Methods: All adult patients who had a bronchoalveolar lavage (BAL) or sputum sample positive for Pneumocystis by PCR at the U. Michigan Hospitals from Feb 2014–Feb 2018 were studied. Before Feb 2017 samples were tested with endpoint PCR followed by agarose gel electrophoresis & after Feb 2017 with RT-PCR. For each patient, a strict case definition based on host factors, clinical presentation, radiological & pathologic findings, was used to classify PCP as proven, probable, possible, & unlikely. Based on this classification, endpoint PCR and RT-PCR results were designated as true positive or false positive presumably colonized (FP).
Results: The number of specimens tested each year was similar, ranging from 751 to 791. 153 patients tested positive: 77/2318 (3%) by endpoint PCR & 76/783 (10%) by RT-PCR. 126 patients had risk factors for PCP: hi-dose steroids (39), hematologic malignancy (38), chemotherapy within 3 months (24), HIV (14), solid organ transplant (12), stem cell transplant (9), & 27 patients had no PCP risk factors. By our definitions, patients were classified as proven (2), probable (70), possible (46) & unlikely (35). RT-PCR gave a higher FP rate (27/76, 35%) than endpoint PCR (8/77, 10%,) p<.0001, especially in those with chronic lung disease, p=.001 & those with no known PCP risk factors, p<.0001. More patients with no risk factors tested positive with RT-PCR (20) than with endpoint PCR (7), p=.006. FP rates RT-PCR were similar in sputum (34%) & BAL (36%).
Conclusion: RT-PCR gave significantly more FP results, likely due to increased detection of Pneumocystis colonization. Pretest probability should be considered when ordering a highly sensitive test such as RT-PCR & positive results must be interpreted in the context of the clinical presentation, radiological findings & risk factors.
M. A. Bachman, None
M. H. Miceli, None