Abstract
Quantitative PCR (qPCR) is now a key diagnostic tool for Pneumocystis pneumonia. However, cutoffs to distinguish between infected and colonized patients according to their HIV status have not yet been determined. According to clinical, radiological, and biological data, we retrospectively classified bronchoalveolar lavage (BAL) samples subjected to qPCR over a 3-year period into four categories, i.e., definite PCP, probable PCP, Pneumocystis colonization, and no infection. Fungal burden was then analyzed according to the HIV status of the patients. Among 1,212 episodes of pneumonia screened in immunocompromised patients, 52 and 27 HIV-positive patients were diagnosed with a definite and probable PCP, whereas 4 and 22 HIV-negative patients had definite and probable PCP, respectively. Among patients with definite or a probable PCP, HIV-negative patients had a significantly lower burden than HIV-positive patients (P < 10(-4)). In both groups, the median fungal burden was significantly higher in patients with definite PCP than in colonized patients. A single cutoff at 1.5 × 10(4) copies/ml allowed to differentiate colonized and infected HIV-positive patients with 100% sensitivity and specificity. In HIV-negative patients, cutoff values of 2.87 × 10(4) and 3.39 × 10(3) copies/ml resulted in 100% specificity and sensitivity, respectively. Using cutoffs determined for the whole population would have led us to set aside the diagnosis of PCP in 9 HIV-negative patients with definite or probable PCP. qPCR appeared to be the most sensitive test to detect Pneumocystis in BAL samples. However, because of lower inocula in HIV-negative patients, different cutoffs must be used according to the HIV status to differentiate between colonized and infected patients.