Key points are not available for this paper at this time.
Background Body composition has a significant impact on the prognosis of cancer patients. However, little is known about its impact on the efficacy and safety of antibody–drug conjugates (ADCs), despite the need for accurate patient profiling to ensure reliable safety data and early signals of activity. Methods All patients treated with ADCs in early-phase clinical trials between March 2015 and March 2023 in our institution were retrospectively included in the analysis. Pre-treatment injected CT scans were acquired for all patients. A deep learning software, Anthropometer3DNet, automatically quantified anthropometric parameters in three dimensions (3D) on the acquired CT scans: skeletal muscle mass (SMM), total adipose tissue (TAT), subcutaneous adipose tissue (SAT), visceral adipose tissue (VAT), and lean body mass (LBM). The effect of these anthropometric parameters on progression-free survival (PFS), overall survival (OS), and time in protocol (TIP) was analyzed. Results A total of 136 patients were included. The median age, Eastern Cooperative Oncology Group Performance Status (ECOG PS), albumin, and number of previous lines of treatment were respectively 60.8 years (30 to 85), 1 (0–2), 42 g/L interquartile range (IQR): 39–44, and 3 (IQR: 0–2). The median PFS and OS were 2.6 and 7.9 months, respectively; 90 (66%) patients had experienced toxicity (of which 46 were grade 3–5). Univariate analyses showed that higher SAT hazard ratio (HR) = 0.67, p = 0.03 and TAT (HR = 0.60, p = 0.01) were significantly associated with longer PFS median PFS (mPFS) = 2.76 vs. 2.3 and 2.76 vs. 1.9, respectively. Higher SAT (HR = 0.66, p = 0.04) and higher VAT (HR = 0.65, p = 0.04) were significantly associated with longer OS median OS (mOS) = 9.34 vs. 7.43 months and 9.27 vs. 6.08 months, respectively. Higher TAT was associated with longer TIP in both univariate and multivariate analyses (HR = 0.56, p = 0.006). A Royal Marsden Hospital (RMH) prognostic score of 2 or more was associated with PFS, OS, and TIP in both univariate and multivariate analyses (HR = 1.78, 1.89, and 1.74, respectively). All anthropometric parameters were significantly associated with all-grade toxicity in the univariate analysis but not in the multivariate analysis. Conclusions Automatic extraction of body composition parameters using artificial intelligence (AI) may help in anticipating the benefits of ADCs in patients included in early-phase clinical trials. Combining anthropomorphic data with clinical and biological data may lead to more refined patient selection.
Delaye et al. (Wed,) studied this question.