Epidemiologic Cutoff Values

Epidemiologic cutoff values, abbreviated ECV (CLSI) or ECOFF (Eucast) , are measures of a drug MIC distribution that separate bacterial populations into those representative of a wild type population, and those with acquired or mutational resistance to the drug. A bacterium with a drug MIC that is greater than the ECV is likely to have an acquired form of resistance, whereas one with a drug MIC lower than or equal to the ECV is likely from the wild type distribution of the bacterium for a particular drug. The ECV is NOT the same as a susceptibility breakpoint, and is difficult to apply clinical decisions regarding antibiotic therapy. Some organisms may have intrinsic resistance to a particular drug, with that drug's ECV being relatively high in terms of what are achievable drug concentrations. In this case, even though a MIC might be less than the ECV, clinical treatment failure rates would likely be high. For example, the tigecycline ECV for Pseudomonas aeruginosa is 64 mg/L; a strain of this organism with a tigecycline MIC of 32 mg/L would have a MIC less than the ECV yet still be considered resistant to tigecycline in vitro, and extremely unlikely to be successfully treated with tigecycline if it were causing an infection. Conversely, a drug MIC greater than the ECV wouldn't necessarily mean that an infection with the bacterium couldn't be successfully treated with that drug, as that might depend on drug pharmacodynamics, site of infection, drug dosage and other factors. For example, the azithromycin ECV for Streptococcus pneumoniae is 0.25 mg/L, with EUCAST and CLSI breakpoints of 0.25 and 0.5 mg/L, respectively.

Why are ECVs listed instead of breakpoints for some bacterium/drug combinations in the CLSI M100 tables? This is because of limited data on the correlation between MIC and clinical responses, lack of robust in vivo and in vitro data and lack of information about drug pharmacodynamic and outcome relationships. This might mean that only MIC distributions are known, but not good data on drug pharmacokinetics and pharmacodynamics, and the correlation between drug MICs and clinical outcomes. ECVs are commonly found in CLSI guidelines for antifungal susceptibility testing, and for several bacteria. The current version of CLSI M100, M100-S27, lists the following ECVS: vancomycin and P. acnes, azithromycin and some Shigella spp, colistin and some Enterobacteriaceae, and azithromycin and N. gonorrhoeae. Appendix G in M100-S27 has a good description of ECVs.

How are ECVs determined? Hundreds of isolates of a particular bacterial species have drug MICs determined, resulting in a known population distribution. This population distribution is then analyzed using a freeware statistical program, "ECOFFinder", which estimates the wild type population, and derives a ECV value. You can find the population distributions of many bacteria, for multiple drugs, on the EUCAST website. Examples are given below of the population distributions from the EUCAST website, and their analyses using the ECOFFinder software.

K. pneumoniae cefepime MIC distribution

This distribution obviously includes both wild type and drug resistant isolates. The ECOFFinder software estimates the MIC breakpoint that separates the population into wild type and non-wild type, with an ECV ≦ 0.0625 mg/L, much lower than the indicated EUCAST and CLSI breakpoints. Only about 47% of the isolates tested had cefepime MICs . It's not a clean separation, with the majority of isolates having cefepime MICs greater than the ECV. The EUCAST and CLSI breakpoints are indicated by arrows.

ECOFFinder input and graphical output

K. pneumoniae meropenem population distribution.

Here you can see that the population is almost all wild type, with 95% of the population being wild type. The ECV, ≦0.125 mg/L, is well below the indicated CLSI and EUCAST breakpoints.

How should you interpret a MIC above or below the ECV, when there are no established breakpoints for a particular bug-drug combination? Apart from knowing whether the MIC is consistent with a wild type population, or a non-wild type population, there's not much that you can infer about the possible utility of a drug for treatment of a particular infection. You really need to focus on the MIC, and integrate that result with what is known about the pharmacokinetics of the drug at the site of infection, its pharmacodynamics, whether the bacterium is likely to develop resistance or tolerance during therapy, whether the MIC result is accurate, and what others have reported about the efficacy of the drug in similar clinical circumstances.


P Edelstein, 3/20/2017