Cumulative incidence and risk factors of trastuzumab deruxtecan–related interstitial lung disease in a multicenter real-world cohort.

Abstract

e15030 Background: Interstitial lung disease (ILD) is a clinically relevant adverse event associated with trastuzumab-deruxtecan (T-DXd). Reported ILD incidence often does not account for competing clinical events, potentially leading to risk overestimation. In parallel, real-world data on patient- and treatment-related risk factors remain limited. We aimed to estimate the cumulative incidence of T-DXd–related ILD using a competing-risk approach and to identify clinical and inflammatory factors associated with increased risk. Methods: We retrospectively collected data from 406 patients with metastatic breast cancer (223 HER2-positive, 162 HER2-low; 21 missing data) treated with T-DXd between October 2024 and January 2026 in 16 Italian Institutions. Competing events included death without ILD and treatment discontinuation due to disease progression or other non-ILD causes. Cumulative incidence functions (CIF) were estimated for each event. For risk factor analysis, 74 candidate variables were evaluated, including 10 clinical characteristics, 20 prior treatment variables, 24 concomitant diseases and drugs, and 20 baseline inflammatory biomarkers. Gray’s test was used for univariable screening, followed by multivariable Fine–Gray regression. Results: Seventy-four patients (18.3%) developed T-DXd–related ILD, with a median time to onset of 5.5 months and a mean of 8.4 months. The CIF of first ILD was 14.5% at 10 months and 16.8% at 20 months, plateauing at 18.8% at 50 months. No significant differences in ILD incidence were observed between HER2-positive and HER2-low disease (Gray test p = 0.83). Disease progression was the most frequent competing event, affecting 47.3% of patients at 10 months and 73.3% at 50 months. In univariate analysis, higher continuous neutrophil-to-lymphocyte ratio (NLR), impaired renal clearance (CLcr), prior lung radiotherapy (RT), smoking, and visceral disease were associated with increased ILD risk, while prior Trastuzumab exposure was protective. No correlation was seen for concomitant disease and drugs. In multivariable analysis, reduced creatinine clearance (sHR 3.53, 95% CI 1.64–7.58; p = 0.001), prior lung radiotherapy (sHR 2.40, 95% CI 1.08–5.33; p = 0.032), and higher baseline neutrophil-to-lymphocyte ratio (sHR 1.14, 95% CI 1.05–1.24; p = 0.002) remained independently associated with increased ILD risk. Conclusions: In this multicenter real-world cohort, the cumulative incidence of T-DXd–related ILD was approximately 19%, accounting for competing risks. The high frequency of competing events underscores the importance of competing-risk methodology for accurate toxicity estimation. Baseline renal impairment, prior lung radiotherapy, and systemic inflammatory status identified patients at increased risk and may support risk-adapted monitoring and earlier clinical intervention during T-DXd treatment.