Pneumocystis pneumonia risk among viral acute respiratory distress syndrome related or not to COVID 19

Lymphopenia, corticosteroids and immunomodulatory therapeutics frequently used in COVID-19 patients with acute respiratory distress syndrome (C-ARDS) may be contributing factors to opportunistic infection such as Pneumocystis jirovecii pneumonia (PCP).

We conducted a retrospective study to compare the prevalence of PCP between patients with C-ARDS and those with non-SARS-CoV-2 viral ARDS (NC-ARDS).

Methods and some data from this cohort have been previously published [1]. There was no systematic protocol to search for PCP but in case of suspicion of PCP (respiratory symptoms with any consistent radiographic features), several analyses were performed on respiratory samples, such as broncho-alveolar lavage (BAL), blind protected sample, or sputum. It included direct examination (using May-Grünwald Giemsa (MGG), or immunofluorescence staining), detection of Pneumocystis jirovecii DNA by real-time polymerase chain reaction (qPCR) [2], and serum (1–3)-β-D-glucan. During the COVID-19 outbreak, immunofluorescence staining was not performed. PCP was defined as per the revised EORTC/MSGERC definition [3] as follows: proven in case of suspicion with positive direct examination; possible in case of suspicion with positive qPCR and positive BDG in ≥ 2 consecutive serum samples provided other etiologies have been excluded. SARS-CoV-2 and other viruses were not considered a priori as host factors. Patients with positive qPCR but lacking the other criteria for possible PCP were classified as colonized.

The primary endpoint was the difference in prevalence of PCP between C-ARDS and NC-ARDS patients.

No statistical sample size calculation was performed a priori, and sample size was equal to the number of patients treated during the study period. All patients were included only once.

Between October 1, 2009, and April 29, 2020, ninety patients had C-ARDS (positive RT PCR test for SARS-CoV-2), while 82 patients had viral NC-ARDS. Our study comprises 120 patients with fungal analyses on respiratory samples obtained from 81 C-ARDS and 39 NC-ARDS patients. NC-ARDS patients had more comorbidities were more often immunocompromised, and had lower lymphocyte counts than C-ARDS patients (Table 1). C-ARDS patient received less steroid than NC-ARDS patients because they were included before randomized trials demonstrating decreased mortality with dexamethasone.

Table 1 Characteristics of patients with Pneumocystis jirovecii research according to C-ARDS and NC-ARDS patients

Pneumocystis analyses were performed on a mean of 3.1 respiratory sample per patient (range 1–15). Direct examination was performed in a total of 72 samples, with two positive cases. qPCR was performed in a total of 368 samples (294 blind protected samples, 72 BAL, and three sputum). All qPCR were negative in C-ARDS patients, while five (13%) NC-ARDS patients had at least one positive PCR, with a median cycle threshold of 36.6 [30–38.3].

Two NC-ARDS patients fulfilled proven PCP diagnostic criteria, with a positive direct examination, a single ß-D-glucan > 80 pg/mL (Table 2), and received treatment for PCP.

Table 2 Characteristics of patients with a positive Pneumocystis jirovecii qPCR

Three other NC-ARDS patients were classified as colonized, while no patient fulfilled possible PCP diagnostic criteria. Time between ICU admission and positive sample for PCP (Table 2) was short (< 2 days) like in other invasive fungal infections (i.e. invasive pulmonary aspergillosis) in severe influenza infection or ARDS.

In this study of patients with viral ARDS, we found a low risk for possible or proven PCP. Our findings are in accordance with two smaller studies in France [4, 5] retrieving a low risk of Pneumocystis colonisation in COVID-19 patients. In our cohort, qPCR was positive in 13% of NC-ARDS. This result is in accordance with a previous study showing 7% of positive qPCR in ICU-admitted influenza patients [6].

The strengths of our study are the analysis of a large ARDS cohort with fungal analyses. Our study also has limitations: monocentric design, NC-ARDS patients more frequently immunocompromised, and a long cohort period.