Prospective Survey of (1→3)-β-d-Glucan and Its Relationship to Invasive Candidiasis in the Surgical Intensive Care Unit Setting

MATERIALS AND METHODS

All patients in the Memorial Hermann-Texas Medical Center Surgical ICU for at least 48 h with an expected length of stay of at least 3 additional days were eligible for inclusion in the study. The study was conducted from April 2003 to February 2007. The study was approved by the Committee for the Protection of Human Subjects, which is the Institutional Review Board for the University of Texas Health Science Center at Houston. A signed informed consent was obtained from all patients.

Baseline demographics and patient characteristics were collected. When patients showed signs and symptoms of a presumed infection which included, but was not limited to, unexplained fever and leukocytosis, a thorough evaluation for the presumed infectious etiology was carried out based on the standardized protocols in the surgical ICU and included a laboratory evaluation; radiographic evaluation; microbiologic evaluation of blood, urine, respiratory secretions, and/or wounds, where appropriate; and an evaluation of sites with foreign body presence, including catheters, chest tubes, and orthopedic devices.

In addition to the clinical and laboratory evaluation for infection, serum was collected twice weekly during ICU stay and tested for BG using a Fungitell kit. Specimens were frozen at −70°C until testing was performed in triplicate and averaged according to the manufacturer's instructions at the Mycology Research Laboratory at the University of Texas Medical School, Houston, TX. According to the kit package insert, BG levels of ≥80 pg/ml were considered positive. Clinicians did not have access to BG data, since the testing was carried out retrospectively. The clinical course of the patients was monitored until 7 days after ICU discharge for evaluation of evidence of invasive candidiasis based on the criteria (1, 18) established in the ICU in which the present study was being conducted (Table 1). The sensitivity and specificity of the BG assay were determined based on the number of positive samples obtained over the ICU stay relative to the diagnostic criteria outlined in Table 1.

RESULTS

A total of 57 sequential patients met the criteria for enrollment and provided informed consent. The baseline characteristics for these patients are shown in Table 2. Based on the clinical diagnostic criteria for the present study, 15/57 (26%) patients developed invasive candidiasis during their ICU stay. A total of 3/15 (20%) of the invasive fungal infections were proven, 6/15 (40%) were probable, and 6/15 (40%) were possible. The three patients with proven invasive candidiasis all had a positive blood culture for C. albicans, and the patients with probable invasive candidiasis had a variety on nonsterile sites, such as urine, peritoneal fluid, and intravenous catheter tips, that were positive for a variety of Candida species with concurrent clinical signs and symptoms.

There were 239 samples obtained from the 57 patients. The median number of samples obtained from each patient was 4 (range, 1 to 11). Diagnostic performance according to the different diagnostic categories based on the total number of positive samples obtained was analyzed (Table 3). On the date of clinical diagnosis of IC, the sensitivity of a positive BG for identifying IC was 87%, with a 73% specificity.

Of the 35 patients without any clinical evidence of invasive candidiasis and with more than 1 sample obtained, 9 had a positive BG at baseline. In 8 patients, the BG level decreased by an average of 237 pg/ml (range, 10 to 983 pg/ml), and 3 returned to normal with the subsequent sample, despite the lack of antifungal therapy. When eliminating the sample obtained in the first 72 h, the sensitivities and specificities of two consecutive positive BG levels for identifying proven, proven-plus-probable, or proven-plus probable plus-possible IC were 100 and 72%, 90 and 80%, and 78 and 86%, respectively.

The median (range) BG level in patients with proven, probable, or possible IC was 171 (5 to 490), and it was 48 (3 to 388) in patients without any evidence of invasive candidiasis (P = 0.02) (Student t test). In patients with proven and proven-plus-probable IC, the first positive BG was detected an average of 6 and 4 days prior to the clinical diagnosis being made, respectively, based on the date in which the B-glucan demonstrated a positive result and the initial culture results grew yeast.

DISCUSSION

This study presents a systematic survey of BG levels in surgical ICU patients. Our survey found a sensitivity and a specificity of 100 and 50%, respectively, for proven IC. When the number of positive samples required to make a diagnosis was increased to two or three, the specificities increased to 59 and 67%, respectively, without experiencing a decrease in sensitivity. Furthermore, when adding probable and possible cases using clinically relevant definitions (recognizing that the gold standard—blood culture—only has a sensitivity of 50 to 70% in autopsy studies), we documented modest decreases in sensitivity and corresponding slight increases in specificity. Our diagnostic performance findings are similar to those of Pazos et al. (19) in the critical care setting and are in general agreement with large recent surveys of BG in other patient populations and autopsy studies (4, 10, 13, 16, 17, 19). As previously reported, BG levels were able to discern between patients that were ultimately diagnosed with IC, and the BG levels tended to be elevated several days before a positive culture or diagnosis of IC was made (16).

An interesting finding is the frequency of positive BG early in the IC admission and the subsequent decrease in levels. It is unknown whether these represent subclinical infection early in the ICU admission or whether this is related to iatrogenic causes such as translocation/leaching, or introduction, of BG into the bloodstream, such as has been described with surgical gauze, transfusions, hemodialysis, and certain drugs (7, 8, 10, 12). The reason for high BG levels on day 3 of ICU stay and the subsequent decrease remains to be studied in detail.

Aside from a limited sample size, our study is limited by the low frequency of cases of proven IC in this data set, as well as by the use of the local clinical definitions of IC. However, we believe it is important to share this information as it represents performance of BG outside of a clinical trial setting with the typical incidence of IC and practice parameters one would see in a surgical ICU.

The present study confirms the diagnostic value of BG to detect invasive fungal infection (IFI) earlier that waiting for culture results in the surgical critical care setting and describes elevated BG levels in patients without a documented IFI early in the ICU admission. The significance of this finding remains to be explored, until then BG levels should be approached with caution in the first 3 days of ICU admission.