Abstract
We compared the specificities and sensitivities of four tests used for the serodiagnosis of candidemia in 39 patients with candidemia, including 10 patients with superficial Candida colonization, 10 patients with deep mycosis, and 20 healthy subjects. The results obtained by the dot immunoblotting assay for detecting the enolase antigen (48 kDa) were compared with those of assays for detecting mannan antigen, heat-labile antigen (a threshold titer of four times), and beta-glucan (> or = 60 pg/ml). Enolase antigen was detected in 28 (71.8%) patients with candidemia, while 30 (76.9%), 10 (25.6%), and 27 (84.4%) patients were positive for the heat-labile antigen by the Cand-Tec assay, the mannan antigen by the Pastorex Candida assay, and beta-glucan by the limulus test, respectively. Ten patients with superficial Candida colonization, 5 patients with invasive pulmonary aspergillosis, 5 patients with cryptococcosis, and 20 healthy subjects were negative for both enolase antigen and mannan antigen. Two patients with superficial Candida colonization, one patient with invasive pulmonary aspergillosis, and two patients with cryptococcosis were positive by the Cand-Tec assay. The beta-glucan concentration was more than 60 pg/ml in all patients with invasive pulmonary aspergillosis; however, it was less than 10 pg/ml in all patients with cryptococcosis. The specificity of enolase antigen in the serodiagnosis of candidemia was 100%, but the sensitivity was 71.8%. The specificity and sensitivity of Cand-Tec, the assay for mannan antigen, and the assay for beta-glucan were 76.9 and 87.5%, 25.6 and 100%, and 84.4 and 87.5%, respectively. Our results demonstrated that antigen detection tests are useful for the diagnosis of candidemia; however, none is satisfactory for the serodiagnosis of candidemia. We suggest that a combination of two assays may increase the accuracy of diagnosis of candidiasis.