C104-20 Proteomic Biomarkers of Lung Function Decline in Interstitial Lung Disease
M V Maddali, S Matson, A Linderholm, C Newton, P L Molyneaux, J Kim, J V Pugashetti, G Liu, G C Goobie, A O Adegunsoye, S Ma, L V Wain, G Saini, I Stewart, S Johnson, R G Jenkins, M Strek, C -H Chen, W Fahy, D Khanna, C J Ryerson, S Madathilparambil, P Diep, A H Hoffmann-Vold, T Maher, C K Garcia, P J Wolters, F J Martinez, I Noth, J OldhamAbstract
Background
Interstitial lung disease (ILD) is associated with progressive lung function decline, but predicting progression is challenging. While circulating protein biomarkers associated with survival have been described, biomarkers that are prospectively associated with longitudinal changes in lung function across ILD subtypes are poorly defined. We sought to identify and validate protein biomarkers of future lung function decline in ILD.
Methods
2,921 circulating plasma proteins were semi-quantitatively measured via Olink in twelve North American and European ILD cohorts (discovery n = 2,063; validation n = 1,112). Patients with idiopathic pulmonary fibrosis (IPF; discovery 46%; validation 33%), connective tissue disease (CTD)-associated ILD (discovery 26%; validation 38%), fibrotic hypersensitivity pneumonitis (fHP; discovery 13%; validation 9.5%), and non-IPF idiopathic interstitial pneumonias (IIP; discovery 14%; validation 19%) were included. All patients had baseline forced vital capacity (FVC) and diffusion capacity (DLCO) measured within 6 months of blood draw with subsequent lung function measured up to 24 months. Ordinary least squares regression was used to identify proteins associated with subsequent annualized FVC or DLCO rate of change (%/year), adjusting for demographics, smoking history, baseline lung function, anti-fibrotic or immunosuppressant use, and center. Proteins meeting a false discovery rate (FDR) threshold of ≤ 0.05 in the discovery cohort were evaluated in the validation cohort. A Least Absolute Shrinkage and Selection Operator (LASSO) regression model trained in the discovery cohort and tested in the validation cohort was used to predict estimated annual FVC decline.
Results
Baseline FVC (70% vs 78%) and DLCO (46% vs 54%) were lower in the discovery compared to validation cohort. Median interval to follow-up lung function measurement was similar: discovery 12.3 months (IQR 10.5-14.2) vs validation 12.2 months (11.0-13.7). Ten and 19 proteins had validated associations with FVC and DLCO annualized change, respectively (Figure A-B). Protein associations were largely consistent across ILD subtypes. AREG, CGB3, ICAM5, ITGB6, KRT19, and TNFSF10 were associated with both FVC and DLCO change. A multivariate LASSO model identified high-risk patients with worse lung function decline (Figure C). When dichotomized at the median risk score, patients in the high-risk group declined 169mL vs 44mL in low-risk group (difference 125mL, 95% CI 92-158mL, p < 0.001).
Conclusion
In the largest study to date linking plasma proteomics with longitudinal lung function in ILD, we identified and validated protein biomarkers of lung function decline across ILD subtypes. These findings may improve prognostication and clinical trial enrichment in ILD.
This abstract is funded by: NIH