Are In Vitro Susceptibilities to Azole Antifungals Predictive of Clinical Outcome in the Treatment of Candidemia?

ANTIFUNGAL SUSCEPTIBILITY TESTING: CLINICAL BREAKPOINTS VERSUS EPIDEMIOLOGIC CUT-OFF VALUES

Clinical breakpoints (CBPs) should differentiate isolates for which there is a high likelihood of treatment success from those for which treatment is more likely to fail. A myriad of data are used to determine CBPs, including pharmacokinetic-pharmacodynamic relationships, clinical outcome data, and wild-type MIC distributions (3). In 2012, the Clinical and Laboratory Standards Institute (CLSI) updated CBPs for Candida species and fluconazole and voriconazole (CLSI document M27-S4) (4). Voriconazole CBPs were further updated in 2017 (CLSI document M60) (5). The European Committee on Antimicrobial Susceptibility Testing (EUCAST) has also defined CBPs for Candida species and azole antifungals, which largely align with those of the CLSI (3, 6). Epidemiologic cutoff values (ECVs) can be used to distinguish wild-type strains (i.e., those without resistance mutations) from non-wild-type strains. ECVs are determined by evaluation of species-specific MIC distributions. The ECV is set at the upper limit of the wild-type distribution, such that an organism with an MIC above the ECV is presumed to express resistance mutations that may impact treatment response. Thus, although not a direct predictor of clinical efficacy, the ECV can provide useful information about the isolate, especially when a CBP has not been established (3, 7). Further information regarding ECVs, as well as an example distribution, are available in a review by Eschenauer and Carver (3). Table 1 provides an overview of established CBPs and ECVs (4, 5, 6, 8).

FLUCONAZOLE SUSCEPTIBILITY AND TREATMENT SUCCESS

In vitro and in vivo models have demonstrated that fluconazole exhibits fungistatic activity against Candida species. In neutropenic murine models of candidemia and candidiasis due to C. albicans, there is generally a good correlation between fluconazole dose, MIC, and outcome (9). For example, among C. albicans isolates with fluconazole MICs of 0.5 to 32 mg/liter, a fluconazole free-drug 24 h AUC to MIC ratio (fAUC/MIC) of 12 to 25 is associated with efficacy. Given the low degree of fluconazole protein binding, total drug exposure is essentially equivalent to free drug exposure (9). In healthy adults with a normal body weight and renal function, the fluconazole 24-hour AUC is virtually equivalent to daily dose (3, 10). As such, dose/MIC can serve as a surrogate for AUC/MIC, and fluconazole doses of 400 to 800 mg should achieve pharmacodynamic targets for Candida MICs up to 32 mg/liter (9). Furthermore, in 1997, Rex and colleagues (11) published composite data for 519 Candida isolates (the majority of which were C. albicans), showing that clinical success for fluconazole and various forms of candidiasis (oropharyngeal, bloodstream, or visceral) was dependent on MIC and dose. For infections caused by Candida isolates with MICs of ≤8 mg/liter, success was achieved in >90% of patients, even with low (100 mg/day) doses of fluconazole (11). However, in infections due to Candida isolates with MICs of 16 to 32 mg/liter, more aggressive dosing was required to improve outcomes, with treatment success ranging from 78% in patients receiving 100 mg/day to 86% in patients receiving a median dose of 400 mg/day. In patients infected with isolates with MICs of ≥64 mg/liter, outcomes were poor (<60% success) (11). These findings provided support for the initial selection of CBPs in 1997 by the NCCLS (now CLSI) for fluconazole and Candida species, which designated isolates with MICs of ≤8 mg/liter as susceptible, 16 to 32 mg/liter as susceptible dose-dependent (SDD, noting that doses of ≥400 mg/day should be utilized), and ≥64 as resistant (12).

These initial breakpoints came under scrutiny for several reasons. First, ∼80% of the outcomes data were obtained from patients with mucosal rather than invasive disease (11). Although the association of an increasing fluconazole dose being necessary to successfully treat recurrent oropharyngeal candidiasis due to C. albicans with progressively increasing fluconazole MICs in AIDS patients was well known, the relationship between MIC, fluconazole dose, and clinical outcome in invasive infections such as candidemia was less evident (12). Second, >70% of the clinical outcomes data were derived from infections due to isolates with MICs of ≤4 mg/liter, limiting confidence in the efficacy of fluconazole in the treatment of infections caused by isolates with elevated MICs, especially those in the SDD category. Finally, the in vitro, in vivo, and clinical data were largely comprised of C. albicans, leaving uncertainty regarding the utility of the breakpoints in treating infections caused by Candida species other than C. albicans; in particular, those caused by C. glabrata, which accounted for only 13% of the Rex et al. data set (11).

In an attempt to validate these initial breakpoints, several studies categorized outcomes by fluconazole susceptibility, as reviewed in reference 12. While these studies allow more robust examination of patients with invasive infections, they continued to suffer from limitations similar to those of the Rex et al. data set, that is, they included small (<15%) numbers of isolates with MICs in the SDD category, they grouped all Candida species together (thus precluding analysis of individual species, such as C. glabrata, for example), and they included patients treated with a wide range of doses (12). Similarly, to explore the role of azole pharmacodynamics, several investigators have investigated correlations between AUC/MIC (or dose/MIC) and outcome in patients with candidemia (Table 2) (10, 13–20). These studies were unable to identify a definitive AUC/MIC or dose/MIC threshold that correlates with treatment success for all patients. Notably, few studies used adjusted analyses to account for differences in severity of illness (18–20). This is an important distinction, as candidemia generally occurs in patients with multiple comorbidities that can significantly influence patient outcomes (21). In a study by Brosh-Nissimov and colleagues, which used a Cox proportional hazards model applied to the entire cohort, AUC/MIC thresholds were not associated with 30-day mortality. Furthermore, use of a classification and regression tree (CART) analysis was unable to determine a significant AUC/MIC or dose/MIC breakpoint associated with mortality (18). Similarly, in a study by Fernandez-Ruiz et al., AUC/MIC thresholds were not significantly associated with fluconazole failure (30-day mortality and/or persistent candidemia) in a multivariate logistic regression model (19).

The failure of these studies to identify a consistent pharmacodynamic target perhaps suggests that a singular approach to establishing CBPs for fluconazole applicable to all Candida species may not be optimal. The argument for species-specific breakpoints is strongest for C. glabrata. C. glabrata, while in the same clade as C. albicans, is actually more closely related to Saccharomyces cerevisiae than to C. albicans. C. albicans is diploid, while C. glabrata is haploid, and the 2 species differ in many aspects of host invasion and in their interactions with the host immune system (22). While drug efflux is the most common mechanism of resistance in both species, molecular mechanisms involved in the development of resistance may differ, and some mechanisms of resistance may result in differential effects on C. albicans versus C. glabrata. Whether these differences are related to or cause the pronounced difference in susceptibility between these species (e.g., the 16-fold higher ECV of fluconazole for C. glabrata than C. albicans; Table 1) is unclear (23). In 2012, the CLSI published revised breakpoints for Candida species and fluconazole, with substantially lower breakpoints for C. albicans, C. parapsilosis, and C. tropicalis compared to those for C. glabrata, and eliminated the “susceptible” category for C. glabrata (Table 1) (4). These revisions were made to be consistent with EUCAST breakpoint recommendations, as well as to better align susceptibility thresholds with ECVs rather than basing them purely on clinical data. In fact, no relationship was found between fluconazole MICs for C. tropicalis and outcome, nor were any clinical data supporting the C. parapsilosis breakpoints provided (24). As shown in Table 2, only two studies have assessed the relationships between pharmacodynamic parameters and outcomes stratified by specific Candida species.

For infections caused by C. glabrata, 2016 IDSA treatment guidelines for candidemia recommend transitioning from an echinocandin to high-dose fluconazole in stable patients with microbiological clearance when the isolate is fluconazole-susceptible (1). This recommendation has resulted in confusion among clinicians, since no C. glabrata fluconazole MICs are designated by CLSI as susceptible (4). However, data by Eschenauer et al. suggest that isolates designated SDD (those isolates with an MIC of ≤32 mg/liter) may be treated successfully when fluconazole is aggressively dosed. In their study of 122 patients with candidemia due to C. glabrata (56, 49, and 17 isolates with MICs of ≤8 mg/liter, 16 to 32 mg/liter, and ≥64 mg/liter, respectively) that were treated with fluconazole, complete response (defined as a composite of clinical success, microbiological success, and survival) at day 14 was attained in 48% (50/105) of patients infected with an isolate with an MIC of ≤32 mg/liter compared to 24% (4/17) for patients infected with an isolate with an MIC of >32 mg/liter (P = 0.07). This 48% response rate was similar to the 52% response rate seen in patients treated with an echinocandin, although echinocandin-treated patients tended to be more ill, both at baseline and at the time of fungemia (17). Regarding fluconazole dose optimization, while a dose/MIC of >12.5 was significantly associated with improved clinical response in an unadjusted analysis (49% versus 20%; P = 0.025) (17), no significant correlations were found between fluconazole AUC, AUC/MIC, or MIC and survival in a subsequent adjusted analysis. On multivariable logistic regression, higher average fluconazole dose (odds ratio [OR], 1.006; 95% confidence interval [CI], 1.001 to 1.010; P = 0.008), average fluconazole dose of ≥400 mg (OR, 3.965; 95% CI, 1.509 to 10.418; P = 0.005), and higher fluconazole dose on day 1 of therapy (OR, 1.007; 95% CI, 1.002 to 1.011; P = 0.002) were identified as independent predictors of 28-day survival, with doses of ≥400 mg/day associated with 28-day survival of 87.7% (20). In a recent study from Korea of 65 patients with C. glabrata fungemia treated with “standard doses” (dosing not elucidated further) of fluconazole for at least 48 h, a fluconazole MIC of 32 mg/liter was independently associated with mortality. However, only 3 fluconazole-treated patients were infected with isolates with MICs of 32 mg/liter, limiting the rigor of this conclusion (25). As such, the current CLSI CBP of ≤32 mg/liter may be appropriate, if fluconazole dosing of at least 400 mg/day is utilized.

Regarding C. albicans, in the study by Brosh-Nissimov et al., in an adjusted analysis, a dose/MIC ratio of ≤400 was associated with 30-day mortality (HR, 9.72; 95% CI, 1.75 to 54.0; P = 0.009) in the subgroup of 36 patients with C. albicans fungemia. However, there were only two isolates with an MIC of >2 mg/liter (i.e., nonsusceptible) (18). A dose/MIC of >400 may be difficult to safely achieve for C. albicans with MICs in the upper range of CLSI susceptible or susceptible dose-dependent breakpoints (≤2 and 4 mg/liter, respectively) (4, 18). In accordance with these findings, a study of 217 patients with C. albicans fungemia (including 8 isolates with MICs of >2 mg/liter) who received treatment with fluconazole monotherapy (with ∼70% receiving doses of ≥400 mg/day) found an MIC of ≥2 mg/liter to be independently associated with infection-related mortality by multivariate analysis (OR, 8.2; 95% CI, 2.3 to 29.7; P = 0.001) (26). Both studies suggest that a breakpoint of ≤1 mg/liter (or perhaps ≤0.5 mg/liter) may be more appropriate for infections due to C. albicans. An analysis of >8,000 clinical isolates of C. albicans found 98.1% of isolates to have fluconazole MICs of ≤0.5 mg/liter, with 1.1% having MICs of 1 to 2 mg/liter; thus, the vast majority of infections due to C. albicans can be treated with standard fluconazole dosing (400 mg/day) (24).

Currently, there are insufficient data to propose CBPs for other Candida species. As stated above, in the absence of such data, ECVs may be used to identify isolates with possible acquired resistance that may impact treatment outcomes. Finally, given that guidelines largely resign fluconazole to stepdown therapy after initial echinocandin treatment, further research is needed to assess optimal dosing in this scenario, as the available data largely consist of patients treated with fluconazole as initial therapy (1).

NEWER TRIAZOLES