The Role of In Vitro Susceptibility Testing in the Management of Candida and Aspergillus

Abstract

At its core, the minimum expectation for susceptibility testing is to reliably identify patients whose infection is likely to respond to or fail a given antifungal agent. This is a very important premise as it has the implication to drive both the use of a particular antifungal over another or, in some cases, preclude the use of a potentially lifesaving drug.

The relative recent development of new antifungal classes and novel antifungals has prompted the parallel development and standardization of antifungal susceptibility testing (AST) in the past few decades as a tool to optimize antifungal therapy [1]. Early attempts to standardize AST were crippled by different groups working with divergent techniques and media, as well as by the relative low frequency of truly resistant organisms. It wasn’t until the early 1990s that the Clinical and Laboratory Standards Institute and the European Committee on Antimicrobial Susceptibility Testing reference methods emerged, bringing standardization and reproducibility to the field; although still significantly behind antibacterial susceptibility testing availability and standardization, AST has ultimately “come of age” [2] and is now a common tool available to clinicians worldwide for clinical management of highly complex fungal infections and to researchers to track the emergence and spread of antifungal resistance. The recent expansion of commercial and automated methods has ultimately brought a new level of sophistication to the management of patients at the bedside in near–real time. Collections of isolates bundled with clinical outcomes and pharmacokinetic/pharmacodynamic (PK/PD) data led to the creation of breakpoints that, at least for azoles, have now been validated with a decade or 2 of clinical correlation [2–4]. Finally, the availability of massive multinational isolate surveys has started to fill the gap for organism/drug combinations for which clinical correlates and PK/PD data are lacking by pointing us in the direction of epidemiological cutoff values (ECVs) to try to distinguish wild-type isolates from those that may harbor a genetic mutation and are less likely to respond to a given antifungal [5, 6].

GENOTYPIC VERSUS PHENOTYPIC RESISTANCE: THE GREAT DISCONNECT?

In vitro susceptibility testing in the clinical laboratory setting is intended to inform the clinician if an antifungal would be expected to result in effective therapy [7, 8]. Thus, in general, if a minimum inhibitory concentration (MIC) is elevated and categorized as resistance for a given bug–drug combination, one would predict treatment failure. The in vitro phenotype of an elevated MIC is commonly explained by 1 or more resistance mechanisms. These genotypic changes are typically either mutations in a drug target or epigenetic changes such as upregulation of a drug efflux pump. For the clinician, the advantage of a quantitative MIC result is the ability to decide if a drug and dosing regimen may provide the drug exposure needed for efficacy. This is typically based on predictions from PK/PD analyses [9–12]. One limitation of susceptibility testing is the relatively long time required for the assay result. This is particularly important for fungi, as the period of time between organism isolation to MIC can be several days. Given the demonstrated importance of early initiation of optimal antifungal therapy, this test performance shortcoming is less than optimal [13–16]. In fact, mortality has been observed to increase 3-fold with delay of appropriate antifungal therapy for invasive candidiasis by as little as 12 hours.

One approach to provide an earlier indication that a therapy would be expected to fail is to assess for genotypic changes responsible for reduced antifungal effect and the elevated MIC phenotype [17–20]. These nucleic acid–based assays have been developed for several of the more common resistance mechanisms due to drug target mutations. For example, investigators have developed assays to detect polymorphisms in the genes encoding the echinocandin glucan synthase enzyme in Candida and the triazole target in Aspergillus, encoded by FKS1/FKS2 and CYP51, respectively [19, 21, 22]. The assays have demonstrated the ability to detect the great majority of isolates exhibiting reduced susceptibility to the respective antifungals. Although these are not yet commercially available, academic laboratory studies suggest that results indicating the presence of a mutation linked to antifungal resistance may be possible within a few hours after organism identification. This advance could lead to earlier appropriate antifungal therapy. However, there are potential limitations even for these genotypic studies.

Most often, the identification of a defined resistance mechanism via genotypic studies correlates with an elevated MIC and subsequently a lower likelihood of clinical success with a given antifungal therapy [23–25]. However, there are factors that impact the predictive value of the genotypic result. First, there may be >1 mechanism that accounts for the resistant phenotype. For example, triazole resistance in Candida albicans has been linked to drug target mutations, upregulation of several efflux pumps, and overexpression of the drug target [26]. Even in the case of echinocandin resistance in Candida species, where the majority of the phenotype seems accounted for by mutations in the glucan synthase genes, resistant strains lacking these mutations have been identified [27]. Second, the quantitative phenotype for certain genotypic mutants can be variable. For example, mutants in both the FKS and CYP51 genes can result in MIC values that can vary up to 100-fold. Animal model PK/PD antifungal studies with these mutants demonstrate treatment efficacy against some strains for which the mutation results in only a modest MIC change [28]. Furthermore, these studies predict the ability for successful therapy with additional dose escalation for certain antifungals with a relatively wide therapeutic window [28–30]. These considerations are potentially important for patients as there are limited second-line antifungal options. For instance, the only viable option for echinocandin and triazole resistance in patients infected with Candida glabrata is a polyene. Unfortunately, several studies have demonstrated the toxicity and associated mortality linked to polyene therapy for invasive fungal infections [31, 32].

Additionally, there is growing evidence that fitness of the organism and the immune state of the host are of equal, if not greater, importance for predicting the outcome of fungal infections. The relevance of the host immune state in modulating treatment efficacy has been widely accepted. For example, searches for independent predictors of outcome in numerous patient populations with invasive fungal infections have identified host immune defects such as neutropenia or immunosuppressive therapies as conferring negative impact [33–35]. The influence of virulence changes in the organism has also been associated with a variety of resistance mechanisms. Specifically, several studies have reported a fitness cost linked to the acquired resistance mechanisms. For example, mutations in glucan synthase in C. albicans have been shown to reduce fitness in animal models [36, 37]. In both the Drosophila and mouse infection models, survival was nearly twice as great in animals infected with FKS mutants compared to wild-type. The impact of these observations in patients remains less clear. Despite these caveats, phenotypic assays have been of clear clinical relevance for clinicians and patients. The availability of more rapidly available genotypic assays that are most often congruent would be a welcome tool for physicians.

An additional issue that is not addressed by either the clinically available in vitro testing or the emerging genotypic assays is the treatment hurdle incorporated during biofilm growth. Biofilm growth typically incurs up to a 1000-fold increase in drug resistance that is not accounted for in the commonly used planktonic assays [38, 39]. Studies in both invasive aspergillosis and Candida device infections have suggested the importance of these biofilm-dependent research mechanisms on patient outcome [31, 40, 41]. Future studies should explore incorporation of phenotypic and genotypic biofilm assays in management of relevant invasive fungal infections.

CASE STUDY: ANTIFUNGAL SUSCEPTIBILITY TESTING IN CANDIDA AND DETECTION OF RESISTANCE

Antifungal resistance emergence is, in general, less common than that reported with antibacterials. However, there have been several notable examples of relatively rapid and impactful resistance emergence in the setting of antifungal therapy. Perhaps the largest historically is triazole-resistant Candida in the setting of mucosal candidiasis in patients with human immunodeficiency virus/AIDS [4, 26, 42]. In retrospect, this epidemic was the perfect storm that combined prolonged antifungal exposure with profound and persistent host immunodeficiency. In large part, this crisis was averted with effective antiretroviral therapy, which reduced patient immunosuppression and the need for prolonged triazole exposure. Over time, multiple genetic mutations have been described and are known know to account for the majority of strains that exhibit azole resistance. These mutations include upregulation of CDR- and MDR-encoded efflux pumps, alteration or upregulation of the ERG11 gene, and mutations of the ERG3 gene as well [43].

A similar but less common scenario has arisen for C. glabrata and echinocandins. Candida glabrata has emerged in North America and several other regions as the second most common Candida species in the setting of invasive candidiasis [44, 45]. The similarities to the earlier epidemic include the relative importance of prior echinocandin exposure [23, 46]. Additionally, the potential importance of the host immune system is also suggested as many patients have had underlying malignancy or organ transplantation [23, 47]. Additionally, research rapidly identified the mechanism underlying the echinocandin treatment failures [48, 49]. As with the triazole resistance scenario, there was robust debate regarding the exact in vitro susceptibility change linked to poor patient outcome [50, 51]. Differences from the prior include the relatively short period of echinocandin exposure needed for resistance compared to that associated with triazole resistance. Treatment durations as short as 7–30 days have been correlated with echinocandin resistance emergence as opposed to the often many months of therapy noted in the setting of triazole resistance. Another distinct feature of the Candida isolates that have emerged in this outbreak is the relatively high rate of resistance to both echinocandins and triazoles despite the lack of a mechanistic or physical link between the involved genes [52, 53]. Recently reported mechanistic studies may help explain this difference. Specifically, the identification of a hypermutable genotype in C. glabrata is very likely responsible for the relatively rapid genetic changes in this species following echinocandin exposures [54]. Initial reports of resistance initially emanated from tertiary care centers, presumably due to their compromised patient populations and the relatively high rate of echinocandin resistance. In some of these medical centers, resistance rates have approached 20%. Conversely, rates have been lower in other centers, presumably due to less drug use and perhaps also the rarity of testing for resistance [55]. However, the emergence of echinocandins at the first-line therapy for invasive candidiasis and the recommendation for routine testing from recent guidelines has uncovered a growing rate of resistance across the United States [31, 56]. The continued development of both phenotypic and genotypic susceptibility assays will be critical for the optimal management of infections with this increasingly important Candida species [57].

CASE STUDY: ANTIFUNGAL SUSCEPTIBILITY TESTING IN ASPERGILLUS AND DETECTION OF RESISTANCE

Azole-resistant Aspergillus was practically considered to be nonexistent. Resistance did not emerge until the 1990s, when azoles started to be used more extensively for both prophylaxis and therapy of invasive aspergillosis, and AST was more standardized. Some centers started to investigate prophylaxis breakthroughs and therapeutic failures as possible cases of resistance with elevated MICs [58].

As with Candida, the first mechanisms of resistance were found to be those related to point mutations in the CYP51A gene, interfering with ergosterol biosynthesis [59]. Depending on the specific mutations, individual isolates can be found to be single or pan–azole resistant [60]. For many years, mutations at this site were believed to be the primary drivers of azole resistance, with a prevalence of approximately 5% within Aspergillus isolates in large collections. However, despite being frequently found and described, mutations in the CYP51A gene do not account for the vast majority of azole-resistant strains. Over time, more cryptic Aspergillus species with intrinsic antifungal resistance, such as Aspergillus terreus to amphotericin B and Aspergillus ustus to azoles, have also been increasingly reported [61].

More recently, and perhaps more disturbing, cases of azole-resistant invasive aspergillosis started to be reported in patients without prior azole exposure [62]. AST was instrumental in identifying these cases with elevated MICs, and very elegant epidemiology and molecular genetics studies have pinpointed the source of these organisms: azole-related agricultural pesticides and fungicides. These agents promote the newly described TR34/L98H and TR46/Y121F/T289A mutations, which are being found both in environmental and patient isolates from Europe, Asia, and, more recently, in the United States [63–66].

PRACTICAL ADVICE AND COMMON PITFALLS IN ANTIFUNGAL SUSCEPTIBILITY TESTING

The Infectious Diseases Society of America (IDSA) guidelines for the management of invasive candidiasis mention that while antifungal susceptibility is predictable if the causative agent is identified to the species level, individual isolates may not follow this logic [56]. Therefore, AST is recommended routinely for C. glabrata against azoles and echinocandins, mentioning that there is less value in routine testing for other species. Nevertheless, we think there is value in routine susceptibility testing of all bloodstream and sterile site isolates of Candida to monitor susceptibility trends and emergence of resistance, locally and regionally. In a resource-constrained environment, susceptibility testing should focus on all bloodstream or sterile site isolates of C. glabrata and other Candida species isolates in the settings of treatment failure, breakthrough infection, or limited therapeutic options due to underlying comorbidities, adverse events, allergies, or previous exposures. While clinical correlation between fluconazole and flucytosine MICs and outcomes is robust, interpretation of MICs and making clinical decisions regarding echinocandins, second-generation triazoles, and amphotericin B is more problematic. An important point to consider for patients who are experiencing clinical response and have seemingly resistant isolates is the possibility of MIC misinterpretation due to artifacts such as 24-hour vs 48-hour reading discrepancies, trailing seen with azoles, and the “paradoxical” effect seen with echinocandins [67]. Finally, as discussed above, PK/PD, host immune status, and site penetration need to be taken into account together with drug MICs when making therapeutic decisions.

For aspergillosis, the IDSA guidelines discourage routine susceptibility testing against azoles during primary therapy, recommending it primarily for patients failing therapy or for epidemiological purposes [68]. This is further compounded by the fact that breakpoints and ECVs have not clearly shown clinical correlation. However, the general thinking is that second-generation triazole MICs <4 μg/mL can be considered amenable to treatment.

THE FUTURE: ANTIFUNGAL SUSCEPTIBILITY TESTING VERSUS EMERGING MOLECULAR METHODS

As discussed above, after many decades, AST is finally standardized and available both as a reference method and in commercial forms/automated systems that have taken it out of reference and research laboratories, and into clinical microbiology laboratories. Current clinical guidelines now recommend and incorporate AST for specific situations during the care of patients with invasive fungal infections, such as for all sterile site isolates of Candida [56] or when a resistant isolate is suspected in aspergillosis [68].

Nevertheless, clinical microbiology is moving away from traditional cultures and susceptibility testing into the realm of rapid molecular diagnostics [69–71]. As research continues to accumulate showing that time to optimal/targeted therapy has a strong correlation with patient outcomes such as mortality rates and length of stay, the probes, panels, and arrays are starting to permeate into clinical laboratories. The next 3–5 years will be characterized by punctual /targeted rapid diagnostics such as polymerase chain reaction for multiple organisms or “syndromic panels” such as those currently available for central nervous system, gastrointestinal, and respiratory infections or for bacteremia/fungemia [72, 73]. The next 5–10 years will, in turn, focus on rapid whole-genome sequencing to directly detect microorganisms in blood or other body fluids or tissues.

As the tide turns to direct genetic material detection of microorganisms for rapid diagnosis, the gap will be susceptibility testing as the organisms will not be cultured or available for traditional susceptibility testing methodologies. However, as detection systems evolve, so will the capability of resistance detection by identifying relevant mutations that are likely to convey phenotypic resistance, so that not only is the organism detected but a resistance profile is produced for the clinician based on known mutations that are likely to translate into clinical resistance. Early examples of such applications include the description of matrix-assisted laser desorption/ionization–time of flight (MALDI-TOF) signatures that not only are species specific but also correlate with resistance, such as has been shown for C. glabrata [74, 75], automated systems that detect FKS mutations in multiple Candida species [20], or detection of Aspergillus mutations directly from respiratory samples [76].

We estimate that as systems automate and gain portability, the next 10 years will bring the transition of these rapid molecular diagnostics from the research laboratory to the reference center, then the clinical microbiology laboratory, and ultimately to the point of care.

Notes

Financial support. L. O.-Z. has received grants and personal fees from Astellas, Pfizer, Scynexis, and Cidara; personal fees from Merck and Gilead; and grants from Meiji and Janssen. D. A. has received grants and other funding from Astellas and Scynexis.

Supplement sponsorship. This work is part of a supplement sponsored by grants from Astellas Pharma Global Development, Inc. and Merck & Co., Inc.

Potential conflicts of interest. Both authors: No reported conflicts of interest. Both authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

References