Rationale for Reading Fluconazole MICs at 24 Hours Rather than 48 Hours When Testing Candida spp. by the CLSI M27-A2 Standard Method^▿

Abstract

Antifungal susceptibility testing for yeasts has been standardized since 1997 when the Clinical and Laboratory Standards Institute (CLSI) published the M27-A method (13). Susceptibility breakpoints for fluconazole and Candida spp. were introduced after exhaustive epidemiological and clinical correlation (12, 14). These breakpoints (susceptible, ≤8 μg/ml; susceptible dose-dependent, 16 to 32 μg/ml; resistant, ≥64 μg/ml) have been validated in numerous studies over time (12, 14).

Currently the breakpoints are established for MICs read at 48 h. However, it is well known that MICs can often be read as early as 24 h (13, 16), quality control limits for reading at 24 h have been developed (3), and research has shown that early/correct antifungal therapy has a positive impact on outcomes of patients with candidemia (2, 7, 10). To provide clinicians with useful susceptibility information sooner, we sought to determine if reading the CLSI M27-A fluconazole MIC at 24 h would produce valid results when interpreted using the existing breakpoints.

Data set.

The original data set used to create the 48-h breakpoints has been described previously (12, 14). Briefly, it integrates dosing and outcome information for 316 patients with oropharyngeal and invasive disease treated with fluconazole with antifungal susceptibility testing information for the 528 isolates obtained from those patients. Antifungal susceptibility testing was performed following the M27-A2 method, and MICs were defined as the lowest concentration of drug producing a prominent growth reduction (50%). MICs were read at 24 and 48 h (11).

Twenty-four-hour MIC distribution and correlation with 48-h MICs.

Table ​Table11 shows the MIC distribution and correlation for the 24- and 48-h MICs. As shown in Table ​Table1,1, in general, the majority of 24-h MICs tend to be 1 or 2 dilutions lower than the corresponding 48-h MICs.

There is good correlation with an r² of 0.50 and with 90% of 24-h MICs being within 2 dilutions of the corresponding 48-h MIC. This corresponds to the accepted 2-dilution margin of error of the method (13). Outliers are difficult to interpret, as the trailing phenomenon could be playing a role (9). However, when analyzing misclassification by the CLSI M23-A methodology (3), we overall found a very major error rate of 1.8%, a major error rate of 0%, and a minor error rate of 9.7%, all rates being within the tolerances required by M23-A.

Correlation with clinical outcomes.

Table ​Table22 shows the clinical success rates (defined as resolution of Candida lesions and symptoms for oropharyngeal disease and resolution of fever and clearance of infected sites for invasive disease) of patients treated with fluconazole doses of >100 mg/day grouped by 48- and 24-h MICs for each patient-episode-isolate event. As seen in Table ​Table2,2, success rates are similar when using the M27-A2 breakpoints for MICs obtained after 48 and 24 h, with both meeting the “90/60 rule” (13). Of note, the response rates for 24-h MICs in the susceptible dose-dependent category are slightly lower than the 48-h rates. This is potentially related to the classification of borderline isolates from resistant to susceptible dose-dependent due to the 1- or 2-dilution shift. Similar results were observed when we analyzed the patients who received ≤100 mg/day of fluconazole (data not shown).

Pharmacodynamic modeling.

Human and animal data suggest that the critical pharmacodynamic parameter for fluconazole is the area under the concentration-time curve/MIC ratio (1, 8, 15). An area under the concentration-time curve/MIC ratio of >25 to 50 accurately predicts therapeutic efficacy. This correlates with success when using 400 mg/day for an isolate with a MIC of 4 to 16 μg/ml or 800 mg/day for one with a MIC of 8 to 32 μg/ml in a 70-kg adult. A surrogate of this measure can be the dose/MIC ratio (1, 8, 15) if one assumes an average weight and normal renal function for all patients in the data set. Table ​Table33 shows the success rates by dose/MIC ratios using different fluconazole doses and the actual 48- and 24-h MICs. The outcomes are similar for the two reading times, suggesting comparable relevance of the pharmacodynamic parameters.

Conclusions.

Reanalysis of the data set that was used to create the original CLSI M27-A fluconazole susceptibility breakpoints for Candida spp. shows good correlation between 48- and 24-h MICs. It also shows that applying the same breakpoints to MICs obtained at 24 h as those applied to MICs obtained at 48 h predicts therapeutic success as accurately as do the 48-h MICs. Stated differently, the 24-h MIC reading considered as a new test method shows a reasonable level of similarity with the reference 48-h method. The caveat for this is the reduced response rate seen in isolates classified as susceptible dose-dependent by the 24-h MIC. As discussed above, this may be related to the reclassification of borderline-resistant isolates in relation to the MIC shift. This is particularly relevant for the species that is most likely to be affected by this change due to its MIC distribution, Candida glabrata (12, 17). Pharmacodynamic parameters predicting therapeutic success are similar when using 48- and 24-h MIC data.

Multiple researchers have investigated 24-h fluconazole readings in different data sets, showing similar results and relevance (9, 13, 16). The same applies to other antifungal susceptibility methods such as Eucast, disk diffusion, Etest, or TREK Sensititre (4, 6, 9, 16), which in fact established 24-h breakpoints early on and have shown very good correlation with clinical outcomes and with the CLSI 48-h reference method (4, 5, 15).

Although limited in size and in the amount of clinical, pharmacological, and microbiological information, as well as by the fact that invasive disease may be underrepresented, this data set proved useful and successful in creating the original CLSI M27-A breakpoints using 48-h MICs. These breakpoints have undergone extensive subsequent validation and are widely accepted (12-14). This data set supports the use of the same breakpoints with the CLSI M27-A2 procedure read after 24 h of incubation, and as such this optional test methodology variation has been included in the latest version of the CLSI M27-A3 method. As noted above, 24-h MICs of species that typically have MICs in the higher end of the range, such as C. glabrata, or MICs near the resistant endpoint, should be reported with a cautionary note and/or not reported until the 48-h MIC is available. Further validation of this concept in other data sets is warranted. Having the interpretative breakpoint at 24 h will provide earlier clinically useful information to clinicians treating patients with invasive candidiasis.

Footnotes

REFERENCES