ABSTRACT
The T2Candida assay is a novel, non-culture-based assay for the diagnosis of candidemia directly from whole blood. The impact of antifungals on the performance of the T2Candida assay and blood culture bottles has not been well described. In this study, the performance of the T2Candida assay was compared to that of blood culture in detecting Candida spp. in spiked blood cultures with or without the presence of antifungals. Clinical bloodstream isolates of Candida spp. were inoculated into human whole blood at low (1 to 5 cells/ml) and high (10 to 50 cells/ml) concentrations with or without the presence of caspofungin and fluconazole. Time to detection (TTD) was assessed for prepared samples using BacT/Alert FA aerobic blood culture bottles or the T2Candida assay. In the absence of antifungals, T2Candida assay sensitivity was comparable to that of blood culture at both the low inoculum and the high inoculum (95% versus 97.5% and 100% versus 100%, respectively) and the assay had an average TTD that was significantly shorter (5.1 h versus 27.2 to 30 h, respectively). Neither caspofungin nor fluconazole was observed to impact the sensitivity or TTD of the T2Candida assay, while fluconazole reduced the overall blood culture sensitivity by 7.5% to 12.5% (at the low inoculum and high inoculum, respectively) and significantly increased the TTD of Candida albicans, C. tropicalis, and C. parapsilosis by 14.8 to 67 h. Neither caspofungin nor fluconazole impacted the performance of the T2Candida assay in vitro, and the assay may be useful for the diagnosis of candidemia in patients receiving antifungal therapy.
INTRODUCTION
Bloodstream infections caused by Candida species are associated with high mortality, and early diagnosis and prompt initiation of antifungal treatment are essential to improving patient outcomes (1, 2). Blood culture is the gold standard method for the diagnosis of candidemia, despite a sensitivity of 50% and a median time to detection (TTD) ranging from 2 days to 3 days or longer depending on the Candida species (3). TTD has historically been shown to be longer in blood cultures obtained from patients on antifungals at the time of blood collection (4). While the addition of adsorbent resins or charcoal to blood culture bottles has led to a mild improvement in the TTD for some organisms, alternative non-culture-based methods are needed to overcome the inherent limitations of culture-based methods (5). Several rapid methods have been developed which can decrease the time to species identification and susceptibility reports by 24 h, including peptide nucleic acid fluorescent in situ hybridization (PNA-FISH), PCR-based methods, and phenotyping-based methods (Accelerate Diagnostics, Inc.). Due to a requirement for a high inoculum (typically >1,000 cells/ml), however, these methods are limited to testing of samples obtained from blood cultures which have been flagged as positive. As such, the time savings of these postculture diagnostics are driven by replacing the need for subculturing on media. Recently, a non-culture-based assay called the T2Candida assay gained Food and Drug Administration (FDA) approval for the diagnosis of candidemia. It has a reported clinical sensitivity of 91.5% and specificity of 99.4%, and early in vitro studies found that the presence of antifungals had no impact on the performance of the assay. Discrepant results between blood culture and the T2Candida assay have been noted in recent clinical studies, however, particularly among patients who had been receiving an antifungal at the time of testing. While the sensitivity and specificity of the T2Candida assay have been previously reported, the impact of antifungals on the performance of the T2Candida assay and of blood culture bottles containing adsorbing agents has not been well described (6, 7). With an increasing number of patients receiving antifungal therapy prior to diagnostic testing for candidemia, a better understanding of the impact on these two diagnostic methods is needed. In this study, the performance of the T2Candida assay was compared to that of blood culture containing an adsorbing agent (BacT/Alert FA) in detecting Candida spp. in spiked blood with or without the presence of antifungals.
MATERIALS AND METHODS
Candida spp. included in antifungal susceptibility testing.Clinical bloodstream isolates of Candida species were obtained from JMI Laboratories and the University of Houston College of Pharmacy fungal biobank. Prior to seeding experiments, isolates were thawed from −80°C storage and grown on Sabouraud dextrose agar (SDA) (Hardy Diagnostics, Santa Maria, CA, USA) for 24 h at 35°C. Susceptibility to caspofungin and fluconazole was assessed by broth microdilution according to CLSI M27-A3 guidelines, and 24-h MIC endpoints were interpreted using CLSI M27-S4 guidelines (Table 1) (8, 9).
Antifungal susceptibility among clinical bloodstream isolates of Candida species utilized in this studya
Antifungals.Analysis-grade caspofungin and fluconazole powder (Sigma-Aldrich) were reconstituted with dimethyl sulfoxide (DMSO) to concentrations of 8 mg/ml and 20 mg/ml, respectively. The stock solutions were aliquoted and stored at −80°C prior to the experiments.
Preparation of spiked blood samples.Fresh colonies of Candida grown on SDA were suspended in phosphate-buffered saline (PBS) to approximately 1 × 106 cells/ml. The suspensions were prepared with 10-fold serial dilutions in PBS and final CFU counts determined by plating 100 μl of the appropriate dilution onto yeast extract-peptone-dextrose (YPD) agar and counting the number of colonies present after 48 h of incubation at 35°C. Recently donated (<5 days) human whole blood (Bioreclamation LLC, Westbury, NY) was inoculated with the Candida suspensions to achieve inoculum levels of approximately 1 to 5 (low inoculum) and 10 to 50 (high inoculum) cells/ml. Caspofungin or fluconazole was then added to spiked blood samples to achieve a clinically relevant concentration of 8 μg/ml or 20 μg/ml, respectively (10).
Using aseptic techniques, BacT/Alert FA aerobic blood culture bottles which contained activated charcoal (bioMérieux, Inc., Durham, NC) (n = 2 replicates for each condition) were filled with 10 ml of spiked blood with or without the addition of antifungals and were incubated in a BacT/Alert 3D machine (bioMérieux, Inc., Durham, NC) at 35°C for up to 5 days. Simultaneously, K2 EDTA tubes (n = 2 for each condition) were filled with 4 ml of spiked blood with or without antifungals for detection and identification of Candida spp. using the T2Candida assay run on a T2Dx instrument (T2Biosystems, Lexington, MA). Negative controls (n = 2 for each antifungal) consisting of blood spiked with only fluconazole or caspofungin were prepared for each method.
Definitions.The TTD for the BacT/Alert FA aerobic blood culture bottles was defined as the time span from the time when the blood culture bottle was entered into the BacT/Alert 3D machine to when it was flagged as positive. The TTD for the T2Candida assay was defined as the time span from the time of initiation of the T2Candida automated assay run on the T2DX machine to when final results were reported.
Statistical analysis.The performance of the BacT/Alert FA aerobic blood culture bottles and T2Candida assay was assessed using descriptive statistics. Differences in the TTD of blood culture bottles based on the Candida sp., the inoculum level, and the presence of caspofungin or fluconazole were assessed using the Student t test. A P value of less than 0.05 was considered to represent statistical significance. Independent predictors of delayed TTD were assessed using multivariate stepwise, linear regression analysis.
RESULTS
Candida species isolates and antifungal susceptibility.A total of 20 strains (n = 4 strains each for C. albicans, C. glabrata, C. tropicalis, C. parapsilosis, and C. krusei) were utilized in this study (Table 1). Caspofungin MICs were within the susceptible range for all strains of C. albicans, C. tropicalis, and C. krusei (MIC = ≤0.25 μg/ml) and C. parapsilosis (MIC = ≤2 μg/ml). Those for C. glabrata were intermediate (MIC = 0.25 μg/ml) but remained below the reported epidemiological cutoff values predictive of FKS hot spot mutations conferring a high risk of clinical treatment failure. Fluconazole MICs were within the susceptible range for all strains of C. albicans, C. tropicalis, and C. parapsilosis (MIC ≤ 2 μg/ml) with the exception of one C. parapsilosis strain which was fluconazole resistant (MIC = 16 μg/ml). All strains of C. glabrata were susceptible to fluconazole (MIC ≤ 32 μg/ml) in a dose-dependent manner, and all C. krusei strains had fluconazole MICs of ≥32 μg/ml, consistent with the intrinsic fluconazole resistance of this species.
Impact of antifungals on the ability of BacT/Alert FA blood culture bottles and the T2Candida assay to detect Candida species.Of the 240 spiked blood samples tested by each method, 100% (240/240) and 99% (238/240) were successfully processed by blood culture and T2Candida assay, respectively. Mechanical error occurred in 2 of the T2Candida assay samples. At the low inoculum level (1 to 5 cells/ml), the overall sensitivities of blood culture and T2Candida in detecting Candida species growth were 97.5% (39/40) and 95% (37/39) in the absence of antifungals, 97.5% (39/40) and 95% (38/40) in the presence of caspofungin, and 85% (34/40) and 95% (38/40) in the presence of fluconazole, respectively (Table 2). At the high inoculum level (10 to 50 cells/ml), the overall sensitivity of blood culture in detecting Candida species growth was 100% in the absence of antifungals (40/40) and in the presence of caspofungin (40/40), while it was 92.5% (37/40) in the presence of fluconazole. In contrast, the sensitivity of the T2Candida assay was 100% in the absence (39/39) or presence of fluconazole (40/40) or caspofungin (40/40). Discordant results occurred in 16/238 (7%) of samples, with 10 being T2Candida positive but blood culture negative and 6 being T2Candida negative but blood culture positive. Among the T2Candida-positive but blood culture-negative samples, 9/10 contained an antifungal (fluconazole, n = 8; caspofungin, n = 1). For the 6 samples which were T2Candida negative but blood culture positive, 2 were C. glabrata samples (without antifungal = 1; with caspofungin = 1), 3 were C. krusei samples (without antifungal = 1; with caspofungin = 1; with fluconazole = 1), and 1 was a C. parapsilosis sample (with fluconazole = 1). None of the negative-control samples were reported to have detectable growth by either blood culture or T2Candida assay.
Impact of caspofungin or fluconazole on the sensitivity and performance of BacT/Alerta FA blood culture bottles and the T2Candida assay in the detection of Candida spp. at low and high inoculum levelsa
Impact of antifungals on the time to detection of BacT/Alert FA blood culture bottles.Overall, the average TTD for the T2Candida assay was 5.1 ± 1.4 h and was not impacted by Candida species, inoculum level, or the presence of antifungals, as would be expected from this automated procedure. In contrast, the average TTD for blood culture was significantly impacted by the Candida species, the inoculum level, and the presence of antifungals (Fig. 1). In the absence of antifungals, the average TTDs for all Candida spp. combined were 32 h (range, 17.2 to 62.4 h) at the low inoculum level and 27.2 h (range, 15.3 to 54.6 h) at the high inoculum level. At the low and high inoculum levels, C. tropicalis had the shortest TTDs (18.2 and 16.1 h, respectively) whereas C. glabrata had the longest TTDs (51.8 and 43.1 h, respectively). Compared to samples containing no antifungals, the average TTD for fluconazole spiked samples was significantly elevated at both the low inoculum (47.3 versus 32 h; P < 0.001) and high inoculum (41.3 versus 27.2 h; P < 0.001). At the species level, the TTD of fluconazole spiked samples was significantly longer at both the low and high inoculum levels for C. albicans (93.1 versus 25.9 h and 62.9 versus 22.1 h, respectively; P < 0.001) and C. tropicalis (36.4 versus 18.2 h and 30.9 versus 16.1 h, respectively; P < 0.001) and at the high inoculum level for C. parapsilosis (44.1 versus 32.8 h; P < 0.001). No significant difference was observed for C. parapsilosis at the low inoculum level or for C. glabrata and C. krusei at both the low and high inoculum levels. In contrast to the fluconazole results, no significant difference in the average TTD was observed overall between caspofungin-spiked samples and those without antifungals at the low inoculum level (33.2 versus 32 h, respectively; P = 0.692) or the high inoculum level (28.8 versus 27.2 h, respectively; P = 0.529). At the species level, however, a marginal but statistically significant increase in the TTD was observed in caspofungin-spiked samples of C. albicans at the low inoculum level (28.9 versus 25.9; P = 0.032) and the high inoculum level (25.9 versus 22.1 h; P = 0.015) and in C. tropicalis at the low inoculum level (20.5 versus 18.2 h; P = 0.003) and the high inoculum level (17.5 versus 16.1 h; P = 0.032) compared to those seen without antifungals. In multivariate analysis, increased TTD of blood culture was associated with fluconazole (21.9 ± 1.4 h; P < 0.0001), certain Candida spp. (C. albicans, 13.7 ± 1.5 h; C. glabrata, 26.2 ± 1.9 h; and C. parapsilosis, 16.7 ± 2 h; P < 0.0001 each), and low inoculum level (5.1 ± 1.4 h, P = 0.0005). No correlation between the MIC and TTP was found for any Candida spp. in samples spiked with antifungals or assayed without antifungals.
The time to detection of BacT/Alert FA blood culture bottles spiked with five different Candida spp. at a low inoculum (1 to 5 cells/ml) (A) and a high inoculum (10 to 50 cells/ml) (B) with or without caspofungin (8 μg/ml) or fluconazole (20 μg/ml). *, P < 0.05 (compared to no antifungal).
DISCUSSION
In this study, the performance of BacT/Alert FA aerobic blood culture bottles and the non-culture-based T2Candida assay in detecting Candida spp. in the presence and absence of antifungals was assessed in vitro using spiked blood samples. In the absence of antifungals, the sensitivity of the T2Candida assay was comparable to that of blood culture at both the low and high inoculum levels (95% versus 97.5% and 100% versus 100%, respectively) and the assay had an average TTD that was significantly shorter (5.1 h versus 27.2 to 30 h at the low and high inoculum, respectively). Neither caspofungin nor fluconazole was observed to impact sensitivity or TTD in the T2Candida assay, while fluconazole reduced overall blood culture sensitivity by 7.5% to 12.5% (at low and high inocula, respectively) and significantly increased the TTD of C. albicans, C. tropicalis, and C. parapsilosis by 14.8 to 67 h. Caspofungin had no impact on blood culture sensitivity and marginally increased the TTD of C. albicans and C. parapsilosis by 3 to 4 h.
Antifungals have been shown in several in vitro studies to reduce blood culture sensitivity by 17% to 50% and to increase the TTD of Candida spp. by 6 to 22 h, depending on the specific brand and type of blood culture media tested (11, 12). As such, new formulations of blood culture media which contain an adsorbing agent (such as charcoal, resins, or polymeric beads) that binds antimicrobials have been developed and are increasingly utilized in routine practice. Data regarding the performance of blood culture containing an adsorbing agent in the presence of antifungals are scarce, however. An in vitro study by Jekarl et al. found that fluconazole at concentrations of 1.9, 4.7, and 6.7 μg/ml (equivalent to the peak plasma levels achieved with oral doses of 100 mg, 200 mg, and 400 mg, respectively) had no impact on the TTD or the sensitivity of two types of blood culture media containing an adsorbing agent (Bactec FX Plus and BacT/Alert FA) and that the TTD was delayed by 20.5 to 27.7 h for blood culture media not containing an adsorbing agent (Bactec Mycosis IC/F and Bactec Mycosis/F Lytic) (11). In contrast, an in vitro study by Riedel et al. found that peak concentrations of the most commonly used antifungals (amphotericin B deoxycholate, caspofungin, fluconazole, and voriconazole) reduced the sensitivity of Bactec FX blood culture bottles (containing an adsorbing resin) by 18.8% and increased the TTD by 5.67 h overall (12). Compared to these previous studies, in our study the concentrations of caspofungin (8 μg/ml) and fluconazole (20 μg/ml) were increased as we aimed to more closely simulate the peak plasma concentrations obtained from intensive care unit (ICU) patients receiving standard intravenous doses of fluconazole (6 mg/kg) and caspofungin (70 mg) (10, 13). Interestingly, we found that caspofungin had no impact on the sensitivity of BacT/Alert FA bottles and increased the TTD only marginally, while a significant reduction in sensitivity and an increased TTD occurred with fluconazole. It is possible that differences in protein binding between caspofungin (highly protein bound; 97%) and fluconazole (low protein binding; 12%%) could possibly account for these findings, as the unbound concentration of caspofungin would be expected to be lower than that of fluconazole and/or could be more readily adsorbed by the charcoal contained in the blood culture media.
In contrast to blood culture, we found that clinically relevant concentrations of fluconazole or caspofungin had no impact on T2Candida assay performance, as was expected from this automated non-culture-based diagnostic. These results largely confirm the findings of an in vitro study by Neely et al. which demonstrated that T2Candida performance was not impacted by the presence of exogenous substances (including antifungals) at concentrations 5-fold higher than those expected to be observed clinically (14).
Real-world data concerning the performance of the T2Candida assay compared to blood culture remain limited, particularly among patients receiving antifungal therapy. In the clinical trial leading to FDA approval of the T2Candida assay for clinical use, discordant results between T2Candida and blood culture occurred in 8 cases (7). Among them, 2 cases were T2Candida negative but blood culture positive and occurred in oncology patients receiving antifungals prior to blood draw. Conversely, 6 cases were T2Candida positive but blood culture negative and occurred in patients who were receiving antifungals but had no clear evidence of fungal infection. More recently, Mylonakis et al. studied the performance of T2Candida and blood culture in monitoring clearance of candidemia (15). Overall, among 31 patients who had been diagnosed with candidemia by blood culture and were receiving antifungal therapy, 13 (41.9%) had at least one positive T2Candida and/or blood culture result during the 7-day surveillance period. All (7/7; 100%) positive surveillance blood cultures had an accompanying positive T2Candida result, while only 7/23 (30.4%) of T2Candida positive results had an accompanying positive blood culture. Overall, the limited clinical data suggest that the T2Candida assay may be more sensitive than blood culture in detecting candidemia in the presence of antifungals. The results of our in vitro study support these findings, as 9/10 samples found to be T2Candida positive but blood culture negative contained an antifungal (fluconazole, n = 8; caspofungin, n = 1). Whereas Mylonakis et al. found that all of their blood culture-positive samples had an accompanying positive T2Candida result, we observed 6 samples which were blood culture positive but T2Candida negative. Among these 6 samples, 2 were C. glabrata (no drug = 1; caspofungin = 1), 3 were C. krusei (no drug = 1; caspofungin = 1; fluconazole = 1), and 1 was C. parapsilosis (fluconazole = 1). All were samples inoculated with a low concentration of cells (range, 2 to 3.2 cells/ml) which approaches the reported limit of detection of the T2Candida assay. Overall, our in vitro data show that antifungals negatively impact BacT/Alert FA blood culture sensitivity whereas they have no impact on the T2Candida assay.
Several limitations to our data need to be acknowledged. First, simulated blood samples created in the laboratory do not capture the significant heterogeneity in host factors or the degree of antifungal exposure that occurs in real-world clinical settings. Second, while the concentrations of fluconazole and caspofungin evaluated in our study were higher and more likely to represent those observed in critically ill patients than those reported previously, higher concentrations may occur in some patients. Whether even higher concentrations would impact the performance of blood culture therefore warrants further investigation. Higher concentrations of antifungals (up to 5-fold higher than can be achieved in the body) have already been demonstrated to have no impact on T2Candida assay performance, however (14).
ACKNOWLEDGMENTS
This work was funded by a research grant from T2Biosystem, Inc. T2Biosystems, Inc., was not involved in the study design, data analysis, or manuscript preparation.
FOOTNOTES
- Received 20 March 2018.
- Returned for modification 9 April 2018.
- Accepted 6 June 2018.
- Accepted manuscript posted online 13 June 2018.
- Copyright © 2018 American Society for Microbiology.