Differential risk of interstitial lung disease with DXd-based versus non-DXd TROP2 antibody-drug conjugates: A systematic review and meta-analysis.

Abstract

e15037 Background: TROP2 directed antibody drug conjugates (ADCs) have demonstrated efficacy across multiple solid tumors. Interstitial lung disease (ILD) has emerged as an important safety signal, particularly with ADCs incorporating the deruxtecan (DXd) payload. The magnitude and consistency of ILD risk across TROP2 ADC platforms remain incompletely defined. Methods: We performed a systematic review and meta analysis of prospective phase II and III clinical trials evaluating TROP2 targeted ADCs in advanced solid tumors. Studies reporting ILD or pneumonitis events were included. ADCs were categorized as DXd based or non DXd based on payload chemistry. ILD incidence was pooled using random effects models, and comparative risk was estimated using risk ratios (RRs), odds ratios (ORs), absolute risk differences, and number needed to harm (NNH). Risk of bias (RoB) was assessed using RoB 2.0 and ROBINS-I, and certainty of evidence was evaluated using GRADE. Results: Eight studies comprising 1,202 patients were included (DXd based: 4 studies, n =802; non DXd: 4 studies, n =400). The pooled incidence of ILD was 5.4% (43/802) for DXd based TROP2 ADCs compared with 0.5% (2/400) for non DXd ADCs. DXd based ADCs were associated with a markedly increased risk of ILD (RR 8.59, 95% CI 2.41-30.57; OR 11.26, 95% CI 2.91-96.48; p <0.001). The absolute risk increase was 4.86 percentage points, corresponding to an NNH of 21. Substantial heterogeneity was observed among DXd based studies (I²=74%), whereas no heterogeneity was detected among non DXd studies (I²=0%). ILD related mortality was rare (0.12%) and occurred exclusively in the DXd based group. ILD incidence was numerically higher in non small cell lung cancer than breast cancer across both drug classes. Sensitivity analyses including restriction to randomized trials, exclusion of small studies, and inclusion of combination therapy yielded consistent results. Meta regression identified payload class as the primary determinant of ILD risk, explaining 65.5% of between study variance. RoB was low or had some concerns in studies contributing 87% of patients, and overall certainty of evidence was rated moderate. Conclusions: DXd based TROP2 ADCs are associated with a substantially higher risk of ILD compared with non DXd TROP2 ADCs, supporting a payload related toxicity effect. These findings underscore the need for vigilant monitoring, careful patient selection, and risk benefit stratification as TROP2 ADC use expands across tumor types. Primary Analysis: Comparison of ILD Risk Between Drug Classes. Outcome DXd-Based Non-DXd Comparison Studies included 4 4 — Total patients 802 400 — ILD events 43 2 — ILD incidence 5.4% 0.5% — Heterogeneity (I²) 74.2% 0% — Risk Ratio (95% CI) — — 8.59 (2.41-30.57) Odds Ratio (95% CI) — — 11.26 (2.91-96.48) P-value — — <0.001 Absolute Risk Difference — — 4.86 pp (3.16-6.57) Number Needed to Harm — — 21 (15-32)