Evaluation of Mass Spectrometry-Based Detection of Panfungal Serum Disaccharide for Diagnosis of Invasive Fungal Infections: Results from a Collaborative Study Involving Six European Clinical Centers

A mass spectrometry (MS) method that detects a serum disaccharide (DS) (MS-DS) was recently described for the diagnosis of invasive fungal infections (IFI). We carried out a European collaborative study to evaluate this assay.

MS-DS detection. Serum and plasma samples were frozen and sent on dry ice to the Laboratory of Clinical Mycology, Lille University Hospital, for processing as described previously (27). After a preanalytical step for the extraction and purification of oligosaccharides, spectra were recorded and analyzed using a 4800 MALDI-TOF/TOF analyzer (Applied Biosystems/MDS Sciex) at a fixed laser intensity within a 300-to 800-m/z range (UMR CNRS 8576, University of Lille).
Detection of circulating poly-and oligosaccharides and DNA in serum or plasma. For the diagnosis of IC and IA, detection of mannan (Man) or galactomannan (GM) and (1,3)-␤-D-glucan (BDG) was performed in the participating centers during routine patient screening or in the Lille Clinical Mycology Laboratory if they had not been tested previously.
BDG was measured using a Fungitell kit (Associates of Cape Cod Inc., Falmouth, MA, USA) following the manufacturer's instructions. The recommended cutoff of 80 pg/ml was used to determine clinical relevance.
Measurement of serum Man and GM was performed using a Platelia Candida Ag Plus test and Platelia Aspergillus Ag test (Bio-Rad, Marnes la Coquette, France), respectively, according to the manufacturer's instructions. The recommended cutoffs of a concentration of 62.5 pg/ml and an index of Ն0.5, respectively, were used.
For the diagnosis of MM, quantitative real-time PCR (qPCR) was performed as described previously (24,31).
Ethics statement. No additional sampling was necessary in any center due to the retrospective nature of the study. In Lille, agreement for the establishment of a biological collection of IFI samples was obtained from the French Ministry of Education and Research under reference number DC2008-642. Institutional review board approval was granted by the Comité de Protection des Personnes Nord-Ouest IV, the ethical committee of the university hospital of Lille.
Statistical analysis. GraphPad Prism (version 6) software was used to compare the distribution of biomarkers in the different groups with the Mann-Whitney two-tailed test and to generate receiver operating characteristic (ROC) curves, derive cutoffs, and construct graphs. A P value of Ͻ0.05 was considered statistically significant.

RESULTS
Invasive candidiasis. (i) Study population. The origin of the IC patients, the delay to serum sampling in relation to the time of the first positive blood culture, and the Candida species isolated are shown in Table 1. The Candida species were representative of the usual epidemiology encountered in southern Europe, with a higher prevalence of Candida parapsilosis complex isolates. The control patients with bacteremia are listed in Table 2. These included the usual panel of patients with community-acquired bacterial infections and two cases of Nocardia infection.
(ii) MS-DS diagnosis of IC in comparison to BDG and Man detection. The distribution of BDG and Man concentrations and MS-DS index values is shown in Fig.  1. All tests significantly discriminated IC patients from bacteremic controls (P Ͻ 0.0001).
Among the 27 IC patients, 9 were positive by all three tests, 12 were positive by two tests (6 by BDG detection and MS-DS; 5 by BDG and Man detection; 1 by Man detection and MS-DS), and 4 were positive by BDG detection alone. Only two patients (patients I19 and I27) infected by C. parapsilosis and C. krusei were negative by all tests. When considering control patients with bacteremia, three were positive by MS-DS, whereas six were positive for BDG, including the patient with Nocardia infection who displayed very high glucan levels. Only one control (patient S5) was positive for two biomarkers (BDG and DS). Figure 2 shows the ROC curves and corresponding sensitivities and specificities for the MS-DS, BDG, and Man detection tests. When considering the results for serum samples ( Fig. 2A and B), application of the cutoff value of 325 for MS-DS showed a sensitivity of 51% and a specificity of 87%, which were intermediate values compared to those obtained by the BDG and Man tests. Analysis of the MS-DS index values for IC diagnosis was then performed for patients ( Fig. 2C and D), and the sensitivity reached 67% without altering the high specificity estimated for serum. Comparison of ROC curves established for the MS-DS and the BDG and Man tests revealed that the diagnostic value of MS-DS was similar to that of the BDG test and positively complemented the high specificity of Man monitoring (revealed by the asymptotic curve). A lack of concordance between MS-DS and the BDG and Man tests was observed for serial serum samples from a given patient. This appeared to be more moderate when considering the global biomarker patterns per patient, since most patients (22/26) aggregated into two groups: those that displayed three positive test and those that displayed two positive tests. Figure 3A shows an example of biomarker kinetics during     Table 3. Except for one patient with invasive sinusitis, all patients presented with invasive pulmonary aspergillosis. "Day 0" indicates the date of the first serum available in the collection relative to the episodes of IA defined according to clinical and radiological arguments.
The characteristics of the controls, consisting of neutropenic patients, are summarized in Table 4. Retrospective analysis of the clinical evolution of IA revealed that  control patients 3, 6, and 7 developed probable IA 1, 2, and 6 months after serum sampling, respectively.
( Fig. 4. The values for the biomarkers were significantly higher in IA patients than in the controls (P Յ 0.0001).

ii) MS-DS diagnosis of IA in comparison to BDG and GM detection. The distribution of BDG concentrations and MS-DS index values is shown in
The sensitivity, specificity, and cutoff values for MS-DS and BDG detection were assessed by establishing ROC curves for serum, as shown in Fig. 5A and B. With a cutoff value of 325 for the MS-DS index, the sensitivity and specificity were 64% and 95%, respectively. The sensitivity values were intermediate between those for GM and BDG detection, which had 100% specificity. Interestingly, the 95% specificity of MS-DS was due to the high MS-DS index observed for control patient 6, who subsequently developed IA. In Fig. 5C and D, an analysis of MS-DS index values for IA diagnosis was performed for the patients, and no difference in terms of sensitivity and specificity was revealed with the data obtained for serum. A comparison of MS-DS and BDG detection ROC curves (according to the manufacturer's recommended threshold) showed that the diagnostic value of MS-DS was better than that of BDG detection, with a maximum sensitivity of 64% for MS-DS and 50% for BDG detection. Figure 3B shows an example of MS-DS and BDG detection kinetic evolution during GM monitoring. For this patient, only one serum sample was positive at the GM cutoff 14 days after the beginning of monitoring, while BDG levels were already positive on day 0, before increasing at unusually high levels. MS-DS was constantly positive during the whole survey, although it decreased at the time when BDG levels and GM index values were maximum.
Mucormycosis. (i) Study population. The characteristics of patients and the level of certainty of a diagnosis of MM according to EORTC criteria are shown in Table 5. The genera/species involved were representative of the usual spectra of Mucorales isolated, as were the risk factors, infection sites, and high mortality.
(ii) MS-DS in comparison with qPCR for diagnosis of MM. In contrast to IC and IA, no biochemical or immunological assays for the detection of circulating fungal poly-or oligosaccharides are available for the diagnosis of MM. We therefore compared MS-DS with qPCR, which is the only test currently available for the diagnosis of MM. The qPCR method used detects species from the genera Mucor, Rhizomucor, Lichtheimia, and Rhizopus.
We     Fig. 6. Among the serum samples tested before a mycological diagnosis was obtained, three (from three patients) were positive by MS-DS and six (from five patients) were positive by qPCR. An example of the kinetic evolution of MS-DS and qPCR in a patient from whom numerous samples were available is shown in Fig. 3C. A steady increase in MS-DS index values was observed during the 2 months of follow-up, whereas DNA circulation could be detected only between day 0 and day 20.

DISCUSSION
In contrast to diagnostic tests for obligate pathogens, whose detection is indicative of disease, diagnostic tests for opportunistic pathogens have to discriminate between the  presence of microbes as endo-or exosaprophytes and their shift to a pathogen (19). This is a kinetic process where the assay has to present the best compromise between specificity and sensitivity over time from disease onset and its evolution to invasive infection. Currently, for IC and IA, only two types of tests have been shown to be clinically useful for disease management (6,21,32,33). On the one hand, the Platelia Candida Ag Plus test and Platelia Aspergillus Ag test, which detect Man and GM, respectively, are considered specific but have a low sensitivity (especially for the detection of Man). On the other hand, the Fungitell test, which detects BDG, is considered more sensitive but less specific. The results of our previous study concerning MS-DS and those of the current study are in agreement with these conclusions. In both studies, MS-DS appeared to provide intermediate results, being more sensitive than the Platelia tests but more specific than the test for BDG (28). It is therefore suggested that MS-DS could be useful in the management of patients at risk of IC and IA. For some other IFI, such as those caused by Mucorales, the lack of glucans in their cell wall and negativity for BDG in the context of host invasion have led to considerable efforts to develop alternative diagnostic methods; a specific PCR is now available (24). qPCR is of great help in the diagnosis of MM, whose emergence is worrying in terms of incidence and severity (34). MS-DS was previously shown to be positive during MM (28). In the present study, a comparison of the results of MS-DS with the results of qPCR showed a similar performance, confirming that a panfungal diagnostic assay is useful in daily practice when an IFI is suspected without mycological evidence. The reproducibility of the sensitivity values in mono-and multicenter studies confirms the robust character of the Platelia and BDG tests, in line with their extensive use worldwide for several decades. A similar reproducibility and, thus, robustness were observed for MS-DS. Additional information regarding MS-DS specificity concerned the absence of false positivity associated with Nocardia infection, in contrast to BDG detection, as reported previously (35) and as recently observed for one of the two patients included in this study. Conversely, among the three neutropenic patients who were included as IA controls and who were revealed to have subsequently developed IA, none were positive for either GM or BDG, while MS-DS index values were positive and above the limit of significance in two patients. Although these results could be considered a coincidence, the long delay before disease development should be considered with caution due to the possible subclinical character of DS circulation. Unpublished data described the structure and function of DS and showed that among the m/z 365 hex-disaccharide signals (27) is trehalose, an important fungal metabolite (36). What the present study, investigating four different IFI biomarkers in comparison with DS, makes particularly obvious is the different kinetics of their circulation, as shown in Fig. 3. From a pathophysiological point of view, the metabolite DS has kinetics of synthesis and release from fungal cells different from those of cell surface-associated Man and GM or BDG, which is thought to be deeply anchored in the cell wall. These fungal polysaccharides are synthesized in situ, and then different processes of degradation take place as a result of either fungal or host carbohydrate catabolism. In parallel, binding to host receptors, catabolism by host soluble enzymes, and circulation as immune complexes make the levels in the circulation completely different. The efficiency of human mannosidases, naturally present for degrading human glycoproteins, and the large amount of antimannan antibodies present in IC patients are responsible for the rapid clearance of mannan. In contrast, mammals are poorly equipped for degrading glucans, which are not self-components, and their poor immunogenicity does not help with their clearance, explaining their longer persistence than mannans (37,38). Despite considerable efforts to solve the problems of DNA extraction and standardization, the lack of knowledge regarding the relationship between clinical outcome and Candida PCR results has prevented clinical recommendations (39). Recent progress has been made by the association of PCR with magnetic resonance detection in the T2MR system, leading to good specificity and increased sensitivity with regard to blood cultures for IC (25,40). For MM and in the absence of other biomarkers, qPCR represents significant progress, as demonstrated by a large collaborative study showing its ability to confirm the diagnosis. Furthermore, the survival rate was significantly higher in patients with an initially positive PCR result that became negative after treatment initiation than in patients whose PCR result remained positive (41). In our study, Mucorales qPCR and MS-DS were complementary since all serum samples were positive by at least one test. Although the significance of DS persistence should be compared to a positive PCR result, these results emphasize the benefit of MS-DS in patient care, principally due to the combination of biomarkers whose kinetics of synthesis and release differ during the pathogenic processes. The panfungal characteristics of MS-DS adapted to the broad diversity of emerging fungal pathogens presents some advantages for first-line screening. This simple, robust physicochemically based technology is easily implementable in the majority of clinical mycology laboratories now equipped with MALDI-TOF MS and can be adapted to single tests or a large series (42). However, as the present collaborative study was retrospective, its promising results have to be confirmed through a prospective study. More generally, this method, which allows the early identification and quantification of fungal glycans, is in its infancy, and studies are in progress to explore its potential. In parallel, studies concerning the contribution of MS-DS complemented with currently recommended tests are exploring the possibility of improving antifungal stewardship based on a better knowledge of the diagnostic and prognostic significance of glycobiomarkers.