Clinical and laboratory features of patients with Candida auris cultures, compared to other Candida, at a South African Hospital

Introduction: Multidrug resistant Candida auris is an emerging threat worldwide. It has been identified in Africa, however, there is minimal data available comparing C. auris to other Candida species in Africa. Methodology: Retrospective, case control study at a tertiary South African Hospital. Clinical and laboratory features of patients with positive C. auris clinical cultures from 1 January 2015 to 31 August 2018 were compared to patients who cultured C. albicans and C. glabrata. Results: Forty-five clinical cases with C. auris cultures were identified. The median age was 32 years (IQR = 26-46). The median duration of hospital stay was 64 days (IQR = 39-88) and median time from admission to diagnosis 35 days (IQR = 21-53). Indwelling devices and previous antibiotic exposure were found to be significant risk factors. All C. auris isolates were susceptible to amphotericin B and micafungin. Patients treated with amphotericin B alone, had a higher mortality (73.33%, n = 11/15) than patients treated with an echinocandin (54.55%, n = 6/11), however this was not statistically significant. All C. auris isolates were healthcare associated with 80% (n = 36/45) acquired in ICU. The 30day all-cause in-patient mortality was 42% (n = 19/45) for C. auris, 36% (n = 16/45) for C. albicans and 53% (n = 24/45) for C. glabrata. Conclusions: C. auris is an emerging multi drug resistant threat in South Africa. Improved access to echinocandins and improvement of infection prevention and control strategies are imperative to prevent further morbidity and mortality due to this pathogen.


Introduction
Candida species are commensals of the skin and mucosal surfaces in healthy individuals. However, in immunocompromised patients, these fungi can cause invasive disease [1,2]. Candida species are a common cause of health care associated bloodstream infections [3,4]. In the past, Candida albicans was the predominant cause of infection, however, there has been a recent shift to non-albicans Candida species (NAC) causing disease [5,6]. Candida auris is an emerging species causing healthcare associated infections (HAI). First identified in 2009, from the external ear canal of a patient in Japan [7], it has since been identified in several countries globally, including South Africa [8][9][10][11][12][13][14][15][16][17][18].
Previously, the commonest Candida species, causing disease in South Africa, was C. albicans followed by C. glabrata [19][20][21]. However, 2016 National institute of communicable diseases (NICD) surveillance, found that C. parapsilosis, was the commonest followed by C. albicans and C. auris ranked fourth [22]. Eleven percent of candidemia in South Africa was due to C. auris in 2016-2017; 94.2% of cases were in Gauteng province [23,24].
Length of hospitalisation prior to infection differs between species. The average number of days between hospitalization and C. auris infection was 24 days [14] and 27.5 days [15], in 2 studies, and 3.5 [25] days for C. glabrata. In intensive care unit (ICU) patients, the median length of stay prior to diagnosis was 25 days for C. auris, 14 days for C. albicans and 16.5 days for C. glabrata [12].
Detection of Beta-D-glucan (BDG) is found to be a useful adjunct for the diagnosis of deep-seated Candida infections due to C. albicans and C. glabrata. A South African study showed 71% sensitivity of BDG in C. auris candidemia [26].
There are currently no established clinical breakpoints for C. auris, but the CDC has provided tentative breakpoints [18]. C. auris is the first Candida species to be classified as multidrug resistant. Studies from multiple countries have demonstrated resistance to fluconazole [9][10][11][12]14,15,28]. The CDC, Public Health England and South African guidelines recommend echinocandins as first line treatment. However, for eye, urinary tract, central nervous system infections and in infants under two months old, amphotericin B is recommended [18,29,30]. The South African public health sector has minimal access to echinocandins and negligible access to liposomal amphotericin B.

Aim of the study
Little information is available on the clinical aspects of C. auris infection and colonisation compared to other Candida species where virulence and antifungal resistance differs. C. auris has been found to be both equally virulent as C. albicans [31] and demonstrates high minimum inhibitory concentrations (MICs) to azoles, similar to C. glabrata [28]. The aim of this study was to determine whether clinical features, risk factors and outcomes of patients who cultured C. auris differed from those who cultured C. albicans and C. glabrata.

Study population and study design
This retrospective, case control study was conducted at Charlotte Maxeke Johannesburg Academic Hospital (CMJAH), a tertiary public hospital in the Gauteng province of South Africa. A deduplicated list of C. auris isolates, during the period 1 January 2015 to 31 August 2018, was obtained from the National Health Laboratory Service (NHLS). A case of C. auris was defined as a positive culture, from any site, during the study period. The first positive specimen was included, and duplicates (cultured within the same admission) were counted as a single case. These cases were matched against patients with C. glabrata and C. albicans cultures. Cases were matched based on ward, age and site of culture. Patients with the closest age (within a 10-year range) were included.

Microbiological methods
Routine culture medium (5% sheep blood agar, chocolate agar and sabouraud dextrose agar) was used to culture C. auris. The Vitek 2® (bioMérieux, Marcyl'Étoile, France) system was used for identification and susceptibility testing of yeasts. The CDC tentative breakpoints were used to interpret susceptibility results for C. auris and the Clinical and Laboratory Standards Institute antimicrobial susceptibility testing guidelines were used for C. albicans and C. glabrata [18,32]. The Fungitell® assay (Associates of Cape Cod, Massachusetts, United States of America) was used for BDG testing and interpreted according to the manufacturer's recommendations. Microscopy of samples using standard microbiological methods were used for bacterial culture. The samples were plated out onto solid agar (including 5% blood, chocolate and MacConkey agar where appropriate). Additional media were dependent on the sample type. Identification of the bacteria was performed using Vitek® mass spectrometry matrix assisted laser desorption ionization-time of flight mass spectrometry (bioMérieux, Marcy-l'Étoile, France) or Vitek® 2 (bioMérieux, Marcy-l'Étoile, France) automated microbial identification and antibiotic susceptibility testing.

Data collection
Inpatient files were retrieved from archives. Information on demographics, clinical features, laboratory investigations within 24 hours of collection of the positive culture, risk factors, treatment and outcomes were entered into an anonymized Microsoft Excel spreadsheet.

Statistical analysis
Standard analyses were performed for the data. Microsoft Excel was used for basic analysis. Categorical variables were reported as frequencies and percentages and continuous variables were reported as median and interquartile range (IQR). Statistica TM (version 14) software was then used for further analysis. Continuous variables were analysed using the Mann-Whitney U Test and categorical variables were analysed using the Chi-Square Test. Confidence interval of 95% were used and p values < 0.05 were considered statistically significant.  [33]. • Septic shock: sepsis with circulatory and cellular/metabolic dysfunction [33]. • Hypotension: systolic blood pressure < 90mmhg

Study definitions
and diastolic blood pressure < 60mmhg.

Ethical considerations
Permission to perform the study was granted by the Human Research Ethics Committee of the University of the Witwatersrand (certificate number: M180307)

Results
From 1 January 2015 to 31 August 2018, C. auris isolates were identified from clinical specimens of 45 patients, C. albicans from 1,959 specimens and C. glabrata from 185 specimens. Patient demographics are summarized in Table 1 and site of culture in Table 2.
Isolates from more than one site were counted separately, and recorded in Table 2, resulting in higher numbers in Table 2. All C. auris infections were HAI, whereas 80% (n = 36/45) of C. albicans and C. glabrata were HAI. Of the 45 C. auris cases, 5 were paediatric (age < 12), and ages ranged from 1 month to 73 years.

Clinical features
Clinical information was available for 41 patients with C. auris and all 45 cases of C. albicans and C. glabrata. Table 3 summarizes clinical features. Fever was more common in the C. auris group than in both the other groups (p < 0.001). Twelve C. auris patients had fever and hypotension. Of these patients, 50% (n = 6/12) had a positive bacterial culture.

Laboratory features
C. auris was cultured from non-sterile sites in 26.66% of cases (n = 12/45). However, 41.67% (n = 5/12) patients concurrently cultured C. auris from a sterile site. Fifteen patients cultured C. auris from a catheter tip. Of these patients, 40% (n = 6/15) also had positive blood cultures, confirming a catheter-related candidemia.     Four of the other nine patients, did not have paired blood cultures done, and five had blood cultures that were negative for C. auris. Nine patients cultured C. auris from urine samples, of these patients, eight had an indwelling urinary catheter. Table 4 summarizes the results of laboratory investigations within 24 hours of culture collection.
The sensitivity of BDG for isolates cultured from sterile sites only, was 84.62% for C. auris, 100% for C. albicans and 81.82% for C. glabrata.
Forty-one patients had susceptibility results available for C. auris, but only 25 patients had MIC data available. All 25 isolates were resistant to fluconazole with high MICs of > 256 µg/mL; MIC 50 and MIC90 for voriconazole were 1 µg/mL and 32 µg/mL respectively. All isolates were susceptible to amphotericin B and micafungin with an MIC50 of 0.25 µg/mL and MIC90 of 1 µg/mL for amphotericin B and MIC50 of 0.06 µg/mL and MIC90 of 0.12 µg/mL for micafungin. There were 29 isolates of C. albicans with MIC results available. All isolates were susceptible to azoles with a MIC50 of 0.12 µg/mL and MIC90 of 0.25 µg/mL for fluconazole and MIC50 and MIC90 both 0.03 µg/mL for voriconazole. MIC results were available for 26 C. glabrata isolates. The MIC50 was 32 µg/mL and MIC90 was 64 µg/mL for fluconazole and 46.15% (n = 12/26) isolates were susceptible to dose dependant fluconazole. The voriconazole MIC50 was 0.25 µg/mL and MIC90 1 µg/mL. All C. glabrata isolates were susceptible to amphotericin B but MICs were not available.

Risk factors
There was an association between indwelling devices and C. auris culture, compared to patients with C. albicans and C. glabrata. Of the C. auris patients with indwelling devices, 18 patients had both haemodialysis catheter and central venous lines. Table  5 details the comparative risk factors between the 3 species.
Antimicrobial exposure data is available for 43 patients. In the 30 days prior to infection or colonisation with C. auris, 97.67% (n = 42/43) patients received antimicrobials. The percentage of patients that received specific antimicrobials are listed in Figure 2   glabrata). Of the patients that presented with community acquired C. albicans and C. glabrata, 66.67% (n = 6/9) of the C. glabrata and 55.56% (n = 5/9) of the C. albicans patients had perforated gastrointestinal organs with intra-abdominal sepsis.

Treatment
Treatment data for 43 patients with C. auris was available. Fifteen patients (34.89%) were not treated. Seven of these patients died within 1-12 days of the positive culture result. Three of these patients had candidemia, two cultured C. auris from CVC tips and two from urine.
Treatment was initiated in 65.12% (n = 28/43) of patients. The median number of days from culture identification of C. auris to initiation of treatment was 5 days (IQR = 3.25-7), in both the micafungin and amphotericin B groups. Micafungin was administered to 39.29% (n = 11/28) of patients, of which 54.55% (n = 6/11) died. Three of the 6 patients who died were initially treated with amphotericin B, then switched to micafungin. Amphotericin B was initiated in 63.64% (n = 18/28) of patients. Of the 15 patients treated with amphotericin B alone, 73.33% (n = 11/15) died. The remaining two patients were treated with fluconazole.
Treatment duration was available for only seven patients who survived. The average number of days of treatment was 16.7 days (IQR = 14-17.5). Of the treated patients who died, 11 died while still receiving treatment, with duration of therapy ranging from 1-12 days of treatment. The difference in mortality rate for C. auris patients, treated with micafungin versus amphotericin B, was not statistically significant (p = 0.32).
In the C. albicans group, 40% (n = 18/45) of patients received treatment, of which 88.89% (n = 16/18) were initiated on fluconazole. Two patients were initiated on amphotericin B then de-escalated to fluconazole. The median time from positive result to treatment initiation was 5 days (IQR = 2-5).
In the C. glabrata group, 48.89% (n = 22/45) of patients received treatment. Fluconazole was initiated in 45.45% (n = 10/22) of patients. Susceptibility data was only available for six of these patients, of which four were resistant to fluconazole, and 20% (n = 2/10) were subsequently escalated to amphotericin B. Amphotericin B was administered to 63.64% (n=14/22) of patients and one patient was switched to micafungin. The median time from positive result to commencement of treatment was four days (IQR = 0. . Of the patients treated with fluconazole, 70% (n = 7/10) died and 71.43% (n = 10/14) of patients treated with amphotericin B died.

Outcomes
From the 28 C. auris patients that were treated, 39.29% (n = 11/28) patients died prior to completion of treatment, 53.57% (n = 15/28) were successfully treated and 2 patients had incomplete outcome data.

Discussion
This study demonstrates that C. auris is indeed an emerging threat in South Africa and is associated with ICU admissions. In this cohort, mortality was 42%, which falls within the range of 28-50% in published studies [10,12,14,15].
In this study, more males than females were infected with C. auris. This is possibly due to the nature of the underlying pathology, as 33.33% (n = 15/45) of patients were admitted for traumatic injuries requiring surgery, and only one of these patients were female.
A longer length of hospital stay, as well as longer duration from admission to diagnosis was statistically significant in the C. auris group compared to the other two Candida species. The median time from admission to diagnosis of C. auris was 35 days, which was approximately a week longer than in previous studies [14,15]. Median time from admission to diagnosis of C. glabrata, in a previous study, was 3.5 days, compared to 8 days in this study [25].
Patients were not routinely screened for C. auris colonization on admission. The total number of infected patients that were colonized is unknown. C. auris was predominantly (84.44%, n=38/45) cultured from sterile sites which suggests infection rather than colonization. However, it is possible that candidaemia was missed in patients with negative blood cultures as a result of poor sensitivity of blood cultures for Candida spp. In addition, biochemical features suggest that, at the time of diagnosis, patients were systemically ill with low median albumin and haemoglobin in all 3 groups.
No specific clinical features were found to be associated with C. auris infection. Many patients had a fever, but it is difficult to conclude if the fever was due to C. auris alone, or due to the concurrent bacterial infection.
The CRP and PCT in patients that cultured C. auris alone (median CRP 61.50 mg/L and PCT 1.45 µg/L) were lower than in patients that cultured bacteria as well (median CRP 75 mg/L and PCT 3.38 µg/L). A study comparing bacterial and fungal infections, found a PCT > 5.5 µg/L was associated with bacterial infections and not candidemia [34]. Median PCT in candidemia was found to be 0.7 µg/L [35], 0.99 µg/L [36] and 0.5 µg/L [37] in 3 different studies. A raised CRP with a low positive PCT may suggest an isolated fungal infection [37,38]. BDG sensitivity was 86.67% for C. auris. BDG can be used as an adjunct to other diagnostic assays for fungal infections but cannot be used to distinguish between different Candida species [26].
The MIC for amphotericin B in our study was lower than other published studies [12,14,15] but the MIC for micafungin is similar to previous studies [10][11][12]14]. There was no resistance to amphotericin B or micafungin, but all organisms were resistant to fluconazole. Despite having a low MIC, patients treated with amphotericin B alone, had a higher mortality (73.33%, n = 11/15) than patients treated with an echinocandin (54.55%, n = 6/11). The causes of mortality in the amphotericin B group were not assessed. We can therefore not conclude, from this study, if the increased mortality was due to poor treatment response, drug toxicity or non C. auris related factors.
It is evident that all indwelling devices predispose patients to C. auris infection. However, the duration of exposure to the indwelling device was not assessed. HIV co-infection was not a significant risk factor and C. auris infection occurred in patients with a high CD4 count.
Immunosuppressant use, specifically glucocorticoids, was documented in many patients but was not a statistically significant risk factor for the development of C. auris.
Surgical procedures and antimicrobial exposure are risk factors for C. auris infections [10,11,15,25]. In this study, the number of repeat surgeries was significantly associated with C. auris infection, as was the exposure to antimicrobials.
In this study, 80% of C. auris infections were acquired in ICU. Since C. albicans and C. glabrata cases were matched for ward-type with C. auris cases, the high number of these organisms found in ICU is not a true reflection of the Candida distribution throughout the hospital.
The findings of this study are limited due to incomplete clinical data, and small sample size. In addition, isolates that were not identified by the microbiology laboratory, and released as 'yeast not Candida albicans' or 'Candida species' may have been missed C. auris isolates. Matching of C. albicans and C. glabrata patients was performed manually and may have resulted in some bias. In addition, cases should have been matched for site of infection first, with a particular focus on bloodstream infections. A control group with no candidemia was not included to compare clinical findings and risk factors, thus the data provided here may be partial. The incidence of candidaemia at this hospital was not calculated. Information regarding inter hospital transfer or recent admission to other hospitals was not collected. Appropriateness of antifungal dosing was not assessed. No data was collected to evaluate source control. There are multiple confounding factors affecting the mortality rate, especially severity of the underlying illness, which have not been addressed in this study.

Conclusions
At CMJAH, C. auris is a healthcare associated pathogen specifically associated with ICU admission, increased length of hospitalisation, broad spectrum antimicrobials and indwelling devices. Diagnosis in this setting may be assisted by non-culture based assays such as the BDG in combination with inflammatory markers such as the CRP and PCT, but may only be valuable in cases where no concomitant bacterial infection is present.
Considering the high mortality rate associated with this MDR pathogen, it is imperative that intensive infection control measures are implemented to prevent horizontal transmission of C. auris. Equally important is a commitment to robust antimicrobial stewardship in order to curb resistance and emergence of further multidrug resistant pathogens.

Authors' contributions
Amirah Parak was the principal investigator and primary author of the protocol and final manuscript. Sarah Stacey and Vindana Chibabhai conceptualised the study and assisted with study design, editing and approving of the final manuscript.