Effects of Corticosteroid Treatment and Antigen Avoidance in a Large Hypersensitivity Pneumonitis Cohort: A Single-Centre Cohort Study

Background: Although the third most frequent interstitial lung disease, hypersensitivity pneumonitis (HP) remains an enigmatic disease without clear diagnostic and therapeutic guidelines. We assessed the effect of the commonly used therapeutic interventions (i.e. exposure avoidance and corticosteroid treatment) in an HP cohort. Methods: We collected clinical data of all HP patients followed at our centre between January 1, 2005, and December 31, 2016. HP patients were stratified according to the presence of fibrosis on chest CT. Survival was analysed using the multivariate Cox proportional hazards model. Forced vital capacity (percent predicted, FVC%) and diffusing capacity of the lung for carbon monoxide (percent predicted, DLCO%) evolution were analysed using linear mixed-effect models. Results: Two hundred and two HP patients were identified: 93 non-fibrotic HP (nfHP) and 109 fibrotic HP (fHP), experiencing a monthly FVC% decline before treatment of 0.93% and 0.56%, respectively. While nfHP had an excellent survival, fHP patients experienced a median survival of 9.2 years. Corticosteroid treatment and exposure avoidance did not result in survival differences. Although nfHP patients showed FVC% and DLCO% increase after corticosteroid initiation, no therapeutic effect was seen in fHP patients. FVC% and DLCO% increased in nfHP patients after exposure avoidance, while a positive numerical trend was seen for FVC% after exposure avoidance in fHP patients (p = 0.15). Conclusions: nfHP patients experienced an excellent survival with good therapeutic effect on pulmonary function tests with both corticosteroid initiation as well as antigen avoidance. In contrast, fHP patients experienced a dismal prognosis (median survival of 9.2 years) without any therapeutic effect of corticosteroid treatment. Whether antigen avoidance is useful in fHP patients is still unclear.


Outcome
Patients were divided in 2 HP subgroups (nfHP and fHP), based on HRCT findings 9,12,13 : fibrosis was defined as the presence of reticulation, traction bronchiectasis and/or honeycombing on HRCT.HP patients without fibrosis on HRCT were included in the nfHP group, HP patients with fibrosis were included in the fHP group.Differences in baseline characteristics, survival and PFT evolution were assessed between the 2 groups.
Effect of corticosteroid therapy initiation on survival and FVC%/DLCO% decline was assessed.For survival analysis a binary parameter was used (ever-vs never-corticosteroid use).FVC% and DLCO% decline before corticosteroid initiation was compared to PFT decline after initiation.To minimize bias, PFT's after withdrawal of corticosteroid therapy were excluded.Survival analyses for corticosteroid dosage (maximal dosage of <40 mg vs >40 mg of prednisolone-equivalent) and duration of therapy (<6 months vs >6 months) was also performed.
Survival and PFT evolution regarding known/unknown exposure, exposure type (using binary parameter: birds vs mold) and exposure avoidance (using binary parameter: antigen avoided vs antigen not avoided) was assessed, comparable to corticosteroid treatment analysis.FVC% and DLCO% decline before antigen avoidance was compared to PFT decline afterwards.
PFTs 5 years before until 1 year after avoidance were used.Baseline characteristics, PFT evolution and survival of patients with known exposure were compared with patients without known exposure similarly to the analyses of the 2 HP subgroups.These analyses were also performed comparing different exposure types.

Statistical analysis
Baseline characteristics: continuous variables were analyzed using Student's t-tests and Mann-Whitney U-tests where appropriate.For discrete variables, chi-square tests and Fisher's exact tests were used, where appropriate.Patients who underwent lung transplantation were censored at the day of transplantation.
Survival analysis: outcome was based on 10-year survival.Data were displayed as Kaplan-Meier curves and analyzed using Cox proportional hazards models.In multivariate analyses, we corrected for age, gender and baseline FVC%.All multivariate analyses are shown in Supplementary Table 1.
PFT evolution: evolution of PFT was analyzed with linear mixed-effects models, using FVC% and DLCO% as outcome measurements (in separate analyses).Subject was corrected for as a random effect, both with random intercept and (independent) random slope.As PFT's were performed in both referring hospitals and the University Hospitals Leuven, the hospital where the PFT was performed was also corrected for as random intercept.In general, time, age, gender were accounted for as fixed effect.
For analysis of the PFT evolution of the different HP subgroups, PFT's from the first year of follow-up were used when untreated and actively exposed.
For the corticosteroid treatment analysis: PFT's from 5 year before until 1 year after the treatment initiation were used.PFTs after stopping corticosteroid treatment (and/or stopping immunosuppression treatment) were excluded.Corticosteroid use was accounted for as fixed effect, both with and without time-varying covariate.Exposure status was corrected for as fixed effect, immunosuppression use was correct for as a time-varying covariate.
For the antigen avoidance analysis: PFT's from 5 year before until 1 year after avoidance were used.
Exposure status as well as corticosteroid use was correct for with and without time-varying covariate, immunosuppression use was corrected for as time-varying covariate.

Table S1 : Report of MDD on cases not complying criteria for inclusion
Patients diagnosed with HP at the University Hospitals Leuven which did not comply with the criteria for validating the HP diagnosis were discussed in MDD.In the table both the reason for not complying, as well as the final decision and reasoning for inclusion were depicted.

Table S2 : univariate and multivariate survival analyses
Results of cox proportional hazard models, both univariate and multivariate (correction for age, gender and baseline FVC%).For details of the specific groups that were analyzed, we kindly refer to the main text.Definition of abbreviation: nfHP = non-fibrotic hypersensitivity pneumonitis, fHP = fibrotic