Time to Positivity Facilitates an Early Differential Diagnosis of Candida tropicalis from Other Candida species

Introduction

C. albicans, C. glabrata, C. tropicalis and C. parapsilosis are four major pathogens causing candidemia worldwide.^1,2 Invasive infections caused by these species have been increasingly seen in patients with immunodeficiency, organ transplantation, cancer patients with chemotherapies, and long-term antibiotics or glucocorticoid treatments.³ Although C. albicans remains the most common pathogen in candidemia, non-C. albicans species are remarkably increasing as well, accounting for more than 50% of invasive candidiasis in multiple regions.^4–6 C. tropicalis is one of the invasive candidiasis notably increasing in recent years and has a sharp spike, especially in Latin America and part of Asian Pacific countries including India and China.^4,7 In addition, the resistance of C. tropicalis to azoles has deteriorated year by year as well.^8,9 It is reported from a Chinese five-year surveillance on invasive candidiasis that the resistance rate of C. tropicalis to fluconazole has increased the prominence from less than 8% in 2010 to over 22% in 2014.¹⁰ It is noteworthy that invasive infection of C. tropicalis has more serious adverse consequences and relatively higher mortality compared with other Candida species,^11,12 and its clinical outcomes largely rely on early effective interventions including susceptible antifungal treatments.^5,13 However, identification of C. tropicalis takes a long time. Therefore, it is extremely important to identify potential pathogens and initiate immediate empiric therapy in patients with candidemia.¹⁴

Time to positivity (TTP) provided by automatic blood culture machines can be used as powerful evidence to determine microbial species,¹⁵ resistance,¹⁶ distinguish real pathogens and pollutants,¹⁷ predict sources and prognosis of bacteremia.¹⁸ Some studies have previously reported that TTP of C. glabrata is longer compared with other Candida species in patients with candidiasis in BacT/Alert¹⁹ and BACTEC 9240 systems.²⁰ Mounting evidence has shown that C. glabrata tend to grow in anaerobic blood culture vials particularly,^19,20 and an exclusive or an earlier growth in anaerobic vials is useful to predict C. glabrata.²¹ However, whether TTP can be applied as a presumptive diagnosis for C. tropicalis is rarely reported.

The purpose of this study was to investigate the diagnostic value of TPP in making an early differential diagnosis of C. tropicalis in patients with candidemia.

Materials and Methods

Retrospective Clinical Study

All patients with bloodstream infections were recorded in the clinical system from the Department of Microbiology Laboratory, the First Affiliated Hospital of Chengdu Medical College from January 1, 2015 to December 31, 2020. After exclusion of bacteremia, polymicrobial, filamentous fungemia, Cryptococcus, no recordings of TTPs, and duplicated strains, only patients with monomicrobial candidemia were enrolled in this retrospective study. Both their demographic information and TTPs were collected for further analysis. The blood culture vials used in the retrospective clinical study were Fastidious Antimicrobial Neutralization Media Aerobic & Fastidious Antimicrobial Neutralization Media Anaerobic (bioMérieux, France, containing activated carbon adsorbent). Around 8 mL −10 mL of blood was inoculated into each vial according to the manufacturer’s instruction, but the exact amount was not recorded for each patient. All blood cultures were performed by BacT/Alert 3D120 system (bioMérieux, France). TTPs were recorded as time intervals between the beginning to an automatic alert signal appeared (indicating growth of organism) during the blood cultures. The shorter one was deemed as the TTP if both aerobic and anaerobic vials were positive. If a patient had persistent candidemia with multiple positive blood cultures, TTPs were recorded only for the first round.

Simulated Blood Cultures

Multiple impactors can affect TTP, including inoculation volume and transportation time. In order to test the accuracy of data obtained from the retrospective analysis, another 314 confirmed Candida spp. strains were applied for simulated blood cultures as an internal control. All these strains were isolated from the Department of Microbiology Laboratory of the First Affiliated Hospital of Chengdu Medical College during the fourth quarter of 2021 (Table S1). Those Candida spp. strains isolated from various clinical specimens, such as blood, urine, abscess, ascites, etc. They were inoculated into blood culture vials with an aerobic and anaerobic environment, respectively. The names of the vials were Fastidious Antimicrobial Neutralization Media PLUS Aerobic/Anaerobic (bioMérieux, France) containing adsorbent of resin. All the performances were followed according to a protocol from Yarbrough et al.²² Briefly, 1000 colony-forming unit (CFU) was used as a final inoculation dose for each vial. Vials for blood cultures were incubated in BacT/Alert 3D120 system. Cultures were stopped until positive growths were detected. The maximum culture time was 120 hours. TTPs were recorded as the same method before. Vials with positive results were consequently sub-cultured to a chocolate agar plate and a sabouraud agar plate, respectively, for further validation of Candida colonies by identification of the pure growth and expected colony morphologies. The recovered growth colonies in the sub-cultures were confirmed by matrix-assisted laser desorption ionization time of flight (MALDI-TOF) mass spectrometry (bioMérieux, France).

Antifungal Susceptibility Testing

Susceptibility tests of Candida spp. to antifungal antibiotics were performed on ATB ™ Fungus 3 stripe (bioMérieux, France). Minimum inhibitory concentrations (MICs) for all antifungal drugs were recorded after 24 hours of incubation. The MIC breakpoints for Candida spp. were referred to the Clinical and Laboratory Standards Institute (CLSI).^23,24 C. albicans ATCC 24433 and C. parapsilosis ATCC 22019 were used for quality controls. All results were tested within determined ranges.

Statistical Analysis

All data analyses were performed by GraphPad Prism 5. TTPs between C. tropicalis and other Candida species were compared by Student’s t-test. Proportions were compared by X² or Fisher’s exact test. The diagnostic values and cut-off values of TTPs for C. tropicalis were evaluated by the receiver operating characteristic (ROC) curve. p<0.05 was considered to be statistically significant.

Results

Demographic Features of 136 Patients with Candidemia in the Retrospective Study

A total of 5031 episodes of bloodstream infection were selected at the beginning of the retrospective study. After excluding bacteremia, polymicrobial, filamentous fungemia, Cryptococcus, no recordings of TTPs, and duplicated strains, a total of 136 episodes of monomicrobial candidemia were finally enrolled in this study (Figure 1). It was shown that C. albicans (n = 45; 33.09%), C. parapsilosis (n = 32; 23.53%), C. glabrata (n = 32; 23.53%), and C. tropicalis (n = 14; 10.29%) were four major pathogens in patients with monomicrobial candidemia.

Figure 1 Flow chart showing the monomicrobial candidemia selection process. Abbreviation: TTP, time to positivity.

The demographic features of 136 episodes of monomicrobial candidemia are listed in Table 1. Patients with urinary tract infection, renal insufficiency, and antibiotic use were significantly different in proportion of C. albicans, C. glabrata, C. tropicalis and C. parapsilosis bloodstream infection (p < 0.05). The results demonstrated that C. glabrata bloodstream infection was more likely to occur in patients with urinary tract infection (73.33%) or renal insufficiency (80.00%). Compared with other Candida species, patients with candidemia caused by C. albicans were less treated with broad-spectrum antibiotics (29.55%).

Table 1 Demographic Features of 136 Patients with Candidemia

Diagnostic Value of TTP for C. tropicalis

As shown in Table 2, the mean TTP of C. tropicalis was 21.26 ± 3.20 h, which was the shortest among Candida species (C. albicans: 34.82 ± 15.55 h, C. glabrata: 66.84 ± 30.87 h, C. parapsilosis: 33.44 ± 8.32 h, C. guilliermondii: 31.00 ± 7.65 h, C. sake: 59.75 ± 19.55, and C. krusei: 30.75 ± 7.76 h). TTPs were statistically different between each type of Candida species (shown in Figure 2A, ** p<0.01, ***p<0.001).

Table 2 TTPs of Different Candida species Included in the Retrospective Study

Figure 2 (A) Scatter plot of time to positivity (TTP) sorted by different Candida species, **p < 0.01, ***p < 0.001. (B) Receiver operating characteristic curves of time to positivity as a diagnostic test for C. tropicalis in patients with candidemia.

The diagnostic value of TTP for C. tropicalis in candidemia and its optimal cut-off values were determined by the receiver operating characteristic (ROC) curve (Figure 2B). Area under the ROC curve (AUC) demonstrated that TTP had a strong diagnostic value to identify C. tropicalis from other Candida species (AUC 0.8896 ± 0.030; 95% confidence interval 0.831–0.949). Based on the results of AUC and ROC curve, TTP < 25.50 h was the best optimal cut-off value to identify C. tropicalis from other Candida species in candidemia. Accordingly, when predicting C. tropicalis with a TTP of <25.50 h, the sensitivity and specificity were 92.86% and 77.87%, respectively; the positive and negative predictive values were 34.15% and 98.96%, respectively.

To verify the accuracy of the retrospective data, another 314 Candida strains were applied as an internal control to exclude other impactors, such as inoculation volume, transportation time, etc. The information regarding the age, gender, sample source, and susceptibility to antifungal agents were shown in Table S1. Strains collected for simulated blood cultures comprised 161 C. albicans (51.3%), 53 C. glabrata (16.9%), 33 C. parapsilosis (10.5%), 51 C. tropicalis (16.2%), and 16 other Candida species. All the results were consistent with the data obtained from our retrospective study (Table S2 and Figure S1). The mean of TTP for C. tropicalis was the shortest compared with other Candida species (18.32 ± 1.70 versus 27.35 ± 6.34, p < 0.0001). The AUC was 0.933 ±0.015, suggesting that TTP had a strong diagnostic value for C. tropicalis (p<0.0001) in simulated blood cultures. According to characteristics of ROC and AUC, the recommended cut-off value of C. tropicalis was <19.99 h, which generated sensitivity and specificity of 92.16% and 92.02%, respectively; the positive and negative predictive values were 69.12% and 98.37%, respectively.

Even though TTPs have been widely used for diagnosing Candida species, there are still some controversial issues. One current controversy is whether different concentrations of Candida species in the blood may have effects on TTP. Thus, we determined the effect of concentration of initial inocula of Candida species on TTP. Twenty Candida strains (selected from the above 314 strains randomly, including 8 C. albicans, 5 C. glabrata, 4 C. tropicalis and 3 C. parapsilosis) were inoculated with different concentrations (1000 CFU/vial, 5000 CFU/vial and 25000 CFU/vial) into vials (including anaerobic and aerobic vials), then the vials were incubated in BacT/Alert 3D120 system. It was found that TTPs tended to be shorter in vials with a higher concentration of initial inocula compared with lower one, but no statistical difference was found in each Candida species by comparisons among three different concentrations (0.06 ≤ p ≤ 0.72, Table S3).

Finally, the relationships were further tested between TTPs and susceptibility of Candida species to common antifungal drugs (amphotericin B, fluconazole, 5-fluorocytosine, and voriconazole). The results showed TTPs were not significantly different between resistant and susceptible strains to those four different antifungal agents (p > 0.05), suggesting that TTPs may not be potential to predict the susceptibility or resistance of each Candida species to amphotericin B, 5-fluorocytosine, fluconazole and voriconazole (shown in Table S4).

Discussion

Candidemia is an extremely dangerous clinical condition and carries a high mortality.^25,26 Timely clinical interventions will prominently save patients’ lives. Therefore, an early and rapid diagnosis of potential pathogens is particularly important. Besides of blood cultures, several other methods can also be employed for the identification of candidemia.^27,28 Mannan antigen and its antibody can be used for diagnosing candidemia, it has a specificity of 90%, but its sensitivity is around 60%.²⁹ Moreover, mannan antigen and its antibody are cleared quickly in the blood and cannot be used to diagnose specific Candida species.³⁰ 1-3-β-D-glucan, a component of the cell wall of many pathogenic fungi, can be employed as an important auxiliary diagnosis of candidemia. However, since it has an extensive expression on various fungi, it has a high sensitivity but a low specificity to diagnose the specific species of Candida in candidemia.³⁰ Therefore, blood culture to isolation then to identification is still a widely accepted and commonly used method for diagnosis of specific Candida species in candidemia. However, since cultures of Candida species take a long time, it is challenging to make a prompt diagnosis of the exact species. As different Candida species have different susceptibilities to antifungal drugs, reasonable prediction of Candida species and choosing proper antifungal therapeutics can be extremely important and prominently reduce the mortality of candidemia.³¹ In this study, data from a retrospective clinical study were firstly used to investigate relationships between TTPs and Candida species in patients with candidemia. It was found that different Candida species had different TTPs. C. glabrata had the longest TTPs, with a mean of 66.84 ± 30.87 h. While C. tropicalis had the shortest TTP, with a mean of 21.26 ± 3.20 h, which was consistent with the previous study.³² In their report, TTPs of C. glabrata and C. tropicalis were 62.68 h and 22.14 h, respectively. However, another study,³³ in which TTPs were slightly different from ours, reported that TTPs of C. glabrata and C. tropicalis were 53.4 h (the longest) and 28.3 h (the second shortest), respectively, and C. krusei had the shortest TTP (23.3 h). The major reason for these differences might be due to a small sample size of C. krusei enrolled in both their and our studies (both were only four strains of C. krusei included).

TTPs might be affected by inoculation volume (volume of blood drawn into vials) and transportation time (from ward to microbiology laboratory) as well. To exclude these effects, another 314 confirmed Candida strains were performed for simulated blood cultures, TTPs were recorded and analyzed accordingly. TTPs of C. glabrata, C. parapsilosis, C. krusei, C. sake, C. albicans, and C. tropicalis were 32.57 ± 5.77h, 28.59 ± 6.13 h, 27.52 ± 5.62 h, 25.91 ± 3.47 h, 25.41 ± 5.66 h, and 18.32 ± 1.70 h, respectively. These TTPs were consistent with but significantly shorter compared to the results obtained from our retrospective study. We speculated many factors could contribute to these differences, such as concentrations of Candida spp. (colony-forming units) in patients’ blood, inoculation volume of blood, transportation time (from ward to microbiological laboratory), and different vials (vials used for our retrospective study were Fastidious Antimicrobial Neutralization Media containing activated charcoal, and vials for simulated blood cultures were Fastidious Antimicrobial Neutralization Media PLUS with adsorbent of resin). All in all, TTP of C. tropicalis was the shortest in both clinical retrospective study and simulated blood cultures.

The major current debate on TTP in patients with candidemia are impactors like the natural growth rate among different Candida species and the number of pathogens in the blood.^32,34,35 The number of pathogens in the blood is closely associated with clinical features (eg, the source or the severity of infection).³⁶ To answer this important question, we systematically evaluated the relationships between TTPs and the concentration of initial inocula. Our results demonstrated that the concentration of initial inocula had no significant effect on TTPs. Therefore, we believed that TTPs were mainly associated with the natural growth rate in Candida species, but not number of pathogens in blood. In addition, we also analyzed the relationships between TTPs and resistance of Candida species. We found no significant correlations between TTPs and resistance of each Candida species.

Secondly, the ROC curve was used to determine the diagnostic value of TTP for C. tropicalis in both retrospective clinical data and simulated blood cultures. In the retrospective study, AUC was 0.8896 ± 0.030 with a 95% confidence interval of 0.831–0.949, which was higher than other reports for making a diagnosis of C. glabrata by TTPs,^19,27,28 suggesting that TTP is a powerful tool to identify C. tropicalis from other Candida species. When TTP < 25.50 h was used as a cut-off value, the sensitivity and specificity of TTP for diagnosing C. tropicalis were 92.86% and 77.87%, and the positive and negative predictive values were 34.15% and 98.96%, respectively. Both specificity and sensitivity were prominently higher than previous reports on diagnosing C. glabrata by TTPs.^27,28 Consistent with our retrospective clinical data, TTPs in simulated blood cultures had a high diagnostic value for C. tropicalis, the AUC was 0.933 ± 0.015 with a 95% confidence interval of 0.904–0.962, the optimal cut-off value was TTP < 19.99 h. Those data suggest that TTP has a higher diagnostic value for C. tropicalis compared with C. glabrata.

There are some limitations in this study as well. Firstly, TTPs might be affected by multiple factors, including blood inoculation volume, culture conditions, time intervals between sample inoculation to reception time in the lab, prior antifungal treatments, and pathogen numbers in patients’ blood, etc. However, our simulated blood cultures confirmed that these factors had little effect on the diagnostic value of TTP for C. tropicalis. Secondly, durations of candidemia were not recorded before blood specimens were collected and cultured, which may also have an impact on TTPs. Finally, this study was a single-center research with limited clinical samples, the results generated in our study might be different from other clinical centers due to different brands of vials and testing machines. Thus, multi-center studies with a large number of samples are required to further validate our results in the future.

Conclusion

In conclusion, our study shows that TTP is an ideal diagnostic tool to facilitate early identification of C. tropicalis from other Candida species in patients with candidemia, which can aid clinicians in timely treating patients with empirical antifungal therapies before final identification of the specific Candida species.