Co-infection with Legionella and SARS-CoV-2, France, March 2020

We describe a March 2020 co-occurrence of Legionnaires’ disease (LD) and coronavirus disease in France. Severe acute respiratory syndrome coronavirus 2 co-infections were identified in 7 of 49 patients from LD case notifications. Most were elderly men with underlying conditions who had contracted severe pneumonia, illustrating the relevance of co-infection screening.


The Study
In France, LD surveillance is based on mandatory notifi cations to Santé Publique France, the national public health agency. To evaluate LD and COVID-19 co-occurrence, we retrospectively studied all LD case notifi cations with symptom onset during March 2020 and included cases in which patients had clinical or radiologic signs of pneumonia combined with Legionella culture, positive Legionella PCR from broncho-pulmonary secretions, or positive Legionella pneumophila serogroup 1 urinary antigen test (UAT) results. There were 65 LD case notifi cations in March 2020 compared with 79 in March 2019. To evaluate the number of UATs, which are performed in 96% of LD cases (7), we contacted the 59 reporting laboratories (in 47 cities), 33 of which sent the relevant data. Among the 65 patients from the case notifi cations, 49 were tested for both LD and COVID-19 and 12 for LD only; no information was available for 4. The frequency of proven LD/COVID-19 co-infection was 14.3% (7/49). This fi nding may be an overestimate because COVID-19 incidence was <5 cases/100,000 persons in the region of residence of the 16 patients not tested at the time of symptom onset; actual coinfection frequency could be from 10.8% (7/65) to 14.3% (7/49).
Most patients (4/7) with co-infection lived in the Grand Est region, the area in France with the most COVID-19 cases during the study period and a region that usually reports a high number of LD cases. Median patient age was 72 years (range 37-83 years); male-to-female ratio was 6:1 (Table), higher than for the overall COVID-19-infected population (8). Of interest, the male-to-female ratio for LD has elsewhere been reported as ≈3:1 (7), similar to the ratio observed in the LD-positive/COVID-19-negative cases.
At hospital admission, co-infected patients had more underlying conditions; 6 (86%) of 7 patients had ≥1, compared with 25 (60%) of 42 LD-positive/ COVID-19-negative patients (Table). For cardiovascular diseases, the proportions were 6 of 7 among co-infected and 1 of 42 among LD-positive/CO-VID-19-negative patients. Sources of LD exposure DISPATCHES 1 Members are listed at the end of this article. We describe a March 2020 co-occurrence of Legionnaires' disease (LD) and coronavirus disease in France. Severe acute respiratory syndrome coronavirus 2 co-infections were identifi ed in 7 of 49 patients from LD case notifi cations. Most were elderly men with underlying conditions who had contracted severe pneumonia, illustrating the relevance of co-infection screening.
were reported in case notifications for 3 (43%) of 7 co-infected versus 14 (33%) of 42 non-co-infected patients; the proportions of exposure sources reported was similar between the 2 groups. Despite the implementation on March 15 of the COVID-19 national lockdown in France, halting travel, 12 (24%) of 49 exposures from the LD notifications were travel associated, a ratio similar to that in a 2017 report (9). Therefore, the decrease in LD cases observed in March 2020 cannot be explained by decreased travel.
Community-acquired LD and COVID-19 were diagnosed at hospital admission in 5 of 7 patients with both infections. For patient 2 (Table), whose symptoms started 48 hours before admission, UAT was not performed until 7 days after admission. Hospital-acquired COVID-19 was suspected in patient 3 because he initially tested negative but was positive after a 4-week hospitalization (Figure 1). All 7 co-infected patients required admission to an intensive care unit (ICU; median stay 13 days,; range 2-34 days). In contrast, only 10 (32%) of 31 of LD-positive/COVID-19-negative patients required ICU, similar to LD-only patients in previous reports (7).
The longitudinal follow-up of patient 1, a 71-year-old man receiving chemotherapy for multiple myeloma, may help decipher the kinetics of each pathogen load. Hospitalized for fever (39°C) and productive cough, he required ICU admission on day 9 because of acute respiratory distress syndrome. A thoracic computed tomography scan found left lobar  atelectasis, multiple ground-glass opacities compatible with COVID-19, and pleural effusion suggesting possible bacterial infection. Results of UAT and nasopharyngeal SARS-CoV-2 reverse transcription PCR were both positive. On day 10, a serum sample was PCR positive for both SARS-CoV-2 and Legionella ( Figure 2, panel A); each pathogen has individually been associated with COVID-19 (11) and LD (12) severity. Beginning on day 10, longitudinal samples of the lower respiratory tract collected every 3-6 days showed a high SARS-CoV-2 viral load (7.5 log 10 RNA copies/100 cells), followed by a decrease to 1.3 log 10 RNA copies/100 cells within 21 days ( Figure  2, panel B). In contrast, lung Legionella DNA load increased and remained high (cycle threshold 21.9) until day 31. To identify potential bacterial co-infections, we performed a lung microbiota analysis on a D19 bronchoalveolar lavage using 16S MinION long-read sequencing technology (Oxford Nanopore Technologies, https://nanoporetech.com). Similar to another study of LD microbiomes (13), we found a predominance of Legionella (61%) and the presence of commensal lung bacteria (Appendix Figure, https://wwwnc.cdc.gov/EID/article/27/11/20-  2150-App1.pdf) but no additional bacterial co-infection. On day 26, while high lung Legionella DNA load persisted, a third chest scan found pseudocavitation. Persistence of culture or PCR positivity in respiratory samples, or both, has been described in patients with Legionella lung abscesses, especially if immunocompromised (14).

Conclusions
Our study found a substantial proportion of patients in LD notifications in France during March 2020, mostly elderly men with underlying conditions, also had COVID-19. They required ICU admission more frequently and had a higher case-fatality rate than patients without SARS-CoV-2 co-infection, but these rates were similar to that for all ICU-admitted LD patients (7,10). Overall, health effects from co-infections were more severe than from single infections, perhaps because of cumulative effects or because patients with co-infections may be more likely to have risk factors associated with poor outcomes. Another possibility is that SARS-CoV-2 infection may be more severe in this population. Longitudinal monitoring of a single co-infected patient found a first phase of predominant SARS-CoV-2 replication followed by a resurgence of Legionella and worsening of respiratory symptoms while SARS-CoV-2 decreased. Initial viral infection could establish pulmonary damage suitable for the bacteria to develop, similar to how bacterial superinfections develop in influenza-infected patients (15). Such co-infections may lead to poor prognoses as demonstrated here and elsewhere (3), underlining the importance of extensive screening for respiratory pathogens in patients with suspected or confirmed COVID-19. Because Legionella and other pulmonary microorganisms share clinical and radiological features with SARS-CoV-2 infection, they should be included in COVID-19 differential diagnoses (3).