The CLSI Approach to Setting Breakpoints Jean B. Patel, PhD, D(ABMM) Deputy Director, Office of Antimicrobial Resistance Division of Healthcare Quality Promotion National Center for Emerging and Zoonotic Infectious Diseases The findings and conclusions in this presentation have not been formally disseminated by the Centers for Disease Control and Prevention and should not be construed to represent any agency determination or policy
Todays Presentation CLSI background information Data needed for breakpoint decisions Examples of recent breakpoint decisions Reporting dose-specific breakpoints Summary 2
CLSI Consensus Process Meetings are open to everyone. Government Meeting materials are publically available. Industry Balance Professions Interests are balanced. Conflicts of interest are fully disclosed. 3
The Antimicrobial Susceptibility Testing Subcommittee What does the subcommittee do? Establishes reference methods for antimicrobial susceptibility testing (Documents M2 & M7) Establishes procedures for setting breakpoints and quality control ranges (Documents M23) Sets breakpoints for interpreting antimicrobial susceptibility testing data M100 - common bacteria M11 - anaerobic bacteria M45 - infrequently isolated bacteria 4
CLSI Antimicrobial Susceptibility Testing Subcommittee We meet twice a year in January & June Meetings are open average 200 people 5
Who Attends the Subcommittee Meeting? Representatives from healthcare Infectious disease and microbiology specialists Representatives from industry Pharmaceutical and susceptibility testing instrument manufacturers Representatives from government agencies CDC, FDA, EMA, EUCAST International partners Japan, China, Central & South America, Canada, Australia 6
MAKING BREAKPOINT DECISIONS 7
Reasons for Making Breakpoint Decisions A new antimicrobial becomes available New data indicates that the breakpoints are not accurate Treatment failures Breakpoints fail to detect important resistance New methods for breakpoint setting are applied to old breakpoints These are the most common reasons for changing breakpoints. It is difficult to get treatment failure data without clinical trials. 8
CLSI vs. EUCAST Question: Why does CLSI change breakpoints more often than EUCAST? Answer: EUCAST is newer than CLSI Resistance mechanisms like ESBLs and carbapenemases were already present when EUCAST set breakpoint Newer methods for setting breakpoints were already developed when EUCAST set breakpoints 9
Data Needed for a Breakpoint Decision MIC and disk diffusion data from bacterial isolates A Monte Carlo Simulation of pharmacokinetic and pharmacodynamic data (PK-PD data) Clinical outcome data for infections caused by bacteria with different MICs 10
Monte Carlo Simulation of PK- PD Data We often need to rely most on this information when setting breakpoints Setting breakpoints using only MIC data overcalls resistance Clinical trials often exclude patients with resistant infections so the data are limited What is a Monte Carlo simulation of PK-PD? 11
Pharmacodynamic (PD) Data PD measures how the drug kills the bacteria Two ways to measure PD % Time > MIC Area Under the Curve (AUC) / MIC Animal models are used to determine which measurement predict how a drug will kill different types of bacteria 12
Pharmacokinetic (PK) Data PK measures the distribution of drug at different body sites and the rate of drug elimination from the body. Examples of differences: Cephalosporin concentrated in the urine Tigecycline concentrates in tissues rather than serum It is important to know that most breakpoints are based upon the concentration of drug in serum 13
Monte Carlo Simulation A model that predicts the percent of patients that would achieve therapeutic levels of drug at different doses of the drug An important decision What is the therapeutic level? A few options: A PK-PD target that results in bacterial stasis A PK-PD target that results in 1 log bacterial kill A PK-PD target that results in 2 log bacterial kill *Most often used * 14
Monte Carlo Simulation of Cefepime Target of 50% Time > MIC Factors that can affect target achievement: Drug dose Infusion time Renal function All of these factors can increase drug concentration. Using a dose of 2g Q8h, 85% of the population will meet the target of 50% T>MIC when the MIC is 8 µg/ml. 15
Setting Breakpoints Using PK- PD Data The general approach: Set the susceptible breakpoint at an MIC where approximately 90% of the patients achieve the therapeutic target CLSI is currently working on a revision of M23 that will provide more guidance for setting breakpoints using PK-PD data 16
TWO RECENT BREAKPOINT DECISIONS 17
Recent CLSI Review of Cefepime Concern: Breakpoints might be too high Current: S < 8 µg/ml, I = 16 µg/l, R > 32 µg/l Breakpoints were set based on the highest approved daily dosage of cefepime (3-6g/day) Evidence of clinical failures for isolates with MICs of 4ug/ml and 8ug/ml, especially when low- to mid-range daily dosages were used Evidence that low- to mid-range daily dosages were commonly used 18
Several Dosing Options for Cefepime Site/Infection Type Dose Frequency Total Daily Dose Mild to Moderate UTI 0.5-1g 12h 1-2g Severe UTI 2g 12h 4g Mild to severe 1-2g 12 2-4g pneumonia Moderate to sever SSTI 2g 12h 4g Complicated IAI 2g 12h 4g Neutropenic fever 2g 8h 6g The non-uti doses for cefepime range from 2 to 6 g/day 19
What doses do people use? 33% of usage was at a dose that is below the current CLSI breakpoint but can be used for nonuti indications. 20
Rationale for Revising Cefepime Breakpoints in 2013 Previous breakpoints were based upon the highest doses of cefepime The rationale for using the highest dose was an assumption that cefepime is used for the most serious infections and will be used at the highest dose. We did not have access to cefepime use data in 2010 when other cephalosporin breakpoints were revised The new cefepime use data indicated that a breakpoint change was needed. 21
Cefepime MIC Distribution of Enterobacteriaceae MIC µg/ml MIC data indicate a cutoff of 0.5 µg/ml 22
Clinical Outcome Data MIC 3g/d 4g/day 6 g/day 0.5 mg/l S S 1 mg/l S F S 2 mg/l S S F F 4 mg/l S F F 8 mg/l F F S F F Best available outcome recent outcome data 17 cases from Lee NY et al. CID 56:488-495, 2013 supplemented by unpublished data from the authors Outcomes: S = Clinical Success S = Success but micro persistence F = Clinical Failure 23
Clinical Outcome Data MIC 3g/d 4g/day 6 g/day 0.5 mg/l 1 mg/l 2 mg/l 4 mg/l 8 mg/l S S S F S S F F S F F F F S F This patient had poor renal function (higher drug concentration) Some evidence that the 3g/d dose is effective when the MIC is elevated No data where we need it the most! 24 F S
Possible Susceptible Breakpoints by PK-PD Data At 2g/day dose: MIC of 2µg/mL At 6g/day dose: MIC of 8µg/mL 25
What breakpoints are suggested by the available data? Data Source Dose Susceptible Breakpoint Micro Data N/A 0.5 µg/ml Comment This breakpoint would overcall resistance. Clinical Outcome 1g 8h 2-4 µg/ml Little data and no data at the highest dose. 1g 12h 2 µg/ml Monte Carlo Simulation of PK/PD Data 1g 8h 2g 12h 2-4 µg/ml 4 µg/ml The susceptible breakpoint could be 2, 4 or 8 µg/ml depending upon the dose. 2g 8h 8 µg/ml 26
The subcommittee decision was A susceptible breakpoint of 2 µg/ml It is supported by clinical and PK/PD data at the 2g/day dose (the lowest, nonuti dose) Problems with 1 µg/ml Clinical data and PK/PD data suggests that this breakpoint is too low Problems with 4 µg/ml Not supported by the 2g/day dose and this dose is commonly used 27
What about intermediate and resistant breakpoints? The usual approach: 2=S, 4=I, 8=R An intermediate breakpoint is important for accommodating MIC testing variability and the lower percent of population meeting the PK- PD target The problem If 4 and 8 µg/ml are supported as a susceptible breakpoints at an approved drug doses should these isolates be categorized as resistant? Should isolates be categorized as intermediate? 28
The breakpoint decision Susceptible Susceptible Dose Resistant Dependent 2 µg/ml 4-8 µg/ml 16 µg/ml 25mm 19 to 24mm 18mm A new interpretive category was used instead of using intermediate. Labs can use intermediate instead, but we think SDD is important 29
A New Interpretive Category SDD: Susceptibility is dose (or dosage) dependent This is an interpretive category that has been used for antifungal susceptibility testing This category communicates what we know a higher dose is indicated SDD is useful for not over-estimating resistance 30
SDD: The Definition The susceptible-dose dependent category implies that susceptibility of an isolate is dependent on the dosing regimen that is used in the patient. In order to achieve levels that are likely to be clinically effective against isolates with MICs or disk zone diameters in this category, it is necessary to use a dosing regimen (i.e., higher doses, more frequent doses, or both) that results in higher drug exposure than the dose that was used to establish the susceptible breakpoint. 31
SDD: The Definition (continued) Consideration should be given to the maximum approved dosage regimen, since higher exposure gives the highest probability of adequate coverage of an SDD isolate. 32
Cefepime Breakpoints CLSI Method S SDD R MIC 2 µg/ml 4-8 µg/ml 16 µg/ml Disk Diffusion 18 mm 15-17 mm 14mm EUCAST Method S Intermediate R MIC 1 µg/ml 2-4 µg/ml 8 µg/ml Disk Diffusion 24mm 21-23 mm 20mm MIC breakpoints are similar.
ANOTHER BREAKPOINT DECISION 34
Carbapenem Breakpoints for Acinetobacter spp. The reason for reconsidering these breakpoints There were no CLSI doripenem breakpoints for Acinetobacter spp. CLSI agreed to not publish doripenem breakpoints until all carbapenem breakpoints could be re-evaluated for Acinetobacter spp. 35
MIC Distributions for Three Carbapenems The distributions for each carbapenem are similar Most resistant isolates demonstrate high MICs (> 32 µgl/ml)
Other Data Some Limitations Clinical outcome data We only had data from one recent clinical trial to evaluate doripenem efficacy PK-PD data We did not have PD data for Acinetobacter spp. Specifically, we don t know what %T > MIC results in optimal killing. 37
Clinical Outcome Data for Acinetobacter spp. Doripenem 500 mg (q8, 1h) MIC (µg/ml) Favorable Clinical Outcome (%) 0.03 2/2 (100) 2/2 (100) 0.06 0 0 0.12 5/5 (100) 5/5 (100) 0.25 4/4 (100) 4/4 (100) 0.5 3/3 (100) 3/3 (100) 1 10/11 (91) 9/11 (82) 2 1/1 (100) 0/1 (0) 4 0 0 8 0 0 16 1/1 (100) 1/1 (100) 32 0 0 64 2/3 (67) 2/3 (67) All Tested 28/30 (93) 26/30 (87) Favorable Microbiological Outcome (%) Source: June 2011 CLSI Agenda Book J & J Very little data for isolates with elevated MICs
Probability of Target Attainment by MIC Value Imipenem 500mg Q6h f %T >MIC 30 f %T >MIC 35 f %T >MIC 40 f %T >MIC 45 1 0.976 0.946 0.914 0.881 2 0.894 0.840 0.780 0.718 Meropenem f %T >MIC 30 f %T >MIC 35 f %T >MIC 40 f %T >MIC 45 1g Q8h 1 0.969 0.927 0.874 0.794 2 0.886 0.806 0.715 0.609 Doripenem f %T >MIC 30 f %T >MIC 35 f %T >MIC 40 f %T >MIC 45 500mg Q8h 1 0.983 0.949 0.889 0.769 2 0.860 0.727 0.565 0.372 *Source: Ambrose and Bhavnani. CLSI Meeting Agenda Book June 2009
Conclusions from PK-PD Analysis The probability of target attainment was similar for all three carbapenems. Therefore, all three carbapenems should have the same susceptible breakpoint. The susceptible breakpoint could be either 1 µg/ml like Enterobacteriaceae or 2 µg/ml like Pseudomonas aeruginosa. We did not have much data to guide us on this decision. To understand the impact of each choice we looked at the MIC distributions. 40
The normal MIC distribution
Conclusions from MIC Distributions The difference between a susceptible breakpoint of 1µg/mL and 2 µg/ml did not result in many more isolates being categorized as nonsusceptible. To improve the accuracy of susceptibility testing methods, it is best to set breakpoints a dilution higher than the normal MIC distribution. This favors a breakpoint of 2 µg/ml.
CLSI The Subcommittee s Decision for Acinetobacter spp. Drug, Dose S I R Imipenem, 500mg q6h or 1g q8h 2 µg/ml 4 µg/ml 8 µg/ml Meropenem, 1g q8h 2 µg/ml 4 µg/ml 8 µg/ml Doripenem, 500mg q8h 2 µg/ml 4 µg/ml 8 µg/ml EUCAST Drug S I R Imipenem 2 µg/ml 4-8 µg/ml 16 µg/ml Meropenem 2 µg/ml 4-8 µg/ml 16 µg/ml Doripenem 1 µg/ml 2-4 µg/ml 8 µg/ml CLSI and EUCAST breakpoints are very similar
What about SDD for these breakpoints? The subcommittee decided not to apply SDD to the carbapenem Acinetobacter spp. breakpoints for because: There are not a wide range of dosing options We did not have sufficient PK-PD data to indicate that SDD should be used instead of intermediate 44
COMMUNICATING DOSE-SPECIFIC BREAKPOINTS 45
Dose-Specific Breakpoints Breakpoints set using PK-PD data are usually based upon a drug dose How should the drug dose information be used within an institution? CLSI dose not recommend reporting a dose with the MIC result. This could be confusing and a physician may interpret this as a recommended dose for their patient. 46
M100-S24, Appendix E A new appendix that describes doses and breakpoints This appendix should be shared with pharmacists and physicians who make dosing recommendations 47
Appendix E What it says All dose-specific breakpoints are listed Proper application of the breakpoints is based upon drug levels at the site of infection for an adult with normal renal function The breakpoints can be applied to patients of different ages, patients with different infections, and patients with renal insufficiency, but dosing should be adequate to achieve the drug levels expected for the susceptible breakpoint 48
SUMMARY 49
Summary CLSI Breakpoint Decision Making We re doing it with all partners in the room We re reorganizing our work process to make decision faster We re looking for innovative ways to address challenging issues 50
Where is the CLSI Subcommittee going in the next year? Working with EUCAST to set colistin breakpoints for gram-negative bacteria Setting azithromycin breakpoints for Salmonella & Shigella Setting azithromycin breakpoints & revising cephalosporin breakpoints for Neisseria gonorrhoeae Working with academic partners to add carbapenemase tests for epidemiological/infection control purposes 51
A few more things Establishing FQ disk diffusion breakpoints for Salmonella & reviewing FQ breakpoints for Enterobacteriaceae Identifying where SDD applies to other breakpoints Updating M23 Data requirements for breakpoint decisions; PK/PD requirements Updating M45 Breakpoints for infrequently isolated bacteria 52
It is an honor to attend your meeting. Thank you. jpatel1@cdc.gov