Characteristic and clinical relevance of Candida mannan test in the diagnosis of probable invasive candidiasis

Transcription

1 Medical Mycology, 2014, 52, doi: /mmy/myu018 Advance Access Publication Date: 12 June 2014 Original Article Original Article Characteristic and clinical relevance of Candida mannan test in the diagnosis of probable invasive candidiasis Bernabé F.F.Chumpitazi 1,, Bernadette Lebeau 1, Odile Faure-Cognet 1, Rebecca Hamidfar-Roy 2, Jean-François Timsit 2,PatriciaPavese 3, Anne Thiebaut-Bertrand 4, Jean-Louis Quesada 5,Hervé Pelloux 1 and Claudine Pinel 1 1 Laboratory of Parasitology and Mycology, Grenoble University Hospital, Joseph Fourier University, Grenoble, France, 2 Clinic of Medical Intensive Care, Grenoble University Hospital, Joseph Fourier University, Grenoble, France, 3 Department of Infectious and Tropical Diseases, Grenoble University Hospital, Joseph Fourier University, Grenoble, France, 4 Department of Hematology, Grenoble University Hospital, Joseph Fourier University, Grenoble, France and 5 Center of Clinic Investigation, Grenoble University Hospital, Joseph Fourier University, Grenoble, France *To whom correspondence should be addressed. BChumpitazi@chu-grenoble.fr Received 17 October 2013; Revised 16 January 2014; Accepted 19 February 2014 Abstract The gold standard laboratory tests used to diagnose invasive Candida infection (ICI) are based on the in vitro culture of blood or samples from other sterile sites. However, these tests have limited sensitivity (Se) and are generally not diagnostic until late in the infectious process. The Serion Candida mannan kit was evaluated for the diagnosis of ICI at Grenoble University Hospital (France) between 2007 and The results were then compared with worldwide data published between 1997 and This retrospective study was based on follow-up from the investigation of 162 patients of whom 91 had proven ICI; 13 had Candida colonization index (CCI) scores 0.42, positive mannan tests, with nonconcomitant infections; and 58 had no evidence of Candida infection. Candida albicans, C. glabrata, C. tropicalis, andc. parapsilosis were the etiologic agents in 104 patients. For patients with or without ICI, the 12-week mortality rates were 35/104 (33.7%) and 6/58 (10.3%), respectively. The mannan diagnostic specificity was 51% and Se was 77%. However, in the metaanalysis (n = 1,536), values were 86% and 62%, respectively. Positive mannan test results may appear early (median 6 days) in the development of candidemia and have moderate diagnostic value for ICI, with a negative predictive value of 83%. In patients at risk of ICI with negative candidemia, the combination of Candida mannan test data with a CCI score 0.42 may improve the diagnosis of probable ICI. Key words: invasive Candida infection, candidemia, mannan, systematic review. 460 C The Author Published by Oxford University Press on behalf of The International Society for Human and Animal Mycology. All rights reserved. For permissions, please journals.permissions@oup.com

2 Chumpitazi et al. 461 Introduction Invasive fungal diseases (IFDs) are usually observed in patients who are immunocompromised and have hematological malignancies, have undergone transplant procedures, and/or have solid tumors. IFDs have also been noted in critically ill, nonimmunosuppressed patients who have had major gastrointestinal or heart surgery [1]. Such infections are usually associated with high mortality. For instance, in North America between 2004 and 2008, the mortality rates for invasive Candida infection (ICI) varied from 37.5% to 75.0% [2]. ICI is also related to significant morbidity as a result of relatively long stays in intensive care units, ranging from 1 to several weeks [3]. Furthermore, the incidence of systemic infections due to Candida spp. has increased in hospitals worldwide [3]. Since clinical signs and symptoms are not specific, therapy with preventative broad-spectrum antibacterial drugs is initiated empirically; if fever persists, an antifungal drug should be administered. In other patients with high-risk factors, antifungal therapy is administered at the start of the hospital stay. The laboratory diagnosis of ICI is mainly based on cultures of blood and samples from other sterile sites. However, both tests may have limited sensitivity (Se), with delays in the development of positive results [4]. A fungal-negative culture may be considered to be inconclusive in certain patients receiving antifungal therapy. Nonculture-based tests, including serology and molecular biology assays, have been developed in order to decrease the incidence of inconclusive results from this phenotypic test [4 10]. Information on circulating biomarkers, such as Candida albicans mannan and/or (1 3)-β-D-glucan, may be obtained prior to receipt of the results of culture-based tests, thus allowing early initiation of adequate antifungal therapy and, in turn, improvement in patient outcomes [11 13]. Mannans, which are involved in innate and acquired immunity, are a main cell wall component of Candida spp. and are used as biomarkers [14 16]. The detection of Candida mannans in a cutaneous biopsy confirmed disseminated candidiasis due to C. albicans in a patient with aplastic anemia and candidemia [16]. While tests for the Candida mannan antigen that use mainly monoclonal antibodies have been well assessed in the literature [17 19], the Serion kit (Serion GmbH; Institute Virion, Wurzburg, Germany), which uses polyclonal antibodies to obtain quantified data, has been evaluated only once [5,17]. The present case-control study was based on common practice at Grenoble University Hospital (France). We hypothesized that the subgroup of patients with negative candidemia, a Candida colonization index (CCI) score 0.42, a positive mannan test, and no other concomitant infections may have probable ICI [20]. This study did not have any epidemiological purpose. The test results were compared with proven and probable ICI results before being integrated into a metaanalysis with data from published studies. Materials and methods Sampling and laboratory tests Clinicians independently prescribed the laboratory analysis and number of sites examined. Biological samples (blood, biopsy, bronchoalveolar lavage fluid, tracheal aspirates, urine) were cultured for fungal detection using the European Community labeled mycosis IC/F medium (Bactec Mycosis IC/F, Becton Dickinson Diagnostic Systems, Sparks, MD, USA) [21,22]. A CCI score was calculated for each patient; stool/rectal swabs were included. Neutrophil and lymphocyte counts and C-reactive protein (CRP) levels were determined. The Serion mannan assay was performed twice, 1 week apart; the initial assay with the fresh serum and the second with the respective serum that had been immediately stored at 20 C after centrifugation of the initial blood samples. The Serion enzyme-linked immunosorbent assay antigen Candida kit was used for detectionof Candida mannan antigen according to the procedure recommended by both the manufacturer and Lunel et al. [5]. Briefly, the kit uses microtiter wells coated with polyclonal antibodies against C. albicans mannans. Every test had blank assays as well as positive and negative controls. The 300-μl serum samples were diluted with 100 μl of sample buffer and incubated at 110 C for 5 min followed by centrifugation at 9000 g for 10 min. The resultant clear supernatant was used for the assay (see manufacturer instructions). A standard curve was plotted to convert the absorbency values into Candida mannan levels, expressed in arbitrary units (U) per milliliter. Per the manufacturer s recommendations, the detection was considered significant when the level was 2.30 U/ml. Patient and control groups Patients who presented with at least four risk factors associated with the development of IFD were included in this retrospective case-control study [3]. The risk factors included premature low-birthweight infants; patients receiving broad-spectrum antibiotics or cancer therapy; and patients with immunosuppression, neutropenia, colonization of several body sites that include candiduria of 10 5 colony-forming units per milliliter, diabetes, renal failure, central venous catheter, abdominal surgery, recent surgery, or suspected ocular candidiasis. The records of 162 patients with these underlying factors who were admitted to Grenoble University Hospital were examined between

3 462 Medical Mycology, 2014, Vol. 52, No. 5 Table 1. Characteristics of fever patients with or without proven or probable invasive Candida infection. Patient/Parameter Proven or probable ICI (n = 104) Without ICI (n = 58) P value Median age, y (range) 57.1 ( ) 19.6 ( ) <0.001 a Gender (men/women) 59/45 43/ b Twelve-week mortality rate (Pearson χ 2 ) a 35/104 (33.7%) 6/58 (10.3%) a Hematological malignancies or solid tumors 51/104 (49.0%) 28/58 (48.3%) b Cirrhosis 10/104 (9.6%) 5/58 (8.6%) b Diabetes mel1itus 15/104 (14.4%) 1/58 (1.7%) b Heart disease 15/104 (14.4%) 2/58 (3.4%) b Renal or lung disease 13/104 (12.5%) 22/58 (37.9%) < b ICI episodes Human immunodeficiency virus+ 3/104 (2.9%) 1/58 (1.7%) Median lymphocytes 10 9 cells /l 0.9 (0 34) 0.8 (0 5.2) Lymphopenia 46/97 (47.4%) 41/75 (54.7%) b Median neutrophils 10 9 cells/l 6.9 (0 30.9) 2.4 (0 31.3) <0.001 a Neutropenia 25/97 (25.8%) 34/75 (45.3%) b CRP >10 mg/l 95/106 (89.6%) 39/67 (55.0%) Median CRP mg/l (range) 88.0 (3 386) 32 (3 342) <0.001 a Candidemia and infection of one sterile site 91/112 (81.2%) 0 Candida colonization 76/112 (67.8%) 33/60 (55.9%) a Median Candida mannan (range) U/ml 3.24 (0 70) 2.19 (0 26) Candida mannan /38 (76.3%) 35/71 (49.3%) a Candida mannan and Candida colonization 25/34 (73.5%) 7/55 (12.7%) < a CRP, C-reactive protein; ICI, invasive Candida infection a Mann-Whitney U test. b Pearson χ 2 test. September 2007 and December 2011 for the presence of mycologically proven or probable ICI (Table 1). The patients were monitored during their hospital stays while in the following wards: intensive care, hematology, pediatry, cancerology, internal medicine, infectious medicine, gastroenterology, cardiology, pneumology, or nephrology. ICI was proven by at least one positive blood culture and/or positive Candida spp. culture from sterile-site biopsies [20]. During the study, eight patients had more than one proven ICI episode; each episode was separated by at least two months with the same species detected. For example, a patient had episodes on the following three days: 10 February 2010 (C. albicans candidemia), 8 July 2011 (C. albicans candidemia), and 10 October 2011 (C. albicans and C. glabrata candidemia). A subgroup of patients was composed of those with negative candidemia, a CCI score 0.42, a positive mannan test, and no other concomitant infections. Patients with no ICI episodes had a CCI score of zero, while this score was <0.42 in patients with possible ICI. Colonization was defined by Eloy et al. [23]. Patients were classified into the following two groups: a control group (possible and without ICI), composed of 58 patients (43 males/15 females; median age, 19.6 years (range, ) in whom at least one Candida mannan test was performed but with a CCI score <0.42, and a group of 104 patients (59 males/45 females; median age, 57.1 years (range, )) with proven (91) or probable (13) ICI (see Table 1) [20]. Statistical analysis of study location Significant differences between the two groups were determined using the Mann-Whitney U test. The mannan levels obtained at a one-week interval were analyzed using the paired t test. A P value 0.05 was considered to be statistically significant. Pearson χ 2 test was used for analysis of Candida mannan and/or colonization prevalence. A receiving operating characteristic (ROC) curve was plotted, and the area under the curve (AUC) was calculated in order to estimate the discriminatory power of the Candida mannan assay between infected and uninfected patients. Metaanalysis To synthesize and quantitatively compare our data with that published worldwide between 1999 and 2011, a hierarchical summary ROC (SROC) curve was drawn. These data were identified using a Medline search for Candida mannan, in which at least the Se and specificity (Sp) of the diagnostic test used were provided [5,6,9 11,18,23 30]. Deeks funnel plot asymmetry test was used to test for publication bias (Dwamena, BA. 2007, a program for

4 Chumpitazi et al. 463 meta-analytical integration of diagnostic accuracy studies in Stata). Stata software, version 12 (StataCorp., College Station, TX, USA), was used to plot the SROC curve. The Se, Sp, and diagnostic odds ratio (DOR = LR + /LR,in which LR + is the positive likelihood and LR the negative likelihood) were calculated for each study and represented in a Forest plot, with a 95% confidence interval (CI). DOR values were considered to give an estimation of test accuracy, as were the Se and Sp. To examine the quality of concordance between ICI diagnosis and Candida mannan contents, Matthews correlation coefficients were calculated. Results Local study The demographic, clinical, hematological, biochemical, and mycological characteristics as well as some of the high-risk factors for ICI of the patients enrolled in our study are shown in Table 1. The laboratory follow-up per patient was from 1 to 124 days, with a median of 29 days. From one to five body sites were tested for Candida colonization, with a median of two sites per patient. The frequency of mycological laboratory examination was from 1 to 10 days (median, 4 days). In total, 173/181 (95.6%; 95% CI, ) species isolated in this study were members of the Candida genus (Table 2). Amphotericin B was used to treat 16/130 (12.3%) patients, azoles (fluconazole, voriconazole) for 84/130 (64.6%), and echinocandins (caspofungin and Table 2. Infecting fungal pathogens and species in the 162 patients. Fungal pathogen and species Isolates, n % Total fungal spp. isolates 181 Candida isolates C. albicans C. glabrata C. tropicalis C. parapsilosis C. krusei C. kefyr C. lusitaniae C. guillermondii Other Candida spp Aspergillus isolates A. fumigatus A. nidulans Other yeasts Saccharomyces cerevisiae Malassezia furfur Geotrichum capitatum Other moulds Trichoderma spp Figure 1. Interassay measurement of Candida mannan levels at 1-week intervals; interassay results for Candida mannan levels during the local study are shown. y-axis: mannan levels in serum samples, expressed in units (U) per milliliter; x-axis: first measurement with fresh serum shown by solid circle ( ) and second measurement with frozen serum samples, 1-week later, shown by empty circles ( ). The levels of Candida mannan were higher for the first measurement than for the second (n = 72; paired t test, P < ). micafungin) for 30/130 (23.1%). In the control group, 23/58 patients had a CCI score equal to zero, while the CCI prevalence among the cases of positive candidemia was 90.1% (73/81; 95% CI, 82.4% 95.4%). The 12-week mortality rates varied significantly among patients with and without ICI as follows: 35/104 (33.7%; 95% CI, 25 42) and 6/58 (10.3%; 95% CI, 5 21), respectively (Pearson χ 2 test: Q = 9.50, P = 0.002). Age, neutrophil count, and CRP content were significantly different between the two groups (Table 1), whereas the prevalence of Candida colonization was not significantly different (Pearson χ 2 test: Q = 2.78, P = 0.095). To optimize the conditions of use of the Serion Candida mannan kit, two quantification tests were performed, one from fresh serum and one from a 1-week-old frozen serum sample (Fig. 1). The first series (median = 2.57 U/ml) showed significantly higher mannan quantities than the second (median = 1.68 U/ml; n = 72; paired t test; P < 0.001), suggesting the metabolic degradation or instability of mannan in serum. Therefore, only the results of the first quantification were used in the present investigation. In the local study, serum mannan prevalence was 76.3% (29/38; 95% CI, 61.6% 87.7%) in the group of patients with proven ICI (Table 1). In our retrospective local study, only 94/162 patients had mannan data. The optimal mannan cutoff obtained from the ROC curve was 2.29 U/ml (Fig. 2a). The difference in serum mannan levels for samples from which C. albicans (median, 2.3 U/ml; n = 37) was successfully recovered and samples that yielded non C. albicans isolates was not statistically different (median, 2.8 U/ml; n = 71; Mann-Whitney U test, P = 0.218). The Candida mannan prevalence was higher in

5 464 Medical Mycology, 2014, Vol. 52, No. 5 Figure 2. (a) Receiving operating characteristic (ROC) curve, (b) summary ROC (SROC) curve with 95% confidence contour and 95% prediction contour, and (c) Deeks funnel plot asymmetry test. (a) The ROC curve (selectivity [Se] vs. 1 specificity [Sp]) for the Candida mannan test used in the local study. The selected cutoff is surrounded by an open circle (2.29 U/ml; area under the curve [AUC] = 0.60; n = 94 patients; 109 assays), which allows for the best discriminatory power of the test. In the SROC plot (b; Se vs. Sp), each numbered circle represents a different study. This SROC plot (solid line: AUC = 0.79; 95% CI, ; 1536 patients) was constructed using data from Table 3 and a logistic regression approach. The 95% confidence (dashed line) and 95% prediction (dotted line) contours are given. An additional empty gray circle ( ) shows the values obtained from the Grenoble University Hospital study. The summary operating point (gray rhombus: Se = 0.62; 95% CI, ; Sp = 0.86; 95% CI, ) was calculated taking into account the χ 2 test of homogeneity and the inconsistency index (% of total variation) among the studies. the group of patients colonized or infected by C. albicans, C. glabrata, and C. tropicalis than in the group of patients colonized or infected by C. parapsilosis, C. lusitaniae, C. kefyr, C. guilliermondii, C. pulcherrima, and C. inconspicua: 49% (30/61; 95% CI, 37 61%) vs. 39% (7/18; 95% CI, 20 61%), respectively. However, this was not statistically significant. In 11/21 (52.4%; 95% CI, 32.2% 71.8%) cases of positive candidemia, serum mannan content tests were positive prior to those that indicated candidemia (median 6 days). In 2/21 (9.5%; 95% CI, 2.9% 29.2%) cases of positive candidemia, serum mannan content was positive when candidemia tests were positive; in 8/21 (38.1%; 95% CI, 20.7% 59.3%), the serum mannan content was positive (median) 9 days after testing positive for candidemia. In the local study, the Se, Sp, and AUC of the Serion Candida mannan test were 77%, 51%, and 0.60, respectively (Fig. 2a). The combination of a positive mannan test with Candida colonization significantly improved (Pearson χ 2 test: Q = 33.73, P < ) the diagnostic Sp of the mannan assay (Table 1; Pearson χ 2 test: Q = 7.45, P = 0.006). Metaanalysis Fourteen studies that covered 1536 patients met our search criteria. To precisely and visually compare our data with other published results, we used a SROC curve (Fig. 2b) with confidence and prediction contours. The metaanalysis values were Se SROC = 62% (95% CI, 52 72), Sp SROC = 86% (95% CI, 76 93), and AUC SROC = 0.79 (95% CI, ). These values were calculated on the basis of mannan detection, the χ 2 test of homogeneity, and the inconsistency index for the studies and the kits used. Figure 2c shows the Deeks funnel plot asymmetry test data for all studies analyzed. The slope of the regression line was significantly different from zero (P = 0.77 > 0.10), and the selected studies did not present any statistical publication bias. Table 3 shows the evaluation of Candida mannan tests for the diagnosis of ICI with the Se, Sp, positive predictive value, negative predictive value, efficiency, and Matthews correlation coefficient. The summary parameters, which are listed on the last line of Table 3, were calculated conventionally, with the exception of Se SROC and Sp SROC, which were estimated as indicated above. The results of the diagnostic accuracy test are presented in Figure 3 in the form of a Forest plot in which the Se, Sp, DOR, and CIs of Candida mannan tests are given for our study and for studies selected from the literature. Concerning these four parameters, the heterogeneity of the results was evaluated graphically using the Forest plot and statistically using the inconsistency index values that were >50%: 74% for Se (95% CI, 61 87), 94% for Sp (95% CI, 92 96), and 99.98% for DOR. The mannan assays used monoclonal antibodies (Platelia, except New Platelia, which uses polyclonal

6 Chumpitazi et al. 465 Table 3. Evaluation and metaanalysis of Candida mannan levels in the diagnosis of invasive Candida infection. Mannan kit Patients Invasive TP FN TN FP Se Sp PPV NPV Efficiency MCC Study Ward c Candida infection events Platelia [23] ICU Serion [5] H, O Platelia [11] H, O Platelia [25] H,O,C Platelia [26] H,O,G Platelia [27] H, O Platelia [28] G, P, H In house [29] G,O,P Platelia [18] G,N,P Platelia [30] NICU Platelia [9] H,O,G New platelia [10] H, O Cica fungi [6] G, P, O Platelia [24] H, ICU Serion a a H, O, G, ICU Overall b 86 b Overall The Candida fungal species, which were isolated from false-negative cases using the Serion kit, included C. albicans (n = 4), C. tropicalis (n = 3), C. glabrata (n = 4), and C. parapsilosis (n = 2). C, cardiology; FN, false negative; FP, false positive; G, gastroenterology; H, hematology; ICI, invasive Candida infection; ICI, Invasive Candida Infection; ICU, intensive care unit; MCC, Matthews correlation coefficient; N, nephrology; NICU, neonatal intensive care unit; NPV, negative predictive value; O, oncology; P, pneumology; PPV, positive predictive value; Se, sensitivity; Sp, specificity; TN, true negative; TP, true positive. a Data from Grenoble University Hospital. b The sensitivity-summary receiving operating characteristic and specificity- summary receiving operating characteristic were obtained by taking into account the χ 2 test of the homogeneity and the inconsistency index (% of total variation) among the studies. c In general, the samples came from several wards. Only the most significant ones in terms of number of cases appear here. antibodies; Cica fungi; and in house ; data from Table 3) appear to have better Sp (86%; [795/( )] 100) than assays that use polyclonal (Serion) antibodies (58%, [59/(59+42)] 100). This difference was statistically significant (Pearson χ 2 test: Q = 5.24, P = 0.022). Discussion The extent of ICI is very difficult to define within the context of IFD. We adapted the classification of patients with IFD from the description by De Pauw et al. [20] of patients with ICI. The notion of proven, probable, and possible IFD was indicated based on the following statement: Cases of probable IFD require that a host factor, clinical features, and mycological evidence be present [20]. Among the indicated mycological criteria are indirect tests (detection of antigen or cell-wall constituents) [20]. Therefore, it is justifiable to add detection of such constituents as a criterion in the definition of probable ICI cases. In the local study, serum mannan prevalence was 76.3% in the group of patients with proven ICI and thus may designate a probability of proven ICI. In the definition of probable ICI, we also introduced the concept of nonconcomitant infections because, in some cases, the detection of mannans may lack Sp. We improved the definition of the probability of occurrence of proven ICI by adding the CCI score. In the local study, the CCI prevalence among the cases of positive candidemia was 90.1%. Taking into account all of these arguments, we can assume that the subgroup formed in material and methods with the summarized characteristics had probable ICI. In the literature, the CCI threshold value varies from 0.4 to 0.5; in our local study, the threshold used for CCI was Thus, the results of our local study suggest that a positive serum Candida mannan test associated with a positivecandida colonization increased the Sp of the laboratory diagnosis. ICI is a potentially fatal disease despite advances in hospital hygiene, therapeutics, and laboratory tests [1,2,9,21,24]. The 12-week mortality rates seen in our hospital corroborate these findings. However, our study was not built for epidemiological purposes as we included only patients at high risk of ICI and with at least one Candida mannan assay. Consequently, the number of patients in the control group was lower than in the invasive Candida group. This does not reflect the fact that ICI patients

7 466 Medical Mycology, 2014, Vol. 52, No. 5 Figure 3. Forest plot showing in the X-axis sensitivity (Se), specificity (Sp), and Log of the diagnostic odds ratio (DOR) of Candida mannan tests and their 95% confidence intervals (CI), respectively. Y-axis: cited study. These graphs show the relative heterogeneity of the measured parameters. Gray symbols show the Se, Sp, and DOR calculated from all studies taken together in the classic way. In the study listing, Our local study indicates data from Grenoble University Hospital obtained between 2007 and represent between one and eight of all patients admitted to hospitals [3]. At our hospital, testing for Candida mannan is not systematic, even for high-risk patients, which accounts for the relatively low number of patients in the control group. We made no attempt to analyze ICI incidence in terms of hospital wards, which would have required a larger study population. Also, we did not evaluate beta-glucans, another important cell wall component of Candida and other fungal species, as they are regarded as pan-fungal biomarkers [13,31]. In our laboratory assays, a significant decrease in mannan levels at 1-week intervals was found, although the serum samples were frozen for storage. Thus, only the first measurements obtained with fresh serum were analyzed in our study. The apparent requirement for fresh serum samples may be a disadvantage of the Serion kit. The lack of reproducibility of the assay between fresh and frozen serum suggests the degradation of mannan epitopes by soluble cytosolic mannosidase activity in C. albicans and humans [32,33]. It is likely that the Serion test also lacks robustness. Together, these considerations may, in part, explain the differences in Se observed in both our study and our metaanalysis [17,18,24,26,27,30]. However, the Se found in our hospital (77%) falls within the range (31% 100%) of Se values mentioned by Mikulska et al. (overall Se 62%) [17]. Our relatively good Se but only moderate Sp is in line with those of the other studies. In contrast with earlier research, the studies selected for our metaanalysis did not show any publication bias for the 1536 patients included [17]. According to Mikulska et al., a known degree of heterogeneity has been observed [17]. Heterogeneity is indeed inherent to ICI for the following reasons: patients of all ages may be affected, from preterm infants to elderly patients; the high-risk factors for ICI can vary among patients and among treatments; different cutoff values are used in the different studies; a SROC curve provides information on the overall performance of the assay; and diverse Candida spp. that cause infection are involved in different parts of the world [1]. The Se of Candida mannan detection tests may vary depending on which infecting Candida spp. is present since each species has a unique mannan structure [14,17]. Furthermore, in our metaanalysis, mannan assays that used monoclonal antibodies (Platelia) may have provided more Sp than assays that used polyclonal (Serion) antibodies. A possible reason for the variations in Se levels of Candida mannan assays may be that the fact that the cell wall mannan of Candida spp. is species specific and the Candida mannan epitopes can involve alpha-1, 2-, alpha- 1, 3-, alpha-1, 6-, and/or beta-1, 2-linked mannopyranose

8 Chumpitazi et al. 467 units [14]. Mikulska et al. noted that Candida spp. affected the mannan sensitivities, which may vary from 0% (C. krusei) to 100% (C. tropicalis) [17]. Another reason for these variations may be that the number of epitopes, the number of which is based on the phosphate-bound beta- 1, 2-linked manno-oligosaccharides, may have fewer hyphal cells than yeast cells, as reflected in the results of Sendid et al. [28]. Finally, circulating Candida mannans bind mannose-binding lectin (MBL), as indirectly observed by the significant decrease in serum MBL levels during the 2 days preceding candidemia and/or mannanemia episodes in ICI patients [14,34]. The cause of the moderate lack of Sp may be due to the Candida mannan antibodies used in the test. These antibodies cross-react with other mannans from other microorganisms such as Saccharomyces cerevisiae and Mycobacterium paratuberculosis [35 37]. The possibility of cross-reaction decreases in patients who do not have bacteremia or bacterial colonization but do who have a positive CCI score ( 0.42) obtained from different body sites. Thus, as seen in our study, positive Candida mannan detection should be combined with a CCI score ( 0.42) in order to improve diagnostic Sp. Several researchers have suggested that positive Candida mannan detection be combined with an assay of anti- Candida mannan antibodies in order to improve ICI diagnosis [11,17,27,29]. Also, in this recommendation [17], the presence of anti-mannan antibodies could improve the Sp of Candida mannan detection. Concentration of anti-mannan antibodies may increase after the mannanemia/blood culture period in ICI patients [34]. Immunocompromised patients have high-risk factors for ICI; thus, the measurements of Candida mannan levels may be useful when combined with low levels of serum anti-candida antibodies, which is usually the case for these patients. However, no Candida spp. have been identified using only this approach; the result is a reduction in optimal therapeutic options. In this study, anti-mannan antibodies were not detected since antibodies to C. albicans wall antigens (indirect fluorescent antibody assays) and precipitins to C. albicans (immunoelectrophoresis) had been measured in our hospital (data not shown). It is important to note that in patients with vulvovaginal candidiasis who do not usually present with a high risk for ICI, Candida mannan may also be detected on vaginal swabs and have Se values ranging from 80.3% to 98.6% [38,39]. In contrast, the serum Se is only 20% (2/10) [18]. Yet, Mokaddas et al. found that none of their colonized patients had positive Candida mannan levels [19]. In patients at high risk for ICI without any other concomitant infection, our data and those from others may back up the argument that serum Candida mannan is a promising biomarker of probable ICI in combination with another specific one such as anti-mannan antibody [3,17,23] or CCI score >0.42. While a sterile-site culture assay remains the gold standard [16] for ICI, a Candida mannan assay may be useful for detecting Candida mannan before a diagnosis of candidemia can be made. In our hospital, about 50% of analyzed cases had early positive results. The advantage of an early diagnosis was shown in five studies reviewed by Mikulska et al. [17]. Likewise, early detection can confirm probable ICI in cases with persistently negative candidemia and a positive CCI score. Moreover, the relatively high negative predictive value of the Candida mannan tests suggests that they are useful in screening patients at risk for ICI [27]. Conclusion Our study and the systematic review of the literature show that the different commercially available Candida mannan tests have moderate to relatively good Se and Sp in patients with ICI or candidemia [17]. In clinically suspected patients whose candidemia results remain negative, a Candida mannan assay could have an adjunctive role [6,38,39] in combination with a CCI score >0.42, providing a probable direct indication of active disease and reducing the possibility of false-positive results. In some cases, faster diagnosis may allow the patient s outcome to be improved thanks to earlier initiation of satisfactory antifungal therapy. Acknowledgments We thank Dr Alison Foote and Sabine Durville for critically editing the manuscript and M. Ali Rahim, Evelyne Bouges, and Sonia Montlahuc-Gallet for their technical assistance. Declaration of interest The authors report no conflicts of interest. The authors alone are responsible for the content and the writing of the paper. References 1. Eggimann P, Garbino J, Pittet D. Epidemiology of Candida species infections in critically ill non-immunosuppressed patients. Lancet Infect Dis 2003; 3: Azie N, Neofytos D, Pfaller M et al. The PATH (Prospective Antifungal Therapy) Alliance R registry and invasive fungal infections: update Diagn Microbiol Infect Dis 2012; 73: Eggimann P, Bille J, Marchetti O. Diagnosis of invasive candidiasis in the ICU. Ann Intensive Care 2011; 1: Hong-Nguyen M, Wissel MC, Shields RK et al. Performance of Candida real-time polymerase chain reaction, β-d-glucan assay, and blood cultures in the diagnosis of invasive candidiasis. Clin Infect Dis 2012; 54:

9 468 Medical Mycology, 2014, Vol. 52, No Lunel FMV, Mennink-Kersten ASH, Ruegebrink D et al. Value of Candida serum markers in patients with invasive candidiasis after myeloablative chemotherapy. Diagn Microbiol Infect Dis 2009; 64: Takazono T, Izumikawa K, Nagayoshi Y et al. Evaluation of the Cica fungi Candida, a novel serum Candida mannan antigen kit, and its comparison with Cand-Tec in candidemia patients. Jpn J Infect Dis 2011; 64: Maubon D, Hamidfar-Roy R, Courby S et al. Therapeutic impact and diagnostic performance of multiplex PCR in patients with malignancies and suspected sepsis. J Infect 2010; 61: Mokaddas E, Khan ZU, Ahmad S, Nampoory MR, Burhamah M. Value of (1 3)-β-D-glucan, Candida mannan and Candida DNA detection in the diagnosis of candidaemia. Clin Microbiol Infect 2011; 17: White PL, Archer AE, Barnes RA. Comparison of non-culturebased methods for detection of systemic fungal infections, with an emphasis on invasive Candida infections. J Clin Microbiol 2005; 43: Lunel FMV, Donnelly JP, van der Lee HAL, Blijlevens NMA, Verweij PE. Performance of the new Platelia Candida Plus assays for the diagnosis of invasive Candida infection in patients undergoing myeloablative therapy. Med Mycol 2011; 49: Prella M, Bille J, Pugnale M et al. Early diagnosis of invasive candidiasis with mannan antigenemia and antimannan antibodies. Diagn Microbiol Infect Dis 2005; 51: Del Bono V, Delfino E, Furfaro E et al. Clinical performance of the (1,3)-β-D-glucan assay in early diagnosis of nosocomial Candida bloodstream infections. Clin Vaccine Immunol 2011; 18: Onishi A, Sugiyama D, Kogata Y et al. Diagnostic accuracy of serum 1,3-β-D-glucan for Pneumocystis jiroveci pneumonia, invasive candidiasis, and invasive aspergillosis: systematic review and meta-analysis. J Clin Microbiol 2012; 50: Shibata N, Kobayashi H, Suzuki S. Immunochemistry of pathogenic yeast, Candida species, focusing on mannan. Proc Jpn Acad Ser B Phys Biol Sci 2012; 88: Saijo S, Iwakura Y. Dectin-1 and dectin-2 in innate immunity against fungi. Int Immunol 2011; 23: Breier F, Oesterreicher C, Brugger S et al. Immunohistochemistry with monoclonal antibody against Candida albicans mannan antigen demonstrates cutaneous Candida granulomas as evidence of Candida sepsis in an immunosuppressed host. Dermatology 1997; 194: Mikulska M, Calandra T, Sanguinetti M et al. The use of mannan antigen and anti-mannan antibodies in the diagnosis of invasive candidiasis: recommendations from the Third European Conference on Infections in Leukemia. Crit Care 2010; 14: R Alam FF, Mustafa AS, Khan ZU. Comparative evaluation of (1, 3)-β-D-glucan, mannan and anti-mannan antibodies, and Candida species-specific snpcr in patients with candidemia. BMC Infect Dis 2007; 7: Mokaddas E, Burhamah MH, Khan ZU, Ahmad S. Levels of (1 3)-β-D-glucan, Candida mannan and Candida DNA in serum samples of pediatric cancer patients colonized with Candida species. BMC Infect Dis 2010; 10: De Pauw B, Walsh TJ, Donnelly JP et al. Revised definitions of invasive fungal disease from the European Organization for Research and Treatment of Cancer/Invasive Fungal Infections Cooperative Group and the National Institute of Allergy and Infectious Diseases Mycoses Study Group (EORTC/MSG) Consensus Group. Clin Infect Dis 2008; 46: Ruhnke M, Rickerts V, Cornely OA et al. Diagnosis and therapy of Candida infections: joint recommendations of the German Speaking Mycological Society and the Paul-Ehrlich-Society for Chemotherapy. Mycoses 2011; 54: Fricker-Hidalgo H, Lebeau B, Pelloux H, Grillot R. Use of the BACTEC 9240 system with mycosis-ic/f blood culture bottles for detection of fungemia. J Clin Microbiol 2004; 42: Eloy O, Marque S, Mourvilliers B et al. Contribution of the Pittet s index, antigen assay, IgM, and total antibodies in the diagnosis of invasive candidiasis in intensive care unit. JMycol Med 2006; 16: Arendrup MC, Sulim S, Holm A et al. Diagnostic issues, clinical characteristics, and outcomes for patients with fungemia. J Clin Microbiol 2011; 49: Fujita S-I, Takamura T, Nagahara M, Hashimoto T. Evaluation of a newly developed down flow immunoassay for detection of serum mannan antigens in patients with candidaemia. JMed Microbiol 2006; 55: Persat F, Topenot R, Piens MA et al. Evaluation of different commercial ELISA methods for the serodiagnosis of systemic candidosis. Mycoses 2002; 45: Ellis M, Al-Ramadi B, Bernsen R et al. Prospective evaluation of mannan and anti-mannan antibodies for diagnosis of invasive Candida infections in patients with neutropenic fever. J Med Microbiol 2009; 58: Sendid B, Jouault T, Coudriau R et al. Increased sensitivity of mannanemia detection tests by joint detection of a- and b-linked oligomannosides during experimental and human systemic candidiasis. J Clin Microbiol 2004; 42: Sendid B, Tabouret M, Poirot JL et al. New enzyme immunoassays for sensitive detection of circulating Candida albicans mannan and antimannan antibodies: Useful combined test for diagnosis of systemic candidiasis. J Clin Microbiol 1999; 37: Oliveri S, Trovato L, Betta P, Romeo MG, Nicoletti G. Experience with the Platelia Candida ELISA for the diagnosis of invasive candidosis in neonatal patients. Clin Microbiol Infect 2008; 14: Ostrosky-Zeichner L. Invasive mycoses: Diagnostic challenges. Am J Med 2012; 125: S14 S Mora-Montes HM, Bader O, López-Romero E et al. Kex2 protease converts the endoplasmic reticulum alpha1,2-mannosidase of Candida albicans into a soluble cytosolic form. Microbiology 2008; 154: Dariusz Szajda S, Waszkiewicz N, Stypułkowska A, Dadan J, Zwierz K. Lysosomal exoglycosidases in serum and urine of patients with pancreatic adenocarcinoma. Folia Histochem Cytobiol 2010; 48: Damiens S, Poissy J, François N et al. Mannose-binding lectin levels and variation during invasive candidiasis. J Clin Immunol 2012; 32:

10 Chumpitazi et al Schaffer T, Müller S, Flogerzi B et al. Anti-Saccharomyces cerevisiae mannan antibodies (ASCA) of Crohn s patients crossreact with mannan from other yeast strains, and murine ASCA IgM can be experimentally induced with Candida albicans. Inflamm Bowel Dis 2007; 13: Mpofu CM, Campbell BJ, Subramanian S et al. Microbial mannan inhibits bacterial killing by macrophages: a possible pathogenic mechanism for Crohn s disease. Gastroenterology 2007; 133: Nnalue NA, Weintraub A, Oscarson S, Lindberg AA. Crossreactivity between the mannan of Candida species, Klebsiella K24 polysaccharide and Salmonella C1 and E O-antigens is mediated by a terminal non-reducing beta-mannosyl residue. Eur J Biochem. 1994; 220: Marot-Leblond A, Nail-Billaud S, Pilon F et al. Efficient diagnosis of vulvovaginal candidiasis by use of a new rapid immunochromatography test. J Clin Microbiol 2009; 47: Matsui H, Hanaki H, Takahashi K et al. Rapid detection of vaginal Candida species by newly developed immunochromatography. Clin Vaccine Immunol 2009; 16: