Syscan3, a Kit for Detection of Anti-Candida Antibodies for Diagnosis of Invasive Candidiasis

Patient samples.

Two frozen (−80°C) deidentified and anonymized serum collections were used. Collection A consisted of 76 subjects: 27 hospitalized patients with proven invasive candidiasis (26 with candidemia and 1 with candidal peritonitis), 6 hospitalized patients with noncandidal fungal infections (4 with cryptococcosis and 2 with invasive mold infections), and 43 healthy control subjects. The Candida species distribution was as follows: C. albicans, 16/27; C. glabrata, 4/27; C. tropicalis, 3/27; coinfection with C. albicans and C. glabrata, 2/27; and C. krusei and C. parapsilosis, 1 each. Sera from patients with invasive candidiasis were drawn within 96 h of the positive culture. Collection B consisted of sera collected twice weekly from 293 patients with acute myelogenous leukemia and myelodysplastic syndrome undergoing induction chemotherapy and receiving antifungal prophylaxis with either itraconazole or caspofungin. Thirteen of these patients were found to have proven or probable invasive candidiasis. The species distribution was as follows: C. glabrata, 7/13; C. albicans, 2/13; C. krusei, 2/13; and C. parapsilosis and C. tropicalis, 1 each. A single sample was selected for testing from each patient. For patients with invasive candidiasis, a sample was selected within 96 h of the diagnosis or the positive culture. For uninfected patients; the first available sample was utilized. Invasive candidiasis was defined according to the EORTC/MSG criteria (3). This study was approved by the UTHSCH Institutional Review Board.

Testing method.

Syscan3 testing kits were provided by Rockeby Biomed Ltd. (Western Australia, Australia). Samples from each collection were mixed with a phosphate-buffered saline diluent (10 μl of serum mixed into 990 μl of serum diluent). One hundred microliters of the respective dilutions was deposited into the wells of a 96-well microdilution tray in which the wells had been coated with a proprietary mixture of purified intracytoplasmic Candida antigens, with enolase as the predominant antigen. Samples were incubated for 45 min at room temperature, washed, and incubated with horseradish peroxidase-conjugated antihuman antibodies for 45 min. After washing the wells with buffer, a peroxidase solution and acid stop solution were added. Sample absorbance was read using a dual-wavelength spectrophotometer at 450 nm with a reference of 650 nm. The kit also included positive, negative, and cutoff controls. Controls and sera were tested in duplicate. The reading for each test sample was determined in arbitrary units as (sample absorbance × 10)/(mean absorbance of cutoff control sample). The negative and positive controls had to fall within predetermined quality control ranges to accept the results as valid. A cutoff of ≥15 U was selected based on a preliminary study by the kit developers and by determining receiver operating characteristic curves with collection A (data not shown). Means of units between groups were compared by t test (SPSS 12.0.1; SPPS, Inc.), and diagnostic test performance was evaluated using standard formulas.

Results.

For collection A, the mean numbers of units ± standard deviation in patients versus controls were 20.78 ± 6.81 U and 11.24 ± 5.94 U (P < 0.0001). For collection B, the mean numbers of units ± standard deviation in patients versus controls were 10.98 ± 6.58 U and 13.85 ± 6.98 U (P = 0.15). Using the cutoff of ≥15 U, for collection A, 20 of the 27 patients with invasive candidiasis were detected by Syscan3, while only 2 of 13 patients in collection B were identified. All of the patients with non-Candida fungal infections had negative results. As shown in Table 1, the sensitivity, specificity, positive predictive value, and negative predictive value for collection A were 74%, 75%, 62%, and 84%, respectively, while for collection B they were 15%, 60%, 1.7%, and 93%. Decreasing or increasing the cutoff for collection B did not improve the performance of the test.

Our evaluation of Syscan3 in both collections A and B yielded relatively high negative predictive values when compared to previously available commercial kits (15), thus offering a good possibility of “ruling out” the disease. The significance of the negative and positive predictive values should be approached with caution in collection A since these values rely on the prevalence of disease in the study population, which in the case of this collection, was artificially constructed (12). In the immunocompromised hematological patients evaluated in collection B, sensitivity and positive predictive values were low. Of note, the mean test results for the cases in collection B were similar to the values of controls in both collections. This might be due to reduced antibody production in the immunocompromised population (6) or the use of antifungal prophylaxis, thus preemptively treating infections in this subject group. Although the negative predictive value appears to be fairly good for collection B, the negative likelihood ratio is 0.38; thus, a negative test may result in only a small decrease in the probability that a patient in this population has an invasive Candida infection. Nevertheless, a statistical artifact related to the low prevalence of the studied disease cannot be excluded (12).

In summary, the Syscan3 kit offers reasonable negative predictive value. A prospective, multicenter evaluation of Syscan3 kit is needed to further evaluate its clinical usefulness in immunocompetent and other immunocompromised patients (e.g., human immunodeficiency virus-infected and organ transplant patients), as well as settings (such as intensive care units) with a high prevalence of invasive candidiasis. The effects of Candida colonization and superficial Candida infections remain to be studied. Also, repeat testing with multiple samples during the course of infection may help in more accurately diagnosing invasive candidiasis, as has been previously noted with other serological tests (2, 6, 17).