Changing trends in epidemiology and antifungal susceptibility patterns of six bloodstream Candida species isolates over a 12-year period in Kuwait

Changing trends in incidence and antifungal susceptibility patterns of six Candida species causing candidemia in Kuwait between 2006–2017 are reported. A total of 2075 isolates obtained from 1448 patients were analyzed. Identity of Candida species isolates was determined by phenotypic methods and confirmed by PCR amplification/PCR-sequencing of rDNA and/or MALDI-TOF MS. Antifungal susceptibility was determined by Etest. C. albicans accounted for 539 (37.22%) cases followed by C. parapsilosis (n = 502, 34.67%), C. tropicalis (n = 210, 14.5%), C. glabrata (n = 148, 10.22%), C. krusei (n = 27, 1.81%) and C. dubliniensis (n = 22, 1.5%). The comparative percent distribution of Candida species causing candidemia between 2006–2011 and 2012–2017 was as follows: C. albicans 41.8% and 33.1%, C. parapsilosis complex 32.01% and 37.04%, C. tropicalis 13.59% and 15.31%, and C. glabrata 8.77% and 11.51%, C. krusei 2.0% and 1.7%, and C. dubliniensis 1.75 and 1.3%, respectively. Three of 371 C. albicans isolates during 2006–2011 and five of 363 during 2012–2017 were resistant to fluconazole. Among C. parapsilosis isolates, one of 310 during 2006–2011 and 21 of 446 during 2012–2017 were resistant to this drug. Furthermore, at an epidemiologic cutoff value (ECV) of ≤0.5 μg/ml, 70.1% C. albicans isolates were wild-type for fluconazole during 2006–2011 as compared to 58.1% during 2012–2017. Likewise, at an ECV of ≤2 μg/ml, 98.0% of C. parapsilosis isolates were wild-type during 2006–2011 as compared to 93.4% during 2012–2017. Clonal spread of fluconazole-resistant C. parapsilosis in one major hospital was documented. An 8.8% shift in favor of non-albicans Candida species with concomitant increase in MICs between the two periods preludes emergence of fluconazole-resistant candidemia cases in Kuwait.

Introduction Candida species are a major cause of healthcare-associated bloodstream infection (BSI) worldwide [1,2]. They are associated with considerable infection-related morbidity and mortality, particularly in intensive care unit (ICU), where at least 50% episodes of candidemia occur [3]. While C. albicans continues to be the most frequently isolated bloodstream pathogen in most studies, there is a gradual shift towards non-albicans Candida species (C. glabrata, C. tropicalis, and C. parapsilosis), which together account for >50% cases of candidemia [1,2,4]. Considerable differences exist in the number of cases caused by individual non-albicans Candida species depending upon the geographic region, patient population, age and prior exposure to antifungal agents [1,5]. Regardless of the geographical area, C. albicans is more frequent in patients aged up to 18 years, the frequency of C. parapsilosis decreases with age and C. glabrata is more common among the elderly [1]. Studies from Northern Europe and the USA [4,6] have reported relatively a higher number of cases due to C. glabrata as compared to Spain and Brazil, where a higher number of cases were caused by C. parapsilosis [1]. On the contrary, C. tropicalis is the most common non-albicans Candida species in Asia [7]. Like many other countries, candidemia is also an important cause of BSI both in adult and pediatric patients in Kuwait [8][9][10][11][12][13][14][15]. In a previous study, we have reported species distribution and antifungal susceptibility profile of bloodstream isolates obtained during a 10-year period (1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005) [8]. Herein, we aimed to identify epidemiological changes that have occurred in the distribution and antifungal susceptibility pattern of six Candida species during a 12-year period since 2006.

Bloodstream yeast isolates and phenotypic methods for identification
Candida species isolates were grown from blood collected after obtaining informed verbal consent from patients for routine diagnosis. The isolates were referred to Mycology Reference Laboratory (MRL) as a part of routine patient care for identification and susceptibility testing. Since this retrospective study did not involve human participants, no specific ethical approval was required. Patients details were fully anonymized and results on de-identified samples are presented in this manuscript. Six Candida species/species complexes viz. C. albicans, C. parapsilosis sensu lato, C. tropicalis, C. glabrata, C. dubliniensis and C. krusei were included. A total of 1448 culture confirmed candidemia cases due to six Candida species were recorded during the 12-year study period (2006 to 2017) across Kuwait. Repeat bloodstream isolates were also obtained from several patients with persistent candidemia. The bloodstream isolates were divided into two 6-year study periods that is 2006 to 2011 and 2012 to 2017 to compare relative shift in the prevalence of Candida species. The isolates were identified by phenotypic methods which included germ tube test, colony characteristics on CHROMagar Candida and VITEK2 yeast identification system.

Antifungal susceptibility testing
The in vitro activity of the antifungal agents was determined by E-test (bioMérieux, Marcy l'Etoile, France). The test was performed in accordance with the manufacturer's instructions. Petri dishes (150 mm diameter) containing 60 ml RPMI agar supplemented with 2% glucose and buffered to pH 7.0 with MOPS were used. The inoculum was applied with cotton swabs using growth suspension prepared in 0.85% NaCl with turbidity adjusted to 0.5 McFarland standard. Plates were incubated at 35˚C and read after 24 h. Reference strains of C. albicans (ATCC 90028) and Candida parapsilosis (ATCC 22019) were used for quality control. Susceptibility breakpoints used for interpretation of susceptible, intermediate/ susceptible dosedependent and resistant strains were those recommended by the Clinical and Laboratory Standards Institute [21]. Due to lack of defined breakpoints for amphotericin B, isolates showing an MIC of �1.0/μg ml were taken as susceptible and those with MIC >1μg/ml as resistant [22]. The CLSI epidemiologic cutoff values (ECV) of fluconazole for C. albicans, C. parapsilosis and C. tropicalis were �0.5 μg/ml, �2 μg/ml and �2 μg/ml, respectively, as also proposed by Pfaller & Diekema [23].

Statistical analysis
Statistical analysis was performed by using chi-square test or Fisher's exact test as appropriate and probability levels <0.05 by the two-tailed test were considered as statistically significant. Statistical analyses were performed by using WinPepi software ver. 11.65 (PEPI for Windows, Microsoft Inc., Redmond, WA, USA).
The year-wise distribution of Candida species isolates during the study period and the annual incidence rate of candidemia are presented in Table 2. Although C. albicans was the most frequently isolated species during 2006 to 2012, it was replaced by C. parapsilosis sensu stricto in the next four years (2013 to 2016) as the most frequently isolated Candida species from candidemia patients (numbers shown in bold). The isolation frequency of C. albicans varied from a low of~29% in 2013 and 2016 to a high value of 52.2% in 2011, however, it showed a general declining trend over the study period. On the other hand, the isolation frequency of C. parapsilosis showed a declining trend during 2006 to 2011 and then an upward trend during 2012 to 2016 ( Table 2). C. tropicalis and C. glabrata were recorded as the third and fourth most frequently isolated Candida species in Kuwait, respectively, and the isolation frequency of both the species generally showed an upward trend during the study period ( Table 2). Consistent with these data, the ratio of isolation of C. albicans versus all six Candida species isolates declined from 0.46 in 2006 to 0.33 in 2017. The annual incidence rate (number of candidemia cases per 100, 000 population) showed a declining trend (from 4.45 to 3.5) in Kuwait over the study period ( Table 2). One-third of candidemia cases (n = 480) occurred among neonates and there was a >2-fold increase in isolation of C. parapsilosis from neonates in Maternity Hospital during 2012-2017 (n = 115) as compared to 2006-2011 (n = 49). C. glabrata was isolated in greater number from elderly immunocompromised patients including those with cancer, that is, 23 isolates in 2012-2017 as compared to only 4 in 2006-2011. None of the bloodstream C. glabrata isolate was identified as C. nivariensis or C. bracarensis by PCR amplification and/or PCR-sequencing of rDNA.
The antifungal susceptibility profiles of Candida species isolates obtained during the study period were determined by Etest against four antifungal drugs, viz. fluconazole, voriconazole, amphotericin B and caspofungin and the data are presented in Tables 3 and 4  fluconazole and cross-resistance to voriconazole was detected more frequently. Only eight C. albicans isolates were resistant to fluconazole and five of these were isolated during 2012-2017. Three of these isolates also showed cross-resistance to voriconazole. Although resistance to fluconazole in C. albicans increased during 2012 to 2017 compared to 2006 to 2011, it was not statistically significant (P = 0.501). However, resistance rate among C. parapsilosis sensu stricto isolates increased over time for both fluconazole and voriconazole and these differences were statistically significant (P = 0.000 and P = 0.021, respectively) ( Table 3). Resistance to fluconazole was also detected more frequently among C. glabrata isolates during 2012 to 2017 compared to 2006 to 201, however, the difference was not statistically significant (P = 0.070) ( Table 3). The MIC 50 , MIC 90 and the geometric mean of the MICs for amphotericin B and caspofungin remained nearly same for different Candida species during the study period ( Table 4). None of the bloodstream isolates of C. albicans, C. parapsilosis, C. tropicalis and C. dubliniensis were resistant to amphotericin B and geometric mean of the MICs remained stable between the two study periods (Table 4) (Tables  3 and 4). Three of these isolates also showed cross-resistance to voriconazole.
The comparative distribution of MIC values for fluconazole for four major Candida species (viz. C. albicans, C. parapsilosis, C. tropicalis and C. glabrata) isolates during the study period is presented in Table 5

Discussion
Invasive fungal infections (IFI) are regarded as the diseases of medical progress. With the changes in clinical practice, the spectrum of fungi capable of causing IFI is also changing, and so is the epidemiology of invasive Candida infections. Candida infections form a major component of healthcare-associated IFIs and are associated with considerable mortality, varying from 29% to 72% in different patient populations representing different geographic regions/ countries [24]. Like-wise, annual incidence of candidemia is also quite variable in different populations and age groups [24]. Surprisingly, the annual incidence rate (number of candidemia cases per 100,000 population) showed a declining trend (from 4.45 to 3.5) in Kuwait over the study period. One likely explanation for this decline is the rapid increase (from 2.377 million in 2006 to 4.136 million in 2017) in the total population of Kuwait particularly during 2010-2014, mostly due to influx of large number of younger (<40 years of age) and healthy expatriate workers who are less likely to have IFIs. In our previous study from Kuwait, we have reported C. albicans as the predominant species accounting for 39.5% of isolates over a 10-year period (1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005), followed by C. parapsilosis (30.0%), C. tropicalis (12.5%) and C. glabrata (5.6%) [8]. While C. albicans remained as the most frequently isolated species from candidemia patients during 2006 to 2012, it was replaced by C. parapsilosis sensu stricto during 2013 to 2016 as the most frequently isolated Candida species. Furthermore, the isolation frequency of C. tropicalis and C. glabrata also showed an upward trend during the study period. Consistent with these data, the ratio of isolation of C. albicans versus all six Candida species isolates declined nearly consistently from a high of 0.46 in 2006 to a low of 0.33 in 2017. Thus, the proportion of Candida species causing candidemia has significantly changed in favor of non-  [19]. Epidemiology of candidemia in Kuwait  Our data are consistent with world reports showing that the prevalence of non-albicans Candida species is gradually increasing and other Candida species such as C. parapsilosis, C. glabrata or C. tropicalis are emerging as more frequent cause of bloodstream infections [1,[25][26][27][28]. C. glabrata has now become the dominant species in some centers of USA, Canada and Europe [4,6] while C. parapsilosis has also become the dominant species in some centers/geographical locations [4,[28][29][30]. Additionally, many less common and newly recognized Candida species including Candida auris are being isolated with increasing frequency [31][32][33][34][35] A comparative distribution of bloodstream isolates of Candida species from the Middle East is presented in Table 6 [8,12,[36][37][38][39][40][41][42][43][44][45][46][47]. Recent studies from Qatar [43], Israel [44] and Iran [45] also indicate predominance of non-albicans Candida species as compared to C. albicans. Generally, the prevalence of C. tropicalis and C. glabrata has been reported to be �20% with one exception from Israel, where a higher rate of C. glabrata (23.7%) was reported in adult patients (Table 6) [8,12,. This shift towards non-albicans Candida species is consistent with many previous reports [1,6,24]. However, this shift is also somewhat species-specific. The largest proportional increase in C. glabrata has been observed in USA [4,29], Australia [48] and Belgium [49]. In contrast, several surveillance and multicenter studies in other countries, such as Spain [50], Latin America [26,28] and South Africa [51] have indicated the emergence of C. parapsilosis as the second most important cause of bloodstream infection after C. albicans. Consistent with many other studies from North America, Latin America, China and Europe [26,28,29,52], C. tropicalis is the third most common cause of bloodstream infection in Kuwait. This is in contrast to countries of Asian Pacific region [53] as well as India [54] and Pakistan [55] where C. tropicalis is the second most common bloodstream pathogen after C. albicans. The epidemiologic factors that might have contributed to predominance of C. tropicalis in these countries are not fully understood. Our data of antifungal susceptibility suggest that resistance is still uncommon (<5%) but it is beginning to emerge against fluconazole particularly in C. albicans and C. parapsilosis. The comparison of MIC data clearly demonstrates that there is a noticeable increase in fluconazole  (Table 5). Notably, none of our fluconazole-resistant isolate came from a neonate. The first fluconazole-resistant bloodstream isolate of C. parapsilosis was detected in April 2010 and gradually resistant strains have spread to other hospitals. In a study of 442 C. parapsilosis sensu stricto isolates, 5 of 11 (45%) fluconazole-resistant isolates had Y132F mutation in ERG11 and originated from a single hospital [56]. Four of these isolates came from blood and one from sputum. It is not known if these patients had fluconazole exposure prior to developing resistance. Since two of the patients were females, it is possible that they might have used fluconazole for vaginal symptoms. In a previous study, Grossman et al. (2015) [57] reported two mechanisms that contribute to fluconazole resistance in C. parapsilosis, that is by an amino acid substitution in ERG11 gene and by overexpression of the efflux pump MDR1, possibly due to point mutations in the MRR1 transcription factor that regulates MDR1. Although there is no clear evidence that prior exposure to fluconazole can be linked to emergence of resistance [51,58], nonetheless, reports of outbreak of candidemia caused by resistant strain following prolonged antifungal therapy have begun to emerge [59,60], thus limiting treatment of choice to amphotericin B or an echinocandin, although the latter drug is relatively less active but still therapeutically used against C. parapsilosis candidemia. Our data seem to suggest that C. parapsilosis may have a greater propensity to develop acquired resistance to azoles than C. albicans or C. tropicalis. The report highlights the need of strict compliance of hand sepsis protocols to prevent transmission of fluconazole-resistant strains of C. parapsilosis through healthcare staff. It is worth mentioning that in a recent study, 11 (52%) of 21 fluconazole-resistant bloodstream isolates came from a single hospital and belonged to a single clone (unpublished data). In most of the published studies resistance to fluconazole has been <10% with few exceptions [24]. In a study from Spain, a high rate of fluconazole resistance (27.6%) has been observed among Candida isolates obtained from hematology and oncology patients [50]. Similarly, among C. parapsilosis isolates from South Africa, only 37% were found susceptible to fluconazole and voriconazole and 44% of fluconazole-resistant isolates also showed cross-resistance to voriconazole. Unlike higher rates of fluconazole resistance among C. tropicalis isolates in China [62,63] and Australia [48], it was uncommon/rare (0.38%) in Kuwait. In India, also where C. tropicalis has emerged as the most common bloodstream pathogen (n = 382; 41.6%), resistance to fluconazole was uncommon (2.6%) and same is true in the Asia Pacific region [63]. C. glabrata is known to be intrinsically less susceptible to fluconazole, however, there was an indication that this species is also becoming less susceptible to this drug. Of 74 C. glabrata isolates tested during 2006-2011, 10.8% isolates were inhibited at MICs of 32 μg/ml or more as compared to 22.6% (n = 115) tested during 2012-2017 and resistance to fluconazole was restricted to only 12%. In Kuwait, caspofungin is in clinical use for several years. However, none of the Candida species isolates developed resistance to caspofungin, so far.
Our study has two major limitations, firstly, because of retrospective nature, clinical details of the patients including antifungal treatment received are not available, and secondly, it is possible that not all Candida species isolates from candidemia cases were referred for identification and antifungal susceptibility testing to MRL by different hospitals.
In conclusion, our data indicate that during the two study periods there is a shift of 8.8% in favor of non-albicans Candida species causing candidemia in Kuwait. Additionally, our antifungal susceptibility data suggest a gradual decrease in susceptibility to fluconazole, particularly among C. albicans and C. parapsilosis, which may be a prelude of emergence of greater number of resistant strains in future. While resistance to fluconazole and other antifungal agents is still uncommon, it calls for a continued need of surveillance and strengthening of antifungal stewardship policies to minimize acquisition of acquired resistance.

Acknowledgments
Authors are thankful to clinical microbiologists and supporting staff of different hospitals for sending us bloodstream Candida isolates and the technical support provided by the Ministry of Health, Kuwait.