Objective: The objective of this study was to define the test characteristics of plasma beta-glucan for diagnosis of Pneumocystis jirovecii pneumonia (PCP) in AIDS patients with respiratory symptoms.
Design: Analysis of baseline blood samples in a randomized strategy study of patients with acute opportunistic infections, limited to participants with respiratory symptoms.
Methods: Participants in the 282-person ACTG A5164 trial had baseline plasma samples assayed for beta-glucan testing. As part of A5164 trial, two study investigators independently adjudicated the diagnosis of PCP. Respiratory symptoms were identified by investigators from a list of all signs and symptoms with an onset or resolution in the 21 days prior to or 14 days following study entry. Beta-glucan was defined as positive if at least 80 pg/ml and negative if less than 80 pg/ml.
Results: Of 252 study participants with a beta-glucan result, 159 had at least one respiratory symptom, 139 of whom had a diagnosis of PCP. The sensitivity of beta-glucan for PCP in participants with respiratory symptoms was 92.8% [95% confidence interval (CI) 87.2–96.5], and specificity 75.0% (95% CI 50.9–91.3). Among 134 individuals with positive beta-glucan and respiratory symptoms, 129 had PCP, for a positive predictive value of 96.3% (95% CI 91.5–98.8). Fifteen of 25 patients with a normal beta-glucan did not have PCP, for a negative predictive value of 60% (95% CI 38.7–78.9).
Conclusion: Elevated plasma beta-glucan has a high predictive value for diagnosis of PCP in AIDS patients with respiratory symptoms. We propose an algorithm for the use of beta-glucan as a diagnostic tool on the basis of the pretest probability of PCP in such patients.