Matthew A. Wikler, Franklin R. Cockerill III, and Paul G. Ambrose
DEFINITION AND CLINICAL UTILITY OF ANTIMICROBIAL BREAKPOINTS
A breakpoint, in its simplest terms, represents the concentration of an antimicrobial agent that separates populations of microorganisms. Breakpoints are used in many ways, and so there may be more than one breakpoint for a specific antimicrobial agent–microorganism combination. It is also of interest that a breakpoint may change from time to time for a variety of reasons, as discussed later. In addition, breakpoints can vary from one country to another and from one official body to another in the same country. For example, in the United States, both the Clinical and Laboratory Standards Institute (CLSI), formerly the National Committee for Clinical Laboratory Standards (NCCLS), and the U.S. Food and Drug Administration (FDA) may provide breakpoints for the same antimicrobial agents. Additionally, breakpoints in the European Union (EU) are set by the European Committee on Antimicrobial Susceptibility Testing (EUCAST). The determination of a specific breakpoint is not a black-and-white decision, as many factors must be considered when selecting breakpoints.
To assist physicians in selecting antimicrobial agents to treat patients, clinical microbiologists categorize clinical isolates as drug susceptible, drug intermediate, or drug resistant. A result of “susceptible” assumes that an infection due to the isolate may be appropriately treated with the dosage of an antimicrobial agent recommended for that type of infection and infecting species. A result of “resistant” assumes that the isolate is not inhibited by the usually achievable concentrations of the agent with normal dosage schedules and/or falls in the range where specific microbial resistance mechanisms are likely and clinical efficacy has not been reliably attained in treatment studies. A result of “intermediate” assumes that an infection due to the isolate may be appropriately treated in body sites where the drugs are physiologically concentrated or when a high dosage of the drug can be used. The category of intermediate is also used as a “buffer zone” to prevent small, uncontrolled technical factors from causing major discrepancies in interpretations (1).
Ultimately, the purpose of breakpoints is to provide clinicians with information to assist in making decisions about antimicrobial treatments for patients with infections. Breakpoints serve many purposes, some of which are important for an individual patient, others for epidemiologic reasons. If breakpoints did not result in better patient care, then there would be little need to determine them other than as an academic exercise.
The remainder of this chapter focuses on breakpoints for bacteria, as these are currently the most advanced; however, many of the principles apply to the setting of breakpoints for other types of microorganisms (e.g., fungi).
HISTORICAL EVOLUTION AND CURRENT CRITERIA FOR ESTABLISHING BREAKPOINTS
CLSI has been providing standards for the testing of bacteria and breakpoints since 1975. Initially, breakpoints were established by examining scatterplots of the distributions of bacterial isolates versus the results of susceptibility testing conducted with antibacterial agents. Such scatterplots would frequently divide the bacterial isolates into two populations, one of which would appear to be more susceptible to the antibacterial agent being tested, the other less susceptible. The breakpoint would be the drug concentration separating the two populations. Establishing breakpoints in such a manner is probably suitable for epidemiologic purposes, as it allows one to easily determine shifts in the populations of organisms and to identify the emergence of resistant populations. From a clinical perspective, however, such an approach does not take into account the clinical implications.
In an attempt to improve the process for establishing breakpoints, CLSI has provided specific guidelines contained in a special document. The first version of this document was published in 1994 (2) and introduced the concept of looking at other types of data, including clinical data, in an attempt to correlate proposed breakpoints with what is likely to occur in the clinical setting. A revision of this document was published in 2001 (3). At the current time, so-called clinical breakpoints (antimicrobial susceptibility test interpretive categories) are determined by CLSI utilizing the following types of information: microbiologic data, animal modeling data, pharmacokinetic (PK) and pharmacodynamic (PD) modeling data, and human clinical data. These data are all considered and compared one against the other. In an ideal world, these data would all correlate with one another so that a breakpoint could be determined with certainty.
The microbiologic data considered consist of distributions of bacterial isolates and their minimum inhibitory concentrations (MICs). Numerous distributions are evaluated including those for a broad spectrum of organisms against which the antimicrobial agent is likely to be utilized and those for select populations of organisms that have specific types of resistance mechanisms. As clinical studies are conducted with a new antimicrobial agent, the susceptibility patterns observed in actual patients enrolled in studies are also reviewed. By utilizing data of this type, one can gain a sense of the various populations of organisms that exist and their relative susceptibility to the antimicrobial agent. Animal studies are quite useful in determining which pharmacokinetic-pharmacodynamic (PK-PD) measure one should be evaluating when trying to predict clinical efficacy. The three most common PK-PD measures are the duration of time the drug concentrations remain above the MIC (T>MIC), the ratio of the maximal drug concentration to the MIC (Cmax:MIC ratio), and the ratio of the area under the concentration time curve at 24 hours to the MIC (AUC0-24:MIC ratio). For most classes of antimicrobial agents, animal studies have demonstrated the ability to predict clinical efficacy by examining specific parameters (4). For example, it has been clearly demonstrated for β-lactam antibiotics that the most critical parameter predictive of clinical outcomes is the time that free drug plasma concentrations remain above the MIC of the causative organism (5–7). This is an example of a class of antibiotics where the predictive parameter is “time dependent.” For cephalosporin antibiotics and Streptococcus pneumoniae, it appears that clinical success is likely if the free drug concentration above the MIC of the causative organism is maintained for 40% to 50% of the dosing interval, while for penicillins, the target appears to be 30% to 40% of the dosing interval (4).
The area under the drug concentration time curve (AUC) is a measure of drug exposure. Mathematically, the AUC is calculated as the integral of the drug concentration time curve. Response to drugs in vivo can usually be linked to the AUC. In some instances, the shape of the concentration time curve can affect in vivo response to a drug, and thus other measures of exposure (e.g., Cmax, Cmin) can also be important. Fluoroquinolone antibiotics have been demonstrated in animal models to be “concentration dependent”; that is, clinical outcomes can be predicted based on the AUC:MIC ratio and/or by the Cmax:MIC. For the fluoroquinolones, it appears that clinical success (8,9) depends on attaining a free drug AUC:MIC ratio of around 30 for gram-positive organisms and around 100 for gram-negative organisms.
Although these general PK-PD targets tend to apply to many types of infections, one must keep in mind that levels obtained in certain tissues and body fluids may result in these targets not being predictive. For example, for drugs that are excreted by the kidneys and where active drug is concentrated in the urine, one would anticipate the ability to successfully treat organisms in the urinary tract with higher MICs. On the other hand, most drugs do not achieve high levels in the cerebrospinal fluid (CSF), and so one would anticipate that higher doses of an antimicrobial agent may be required to adequately treat an infection in that site. It would not be sufficient only to determine PK-PD targets based on responses in animal models, and so it is important that the results of clinical studies be correlated with these targets. Such work has been done for many classes of antimicrobial agents (4,8,10–14). Unfortunately, for new classes of antimicrobial agents, including LpxC inhibitors (deacetylase inhibitors of endotoxin biosynthesis), topoisomerase type-B subunit inhibitors, β-lactam–β-lactamase inhibitors, the correlations between the targets and the clinical outcomes in humans have yet to be well studied.
Once the PK-PD measure and target predicting clinical success has been identified, one can integrate this information with human PK data to estimate the probability of attaining drug exposures sufficient across a range of MIC values. One of the best ways to conduct such an analysis is by using Monte Carlo techniques (15,16). Basically, the strategy is to take (a) the PK parameters along with anticipated variability and (b) the MIC values of organisms likely to cause an infection along with the proportion of time; a specific MIC value is achieved or exceeded by in vivo concentrations of antimicrobial agent and then model patients by randomly matching up PK profiles and MICs. By using such a technique, one can easily simulate 5,000 or 10,000 patients and make predictions as to what MICs one is likely to be able to treat successfully with various dosing regimens (17,18). Ideally, this process should be accomplished in the earliest stages of drug development, as this allows one to determine the optimal dosing regimen likely to result in a successful clinical outcome while minimizing the potential for toxicity. Such PK-PD modeling can also be utilized to justify the initial breakpoint for a new antimicrobial agent prior to the availability of a large amount of clinical data (19,20).
Ultimately, the purpose of breakpoints is to provide information to the clinician for the selection of optimal antimicrobial therapy. Because of this, it is critical to evaluate clinical data from well-designed clinical studies which correlate breakpoints with clinical outcomes. Unfortunately, this is far from an exact science, as there are many factors that determine clinical outcomes other than the antimicrobial agent used. Consequently, clinical correlations are generally used to confirm susceptibility breakpoints predicted by the previously mentioned techniques and data.
For many reasons, the true limits of an antimicrobial agent are rarely tested in clinical studies conducted for the purpose of gaining regulatory approval. First, most of these studies exclude or discontinue patients whose infections are caused by organisms with an MIC above a tentative breakpoint. As a result, even if clinical studies could demonstrate a breakpoint, the probability of this happening is greatly reduced. In most cases, clinical studies can be ethically designed in a manner that would blind the investigator to susceptibility test results, allowing the decision to continue or discontinue therapy with the study drug to be determined by clinical and microbiologic responses. Clinical studies so designed are more likely to aid in determining a clinical breakpoint. Another reason that breakpoints often fail to be determined by clinical data is that many of the newer antibiotics being developed are quite potent and only a small percentage of organisms will have MICs high enough to truly test their limits. In fact, few or no patients may be enrolled who have infections due to organisms with MICs at a sufficient level to uncover the limits of the antibiotic being evaluated.
ESTABLISHING BREAKPOINTS BY SITE OF INFECTION
It is sometimes necessary to consider the need for different breakpoints based on the site of the infection for which the antimicrobial agent is intended. For example, the ability of a drug to concentrate within the CSF is frequently quite different from the ability of the same drug to be concentrated in the urine or into tissues such as lung. As a result, it is reasonable to anticipate the requirement for a different breakpoint for the treatment of meningitis, urinary tract infections, or pneumonia.
As an example, CLSI recommends different penicillin breakpoints for S. pneumoniae isolated from spinal fluid as compared with respiratory secretions. Due to PK/PD and outcome data, lower (more stringent) breakpoints are provided for CSF isolates versus higher (more lenient) breakpoints for respiratory isolates (13).
ESTABLISHING BREAKPOINTS WHEN THERE ARE NO RESISTANT BACTERIAL STRAINS
As previously noted, many of the newer antimicrobial agents being developed are relatively potent and few resistant bacterial strains exist. When there is a dearth of resistant strains, it is difficult to obtain animal or clinical data to determine the true breakpoint. In these circumstances, the breakpoint is generally set at one or two 2-tube dilutions above the known susceptible population of strains for that organism. In such a situation, only a susceptible and not a resistant breakpoint is usually published, along with a notation that any strains isolated with a higher MIC should be sent to a reference laboratory to confirm the results. With time, a population of less susceptible strains will frequently emerge, and additional animal and clinical data may become available. At that point, it may become possible to establish a breakpoint reflective of the new situation.
CORRELATION BETWEEN MINIMAL INHIBITORY CONCENTRATION AND DISK ZONES
Many laboratories do not do MIC testing but rather depend on disk diffusion susceptibility testing methods. In order to meet the needs of these institutions, breakpoints are set for disk diffusion methods by correlating MIC results and disk diffusion results. Once again, various types of MIC distributions of clinical isolates of bacteria are reviewed, including distributions for a broad spectrum of organisms against which the antimicrobial agent is likely to be used and for select populations of organisms that have specific types of resistance mechanisms. Statistical methods are generally utilized to determine the best correlation between disk zones and MICs. Once the disk zone breakpoint is statistically determined, the rates of discrepant results are evaluated (i.e., where one method predicts susceptible and the other predicts resistant or intermediate). The number of discrepancies that occur within one twofold dilution of the intermediate MIC is less important than the number of discrepancies that occur at other MICs. After analysis of such discrepancies, the disk test may be adjusted to make it more predictive of the MIC test (i.e., by reducing the number of discrepancies). In some cases, it is impossible to develop a disk test that correlates with the MIC test. A recommendation is then made to not perform disk testing.
OVERRIDING THE BREAKPOINT
There are circumstances in which the breakpoint determined utilizing the standard methods is known to be inaccurate. When this occurs, the laboratory is instructed to override the results of the test and to adjust the report to the physician. For example, in recent years, gram-negative organisms that produce extended-spectrum β-lactamases (ESBLs) emerged. The standard testing methods that were used produced results indicating that the organism was susceptible to cephalosporins; however, because of the presence of an ESBL, these drugs were ineffective against such strains. As a result, CLSI developed specific testing methods to detect ESBL-producing strains. When such strains were detected using these specific methods, laboratories were instructed to override all MIC results previously interpreted as susceptible with the interpretation of resistant (21). Recently, CLSI has modified the MIC breakpoints for these drugs so that ESBL-producing strains are essentially captured. However, until these new breakpoints are adopted by laboratories, the specific ESBL tests must still be used with “overriding of results” as directed. As new mechanisms of resistance develop, it is critical that organizations and agencies that produce standardized antimicrobial susceptibility testing methods and interpretive criteria be diligent in looking for circumstances where the results of such tests are not accurate.
THE USE OF SUSCEPTIBILITY TESTS OF ONE ANTIBIOTIC TO PREDICT THE SUSCEPTIBILITY OF ANOTHER ANTIBIOTIC
Laboratories often use testing systems developed by antimicrobial susceptibility testing manufacturers. Such testing systems may contain a panel of antibiotics that do not replicate the available agents in a particular hospital or within the formulary of a particular health care system. In such a case, a laboratory may wish to use the results for one antibiotic to predict the susceptibility of an organism to another similar antibiotic. This has been a common practice, for example, with various cephalosporin antibiotics. In Table 1 of CLSI document M100 (21), there are suggestions as to when this may be possible. One must be aware that the correlations for certain types of organisms may be much better than for others within the same antibiotic class and that depending on one antimicrobial agent to predict another will invariably lead to some reporting errors.
WHY ARE THERE DIFFERENT BREAKPOINTS IN VARIOUS PARTS OF THE WORLD?
It is not uncommon to find that the clinical breakpoints for an antimicrobial agent–microorganism combination are different in different parts of the world. The reasons for this should become clear when one examines the variables involved in determining breakpoints. Breakpoints are set based on results achieved using a standardized testing method. If everyone used exactly the same testing method with adequate controls, the results would be expected to be similar. Unfortunately, testing methods are not currently standardized around the world, and thus there is the potential for different breakpoints to be established using different methods. Second, an antimicrobial agent might be used differently in different geographic areas. If it is customary to use an antimicrobial agent at a higher dose or to dose more frequently (including constant infusion) in one geographic area, the breakpoint will likely be higher in that area. Third, different microorganisms are encountered in different parts of the world. If resistant strains of bacteria are present in one geographic area but not other areas, it may be necessary to regionally adjust the breakpoints to ensure that the new resistant strains are being properly reported. There are also public health reasons why a breakpoint may vary. Public health authorities in one geographic area may want to avoid a resistance problem, deal with a resistance problem, or promote the use of certain antimicrobial agents over others. One way to impact antimicrobial use is to adjust breakpoints.
CURRENT EFFORTS TO DEVELOP STANDARDIZED METHODS AND BREAKPOINTS IN OTHER PARTS OF THE WORLD
Currently, efforts are being undertaken in various parts of the world to develop standardized methods for susceptibility testing. Ultimately, it would be ideal if one standardized testing method was accepted worldwide, as this would allow the direct comparison of results from one part of the world to another. The lack of harmonization of methods can create significant problems when evaluating epidemiologic trends. For example, suppose the goal is to examine the development of resistance for a particular organism and its spread in various parts of the world. The use of different methods makes it impossible to ascertain the true level of resistance. Even if one standardized testing method was accepted and utilized throughout the world, it is likely that there would still be different clinical breakpoints, for the reasons noted previously.
Another problem resulting from the lack of standardized methods for susceptibility testing concerns the development of new antibiotics. When conducting clinical trials and looking for correlations between outcomes and MICs or disk zones, it is necessary to use the same methods wherever the drug is tested.
The EUCAST is a standing committee of the European Society of Clinical Microbiology and Infectious Diseases (ESCMID). EUCAST was set up to standardize susceptibility testing in Europe so that comparable results and interpretations would be produced. It has both a general committee, whose membership includes representatives from all European countries, from the pharmaceutical industry, and from the in vitro media and device industries, and an ESCMID-appointed steering committee, which consists of a chair, a scientific secretary, six National Breakpoint Committee representatives, and two representatives from the general committee. Decisions are made by the steering committee after consultation with the general committee (ESCMID Web site, https://www.escmid.org). Unlike the CLSI, which develops only clinical breakpoints, EUCAST is in the process of developing both epidemiologic and clinical breakpoints. Epidemiologic breakpoints are breakpoints that differentiate the wild-type strains from strains that have developed resistance. Epidemiologic breakpoints can be extremely valuable when one wants to evaluate the emergence of resistant populations of organisms. The methods used by EUCAST (22,23) are in general agreement with those of the CLSI. EUCAST has been collecting MIC distribution data from worldwide sources for the purposes of establishing epidemiologic breakpoints. This extensive database is available on the EUCAST Web site (www.eucast.org).
WHY THE BREAKPOINTS OF CLSI AND THE FDA MAY DIFFER
At times, breakpoints contained in the FDA-approved package insert for an antimicrobial agent do not agree with the breakpoints published in CLSI documents. There are many reasons why this may occur, including differences in the interpretation of data and differences in how the two organizations function. When a new antimicrobial agent is approved by the FDA, interpretive criteria utilizing the CLSI standardized methods are approved and included in the product label. Most pharmaceutical sponsors will also submit a package of data to CLSI within a year of a new drug approval requesting breakpoints. Although there is general agreement between the FDA and CLSI, at times there are differences based on interpretation of the data. It is common for differences to occur after an antimicrobial agent has been on the market for a few years, as new resistance mechanisms become apparent requiring a reevaluation of the breakpoint. Unlike the FDA, CLSI has the ability to review breakpoints for any antimicrobial agent when there appears to be a need to do so. As a result, CLSI will make changes in single drugs or will frequently evaluate a class of drugs at the same time and make whatever adjustments seem necessary. Currently, the FDA generally considers a change in a breakpoint only when the sponsor submits a package to the FDA requesting a change. FDA staff participate in CLSI meetings as advisors and reviewers. Sponsors are encouraged to submit new data to the FDA to allow for updating the product label.
The obvious question arising from this is which breakpoints will clinical laboratories in the United States use? When there are published CLSI breakpoints, US laboratories use these breakpoints when testing organisms and providing reports to physicians. In fact, US laboratories are tested and accredited based on their compliance with CLSI methods and interpretive standards. If US laboratories use devices provided by in vitro diagnostic (IVD) manufacturers, those manufacturers must receive FDA approval at the breakpoints the FDA specifies before marketing these devices.
THE IMPORTANCE OF QUALITY CONTROL
The ability of the laboratory to follow the standardized testing methods utilized for setting breakpoints and the ability of the automated testing system to replicate the results that would be obtained utilizing the standard methods are critically important. Clearly, if such methods are not followed and well controlled, then the MIC or disk zone reported may be inaccurate, potentially resulting in a misinterpretation and inaccurate reporting to the health care provider. For this reason, care must be taken in performing these tests, and proper quality control must be implemented. In order to help the laboratory, CLSI and other organizations that produce documents outlining standardized methods provide quality control ranges for various standard bacterial strains tested against specific antibiotics. It is critical to ensure that the test performed on the quality control strains produces results that are within the accepted ranges. In addition, the laboratory must make certain that the breakpoints applied are those based on the methods that are being utilized. It is also critical that growth of the bacterial strain is sufficient for an accurate MIC or disk zone to be obtained. That is why it is necessary to have a control well or area on the test plate where the bacterial strain can grow uninfluenced by the antimicrobial agent.
HOW TO REPORT AND USE BREAKPOINTS
Although breakpoint information is valuable when used for a specific patient, it can also have an impact on antimicrobial agent selection for a much larger group of patients. Most antimicrobial agent use is empiric; that is, a patient appears with what seems to be a bacterial infection, and the physician prescribes an antimicrobial agent without knowledge of the causative organism or its susceptibility. If the laboratory periodically collects its susceptibility testing data, summarizes such data, and distributes them to its physicians, then physicians are in a better position to prescribe antimicrobial agents likely to be successful. Most hospitals publish an antibiogram once or twice a year just for this purpose. There are numerous things one must consider when constructing antibiograms, and CLSI document M39-A3 (24) provides a guideline to help laboratories in developing them.
Although epidemiologic breakpoints tend to be static, clinical breakpoints are not. There are numerous reasons why a breakpoint may need to be changed, and many of them are outlined in CLSI document M23 (3):
1. Strains less susceptible and/or more resistant to an antimicrobial agent may evolve.
2. Organisms with new mechanisms of resistance may develop.
3. New dosages or formulations of an antimicrobial agent and/or new clinical uses may require a change.
4. New clinical and/or pharmacologic data may suggest the need for reassessment.
5. Actions by and/or data from the FDA or other regulatory authorities, the Centers for Disease Control and Prevention (CDC), the College of American Pathologists, or other sources may suggest the need for reassessment.
6. Changes in CLSI-approved reference methods may have an impact on interpretive criteria and/or quality control parameters.
7. Other in vitro testing may suggest the need for reassessment.
8. Changes may also be made when public health concerns require action in situations where clinical information is limited.
As a recent example, in January 2010, CLSI changed the breakpoints for cefazolin against Enterobacteriaceae to reflect the emergence of resistance caused by ESBLs. However, after further review of common dosing regimens, MIC distributions and PK-PD data, this correction was determined to be too severe. Therefore, in January 2011, CLSI increased the MIC concentration interpretation for resistance by one twofold dilution (25).
PUBLIC HEALTH CONSIDERATIONS
The setting of breakpoints has an impact not only on individual patients but also on public health. The breakpoints will determine how antimicrobial agents will be perceived to work against specific organisms. As a result, when a physician receives an antibiogram of the susceptibility patterns of the organisms in his or her hospital and/or the local community, the physician’s antimicrobial agent use patterns may be affected. Because the vast majority of bacterial infections are treated empirically, physicians depend on the antibiogram to direct their selection of initial antimicrobial therapy. If a breakpoint is changed and the change results in a commonly used antimicrobial agent no longer appearing to be efficacious, this may lead physicians to alter their prescribing habits. As a result, one class of drugs may end up being substituted for another. This shift in antimicrobial use can have an impact on future resistance patterns in the hospital and the community. For example, a previously low MIC that defined resistance for penicillin against S. pneumoniae for nonmeningitis infection resulted in a greater use of vancomycin. The increased use of vancomycin in hospitals appears to have led to an increase in the incidence of vancomycin-resistant enterococci. In addition, if a change in a breakpoint leads to the use of more expensive antimicrobial agents, there is an economic impact on the health care system. Clearly, there are important economic and health consequences when a breakpoint change results in the development of more problematic and difficult-to-treat organisms. These potential issues must be carefully considered when setting and/or changing breakpoints.
NEED FOR GREATER UNDERSTANDING OF BREAKPOINTS
As stated previously, the primary reason for breakpoints is to provide information to health care providers that will allow for the selection of antimicrobial agents likely to successfully treat infections. If a health care provider neither understands what the susceptibility report means nor understands the assumptions that underlie the report, then the actions taken may not be optimal for the patient. For example, if the breakpoint is set based on a specific dose of an antimicrobial agent being used, and if the physician uses a lower dose, then the actual clinical result may not be as anticipated. It is critical that efforts be made to properly communicate to health care providers what breakpoints mean and the assumptions that go into these interpretive standards. Information concerning some of the assumptions made in selecting breakpoints is contained in the documents and tables developed by CLSI and other standards-setting organizations. Unfortunately, this information rarely is communicated to physicians. If breakpoints are to be optimally utilized to maximize patient care, greater communication and education must occur. The education process could involve scientific publications that specifically discuss the decisions made by CLSI and other standards-setting organizations and the rationales for the decisions and the assumptions made. Such publications would likely be of interest primarily to microbiologists, infectious disease physicians, and hospital epidemiologists. These health care professionals should then convey the information they acquire to physicians through local educational activities.
In summary, breakpoints allow microbiology laboratories to provide valuable information to clinicians for the optimal selection of antimicrobial therapies. Epidemiologic breakpoints make it easier to detect the emergence of resistant populations of bacteria. Clinical breakpoints may vary due to differences in testing methods and in how antimicrobial agents are used in different parts of the world. Health care providers must be knowledgeable about the assumptions that go into the setting of breakpoints if they are to utilize such information to optimize patient care.
Ms. Tracy Dooley is thanked for her review and helpful comments for this manuscript.
1. Clinical and Laboratory Standards Institute. Methods for dilution antimicrobial susceptibility tests for bacteria that grow aerobically; approved standard—ninth edition. Wayne, PA: Clinical and Laboratory Standards Institute, 2012. CLSI document M07-A9.
2. National Committee for Clinical Laboratory Standards. Development of in vitro susceptibility testing criteria and quality control parameters. Villanova, PA: National Committee for Clinical Laboratory Standards, 1994. NCCLS document M23-A.
3. Clinical and Laboratory Standards Institute. Development of in vitro susceptibility testing criteria and quality control parameters; approved guideline—third edition. Wayne, PA: Clinical and Laboratory Standards Institute, 2008. CLSI document M23-A3.
4. Craig WA. Pharmacodynamics of antimicrobials: general concept and applications. In: Nightingale CH, Murakawa T, Ambrose PG, eds. Antimicrobial pharmacodynamics in theory and clinical practice. New York: Marcel-Dekker, 2002:1–22.
5. Andes D, Craig WA. In vivo activities of amoxicillin and amoxicillin-clavulanate against Streptococcus pneumoniae: application to breakpoint determinations. Antimicrob Agents Chemother 1998:2375–2379.
6. Eagle H, Fleischman R, Musselman AD. Effect of schedule of administration on therapeutic efficacy of penicillin: importance of aggregate time penicillin remains at effective bactericidal levels. Am J Med 1950;9:280–299.
7. Eagle H, Fleischman R, Levy M. Continuous vs. discontinuous therapy with penicillin. N Engl J Med 1953;238:481–486.
8. Ambrose PG, Grasela DM, Grasela TH, et al. Pharmacodynamics of fluoroquinolones against Streptococcus pneumoniae in patients with community-acquired respiratory tract infections. Antimicrob Agents Chemother 2001;45:2793–2797.
9. Craig WA, Andes DR. Correlation of the magnitude of the AUC24/MIC for 6 fluoroquinolones against Streptococcus pneumoniae with survival and bactericidal activity in an animal model. In: Program and abstracts of the 40th Interscience Conference on Antimicrobial Agents and Chemotherapy; September 2000; Toronto, Canada.
10. Bodey GP, Ketchel SJ, Rodriguez N. A randomized study of carbenicillin plus cefamandole or tobramycin in the treatment of febrile episodes in cancer patients. Am J Med 1979;67:608–616.
11. Forrest A, Chodosh S, Amantea MA, et al. Pharmacokinetics and pharmacodynamics of oral grepafloxacin in patients with acute exacerbation of chronic bronchitis. J Antimicrob Chemother 1997;40(Suppl A):45–57.
12. Forrest A, Nix DE, Ballow CH, et al. Pharmacodynamics of ciprofloxacin in seriously ill patients. Antimicrob Agents Chemother 1993;37:1073–1081.
13. Preston SL, Drusano GL, Berman AL, et al. Pharmacodynamics of levofloxacin: a new paradigm for early clinical trials. JAMA 1997;279:125–129.
14. Ambrose PG, Bhavnani SM, Rubino CM, et al. Pharmacokinetics-pharmacodynamics of antimicrobial therapy: it’s not just for mice anymore. Clin Infect Dis 2007;44:79–86.
15. Drusano GL. Antimicrobial pharmacodynamics: critical interactions between “drug and bug.” Nat Rev Microbiol 2004;2:289–300.
16. Dudley MN, Ambrose PG. Pharmacodynamics in the study of resistance and establishing in vitro susceptibility breakpoints: ready for primetime. Curr Opin Microbiol 2000;3:515–521.
17. Ambrose PG, Craig WA, Bhavnani BM, et al. Pharmacodynamic comparisons of different dosing regimens of penicillin G against penicillin-susceptible and -resistant pneumococci. In: Program and abstracts of the 42nd Interscience Conference on Antimicrobial Agents and Chemotherapy; September 27–30; San Diego, CA. Abstract A-635.
18. Dudley MN, Ambrose PG. Monte Carlo simulation and new cefotaxime, ceftriaxone, and cefepime breakpoints for S. pneumoniae, including strains with reduced susceptibility to penicillin. In: Program and abstracts of the 42nd Interscience Conference on Antimicrobial Agents and Chemotherapy; September 27–30; San Diego, CA. Abstract A-1263.
19. Ambrose PG, Grasela DM. The use of Monte Carlo simulation to examine the pharmacodynamic variance of drugs: fluoroquinolones against Streptococcus pneumoniae. Diagn Microbiol Infect Dis 2000;38:151–157.
20. Drusano GL, Preston SL, Hardalo C, et al. Use of preclinical data for selection of a phase II/III dose for evernimicin and identification of a preclinical MIC breakpoint. Antimicrob Agents Chemother 2001;45:13–22.
21. Clinical and Laboratory Standards Institute. Performance standards for antimicrobial susceptibility testing; twenty-third informational supplement. Wayne, PA: Clinical and Laboratory Standards Institute; 2013. CLSI document M100-S23.
22. EUCAST Definitive Document E. Def 2.1. Determination of antimicrobial susceptibility test breakpoints. Clin Microbiol Infect 2000;6:570–572.
23. EUCAST Definitive Document E. Def 3.1. Determination of minimum inhibitory concentrations (MICs) of antibacterial agents by agar dilution. Clin Microbiol Infect 2000;6:509–515.
24. Clinical and Laboratory Standards Institute. Analysis and presentation of cumulative antimicrobial susceptibility test data; approved guideline—third edition. Wayne, PA: Clinical and Laboratory Standards Institute; 2009. CLSI document M39-A3.
25. Turnidge JD. Cefazolin and Enterobacteriaceae: rationale for revised susceptibility testing breakpoints. Clin Infect Dis 2011;52:917–924.