This was a single-centre retrospective study performed at the Lung Transplantation Centre of Bichat Hospital, Paris, France. We included all consecutive adults (>18 years old) who underwent LTx for chronic fibrosing ILD from April 2011 to June 2024. We excluded patients with sarcoidosis, interstitial diseases not classified as chronic fibrosing ILD (pulmonary alveolar proteinosis, hemosiderosis, acute respiratory disease syndrome), lung retransplantation for restrictive allograft syndrome (RAS), and multiple organ transplantations or if essential data were missing.

We relied on a local prospective database of all patients undergoing LTx and the electronic health records system at our institution. Data collected included patient demographics and pre- and post-LTx information.

Recent IA exposure was defined as IA treatment within the 12 months before LTx, as a dichotomous variable. IAs included biologic IAs (rituximab, tocilizumab, abatacept, anti-tumour necrosis factor) and non-biologic IAs (mycophenolate mofetil, azathioprine, cyclophosphamide, methotrexate, tofacitinib or tacrolimus, leflunomide, anti-JAK therapy). Systemic corticosteroids were not considered as IA but were recorded separately. The 12-month threshold was chosen to capture the clinically relevant peri-transplant immunosuppressive burden, encompassing agents with prolonged biological effects such as rituximab (B-cell depletion persisting months after the last infusion) while excluding remote exposures unlikely to affect peri-operative immunity.

Several variables were defined as follows: severe pulmonary hypertension (PH) as mean pulmonary artery pressure >20 mmHg and pulmonary vascular resistance (PVR) ≥5 Wood units (WU), or in the absence of PVR measurement, cardiac index <2 L/min/m²; pneumonia episode as a clinical suspicion of respiratory infection and evidence of new pulmonary infiltrates on chest imaging (chest X-ray or CT-scan), which was reviewed by the investigator; bronchovascular fistula as an abnormal connection between the bronchial lumen and adjacent thoracic blood vessels on bronchoscopy and CT-scan; acute cellular rejection (ACR) by a clinical suspicion supported by histological evidence or a favourable clinical and functional response to corticosteroid pulse therapy; allo-immunization as de novo donor-specific antibody detection with mean fluorescence intensity >1000; and acute antibody-mediated rejection (AMR), chronic lung allograft dysfunction (CLAD), bronchiolitis obliterans syndrome (BOS), RAS and primary graft dysfunction (PGD) as consensus reported [11, 12]. We defined grade 3 PGD as persistent PaO2/FiO2 ratio <200 or extracorporeal membrane oxygenation requirement (ECMO) for 72 h after LTx or death occurring earlier than 72 h with grade 3 PGD at the time of death. The LTx protocol at our institution is presented succinctly in Supplementary Methods, including bronchoscopy surveillance, maintenance immunosuppression and antimicrobial prophylaxis.

The primary outcome was 12-month all-cause mortality or retransplantation. Retransplantation was included in the composite primary endpoint as an event equivalent to death, representing definitive graft failure. Secondary outcomes were the occurrence of grade 3 PGD, occurrence of bronchovascular fistula, number of pneumonia episodes in the first 6 months, number of antibiotics courses in the first 6 months, invasive fungal infection in the first 6 months, occurrence of cytomegalovirus (CMV) viremia under prophylaxis and after prophylaxis withdrawal, occurrence of ganciclovir-resistant CMV, ACR, acute AMR within the first 12 months, forced expiratory volume in 1 s (FEV1) trajectory within the first 24 months, and all-cause mortality and CLAD-free survival at 36 months.

To evaluate infectious events during the first 6 months, only patients followed for at least 6 months were analysed. CMV donor-negative and recipient-negative (D-/R-) patients were excluded from CMV outcomes analysis. CMV viremia was defined as DNAemia over 3 log units/mL, which is the threshold for beginning curative treatment at our centre. FEV1 trajectory was analysed at each time point according to procedure laterality (single/double LTx).

Variables are reported as median with interquartile range (IQR). We compared means for quantitative variables by unpaired Student’s t-test or Mann Whitney U test depending on the normality of the data distribution and chi-squared test or Fisher’s exact test for categorical variables (the normal distribution of variables was assessed with the Shapiro-Wilk test and by examining QQ plots). For multiple comparisons, we used ANOVA or the Kruskall-Wallis rank sum test. Survival analysis involved the log-rank test, Kaplan Meier curves and a Cox model for multivariate analysis. For multivariate analysis, covariables were selected according to clinical relevance, statistical significance (p < 0.1) and number of events to avoid overfitting. Preoperative patient characteristics were prioritized for survival analysis adjustment considering the focus of our study on pre-LTx risk factors and the expected collinearity between post-operative outcomes and early death. As a sensitivity analysis to assess the robustness of the primary finding to group imbalance, an inverse probability of treatment weighting (IPTW) analysis was performed. A propensity score for IA exposure was estimated using a logistic regression model including ILD subtype, severe pulmonary hypertension, acute exacerbation of ILD in the preceding 12 months, high emergency transplantation status, annual corticosteroid dose, age, and BMI. Stabilised average treatment effect (ATE) weights were derived and applied to a weighted Cox model with robust sandwich standard errors.

We conducted univariate logistic regression to assess the risk associated with early complications of LTx such as PGD and bronchovascular fistula. Competitive risk analysis was performed to analyse acute rejection risk and CMV viremia occurrence (with death as a competing risk), calculating cumulative incidence functions to describe the probability of events, with univariate regression analyses involving the Fine and Gray proportional-hazards model, estimating hazard ratios (HRs) and 95% confidence intervals (CIs). A multistate model was constructed, and transition probabilities were calculated for the analysis of CLAD-free survival at 36 months. A Wald-type Z-test was used to determine the statistical significance of differences in probability of events. Linear mixed models were fitted with FEV1 as a dependent variable; time (continuous), type of LTx (single/double) and optionally IA exposure as fixed effects; and individual as a random effect. Comparing these models allowed for assessing the effect of IA exposure on FEV1 trajectories. Correlation tests (Pearson or Spearman) were used to assess correlation between continuous variables. P < 0.05 was considered statistically significant. Statistical analyses were performed using the R environment R v4.2.0, The R Project for Statistical Computing. ¹

This observational study was designed, conducted, and written following the STROBE guidelines, and it received ethical approval from the French society of respiratory medicine (Société de Pneumologie de Langue Française) review board (CEPRO no. 2025-006).

Patients or members of the public were not involved in the design, conduct, reporting, or dissemination plans of this research.

From April 2011 to June 2024, 559 LTx procedures were performed; 241 (43%) were for ILD. A total of 209 patients were included in this study; 76 (36%) had received IAs within 1 year of LTx (Figure 1). The most frequent IAs administered were cyclophosphamide (n = 25, 33%), mycophenolate mofetil (n = 33, 43%), azathioprine (n = 15, 20%), and rituximab (n = 15, 20%), with 17 (22%) patients having received two or more of these treatments during this interval (Supplementary Table S1). In the IA group, 49 (62%) patients were in the inter-cure phase, meaning they were within 6 months of their last dose of rituximab (n = 9, 12%), 28 days of cyclophosphamide (n = 4, 5%), and 2 weeks of tocilizumab (n = 2, 3%) and/or were receiving mycophenolate mofetil (n = 26, 34%), azathioprine (n = 10, 13%), tacrolimus (n = 2, 3%), tofacitinib (n = 2, 3%) or abatacept (n = 1, 1%) until the week of transplantation.

Pre-LTx baseline characteristics are summarized in Table 1. Patients with recent IA exposure were younger, more frequently women and had a higher body mass index. They more frequently had a diagnosis of connective tissue disease-ILD (CTD-ILD) or idiopathic non-specific interstitial pneumonia as opposed to idiopathic pulmonary fibrosis. They also had higher cumulative 12-month corticosteroid exposure (median 3.67 g [IQR 3.27–5.47] vs. 0 g [0–3.65], p < 0.001) and less frequently had received antifibrotics (28% vs. 62%, p < 0.001).

More than half of procedures were bilateral (55%) (Table 1). LTx was performed with the French high emergency allocation protocol for 64 (31%) patients. Patients with recent IA exposure more frequently had bilateral LTx (66% vs. 48%, p = 0.014), had high emergency transplantation (41% vs. 25%, p = 0.016) (Table 1) and received more packed red blood cells intraoperatively (median 3 packs [1–5] vs. 2 [0–3], p < 0.001) than those not exposed (Table 2).

Overall median survival was 43 months (95%CI [33.7–64.4]), and 12-month retransplantation-free survival was 74% (Supplementary Table S2). Four patients (1.9%) underwent re-transplantation, 3 (1.4%) within the first 12 months after LTx (1 in the IA group).

Twelve-month re-transplantation-free survival was reduced with recent IA exposure on univariate analysis (62% vs. 80%, HR 2.21, 95% CI [1.30–3.76], p = 0.003) (Figure 2) and multivariate analysis (HR 1.99 [1.11–3.56], p = 0.022) after adjusting for severe PH, cumulative dose of corticosteroids, high emergency status and recent ILD acute exacerbation (Table 3). Preoperative ECMO was not included in the multivariable model, as it is collinear with high emergency status. In the IPTW sensitivity analysis (n = 204 with complete covariate data; 10 of 12 covariates achieved standardised mean differences <0.10 after weighting; Supplementary Table S3), the weighted HR of IA exposure for 12-month retransplantation-free survival was 2.44 (95% CI 1.27–4.70, p = 0.007), consistent with the primary multivariable estimate.

Notably, patients exposed to a non-biologic IA alone (n = 56) versus a biologic IA alone (n = 10) in the year before LTx did not differ in 12-month re-transplantation-free survival on multivariate analysis (n = 58) (HR 1.53 [0.57–4.14], p = 0.398) (Supplementary Table S4; Supplementary Figure S1). Considering the span of our study included the recent start of a systematic immunological induction protocol at our centre, we performed a sensitivity analysis excluding the patients who had received induction (Basiliximab n = 8, anti-thymocyte globulin n = 2), and found a similar impact of IA exposure on survival (HR 2.29 [1.22–4.30], multivariate p = 0.010). To account for other potential changes in practice over the years, we performed univariate Cox regressions stratified on LTx era. The prognostic effect of recent IA exposure was preserved in both early (2011-2018, n = 97, HR 2.38 [1.08–5.22], p = 0.031) and recent eras (2019-2024, n = 122, HR 2.08 [1.01–4.26], p = 0.046).

Finally, longer-term analysis using multi-state modelling incorporating CLAD risk demonstrated a higher probability of death at 36 months for patients with recent IA exposure before LTx (48% vs. 35.5%, p = 0.034) (Supplementary Table S5; Supplementary Figure S2).

Patiencs with recent IA exposure more frequently required postoperative veno-arterial ECMO (VA-ECMO) after surgery (29% vs. 16%, p = 0.024) and had grade 3 PGD (22% vs. 8.3%, p = 0.004) and acute kidney injury (41% vs. 24%, p = 0.011) (Table 2). Univariate logistic regression analysis revealed the following risk factors for grade 3 PGD: recent IA exposure (OR 3.20, 95% CI [1.42–7.45], p = 0.005), perioperative blood transfusion volume (OR 1.19, 95% CI [1.07–1.35], p = 0.003), and double LTx (OR 4.55 [1.78–1.04], p = 0.003) (Supplementary Table S6).

Bronchovascular fistula was diagnosed in 6 patients (2.9%), mainly in the IA group (5/76 (6.7%) versus 1/133 (0.8%), p = 0.024) and recent IA exposure was the only factor associated with this severe complication in univariate analysis (OR 9.43, 95% CI [1.48–183], univariate p = 0.042) (Supplementary Table S7). Conversely, there were similar rates of bronchial stenosis requiring dilatation during the 12 months after LTx in both groups (Table 2).

We included 156 patients with follow-up data for at least 6 months in the analysis. IA exposure was associated with increased number of pneumonia infections (median 2 [1–4] vs. 1 [0.5–2.5], p = 0.001) and antibiotics courses (4 [3–5] vs. 3 [1–5], p = 0.010) within 6 months post-LTx (Supplementary Table S8). The groups did not differ in Clostridium difficile, Pseudomonas aeruginosa or invasive fungal infections.

Overall, 20 of 150 recipient-positive (R+) and donor-positive recipient-negative (D + R-) patients (13% of those at risk) experienced CMV viremia under prophylaxis and 60 (40%) after prophylaxis withdrawal. In a survival analysis with death as a competing risk, the cumulative incidence of CMV viremia under prophylaxis during the first 12 months after LTx was higher for patients with IA exposure (21% vs. 5.2%, p = 0.005) (Supplementary Figure S3; Supplementary Table S9). Also, ganciclovir-resistant CMV was more frequent in patients with recent IA exposure (14% vs. 3.2%, p = 0.021) (Supplementary Table S10).

Preformed DSA was present in 14 patients (18.7%) in the IA group versus 34 (26.8%) in the non-IA group (p = 0.3), while 35 (46%) patients in the IA group (50%) exhibited allo-immunization after LTx vs. 70 (53%) (p = 0.4), and acute AMR and ACR affected 22 (11%) and 62 patients (30%) in the first year, with no difference between the 2 groups (Supplementary Figure S4). CLAD, mainly BOS (74% of all CLAD diagnoses), was diagnosed in 48 patients (23%) (Supplementary Tables S11-S13). Multi-state modelling did not demonstrate a difference in probability of CLAD occurrence between the 2 groups at 36 months (Supplementary Figure S2).

Patients reached their best median FEV1% of 76 (IQR 62–88) % predicted after a median of 177 (87–405) days post-LTx. On time point analysis, FEV1% during the first 3 months was lower for patients with recent IA exposure (Supplementary Figure S5), and linear-mixed modelling of FEV1 during the first year confirmed that IA exposure negatively affected FEV1 evolution over time (effect size: 0.77, 95%CI [0.63–0.96], random effect standard deviation: 0.04). ANOVA showed that including IA exposure status as a fixed effect provided improved the model performance (p = 0.027).

To explain this difference in early evolution in FEV1, we sought to analyse the relation between perioperative infections and respiratory function. The number of pneumonia episodes in the first 6 months was negatively correlated with FEV1 at 6, 12, and 24 months (r = −0.45 to −0.57, p < 0.001 for all) (Supplementary Figure S6).