We retrospectively included all adult LTx recipients with at least one lower-respiratory-tract specimen in which a Corynebacterium spp. was isolated between July 2014 and December 2018 in the Paris-Bichat Lung Transplant Program. Cases were identified in the local microbiology department database, where all respiratory samples are recorded. Each case was matched to a non-exposed control, selected as the next LTx patient in chronological order. The matching was according to age at LTx ±5 years, mono- or bipulmonary status, underlying respiratory disease (defined in four categories: chronic obstructive pulmonary disease [COPD]/emphysema, interstitial lung disease, bronchial dilatation, miscellaneous). Data were collected anonymously, and the electronic files were used according to French law (Informatique et Libertés). The Evaluation Committee for observational research protocols of the French Respiratory Society (SPLF, CEPRO 2020-044) approved the study and waived informed consent.

All LTx candidates and recipients have been prospectively included in Paris-Bichat Lung Transplant database since 2006. This database includes demographical and anamnestic data, details on LTx surgery and post-operative course, bronchoscopic findings, and respiratory function.

All lower-respiratory-tract samples (sputum, tracheal aspirate, bronchoalveolar lavage [BAL], or protected distal aspiration) taken during usual care were immediately sent to the bacteriology laboratory. They were inoculated onto routine agar plates incubated for 48 h at 35°C under aerobic and anaerobic conditions. Bacteria were identified at the species level by matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) Microflex LT Biotyper (Bruker Daltonics, Bremen, Germany). Bacterial susceptibility to antibiotics was determined with the disk-diffusion method according to EUCAST guidelines (www.eucast.org).

From the bacteriology laboratory database, we retrieved all cases of LTRs in whom Corynebacterium spp. had been isolated in at least one lower-respiratory-tract specimen. In patients with sputum and tracheal aspirates, we included only specimens showing ≥25 leukocytes/field and ≤10 upper respiratory epithelial cells/field, as assessed by the scoring system of Murray and Washington (10). The usual thresholds were applied for interpreting quantitative cultures (i.e., ≥10⁴, ≥10⁵ and ≥10⁷ colony formation units/mL for BAL specimens, tracheal aspiration and sputum culture, respectively). Microbiological data are described in terms of the first Corynebacterium spp. isolation.

Usual management of LTx in our center is highly protocolized and reported elsewhere (11). In brief, intraoperative veno-arterial extra-corporeal membrane oxygenation (ECMO) was initiated according to hemodynamics and respiratory findings during surgery (12), with peripheral cannulation. All patients receive the same initial immunosuppressive regimen (mycophenolate mofetil, corticosteroids and tacrolimus). All patients receive life-long proton pump inhibitors. Antibiotic prophylaxis with cefazoline is administered for 48 h, then adapted to postoperative microbiological analysis. A first bronchoscopy is systematically performed within the first hours after LTx. During the post-operative course, surveillance bronchoscopy and BAL are performed in case of clinically suspected respiratory infection. In case of abnormalities in bronchial healing, these bronchoscopies are repeated, and microbiological samples are taken if an infection is suspected, thus allowing for longitudinal study of colonization.

Transbronchial biopsies are performed in case of clinically suspected acute cellular or antibody mediated rejection (AMR). Acute cellular rejection (ACR) was defined according to established criteria (13), as was AMR (14).

We defined an SBC as the presence of a bronchial fistula diagnosed by chest CT-scan or bronchoscopy or the need for interventional bronchoscopy for dilatation or bronchial stenting (2). Persistent respiratory colonization was defined by isolation on a respiratory sample on at least three occasions at least 1 month apart in less than 1 year (15). Chronic lung allograft dysfunction (CLAD) was defined according to ISHLT recommendations (16) as a decline in forced expiratory volume per second (FEV1) ≥ 20% from baseline, persisting at 3-month intervals, excluding other causes. Baseline FEV1 was the mean of the 2 best post-transplant FEV1 measurements. The diagnosis of infection (pneumonia or bronchitis or colonization) was retrospectively defined according to Centers for Disease Control and Prevention definitions (17) by the review of all the medical files by a blinded adjudication committee.

The primary outcome was the occurrence of an SBC. The date of the event was the date of the first SBC. Secondary outcomes were 1) persistent respiratory colonization with Corynebacterium spp., 2) CLAD occurrence and its delay after transplantation, 3) all-cause mortality during follow-up, and 4) Corynebacterium species and their distribution.

First, a descriptive analysis was performed in the entire cohort population and according to exposed or unexposed status. Categorical variables were summarized as counts (percentage) and frequency distributions were compared with the Mac Nemar test. Continuous variables were expressed as median (IQR) and differences were tested with the Wilcoxon test.

Second, we compared occurrence of SBC between exposed and unexposed patients in the subset of patients with respiratory Corynebacterium spp. isolation before the occurrence of SBC and their matched controls, using the Mac Nemar test. We also searched for factors associated with SBC among the following variables: Corynebacterium spp. isolation, underlying respiratory disease, invasive mechanical ventilation duration, ECMO necessity, by a univariate analysis. Third, factors associated with with Corynebacterium spp. isolation were investigated by univariate then multivariate logistic regression in the entire cohort population, using the same approach.

Time between LTx and occurrence of CLAD or death was compared between the exposed and unexposed groups by means of survival curves (Kaplan-Meier method) and tested by means of a log-rank test. If the patient was alive or without chronic rejection at the end of the study, the patient was censored at the study end date (December 31, 2018). Those analyses excluded patients with CLAD before isolation of Corynbacterium spp. Statistical tests were 2-sided with a significance level of 0.05. All analyses were performed using R software (version 4.0.3).

Over the study period, the cohort of LTx recipient represented 367 patients (Figure 1). Corynebacterium spp. were isolated in 60/367 LTRs (16.3% of the cohort) after a median of 128 days [interquartile range 38–503] after LTx. One patient was excluded from the analysis because Corynebacterium spp. had also been isolated before LTx on a systematic bronchoscopy, which left 59 LTR as the “exposed” cohort; these were matched to 59 non-exposed LTRs. Patient characteristics are in Table 1.

Strict matching for the underlying respiratory disease leading to transplantation was not possible for 3 cases with rare lung diseases (histiocytosis, lymphangioleiomyomatosis and pulmonary graft-versus-host disease) who were matched with patients transplanted for COPD. Matching on mono- or bipulmonary status was favored whenever possible, to eliminate the risk of confounding colonization or infection of the native lung on isolation of Corynebacterium spp.

All patients were receiving systemic corticosteroid therapy, with no difference between exposed and non-exposed patients in median dose of corticosteroids or use of antimetabolites or anticalcineurins. Patients with Corynebacterium spp. isolated significantly more frequently received other types of immunosuppressive therapies than non-exposed patients (7/59, 11.9% vs. 1/59, 1.8%, p = 0.04; mammalian target of rapamycin inhibitors for 4 patients, belatacept for 2, and rituximab in the previous 6 months for 1 vs. belatacept for 1) (Table 1).

A history of AMR was more common in the exposed than non-exposed group (16/59, 27.1% vs. 5/59, 8.47%, p = 0.015) (Table 1). None of the included patients underwent fundoplication surgery.

Presence of a SBC was significantly more frequent in the exposed than non-exposed group: 35/59 (59.3%) and 12/59 (20.3%) (p < 0.001). Likewise, an interventional bronchoscopy procedure and placement of a bronchial stent were more frequently required in exposed patients (31/59, 52.5% vs. 10/59, 16.9% and 19/59, 32.2% vs. 2/59, 3.4%; both p < 0.001). Among the patients with bronchial stents, 12 in the exposed group had a mechanical stent obstruction requiring bronchoscopy and no patient in the non-exposed group.

We analyzed data for 74 patients (Table 2) to evaluate whether Corynebacterium spp. was a risk factor for SBC: 21 patients in the exposed group had an SBC before Corynebacterium spp. isolation and were excluded from the analysis with their matched control. One non-exposed patient had an SBC before the matched exposed counterpart had Corynebacterium spp. isolated. Therefore, he was excluded. The respiratory isolation of a Corynebacterium spp. was not associated with increased frequency of further SBC (OR 2.33, IC95 0.60–9.02, p = 0.220). The time to onset of SBC did not significantly differ between the two groups. None of each SBC type was significantly more frequent in patients with an Corynebacterium spp. isolated.

We compared data for exposed and unexposed patients by univariate then multivariate logistic regression (Table 3).

The presence of a previous SBC and history of corticosteroids pulses in the last 3 months were the only factors associated with an increased risk of Corynebacterium spp. isolation in multivariate analysis (OR 12.38, 95% CI [3.92–39.11]; p < 0.001, and 6 [1.14–31.49]; p = 0.0034). Prior antibiotic therapy exposure and history of chronic allograft dysfunction were not associated with isolation of Corynebacterium spp.

The presence of Corynebacterium spp. in a respiratory sample was associated with a lower respiratory-tract-infection pattern or functional decline in 34/59 (57.6%) LTRs, including 9 (15.3%) with a monomicrobial isolation (Supplementary Table S1). Nine (15.3%) had radiological pneumonia, 2 (3.4%) with monomicrobial isolation. Overall, 20 (33.9%) patients had signs of bronchitis: cough in 14 (23.7%), sputum or bronchoscopic evidence of purulent secretion in 13 (22%) and both in 7 (11.9%). Five (8.5%) had functional decline associated with dyspnea. Corynebacterium spp. isolation was associated with functional decline in 14 (24.6%) patients, 5 (8.5%) with monomicrobial isolation. In total, 25 (42.4%) patients had no clinical or biological sign of infection and were therefore considered colonized.

In 42/59 (71%) patients, the first Corynebacterium spp. respiratory isolation occurred during the hospital stay, including 14 (24.7%) in the intensive care and 7 during the immediate post-LTx stay. Among the 7 patients admitted to the intensive care unit, 3 (21.4%) had acute respiratory failure and 4 (28.6%) respiratory-related sepsis. Oxygen therapy was needed for 19 (32.2%) patients, invasive mechanical ventilation for 11 (18.6%) patients, and non-invasive ventilation for one patient.

C. striatum was the most frequently retrieved species, accounting for 71.2% of patients (n = 42), followed by C. amycolatum (14 patients, 23.7%) (Supplementary Table S2); C. pseudodiphteriticum, C. accolens, and C. propinquum were isolated from one patient each. In 41/59 (69.5%) patients, at least one other bacterium was isolated from the respiratory sample, with 25 (43.1%) above the significance threshold. P. aeruginosa was the main bacterium isolated from the plurimicrobial samples, in 18 (30.5%) patients.

In total, 18 (30.5%) patients received effective antimicrobial therapy against Corynebacterium spp. infection based on antibiotics susceptibility testing.

Only 31 patients (52.5%) received antibiotic therapy on first isolation; only 18 (58.1%) of these received effective antibiotic therapy for Corynebacterium spp. based on antimicrobial susceptibility testing.

Forty-two patients (71.2%) had persistent Corynebacterium spp. colonization; among them, 16/18 (88.9%) patients who received an antibiotic course that was deemed effective on in vitro data. The antimicrobial therapies are detailed in the Supplementary Table S3. The median duration of Corynebacterium spp. carriage was 58 days [interquartile range 7–412]. We found no association between the administration of an effective antibiotic therapy and the presence of persistent colonization.

The frequency of ACR episodes did not significantly differ between the exposed and non-exposed groups. An AMR episode occurred in 16/59 exposed patients, significantly more often than in the non-exposed patients (27.1% vs. 8.5%, p = 0.015). The frequency of CMV reactivations did not differ according to Corynebacterium spp. exposure (exposed vs. non-exposed 18.6% vs. 8.5%, p = 0.178). The length of follow-up did not significantly differ between the groups, but the occurrence of CLAD at 5 years of follow up was significantly higher in exposed than non-exposed patients (27.1% vs. 6.9%; p = 0.02122.0% vs. 5.1%, p = 0.0135). In the exposed group, CLAD was diagnosed after a median time of 497 days [346–888] of Corynebacterium spp. first isolation. Conversely, survival without CLAD differed significantly between the two groups (Figure 2; p = 0.012), with earlier onset in exposed versus non-exposed patients, but all-cause mortality did not significantly differ.