To the Editor: The epidemiology of yeast infections is evolving, and species in the Candida haemulonii complex have been identified as a cause of candidiasis (1). In 2012, C. haemulonii complex was reclassified as 2 species and 1 variety: C. haemulonii (former group I), C. duobushaemulonii (former group II) and C. haemulonii var. vulnera (1).

Despite the growing knowledge about the biology and clinical relevance of these pathogens, species-specific data comparing clinical and microbiological aspects are lacking. We describe the clinical and microbiological characteristics of patients from 5 hospitals in São Paulo, Brazil, whose cultures were positive for the C. haemulonii complex species.

During January 2010–March 2015, samples from case-patients in 5 hospitals affiliated with the University of São Paulo were cultured; samples positive for C. haemulonii were further analyzed. Clinical and epidemiologic data were retrospectively collected. Species identification of the first isolate from each patient was made by sequencing the internal transcribed spacer region of the rRNA gene (2). Sequence similarity searches were done by using BLAST (http://www.ncbi.nlm.nih.gov/blast). Antifungal susceptibility testing was performed by using the Clinical and Laboratory Standards Institute reference method for susceptibility testing of yeasts (3) for amphotericin B (AMB), fluconazole, voriconazole, caspofungin (all from Sigma, St. Louis, MO, USA), and anidulafungin (Pfizer, New York, NY, USA).

Among the 14,642 specimens that showed positive yeast cultures, 40 (0.3%) isolates from 31 patients belonged to the C. haemulonii complex. Most sample sources were bone and soft tissue samples from lower extremity chronic wounds (n = 17, 42%) and blood cultures (n = 11, 32%). Other positive sources were central venous catheter (CVC) tips (n = 3), toenail scrapings (n = 3), vaginal discharge (n = 2), bile (n = 1), peritoneal fluid (n = 1), pleural effusion (n = 1), and purulent fluid from the mediastinum (n = 1).

Molecular identification characterized 14 isolates as C. haemulonii (2 alleles), 8 as C. haemulonii var. vulnera, and 9 as C. duobushaemulonii (Technical Appendix Table 1). Clinical and microbiological features of the 31 patients who tested positive are summarized in the Table. Diabetes mellitus was found substantially more frequently among patients with C. duobushaemulonii (66% vs. 25%–28% for the other 2 species), but rates for other underlying conditions were similar for all 3 species.

Susceptibility testing results varied by drug and species (Table). C. duobushaemulonii showed higher MICs for AMB than C. haemulonii and C. haemulonii var. vulnera , but all isolates showed high MICs for fluconazole and voriconazole. Conversely, MICs were low for caspofungin and anidulafungin. However, 1 isolate of C. duobushaemulonii showed high MICs of 8 μg/mL for caspofungin and 0.5 μg/mL for anidulafungin.

Of the 31 patients investigated, 11 had chronically infected wounds of lower extremities with positive surgically collected bone or soft tissue cultures, or both (Table). Samples for 4 of those patients had positive cultures for C. haemulonii, 3 for C. haemulonii var. vulnera, and 4 for C. duobushaemulonii. In most patients (n = 9, 82%), samples showed polymicrobial growth; Staphylococcus spp. (n = 7) were the most common concomitant microorganisms. All patients were treated by surgical debridement.

Samples from 8 (25%) of the 31 patients were positive for candidemia; 7 had C. haemulonii (3 var. vulvera) and 1 C. duobushaemulonii (Technical Appendix Table 2). Five (62%) patients had received antimicrobial drugs before the infection. Drug therapy failed in 5 (62%) that had positive cultures during deoxycholate AMB (n = 4) or fluconazole (n = 1) therapy. Among the 7 patients with CVC-associated candidemia, 4 had the CVC removed; 3 of those survived. The 30-day all-cause mortality rate was 50%.

Our study showed a prevalence of 0.3% C. haemulonii among yeast isolates, which was much higher than previously reported (4). Older commercial methods are unable to correctly identify C. haemulonii species, contributing to this underestimation (4). More closely related species such as C. auris, mainly found in South Africa, Asia, and the Middle East, have been misidentified as C. haemulonii and C. famata by using older systems. Thus, matrix-assisted laser desorption/ionization–time of flight mass spectrometry and internal transcribed spacer rRNA sequencing are necessary to provide the correct identification (5–7).

The data we document suggest that patients with diabetes mellitus are more likely to have positive cultures for C. duobushaemulonii than for the 2 C. haemulonii species. Moreover, C. duobushaemulonii isolates have higher AMB MICs than the C. haemulonii species. As previously reported (8), echinocandins showed better in vitro activity than azole compounds.

In summary, we demonstrated that C. haemulonii species complex are critical pathogens of chronic lower extremity wounds and that fungemia by such species remains a rare event. The 30–day all-cause mortality rate among patients with candidemia was 50%, lower than previously reported in our institution (9) and other centers in Brazil (10). We believe that in cases of candidemia by C. haemulonii spp. that 1) empirical use of AMB or azole compounds should be avoided; 2) removal of CVC should be performed; and 3) antifungal susceptibility testing should be done to guide antifungal therapy.

Top

João Nobrega de Almeida , João Guilherme Pontes Lima Assy, Anna S. Levin, Gilda M.B. Del Negro, Mauro C. Giudice, Marcela Pullice Tringoni, Danilo Yamamoto Thomaz, Adriana Lopes Motta, Edson Abdala, Ligia Camara Pierroti, Tania Strabelli, Ana Lucia Munhoz, Flávia Rossi, and Gil Benard

Author affiliations: Universidade de São Paulo, São Paulo, Brazil

Related Links

• More Letter Articles