Abstract 6364: Serum proteomics analysis: Potential biomarkers and mechanisms of trastuzumab deruxtecan (T-DXd)-related interstitial lung disease/pneumonitis (ILD)

Abstract Background: ILD, an important safety risk of T-DXd across various cancer indications, requires regular monitoring in patients (pts) receiving T-DXd. Diagnosing ILD is complex and requires physiological, clinical, and radiologic assessments, and mechanisms driving ILD onset are not well understood. We performed longitudinal proteomics analysis in serum samples to identify early diagnostic biomarkers and characterize biological mechanisms associated with T-DXd-related ILD. Methods: The SomaScan™ and Olink platforms were used to perform proteomics analysis on 1247 baseline and longitudinal serum samples from 189 selected pts treated with T-DXd across DESTINY-Breast01 (n = 56; NCT03248492), DESTINY-Breast04 (n = 78; NCT03734029), and DESTINY-Lung01 (n = 55; NCT03505710). All dosed and consenting pts with ILD (n = 95) were included and classified based on their worst adjudicated grade (74 mild/moderate [grades 1 or 2] and 21 severe [grades 3-5] ILD cases), and 94 dosed matching controls, who did not develop ILD, were then selected based on pt characteristics. Variance decomposition identified the main confounding factors by assessing protein variance explained by each potential covariate. Proteins associated with ILD onset were then identified via longitudinal statistical analysis, conducted using linear mixed modeling following time alignment of ILD diagnosis across pts and adjustment for personal baseline and the top covariates (age, body mass index, and tumor burden). Pathway enrichment analysis was performed across a range of databases (eg, Gene Ontology, KEGG, Wikipathways, and Reactome). Results: Using linear mixed models, 796 differential proteins were identified in longitudinal samples from patients with severe ILD (false discovery rate [FDR] <0.05), many of which were related to ILD pathophysiology, including alveolar damage (ie, SP-D, SFN), inflammation (ie, CXCL9/10/11), and fibrosis (ie, IL-18BP). Differential signals were typically more pronounced in severe ILD cases. Pathway analysis revealed that ILD onset was associated with biological processes involving the inflammatory response and extracellular matrix (FDR <0.05). Longitudinal trajectory analysis identified CXCL11 as a promising ILD diagnostic biomarker with abundance increases observed up to 2.5 months before ILD diagnosis. Results were confirmed with a multiplexed targeted quantitative Luminex™ assay. Conclusions: Proteomics analysis of longitudinal samples identified key biological processes associated with ILD onset, and CXCL11 emerged as a candidate diagnostic biomarker. Validation in additional cohorts is underway to strengthen confidence in the findings. This work could enable early ILD detection in patients treated with T-DXd and help design therapeutic interventions to improve pt management, care, and outcomes. Citation Format: Kevin Contrepois, Zenta Tsuchihashi, Kazuishi Kubota, Nuria Folguera Blasco, Vinit Kumar, Jing Zhao, Xuya Wang, Brad Jakubison, Sonia Terrillon, Yingkai Cheng, Tomohiro Kaji, Miho Nakamura, Charles A. Powell, Egbert F. Smit, Shanu Modi, Peter Newham. Serum proteomics analysis: Potential biomarkers and mechanisms of trastuzumab deruxtecan (T-DXd)-related interstitial lung disease/pneumonitis (ILD) [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 6364.