Mortality rates of IC episodes in the critical care setting remain excessively elevated due to diagnosis difficulties and, subsequently, inappropriate treatment. IC diagnosis usually requires a high index of suspicion and is difficult because the infection lacks pathognomonic signs, blood cultures are often negative and, in many instances, it is not possible to obtain specimens for histology; therefore, serology could be an aid in its diagnosis, as it provides specific biomarkers [2, 24, 25]. Demonstration of IC may not be apparent because the infectious burden is beneath the analytical sensitivity of the chosen diagnostic modality, or infection remains limited in the context of neutrophil recovery or follows administration of prophylactic or empirical antifungal therapy. Serological tests for the diagnosis of IC by detecting antibodies against different antigens of Candida must differentiate Candida colonization or superficial infection from tissue invasion and candidemia requiring antifungal therapy. Sera from patients with IC recognize a mannoprotein of 230-250 kDa located on the germ tube cell wall surface. Pontón et al. [26, 27] have developed an indirect immunofluorescence assay to detect antibodies (CAGTA) against this antigen present in C. albicans germ tubes, which has been a useful biomarker for the diagnosis of IC in different groups of patients, including intravenous heroin users, BMT recipients, patients with haematological disorders and ICU patients [16, 18, 28, 29]. The test has shown an overall sensitivity of 77-89% and a specificity of 91-100%. Although the titres found in immunocompromised patients are lower than those found in immunocompetent patients, the overall performance of the test is similar. Detection of CAGTA in patients with invasive infections caused by Candida species other than C. albicans (Candida tropicalis, Candida parapsilosis, Candida glabrata, Candida dubliniensis, Candida guilliermondii and Candida krusei) may also be positive, although titers are lower than in candidiasis by C. albicans [16, 27, 29–31].
Candida albicans IFA IgG has been recently commercialized for CAGTA detection. This test has been compared in a retrospective study to the standard test using 172 sera from 51 haematological and intensive care patients . The commercially available test was similar to the standard test and provided faster and easier diagnosis of IC in the clinical microbiology laboratory.
To our knowledge, this is the first time that the new commercial kit for CAGTA detection has been studied in the ICU setting. It is well known that the global sensibility rate of blood culture is not too high (50%) in the hospital population . Moreover, this ratio is considerable lower in the ICU setting as León et al.  observed in the Candida score study, where the ratio of proven IC was only 5.7%. However, until now this procedure is still the gold standard for candidemia diagnosis. The reasons for the low number of positive blood cultures in critical care setting are unknown but are likely to be related to the prophylactic and empiric treatments administered to these patients [33, 34]. In our study, no blood cultures were positive. The low ratio of positive blood cultures in critically ill patients obtained in all studies in this setting, including this one, confirms the compelling necessity of finding another sound gold standard for validating new diagnostic tools [12, 35, 36]. Then, serological data may be used as evidence to implicate Candida spp. when mycological and/or histological data are negative or cannot be acquired. The high prevalence of CAGTA-positive results in the population studied (41.5%) corroborates the adequacy of the inclusion criteria used in this study as a predictive biomarker of Candida infection and the need of consolidating the CAGTA determination in a delimited group of ICU population. For these reasons and for the excellent sensitivity and specificity previously reported , all the CAGTA-positive patients could be considered as "probable" IC. The preliminary results presented in this study evidence the need for a large multicentre study to validate the CAGTA detection as an alternative diagnosis tool in the ICU setting [16, 30].
This study identifies the principal risk factors associated with serological proven candidiasis (Table 1) and, probably, with hidden invasive Candida infection in the ICU population due to the absence of other microbiological positive tests. Only previous surgery was statistically more frequent in patients with CAGTA-positive results. Furthermore, in our study previous surgery is a determinant risk factor for candidemia as has been described by León et al. ; however, although diabetes mellitus has been previously described as a risk factor for developing IC in critically ill patients , our findings do not support the inclusion of diabetes mellitus in clinical scores to predict IC in ICU patients. Recently published clinical scores such as the "Candida score"  or the Ostrosky-Zeichner et al. rule  have not included this risk factor. Consequently, periodical CAGTA determination should be mandatory especially in ICU surgical patients to achieve an early diagnosis and, therefore, to improve the candidemia prognosis, especially in those patients with patterns of increasing CAGTA titres . This assay could improve the predictor value of these previously described prediction scores and rules.
The rate of highly colonized patients was very high as has been previously described  and no statistical differences were found between CAGTA-positive and -negative patients at the end of the study. This finding could corroborate the CAGTA technique as a key biomarker to distinguish between colonization and infection in these patients.
In our study, the decision to add empiric antifungal therapy for patients with suspected IC was at the discretion of the prescribing physician based on clinical criteria, but was never guided by CAGTA results. The no difference observed neither the global administration of antifungal treatment nor antifungal agent used between CAGTA-positive and -negative give more value to this technique avoiding false-negative results in presence of any antifungal therapy.
Several limitations must be noted in this study. First, the small number of patients due to the difficulty to enrol this kind of patients with predefined criteria. Moreover, there was no possibility to establish the sensibility and specificity of CAGTA technique due to the absence of microbiological and histological documentation of IC in this study.