Epidemiology and antifungal resistance in invasive candidiasis

Abstract

The epidemiology of Candida infections has changed over the last two decades: The number of patients suffering from such infections has increased dramatically and the Candida species involved have become more numerous as Candida albicans is replaced as an infecting agent by various non-C. albicans species (NAC). At the same time, additional antifungal agents have become available. The different Candida species may vary in their susceptibility for these various antifungals. This draws more attention to in vitro susceptibility testing. Unfortunately, several different test methods exist that may deliver different results. Moreover, clinical breakpoints (CBP) that classify test results into susceptible, intermediate and resistant are controver- sial between CLSI and EUCAST. Therefore, clinicians should be aware that interpretations may vary with the test system being followed by the microbiological laboratory. Thus, knowledge of actual MIC values and pharmacokinetic properties of individual antifungal agents is important in delivering appropriate therapy to patients

Introduction

In 2001, McNeil et al. reported that (in the USA) "from 1980 through 1997, the annual number of deaths in which an invasive mycosis was listed on the death certificate (multiple-cause mortality) increased from 1557 to 6534" (Table 1, Figure 1) [1]. Augmentation of fungal infections had been published earlier i.e. by Beck-Sague and Jarvis [2] and Edmond et al. [3] for the USA and by Lamagni et al. [4] for England and Wales (Figure 2). There are several possible reasons for this change. An important one might be the increase in lifespan in the populations of the developed world and the age related loss of immune-competence. An increase of systemic fungal infections is probably also due to more intensive treatment schemes for hematological and oncological patients causing prolonged neutropenic phases. Finally, more effective antibacterial treatments allow patients with infections to survive longer without necessarily overcoming the underlying diseases and thus leaving them susceptible to other opportunistic infections. Eggimann et al. [5] summarized prior surgery, acute renal failure, previous yeast colonization, neutropenia, antibacterial therapy, parenteral nutrition, and central venous catheters as risk factors for invasive Candidia infections.

Table 1 Ranking of underlying causes of deaths due to infectious diseases in the united states in 1980 and 1997 [1].

Figure 1

Mortality in the united states, 1980-1997, 'due to candidiasis, and other mycoses in persons infected and persons not infected with HIV [1].

Figure 2

Annual rates of candidosis laboratory reports, by sex (England and Wales: 1990-9) [4].

Change in Epidemiology

The increase in incidence of Candida infections barely preceded the introduction of fluconazole in 1990. This azole agent combined good activity against Candida albicans with reduced toxicity as compared to i.e. polyene anti-fungals. It is orally and parentally available and has a reliable that means linear pharmacokinetic profile [6], which makes it easy to handle. Not surprisingly, fluconazole became the agent of choice for many fungal infections as well as for prophylactic purposes, at that time often being applied in rather low doses. Although there is inconclusive evidence, many experts in the field believe that it was the selective pressure exerted by this therapeutic concept that caused changes in the epidemiology [7]. While in earlier years, C. albicans was responsible most of the invasive fungal infections (Table 2) [8], gradually more and more non-C. albicans species (NAC) were found as offending agents (Table 3) [9–16]. While the data show similar tendencies in the prevalence of various Candida ssp. worldwide, considerable differences can be observed as well. However, these do not lend themselves to further interpretation since significant differences in the demographics of the patients observed seem to be obvious. This change in prevalence of various Candida spp. is nevertheless of clinical importance, since individual species vary in their susceptibility to various antifungal agents. While national and international surveillance is important to recognize trends in epidemiology it is, however, of utmost importance to gain knowledge about the local epidemiology as this information should guide the empiric therapy of patients.

Table 2 Species distribution of Candida from cases of invasive candidiasis^a [8].

Table 3 Species distribution of Candida in blood stream infections in various studies.

Susceptibility Testing

Some of the already cited and many other studies have also reported on the in-vitro susceptibility of Candida spp. For various reasons it is difficult to assess the various results. There are currently several test methods for performing these assays. Standardized methods have been published by the Clinical and Laboratory Standards Institute (CLSI) of the USA [17, 18] and by the European Committee on Antimicrobial Susceptibility Testing (EUCAST) [19]. The two broth dilution methods are not identical as methodological differences include glucose concentration, inoculum size, shape of microtitration wells (flat or round), and end-point reading (visual or spectrophotometric). However, it appears that they result in similar MIC levels for polyenes, azoles and echinocandins for identical isolates with a few noted exceptions [20–23]. This is especially true if isolates with defined resistance mechanisms are being tested [22, 23]. However, some "drug/bug" combinations seem to offer particular test problems i.e. caspofungin and C. glabrata [22]. Moreover, EUCAST (and for that matter Etest) results have a tendency for one to two dilution steps lower MIC values [20, 23]. There are also some commercially available test devices that have been tested in their performance. A high degree of correlation with the reference methods was found for the Etest by various authors [22, 23]. The percentage of strains classified as resistant in vitro by the EUCAST procedure and as susceptible in vitro by the VITEK 2 system was 2.6%, and as resistant by the CLSI method and as susceptible by the VITEK 2 was 1.6% (very major error) [24]. The difference observed for the Vitek 2 results and CLSI and EUCAST is driven by the fact that there are differences in clinical breakpoints (CBP) suggested by the two organizations. However these results indicate that although the CBP of EUCAST and CLSI are significantly different, currently only few strains are affected by this difference.

CBP are used to classify MIC results into susceptible (S), intermediate (I) (CLSI for some strange reason "susceptible dose dependent"; S-DD), and resistant (R), respectively. This has complicated to assess and compare studies were only percentages of S, I and R are published. The same is true for published MIC50/90 values, as very different MIC distributions might be behind these numbers. To illustrate this further, two very different fictitious distributions resulting in the same MIC50/90 values are given with Figure 3. Therefore, meaningful surveillance data should be published as MIC distributions. That allows for evaluation of the results even at a later date when i.e. CBP had been changed as has happened in the past. To circumvent the problem with differences in CBP published by various organizations, it has been suggested to use epidemiological cut -off values to distinguish between wild type (WT) organisms without any resistance mechanisms and non-wild type (N-WT) strains with higher MIC values that are thought or known to possess a resistance mechanism [22, 25]. This would put in vitro test results on the safe side as long as the particular species is a target for the antifungal agent in question. Obviously, CBP should not divide WT distributions as this will cause arbitrary test results. Monitoring the development of N-WT strains in surveillance studies allows one to analyze the spread of resistance mechanisms.

Figure 3

Two very different fictitious MIC distributions result- ing in the same MIC50/90 values: 2 and 32 mg/L.

Resistance Mechanisms

A number of different resistance mechanisms have been described in Candida spp. Often, several of these mechanisms are combined to result in a stepwise development of clinically relevant resistance. Resistance to i.e. fluconazole can be caused i.e. by alterations in sterol biosynthesis, by mutations in the drug target enzyme, sterol 14α-demethylase, which lowers its affinity for fluconazole, by increased expression of the ERG11 gene encoding for this enzyme, or by overexpression of genes coding for membrane transport proteins of the ABC transporter (CDR1/CDR2) or the major facilitator (MDR1) superfamilies [26].

Similarly Candida isolates were found with reduced susceptibility to echinochandins that showed mutations in selected regions of fks1, the gene encoding the echinocandin target enzyme 1-3-b-D-glucan synthase [27]. In particular, mutations of the serine at position 645 and also, in some cases, at position 641 have been associated with decreased susceptibility to echinochandins [28].

It should be noted that there is at least for fluconazole a clear relation between MIC values, pharmacokinetics (expressed in serum AUC) and outcome (Figure 4) [29]. Since resistance development against i.e. echinocandines so far is very limited, it is obviously very difficult to establish such a correlation. This problem is further aggravated by the fact that current antifungals still leave much to be desired as they i.e. do not reach the cure rates of antibacterial agents for susceptible bacteria. Moreover, the described resistance mechanisms were at least in individual cases associated with clinical failure of the involved patients [30–35]. This has raised concern over the CBP as suggested by the CLSI. For a discussion of this subject see [36, 37].

Figure 4

Mortality rate stratified by tertiles and fluconazole Auc/Mic at 24 h (P = 0.09 using logistic regression controlling for time to initiation of fluconazole therapy) [29].

Resistance surveillance

Among the worldwide largest and long run surveillance systems is the ARTEMIS program. In a recent publication [9], comparative susceptibility data for fluconazole and voriconazole for more than 190,000 isolates collected from 2001 to 2007 were provided and analysis of resistance rates by year, geographic location, hospital location, and specimen type for selected species were included. The data were collected employing the CLSI disk diffusion method and are summarized in Table 4. They show that fluconazole resistance has to be expected especially in C. glabrata, krusei, guillermondii, famata, inconspicua, rugosa, norvegensis and some other rarer species. There seem to be some but no complete crossresistance with voriconazole which leaves the latter as an option in appropriate cases. A trend toward increased resistance over the most recent 3 years (2005 to 2007) was observed for voriconazole and some species with low prevalence such as C. famata (1.1% to 5.7%), C. norvegensis (0.0% to 6.9%), C. lipolytica (0.0% to 11.1%), and C. pelliculosa (14.3% to 16.7%). However, there was no trend toward increased resistance to voriconazole among the fluconazole-resistant species C. glabrata, C. krusei, C. guilliermondii, C. rugosa, and C. inconspicua.

Table 4 In vitro susceptibilities of Candida spp. to fluconazole and voriconazole as determined by CLSI disk diffusion testing^a [9].

MIC distributions for echinocandins have been published by Pfaller et al. (Table 5) [39]. The results of this study demonstrate the comparable spectrum and potency of all three available echinocandin antifungals against a large collection of clinically important Candida spp. It also highlights the fact that species such as C. parapsilosis and C. guilliermondii exhibit decreased susceptibilities to all three echinocandins. The clinical relevance of these elevated MICS currently remains doubtful.

Table 5 In vitro susceptibilities of 5,346 clinical isolates of Candida spp. to anidulafungin, caspofingin, and micafungin [38].

The data collected by the SENTRY surveillance program were published as MIC50/90 values and interpreted results according to CLSI, only [39, 40]. However, this study included most important antifungals. An excerpt for C. parapsilosis is given with Table 6. German data were published in the same way [15], not offering significant differences to the SENTRY results with the exception of a high degree of flucytosine resistance in C. krusei and C. tropicalis. Finally, susceptibility data for rare Candida isolates have been collected by Diekema et al. [41] (Table 7) and chen et al. [42] that might help to guide therapy in these cases.

Table 6 In vitro antifungal agent susceptibilities of Candida and Cryptococcus isolates collected by the SENTRY Program in 2006 to 2007 [39].

Table 7 Antifungal susceptibilities of rare Candida bloodstream isolates [41].

Conclusion

Meanwhile in vitro methods are available to assess reliably the susceptibility of fungal isolates. There are, however, considerable differences in the evaluation of the results, as CLSI and EUCAST breakpoints vary. If isolates with known resistance mechanisms that have been shown to be clinically relevant at least in individual cases shall not be categorized as susceptible, some CLSI CBP need to be reconsidered. Despite the fact that a number of new antifungals are nowadays available, clinical results of antifungal therapy leave much to be desired. Hence, optimization of empiric therapy according to the local epidemiological situation and reevaluation of the therapeutic regimen when susceptibility results become available should carefully be followed. With our expanded knowledge on pharmacokinetics of antifungal compounds, MIC data could be valuable at least when treating invasive fugal infections. More information about MICs of clinical isolates and outcome of the particular patients would be helpful to establish further and validate current CBP.