30s and 50s; or 70s) also tend to be bactericidal. In contrast, drugs that target a single ribosomal subunit (tetracyclines, macrolides, lincosamides) or metabolic pathway (sulfonamides) tend to act bacteriostatic. Combinations of two bacteristatic antimicrobials that act in an additive or synergistic fashion may also result in bactericidal effects (e.g., a sulfonamide combined with a potentiating dipyrimidine). However, the distinction between bactericidal and bacteriostatic effects of a drug depends on the concentration; a bactericidal drug can be rendered nonbactericidal if concentrations sufficient to kill the organism are not reached at the site of infection the site of infection, or under conditions that slow the growth of the target organism (e.g., hypoxic environment or if used in combination with drugs that antagonize bactericidal actions). In such instances, the bactericidal drug will act in a bacteriostatic fashion. On the other hand, drugs classified as bacteriostatic may act bactericidal if high enough concentrations can be achieved, as might occur if the drug is accumulated in an unbound form (e.g., WBC or other tissues).

Table 6-5 Bactericidal Versus Bacteriostatic Drugs

Because host defenses must be effective to kill those organisms whose growth is merely inhibited, achieving bactericidal concentrations of an antimicrobial drug are paramount to therapeutic success in immunocompromised hosts (e.g., viral infections, granulopoietic patients, use of immunohibiting drugs) or immunocompromised sites (septicemia, meningitis, valvular endocarditis, and osteomyelitis).^23,26

Integration of Pharmacokinetics and Pharmacodyamics: Pharmacodynamic Indices

Although the MIC of a (presumed) infecting organism offers a target concentration for antimicrobial therapy, simply achieving the MIC of the organism in plasma may not be sufficient to ensure efficacy. Among the relationships that affect efficacy is the PK/PD relationship—that is, the dynamic relationship between the drug concentration to which the organism is exposed throughout the dosing interval (PK) and the response of the infecting organism to the drug, as estimated, for example, by the MIC (PD).^55-57 This relationship is affected by many host and microbial factors. Definitions of terms used to describe the integration of PK and PD (PD indices or PDI) are varied, depending on the author. For the purposes of this text, definitions will be drawn from Mouton.⁴⁷ It is important to note that many of the terms are based on parameters determined through in vitro testing. Therefore host and microbial factors still need to be considered. Further, most PDI are based on a 24-hr-dosing interval, thus modifications in dosing regimens should be based on a 24-hr period (Figure 6-12). The relevance of PDI to drugs with half-lives longer than 24 hr (e.g., azithromycin, cefovecin) is not clear.

Postantibiotic Exposure

Antimicrobials may continue to exert an effect even though the drug is no longer present at concentrations that exceed the MIC. The term postantibiotic exposure has been promoted to refer to the combined definitions that have emerged experimentally. Among the terms is the postantibiotic effect (PAE), which has both an in vitro and an in vivo definition.⁴⁷ The PAE is exhibited by drugs, and is defined in vitro as the period of suppression of bacterial growth after a short exposure of the organism to the antimicrobial.⁴⁷ The PAE for a drug, is determined in vitro by exposing a standard inoculum to it, removing the drug and determining the time that elapses (in hours) before the culture CFUs increase by tenfold. In vivo, the PAE is the time it takes for the number of CFUs to increase tenfold in treated animals after concentrations drop below the MIC at the tissue site.⁴⁷ Clinically, the PAE indicates the ability of a drug to inhibit bacterial growth after the drug is no longer present or is below the MIC of the infecting microbe.^49,58-60 As such, it also takes into account an effect a drug might have at subinhibitory concentrations. The impact of the PAE on antimicrobial efficacy can be profound, particularly for concentration-dependent drugs. It is the PAE that allows some drugs to be administered at long intervals despite short half-lives.^{41,50,52,59,61,62} The PAE may be absent for some organisms or some patients (e.g., some immunocompromised patients).⁴⁹ The duration of PAEs varies with each drug and each organism and the relationship between PDC and MIC (Table 6-6).⁶³ In general, concentration-dependent drugs appear to exhibit longer PAEs, with the duration of the PAE being proportional to the magnitude of the peak PDC (i.e., longer with higher PDC).⁶⁴ However, for each drug, and within drug classes, the PAE is markedly variable, depending on the organism.⁶⁵ Whereas beta-lactams exhibit a substantial PAE toward selected streptococci (i.e., thus making treatment less time dependent for streptococci), their PAE toward gram-negative organisms is minimal.⁶⁶ Applying information regarding the PAE to clinical patients is complicated by variable results (reflecting marked variability in methods) among investigators. The PAE is enhanced by combination antimicrobial therapy.^67-69 The duration of the PAE should be included in estimates of doses or dosing intervals. Some antimicrobials also have been associated with a postantibiotic sub-MIC effect (PASE) that may further prolong the dosing interval^70,71; further, a postantibiotic leukocyte enhancement effect (PALE) has been described for some antimicrobials. These are incorporated in in vivo estimates of PAE. However, clinical relevance of measurements of PAE, PASE, and PALE based on in vitro observations is not clear.^66,72 These studies do point out the reasons that some antimicrobials are effective at long intervals and indicate the need for a better understanding of the relationship of PDC, MIC, and PAE in the clinical patient.

Table 6-6 The Duration of the Postantibiotic Effect Demonstrated by Selected Drugs Toward Selected Organisms^63,65,66

Time- Versus Concentration-Dependent Drugs

The relationship among efficacy, MIC, and the magnitude and time course of PDC can be categorized, in vitro, as either concentration-dependent (sometimes referred to as dose-dependent) or time-dependent (sometimes referred to as concentration-independent) (see Figure 6-13; and Table 6-4).⁴¹ A third classification has emerged with characteristics from each of these classes. (e.g., as shown by fluoroquinolones). Although studies that categorize drugs are largely in vitro, the categorizations generally are supported by in vivo studies that include animal models and human clinical trials.⁵⁷ Concentration-dependent drugs, best represented by the fluoroquinolones and aminoglycosides, are characterized by efficacy that is best predicted by the magnitude of PDC (C_max) compared to the MIC of the infecting organism (see Figures 6-11 to 6-13)^73-79 For such drugs the magnitude of C_max/MIC (or C_max/MIC₉₀) generally should be 10 to 12; for more difficult infections (e.g., P. aeruginosa, or infections caused by multiple organisms), the higher index should be targeted.⁵⁷ The time that PDC is above the MIC—that is, the duration of exposure, (T > MIC or T > MIC₉₀)—is not as important as is the C_max/MIC; in fact, efficacy may be enhanced (e.g., for the aminoglycosides) by a drug-free period (i.e., a long interval between doses; see Figure 6-13).^{61,73,74,80-82} This may reflect, in part, the phenomenon of adaptive resistance.⁸³ Adaptive resistance refers to a reversible refractoriness to the bactericidal effects of an antibacterial agent. This phenomenon has been documented particularly for gram-negative organisms and the aminoglycosides, but it appears to occur with the quinolones as well. The resistance appears to reflect a protective phenotypic alteration in the bacteria, such as reversible downregulation of aminoglycoside active transport. Adaptive resistance occurs rapidly (within 1 to 2 hours) of antimicrobial therapy; duration reflects the elimination half-life of the drug. In humans adaptive resistance to aminoglycosides may last for up to 16 hours after a single dose of aminoglycoside, with partial return of bacterial susceptibility at 24 hours and complete recovery at approximately 40 hours.⁸³

Figure 6-13 The relationship between plasma (tissue) drug concentration, and the minimum inhibitory concentration (MIC) of the organism may determine drug efficacy. The efficacy of concentration-dependent drugs (e.g., aminoglycosides; fluorinated quinolones; and, in some cases, azithromycin or other azalides) depends on a high C_max to MIC ratio. Doses might be increased to ensure sufficiently high plasma drug concentrations to achieve the target ratio. In contrast, for time-dependent drugs, such as beta-lactams, sulfonamides, and nonaminoglycoside ribosomal inhibitors, efficacy is maximized by ensuring that plasma drug concentration remains above the MIC for most (50% to 75%) of the dosing interval.

For concentration-dependent drugs, a dose that is too low is particularly detrimental. In a mouse model of E. coli peritonitis, the antibacterial efficacy of ciprofloxacin, but not imipenem (or meropenem), was improved by doubling the dose. For some concentration-dependent drugs, efficacy may be both dose- and time-dependent, with the best predictor of efficacy being AUC/MIC. For example, efficacy of fluorinated quinolones can be predicted by both a C_max/MIC (target 10 to 12) or AUC/MIC (target 100 to 125) (see Figure 6-13).^41,84-86 The AUC/MIC may be particularly predictive for fluoroquinolones when treating gram-positive isolates; adding a second full dose may be prudent for some infections. Otherwise, concentration-dependent drugs generally can be administered at longer intervals (e.g., once a day).

In contrast to concentration-dependent drugs, efficacy of time-dependent drugs (e.g., beta-lactams) is best predicted by the time that PDC remains above the MIC. For such drugs PDC should be 2 to 4 times the MIC of the infecting microbe and should be maintained above the MIC (T > MIC) throughout a significant portion of the dosing interval (see Figures 6-12 and 6-13). However, the recommended duration of T > MIC varies from a low of 25% for carbapenems to 50% to 70% for extended-spectrum penicillins to potentially 100% for penicillin and aminopenicillins (an exception being treatment of streptococci).^{56,57,61,74,87,88} With time-dependent drugs, increasing the dose may be necessary to ensure that PDCs are above (ideally severalfold) the MIC.⁵⁷ Maintaining T > MIC may be problematic for drugs with a short half-life unless the isolate is extremely susceptible (e.g., most beta-hemolytic streptococci for penicillins). Given that drug concentrations decrease by 50% every drug half-life, a C_max/MIC of 2 will result in a PDC that will reach the MIC in 1 half-life. The duration of the dosing interval then depends on the desired target duration (i.e., 25% to 100%) of T > MIC. If T > MIC is 100%, then the dosing interval would be 1 half-live. If T > MIC is 50%, the dosing interval would be 2 half-lives, and if T > MIC is 25%, the dosing interval could be as long as 4 half-lives. For each additional half-life to be added to the duration that T > MIC, concentrations, and thus dose, must be doubled again (i.e., quadrupuled if T > MIC = 2 half-lives, increased eightfold if T > MIC=3 half-lives, and so on). Table 6-4 demonstrates the impact of C_max:MIC and half-life on time-dependent drugs. Although efficacy of time-dependent drugs requires T > MIC for a sufficient time, efficacy might be enhanced by increasing the dose for drugs with a sufficiently long half-life, shortening the interval for drugs with a short half-life, or both. Constant-rate infusion,⁸⁹ or slow-release products,⁹⁰ might be ideal for time-dependent drugs with short half-lives. Drugs characterized by longer elimination half-lives might be preferred (e.g., cefovecin^43,44). Efficacy also should be enhanced for time-dependent drugs that persist to accumulate in the unbound state in selected tissues (i.e., macrolides, clindamycin,⁹¹ or drugs that accumulate in phagocytes.) The downside to using antimicrobial drugs with a very long half-life is that the time to steady-state concentrations and thus peak effects might be prolonged. Moreover, a “hit hard, get out quick” approach to therapy is difficult to implement with such antimicrobials.

The relationship between PDC, MIC, and time- versus concentration-dependency might be explained, in part, by the mechanism of antimicrobial action. Efficacy of the aminoglycosides or fluorinated quinolones depends on drug binding to the target (ribosome and topoisomerase or DNA gyrase, respectively); once sufficient binding occurs, protein synthesis or DNA activity, respectively, is prevented and does not re-initiate. However, beta-lactams substitute as a substrate for cell wall synthesis, and, as long as the organism is growing, it is synthesizing cell wall. Thus, the drug needs to be present as long as the organism is growing. The glycopeptides (e.g., vancomycin), which also target the cell wall, are also time-dependent drugs.

Increasingly, CLSI is using PDI as the basis for determination of breakpoint MIC. However, not surprisingly, the optimal relationship between PDC and MIC that determines efficacy of a drug is not so simple and varies with organisms and drugs. The optimal relationship between PDC and MIC and the parameter that best predicts antimicrobial efficacy (e.g., peak PDC; the ratio of area under the drug concentration versus time curve to the organism’s MIC; duration of PDC above MIC) have not been established definitively for all antimicrobials.^52,61,84,85 However, for drugs characterized by inhibition (bacteriostatic drugs), T > MIC may best predict efficacy. For some fluoroquinolones, and particularly for gram-positive organisms, efficacy is best predicted by the ratio of AUC (which is influenced by both dose and interval) to MIC rather than simply the C_max/MIC. The optimal AUC/MIC also varies with the organism, ranging from as low as 30 to 40 for S. pneumoniae and levofloxacin to greater than 350 for P. aeruginosa and ciprofloxacin. The area under the inhibitory curve (AUIC) reflects the integrated AUC above the MIC during the dosing interval. This parameter is similar to but varies from AUC/MIC in that it is the AUC that is above the MIC (in contrast, AUC/MIC involves the complete area). The AUIC should exceed 125 for fluorinated quinolones to achieve bacterial killing; an AUIC that exceeds 250 results in rapid killing.⁵⁷ Thus for treatment of some infections with concentration-dependent drugs, the dosing regimen might be designed to maximize both the C_max/MIC and the AUC/MIC. Some drugs (e.g., macrolides) are characterized by time-dependency for some organisms but concentration-dependency for others.

Table 6-4 offers examples of PDIs that are achieved using current recommended dosing regimens for selected drugs and selected PD data from selected pathogens that cause infection in dogs. The PD data are based on MIC₉₀ obtained from package inserts of drugs approved in dogs or literature that provided PD data specific for canine pathogens that was more recent than package insert data. For some drugs PD was not available for canine pathogens, so PD from human pathogens was used. Doses were chosen from Table 7-1 and generally reflected the highest dose for which PK was available. When possible, nonintravenous routes were chosen because C_max from intravenous data may not represent C_max following distribution. Table 6-6 is intended to demonstrate how PDI might be used to assess a dosing regimen. It is not unusual to find that the target C_max/MIC (>10) often is not reached for concentration-dependent drugs) or T > MIC (50% for most) for time-dependent drugs. One could argue that the MIC₉₀ is an unreasonable PD target; indeed, for some isolates it may be. The preferred PD statistic would be the MIC from the isolate infecting the patient; it can be substituted in this table for the MIC₉₀. Likewise, the 95th lower confidence interval is preferred to the mean Cmax or half-life for the PK component of PDI. For concentration-dependent drugs, doses can be increased (whenever safety permits) to achieve the target C_max/MIC; for time-dependent drugs, both the dose (increase) and interval (shorten) might be modified. Alternatively, or perhaps in addition to, combination therapy might be considered. Note that the PDIs are based on plasma drug concentrations (PDCs). For some drugs, PDC underestimates concentrations in extracellular fluid. However, for others, PDC frequently overestimates by 25% to 50% or more extracellular fluid drug concentrations. Doses may need to be increased by 25% to 50% to adjust for this difference. As important as PK/PD integration is to the design of the dosing regimen, its application to the clinical patient will be facilitated by an understanding of the microbial and host factors that influence response to the drug.