Retrospective analysis of the clinical characteristics of Candida auris infection cases over a 10-year period

Candida auris is an emerging multidrug-resistant fungus with a high mortality rate. The first case of Candida auris infection was reported in 2009 and since then infections have been reported in nearly 40 countries. The fungus now represents a major global public health threat. We analyzed cases from the emergence of Candida auris infections up until the end of 2019. PubMed and Web of Science databases were searched for all papers related to Candida auris infections up until 31 December 2019. We organized these data into the following categories: date of publication, patient age and gender, underlying diseases, risk factors for infection, patient mortality information, drug sensitivity information of Candida auris isolates, and genetic classification. The {chi}2 test was used to screen for factors that may affect patient mortality before logistic regression analysis was used to further assess the suspected influencing factors to determine if they represent independent factors for patient mortality. Information pertaining to 542 patients was included. There were more male patients than female patients and the mortality rate was higher in males than females. A high proportion of patients were premature babies and elderly people. The proportions of patients with underlying diseases such as diabetes, kidney disease, and ear disease were also high. 65% of patients had a history in ICU and 60% were given broad-spectrum antibiotics. Logistic regression analysis revealed that kidney disease (P<0.05) and tumors (P<0.05) are independent factors that affect mortality in Candida auris infected patients. Patients infected with echinocandin-resistant Candida auris ultimately die.

CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 22, 2020. . https://doi.org/10.1101/2020.05.19.20105817 doi: medRxiv preprint The isolates detected by various countries were classified by whole genome sequencing data 180 into South Asia strains (clade I), East Asia strains (clade II), South Africa strains (clade III), and 181 South America strains (clade IV). The South Asia strains (clade I) were the most prevalent (17 182 countries), followed by the South Africa strains (clade III) which were observed in 8 countries. 183 Only 4 and 3 countries reported the presence of the East Asia strains (clade II) and the South 184 America strains (IV), respectively. Both South Asia and South Africa strains simultaneously 185 occurred in China, Germany, UK, and Saudi Arabia. All 4 clades appeared in the US, a 186 phenomenon that may be due to large population movements. In addition, an isolate identified in 187 Iran in 2019 may represent a potential clade V strain (shown in Figure 1) (17). 188

Patient gender and age distribution 189
For the included patients, gender was mentioned for 509 patients, of which 296 (58%) were 190 male and 213 (42%) were females. The male to female ratio was around 3:2 and infection resulted 191 in significant gender differences. This is shown in Figure 2a. 192 The World Health Organization age group classification criteria (20) were used to divide 193 patients into children (0-18 years), adolescents (19-44 years), middle-aged (45-59 years), and 194 elderly (≥60 years). Figure 2b shows the age distribution of 195 cases in which age was mentioned. 195 Most patients, approximately half of the population (n=93, 48%), were elderly. Fifteen of the 196 patients were infants less than 1 month old; 10 of the 15 (2/3 of the infant population) patients were 197 premature infants. Therefore, the possibility of Candida auris infection should be considered 198 when premature infants present with infection symptoms due to unknown causes. 199 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

Risk factors that may lead to infection and infected population 212
. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 22, 2020. Drug-resistant Candida auris strains 220 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. sensitive, 45 isolates (38%) were resistant to one antifungal, 36 isolates (31%) were resistant to 236 two antifungals, 19 isolates (16%) were resistant to three antifungals, and 1 isolates (1%) was 237 resistant to five antifungals. An analysis of these data revealed that there was no statistical 238 correlation between the degree of multidrug resistance in Candida auris and patient mortality 239 (Table 3). It should be noted that sensitive ones were isolated from 5 dead patients and the only 240 sensitive isolate was from only a surviving patient. Only one isolate was resistant to 5 drugs (FLC, 241 VRC, ITC, AMB, and CAS) and this isolate was obtained from a dead patient from Malaysia. 242 Currently, echinocandin is the only first-line drug for Candida auris infection (14,15). When 243 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 22, 2020. Among the isolates reported to date, the major drug resistance loci for azoles are ERG11, 251 CDR1, and MDR1 (27,28,29); the major drug resistance loci for AMB are ERG 2, 3, 5, 6, or 11 252 (30); and the major drug resistance locus for echinocandin is FKS11 (31,32). As part of a 253 mechanistic analysis of drug resistance towards 5FC, Rhodes et al (16). sequenced the entire 254 genome of 5FC-resistant Candida auris and found an amino acid substitution in F211I in the 255 FUR1 gene. However, further studies are required to determine whether this FUR1 mutation is the 256 cause of 5FC resistance in Candida auris (16). 257

Mortality rate 258
. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 22, 2020. . https://doi.org/10.1101/2020.05.19.20105817 doi: medRxiv preprint A total of 218 cases reported whether patients died or survived upon discharge. Among the 259 latter cases, 97 died and the mortality rate was 44%. Gender information was available for 88 dead 260 patients, of which 29 out of 53 males died and the mortality rate was 53%; 14 out of 35 females 261 died and the associated mortality rate was 40%; the mortality rate for males was slightly higher. 262 Major causes of death that were mentioned included sepsis, septic shock, and multiorgan failure. 263

Influencing factor for mortality 264
Among the patients included in the analyzed cases, the underlying disease of 100 patients 265 corresponded to pre-discharge mortality and could be included in the analysis (in addition, many 266 papers compiled patients' information so that information could not be pertaining to each 267 individual patient, which could not be included in the analysis). Gender, age, underlying disease, 268 and drug resistance of isolates were used to screen for factors that affect mortality to discover 269 influencing factors for mortality. The types of underlying diseases in dead and surviving patients 270 were similar. First, the χ 2 test was used to screen for single factors that may affect patient mortality 271 by category (shown in Table 3). Among the factors screened, kidney disease had a P<0.05 and was 272 analyzed by multivariate analysis. As the number of factors with P<0.05 was low, the inclusion 273 criteria were relaxed to P<0.1 and tumor (P=0.076) was included in multivariate analysis as a 274 suspected influencing factor (19,20). 275 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 22, 2020. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 22, 2020. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 22, 2020.  The following conclusions were ultimately drawn from the multivariate analysis: 287 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 22, 2020. . In this analysis of collected literature, incomplete information was available for many 303 patients. The main reason for these gaps in information relate to the fact that Candida auris 304 isolates from these patients were not isolated during hospitalization but a few years later by public 305 health institutions and research institutions during retrospective screening of samples. 306 Furthermore, although Candida auris is difficult to culture many medical institutions still rely on 307 culture to identify microorganisms (34). In addition, Candida auris is closely related to Candida 308 haemulonii and Candida pseudohaemulonii. These three fungi are extremely close in phylogeny 309 and it is difficult to distinguish them by phenotype. Thus, traditional identification methods such 310 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 22, 2020. . https://doi.org/10.1101/2020.05.19.20105817 doi: medRxiv preprint as Vitek 2 and API 20CAUX tend to wrongly identify Candida auris as these two closest 311 relatives. 12 Although the development of matrix-assisted laser desorption/ionization time-of-flight 312 mass spectrometry (MALDI-TOF MS) (35) and sequencing (36) strategies have helped in the 313 rapid and accurate diagnosis of Candida auris, most parts of the world do not have the 314 infrastructure to carry out these techniques. In addition, it is likely that there is a large volume of 315 unpublished data pertaining to Candida auris infections and the number of infected patients may 316 be far higher than that reported in the literature. Therefore, this is an "invisible pandemic". 317 In this retrospective analysis, we found that the male to female ratio for patients was nearly 318 3:2 and mortality data revealed that the male to female ratio for mortality was close to 4:3. Similar 319 ratios have not been reported for other Candida infections (37). The proportion of Candida auris 320 patients with diabetes, kidney disease, and ear disease is higher than that reported for other 321 Therefore, otolaryngology outpatient departments should be highly vigilant for Candida auris 327 infections. Among the Candida auris patient population, 21% had kidney disease; most of the 328 latter patients had chronic kidney disease or nephrotic syndrome. More importantly, kidney 329 disease is also an independent influencing factor that affects mortality in Candida auris infection 330 patients. Therefore, nephrologists should be extremely vigilant for Candida auris infection. As the 331 high infection rate in kidney disease patients is specific to Candida auris compared with other 332 Candida species, nephrologists in many regions do not have high awareness for Candida auris 333 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 22, 2020. Candida auris infections. In addition, many Candida auris patients exhibit severe symptoms and 337 although ICUs have been a focus of attention since Candida auris was first reported, overall 338 vigilance in ICUs is still not very strong and there is a need to strengthen knowledge gaps 339 pertaining to prevention, diagnosis, and treatment knowledge. 340 FLC exhibits broad-spectrum antifungal activity, good efficacy, a low incidence of adverse 341 reactions, and appropriate plasma concentrations for long durations (38). Therefore, this drug is 342 the most frequently utilized empirical therapy in patients suspected of fungal infection. Candida 343 auris has a high FLC resistance rate and the overall drug resistance ratio has been as high as 72% 344 over the last 10 years. Thus, FLC is not recommended as an empiric drug therapy when treatment 345 is urgent and drug sensitivity results have yet to be released for patients who either have been 346 infected with Candida auris or are suspected to have been infected with Candida auris. Clinicians 347 need to develop an awareness when considering whether FLC should be used as empiric therapy in 348 patients. 349 Although the data generated by this study reveal that there is no overall statistical correlation 350 between the degree of multidrug resistance in Candida auris and patient mortality, we did identify 351 some areas for attention after an in-depth analysis. Among the isolates for which patient mortality 352 could be determined, sensitive ones were isolated from 5 dead patients and the only sensitive 353 isolate was from only a surviving patient. We analyzed the possible reasons for the dearth of 354 information pertaining to isolates causing mortality and speculated that virulence and mortality 355 rates are higher when the Candida auris isolate is sensitive (39, 40). However, because sensitive 356 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 22, 2020. .
isolates can be treated with antibiotics, clinical outcomes can be changed if the patients are 357 diagnosed early and treated in a timely manner. Because candidiasis patients do not have apparent 358 clinical characteristics at an early stage, early laboratory tests are key to reducing the mortality rate 359 (11). Most Venezuelan and US patients exhibited VRC resistance. As these two countries are in the 360 Americas, this phenomenon suggests that there is a relationship between antifungal treatment 361 habits in different regions in the world. 362 In addition, among the acquired data, only one isolate that was resistant to 5 drugs was 363 obtained and this was isolated from a Malaysian patient where the outcome was death. Therefore, 364 when multidrug resistance in Candida auris gradually evolves, there will be situations where 365 patients cannot be treated by drugs and death will result. However, because the outcome for 366 patients infected with strains exhibiting only echinocandin resistance is death, vigilance is still 367 required for instances where the fungi isolated from patients do not show severe drug resistance 368 (e.g. resistance to 5 drugs). Echinocandin is currently the recommended treatment for Candida 369 auris. However, if Candida auris gradually evolves, more and more strains will develop 370 echinocandin resistance and this may lead to an increase in the mortality rate of patients with 371 Candida auris infections. Although the β-glucan synthesis inhibitor SCY-078 (41) and other new 372 drugs have shown some promise, they have not been widely used in clinical practice. Therefore, it 373 is especially important to develop new drugs to combat high levels of echinocandin resistance. The 374 appearance of antifungal drug resistance and the associated global range will continuously 375 threaten global public health (42) and cannot be ignored. 376 There are some limitations pertaining to this study. As this is a retrospective analysis of 377 published studies, there may be some original studies where the authors believed the associated 378 conditions were not important and some of the influencing factors were not mentioned. Thus, these 379 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)   . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 22, 2020. . https://doi.org/10.1101/2020.05.19.20105817 doi: medRxiv preprint