Time to Positivity and Detection of Growth in Anaerobic Blood Culture Vials Predict the Presence of Candida glabrata in Candidemia: a Two-Center European Cohort Study

Abstract

This study shows the accuracy of exclusive or earlier growth in anaerobic vials to predict Candida glabrata in a large series of candidemic patients from two European hospitals using the Bactec 9240 system. Alternatively, C. glabrata can be predicted by a time to positivity cutoff value, which should be determined for each setting.

TEXT

The time to positivity (TTP) has been successfully exploited as a tool for the presumptive identification of the species or type of microorganism involved in bacteremia (1–5). In patients with candidemia, the TTP has been reported to be longer for Candida glabrata than for other Candida species in BacT/Alert (6, 7) and Bactec 9240 systems (8–10). Previous works showed that C. glabrata is particularly prone to grow in anaerobic blood culture vials (11, 12), and two clinical studies reported that an exclusive or earlier growth in anaerobic vials is useful for predicting C. glabrata (8, 13). However, while both clinical studies showed a high specificity, the sensitivity was 37% in one and 95% in the other. This raises a concern about the usefulness of this rule in different clinical settings.

Data on the TTP of consecutive patients with monomicrobial candidemia were collected in two European 700- and 1,400-bed university hospitals located in Barcelona (Spain) and Cologne (Germany), respectively. The Barcelona center provided data from 1 January 2008 to 31 March 2012 and the Cologne center from 1 August 2006 to 31 December 2011. The ethics committee of Hospital Clinic de Barcelona approved the study. Only the first set of positive blood cultures was included, and the shortest TTP (measured in hours) registered for any positive aerobic and anaerobic vials were analyzed.

In Barcelona, blood cultures were processed by the Bactec 9240 system (Becton-Dickinson, MD, USA), and vials were loaded into the machine around the clock. In Cologne, blood cultures were processed by the same system, but vials were placed into the system only during the working hours of the laboratory. The vials used in Barcelona were the resin-containing Bactec Plus Aerobic/F and Bactec Plus Anaerobic/F or the non-resin-containing Bactec Standard 10 Aerobic/F and Bactec Lytic/10 Anaerobic/F. In Cologne, only resin-containing vials were used.

Comparisons of TTP between species were carried out using the Student t test. Proportions were compared by Fisher's exact test. A receiver operating characteristic (ROC) curve was used to assess the diagnostic accuracy of TTP for C. glabrata and to identify the optimal cutoff value.

A total of 379 episodes of monomicrobial candidemia were analyzed; 157 (41%) occurred in Barcelona and 222 (59%) in Cologne. The prevalence of C. glabrata in each center was similar with 19 cases (12%) in Barcelona and 39 (17.5%) in Cologne (P = 0.18). Resin-containing vials were used in 17 of 19 episodes of C. glabrata candidemia in Barcelona and in all episodes in Cologne.

The numbers of episodes where yeast was recovered from aerobic and/or anaerobic vials are presented in Table 1, and the TTP for each species is shown in Table 2. The TTP for C. glabrata was longer in the Cologne series (80.8 h versus 53.4 h; P = 0.004), and therefore data from each center were analyzed separately. A TTP of >56.5 h for the Barcelona series and a TTP of >79 h for the Cologne series were calculated as the optimal cutoff values for predicting C. glabrata. The areas under the curve were 0.66 (95% confidence interval [CI], 0.5 to 0.8) in Barcelona and 0.77 (95% CI, 0.7 to 0.8) in Cologne, indicating poor to fair accuracy of TTP as a diagnostic test.

TABLE 1.

Detection of growth in aerobic, anaerobic, and both types of blood culture vials by study center and Candida sp.^a

Species	No. of vialsb		Growth (no.[%]) in:
			Aerobic and anaerobic vials		Aerobic vials only		Anaerobic vials only		Anaerobic vials exclusively or earlier
	B	C	B	C	B	C	B	C	B	C
C. albicans	81	155	10 (12)	8 (5)	70 (86)	146 (94)	1 (1)	1 (1)	3 (4)	2 (1)
C. parapsilosis	31	15	3 (10)	0	28 (90)	15 (100)	0	0	2 (7)	0
C. tropicalis	20	8	4 (20)	3 (20)	16 (80)	5 (63)	0	0	0	0
C. glabrata	19	39	5 (26)	11 (28)	10 (53)	25 (64)	4 (21)	3 (8)	7 (37)	12 (31)
C. krusei	4	2	1 (25)	1 (50)	3 (75)	1 (50)	0	0	0	0
Other	2	3	0	1 (33)	2 (100)	2 (67)	0	0	0	0
Total	157	222	23 (15)	24 (11)	129 (82)	194 (87)	5 (3)	4 (2)	12 (8)	14 (6)

^aInformation is based in part on data from reference 8.

^bB, Barcelona, C, Cologne.

TABLE 2.

Time to positivity of Candida spp. included in the study^a

Species	Results from Barcelona:			Results from Cologne:
	No. (%) positive	Mean (SD) TTPb	Median (IQRc) TTP	No. (%) positive	Mean (SD) TTPd	Median (IQR) TTP
C. albicans	81 (51.6)	41.5 (19.4)	39.1 (28.4–49.9)	155 (69.8)	38.8 (25)	34 (22–48)
C. parapsilosis	31 (19.7)	36.2 (17)	34.8 (25.1–46.5)	15 (6.8)	38.2 (18)	37 (24–49)
C. tropicalis	20 (12.7)	28.3 (19.9)	22 (16.6–31.8)	8 (3.6)	15 (6.8)	17.5 (17.5–19)
C. glabrata	19 (12.1)	53.4 (26.8)	56.5 (29.7–77.3)	39 (17.6)	80.8 (40.8)	83 (46–113)
C. krusei	4 (2.5)	23.3 (17.1)	21.3 (8.3–40.3)	2 (0.9)	24 (4.2)	24 (21–24)
Other	2 (1.3)e	19.9 (4.6)	19.8 (16.5–19.9)	3 (1.4)f	30 (14.8)	23 (20–23)
Total	157 (100)	39.12 (21.4)	34.8 (25.1–48.7)	222 (100)	45 (32.5)	36 (23–55.3)

^aInformation is based in part on data from reference 8.

^bThe TTP of C. glabrata was significantly longer than those of C. albicans (P = 0.03), C. parapsilosis (P = 0.008), C. tropicalis (P = 0.002), and C. krusei (P = 0.04).

^cIQR, interquartile range.

^dThe TTP of C. glabrata was significantly longer than those of C. albicans (P < 0.001), C. parapsilosis (P = <0.001), C. tropicalis (P = <0.001) and C. krusei (P = 0.05).

^eC. guillermondi and C. kefyr.

^fC. dubliniensis, C. famata, and C. rugosa.

The test characteristics for predicting C. glabrata either by the exclusive or earlier growth in anaerobic vials (primary test) or by TTP cutoff values (alternative test) are shown in Table 3. In regard to the primary test, there were no significant differences between Cologne and Barcelona in either sensitivity (31% versus 37%, P = 0.6) or specificity (99% versus 96%, P = 0.14). Given its low sensitivity, the alternative rule may be appropriately used in patients in whom the first test is negative. When yeast was detected exclusively or first in aerobic vials, a TTP of >56.5 h in Barcelona and a TTP of >79 h in Cologne had positive predictive values (PPVs) for C. glabrata of 30% and 69%, respectively. In the two sites, the corresponding negative predictive value (NPV) was 96%. A two-step diagnostic approach using the primary test in the whole population and then the alternative test in cases with exclusive or first detection in aerobic vials had 79% sensitivity, 90% specificity, 59% PPV, and 96% NPV.

TABLE 3.

Performance characteristics of TTP and exclusively or earlier growth in anaerobic vials of each hospital^a

Test characteristic	Results (%) from:
	Exclusively or earlier growth in anaerobic vials			Alternative test (TTP cutoff)		Exclusively or earlier growth in aerobic vials
	Global	B	C	TTP > 56.5 hb	TTP > 79 hc	TTP > 56.5 hd	TTP > 79 he
Sensitivity	33	37	31	47	54	58	74
Specificity	98	96	99	88	95	88	95
PPV	73	58	86	36	70	30	69
NPV	89	92	87	92	89	96	96

^aB, Barcelona, C, Cologne; AUC, area under the curve; PPV, positive predictive value; NPV, negative predictive value.

^bIn Barcelona: AUC, 0.66 (95% CI, 0.5 to 0.8).

^cIn Cologne: AUC, 0.77 (95% CI, 0.7 to 0.8).

^dn = 145 in Barcelona.

^en = 208 in Cologne.

This study shows that exclusive or earlier detection of growth in anaerobic vials by the Bactec 9240 system can predict the presence of C. glabrata with high specificity. The sensitivity found in the present study (31% to 38%) was much lower than the 95% described by Foster et al. (13). This increased sensitivity was probably due to the systematic use by Foster et al. of the Lytic/10 Anaerobic/F culture vials instead of the regular Bactec anaerobic vials used in the present study. However, earlier or exclusive detection in anaerobic vials increased the likelihood of C. glabrata from pretest probabilities of 12% in Barcelona and 17% in Cologne to posttest probabilities of 58% and 86%, respectively.

An unexpected finding of this study was the difference in the mean TTP of C. glabrata between the two sites. Potential explanations are an unequal distribution of catheter-related candidemia between sites and different proportions of patients undergoing antifungal therapy at the time when blood samples for culture were drawn. In any case, our data indicate caution about the use of absolute cutoff values for TTP in predicting C. glabrata as has been proposed in other studies (8–10).

Our results suggest that a two-step approach, using first the test less prone to be affected by confounders (earlier or exclusive detection in anaerobic vials) followed by the site-specific TTP cutoff test in those cases with earlier or exclusive detection in an aerobic vial, has the highest sensitivity (79%) and NPV (96%).

This study had several limitations. First of all, we did not assess variables that may have influenced TTP, such as ongoing antimicrobial therapy, the source of candidemia, and the clinical conditions of the patients (2). Furthermore, the clinical impact of changing the empirical treatment with the result of the Gram stain could not be assessed. Finally, as the BacT/Alert system may have better yields and shorter TTP for C. glabrata (12), our results cannot be extrapolated to other incubation systems or type of vials.