A prospective comparison of fungal load in paired sputum and bronchoalveolar lavage fluid for Pneumocystis pneumonia diagnosis: Clinical thresholds and diagnostic accuracy

Abstract

Background: To compare diagnostic performances and accuracy of non-invasive specimens and bronchoalveolar lavage fluid (BALF) for diagnosing Pneumocystis pneumonia (PCP) in immunocompromised pneumonia patients. Methods: A prospective study of 112 immunocompromised patients to evaluate P. jirovecii fungal loads in paired sputum and BALF from the same patients. Patients were classified as definite PCP, probable PCP, and non-PCP based on criteria blinded to qPCR results. Optimal diagnostic cut-offs were derived, and agreement between specimen types was analyzed. Results: A BALF fungal load threshold of 2,613 DNA copies/μL demonstrated high sensitivity (82.6 %, 95 % CI: 62.9–93.0) and specificity (96.7 %, 95 % CI: 90.8–99.1) for PCP diagnosis. In sputum, a cut-off of 474 DNA copies/μL offered moderate sensitivity (65.2 %, 95 % CI: 44.9–81.2) and high specificity (89.0 %, 95 % CI: 80.9–93.9), enabling the exclusion of PCP in patients with low clinical suspicion. Fungal loads in sputum correlated well with BALF but showed greater variability in confirmed PCP cases, highlighting limitations for non-invasive diagnosis in high-risk patients. Conclusion: Non-invasive fungal load in sputum provides a valuable tool for excluding PCP in immunocompromised pneumonia patients, potentially reducing the need for bronchoscopy.