754. Performance Characterization of a Real Time PCR Assay for Pneumocystis jirovecii in Bronchoalveolar Lavage Fluid and Sputum

Abstract

Background Pneumocystis jirovecii pneumonia (PJP) affects immunocompromised patients and contributes significantly to mortality. Outcomes depend on early treatment, making timely and accurate diagnosis critical. Typically, PJP diagnosis is through identification of trophozoite or cyst forms in bronchoalveolar lavage (BAL) fluid or sputum, a labor-intensive and insensitive process. Options for more accessible and sensitive molecular detection are limited. It is known that patients may be colonized, which can cast doubt on the clinical significance of low levels of DNA amplification in qualitative result reporting. In this study, we describe a real time (rt) PCR assay utilizing analyte specific reagent primers targeting the mtLSU gene of P. jirovecii and correlate amplification with morphological PJP identification.MethodsIUHPL Clinical Microbiology assessed sputum or BAL fluid from 109 patients with clinical concern for PJP microscopically via fungal stains (GMS, calcofluor white). Comparative rtPCR was conducted as follows. First, 2µL of residual specimen or control were mixed with an 8µL combination of rtPCR mastermix, control DNA, and primer pairs (Simplexa). No nucleic acid extraction was performed. Real time PCR was executed and analyzed on the LIAISON MDX (DiaSorin) platform. Qualitative amplification results and cycle threshold (CT) values were correlated with microscopic methods to establish performance. Chart review was performed to assess the clinical impact of this assay.Results P. jirovecii was microscopically detected in 26% (29/109) of samples, while 31.1% (34/109) exhibited amplification by rtPCR. Agreement between the two methods was 95.4%; rtPCR demonstrated 100% sensitivity and 93.8% specificity in comparison.ConclusionOur results indicate that this assay has exceptional negative predictive value (100%), and therefore may be valuable as a screening test. Considering this data alone, the positive predictive value is lower (85.3%). Further examination of the data, however, revealed that 80% (4/5) of discrepant results demonstrated CT values of >34, while the highest CT for a microscopically positive sample was 31.2. Further clinical correlation may establish a CT cutoff that will reduce false positive and potentially clinically insignificant cases.Disclosures All Authors: No reported disclosures