Molecular epidemiology of azole-resistant candida species in South Africa

Abstract

Introduction With an increase in the prevalence of non-albicans Candida species associated with candidaemia and emergence of multidrug-resistant Candida species, we conducted a number of studies to address the epidemiology, genetic relatedness, and possible mechanisms of antifungal resistance in two Candida species (Candida parapsilosis and Candida auris) causing candidaemia in South Africa. Methods Study 1: C. parapsilosis From February 2009 through to August 2010, cases of candidaemia were reported through national laboratory-based surveillance from public-sector hospital neonatal units in South Africa. Bloodstream Candida isolates were submitted to the National Institute for Communicable Diseases (NICD) in Johannesburg with corresponding laboratory reports containing patient demographic data such as age, sex, hospital and hospital unit. An incident case of candidaemia was defined as a patient of any age admitted to a sentinel hospital with first isolation of any Candida species from blood culture. A neonate was defined as a patient aged ≤28 days admitted to neonatal intensive care unit. Species identification was confirmed using ITS and D1/D2 sequencing, followed by genotyping using polymorphic microsatellite assay. Antifungal susceptibility results were analyzed using CLSI guidelines. To screen for mutations that confer resistance to azole antifungals, we sequenced the ERG11 gene and transcription factor gene MRR1 of C. parapsilosis isolates. The deduced nucleotide sequences were then compared to C. parapsilosis 22019 ATCC wild-type ERG11 and CDC317 wild type MRR1 strain. Study 2: C. auris We investigated the genetic relatedness of clinical C. auris isolates from patients admitted to either public- and private-sector hospitals in SA, which were submitted to a reference laboratory from 2012 through to 2015. Patient demographics and clinical details were recorded. We performed antifungal susceptibility testing, sequencing of the hotspot 1 and 2 regions of the FKS1 and FKS2 genes for all isolates with an echinocandin MIC of ≥1 µg/mL and cluster analysis using multilocus sequence typing. To address the epidemiology of C. auris in South Africa, we reviewed data from public- and private-sector diagnostic laboratories that reported confirmed and probable cases of invasive disease and colonization for October 2012-November 2016. We defined a case as a first isolation of C. auris from any specimen from a person of any age admitted to any healthcare facility in South Africa. We defined probable cases as cases where the diagnostic laboratory had used a nonconfirmatory biochemical identification method and C. haemulonii was cultured. Results Study 1: C. parapsilosis Of 1671 cases with a viable Candida isolate, 393 (24%) occurred among neonates. Isolates from 143 neonatal cases were confirmed as C. parapsilosis sensu stricto. A large proportion of C. vii parapsilosis isolates were resistant to fluconazole (77/143; 54%) and voriconazole (20/143; 14%). Of 79 closely-related genotypes, 18 were represented by ≥2 isolates; 61 genotypes had a single isolate each. Seven clusters, comprised of 82 isolates, were identified at 5 hospitals in 3 provinces. Isolates belonging to certain clusters were significantly more likely to be fluconazole resistant: all cluster 7 isolates and the majority of cluster 4 (78%), 5 (89%) and 6 (67%) isolates (p<0.001). A total of 73 isolates with antifungal susceptibility results from a single neonatal unit were analysed for mutations. Of these, 57 (78%) were resistant, 11 (15%) susceptible-dose dependent and 5 (7%) susceptible. The most commonly identified amino acid substitution within the ERG11 gene was Y132F in 68% (39/57) of fluconazole-resistant isolates and none in susceptible isolates. Three amino acid substitutions (R405K, G583R, and A619V) and 1 nucleotide deletion at position 1331 were identified within MRR1 gene in 19 (26%) isolates. Microsatellite genotyping of 73 cases from a single neonatal unit grouped isolates into four clusters (50 isolates). Cluster 1 accounted for 23% (17/73) of all cases, cluster 2 for 22% (16/73), cluster 3 for 14% (10/73) and cluster 4 for 10% (7/73). We found an association between cluster type and fluconazole resistance (p-value=0.004). Isolates harboring the Y132F substitution were more likely to belong to a cluster than non-Y132F isolates. Study 2: C. auris Eighty five isolates were confirmed as C. auris. The median patient age was 59 years (IQR: 48- 68 years), with males accounting for 68% of cases. Specimen types included: urine (29%), blood (27%), central venous catheter tips (25%), irrigation fluid (7%), tissue (5%), respiratory tract specimens (4%) and other (3%). Ninety seven per cent of isolates were resistant to fluconazole; 7% to both fluconazole and voriconazole; 8% to both fluconazole and echinocandins (considered multidrug-resistant) and all were susceptible to amphotericin B. Of 15 randomly-selected fluconazole-resistant isolates, 14 had an isavuconazole MIC ≤1 µg/mL. No FKS mutations were detected. MLST analysis grouped isolates into two clusters: cluster 1 and cluster 2 comprising of 83 and 2 isolates respectively. To address the epidemiology of C. auris in South Africa, we analyzed 1,692 cases; 93% were from private-sector healthcare facilities, and 92% of cases from known locations were from Gauteng Province. Of cases with available data, 29% were invasive infections. The number of cases increased from 18 (October 2012-November 2013) to 861 (October 2015-November 2016). Conclusion We found that C. parapsilosis-associated candidaemia in public-sector neonatal units was caused by closely related genotypes, and there was molecular evidence of undetected outbreaks as well as intra-hospital transmission. Fluconazole resistance in C. parapsilosis strains from a single South African neonatal unit was predominantly driven by Y123F amino acid substitutions in the ERG11 gene. Azole-resistant C. auris strains circulating in South African hospitals were related by MLST but the possibility of nosocomial transmission should be explored using a more discriminatory technique such as whole genome sequencing. Our results show a large increase in C. auris cases during the study period, centered on private hospitals in Gauteng Province