Prognostic value of body composition analysis in cancer patients treated with immunotherapy: a systematic evaluation...

Body composition parameters offer objective, imaging-derived prognostic biomarkers, valuable in the context of oncological diseases. This study evaluated the predictive value for overall survival of body composition parameters automatically derived via TotalSegmentator in cancer patients undergoing immunotherapy.

We retrospectively included pretreatment CT of chest, abdomen, and pelvis (CT CAP) of patients with advanced non-small cell lung cancer or melanoma. Following TotalSegmentator v2.8.0 segmentation, skeletal muscle (SM), visceral adipose tissue (VAT), subcutaneous adipose tissue (SAT), intramuscular adipose tissue (IMAT) volumes from T1 to the sacrum and areas at the T4, L3 levels were quantified, alongside soft tissue densities. Pearson correlations, height-normalized indices, and tissue-to-bone ratios were calculated. Survival was analyzed using Kaplan-Meier curves, log-rank tests, and multivariable age/sex-adjusted Cox regression and the c-index.

Of the 357 patients, the one-year and five-year mortality rates were 52% (95% CI 47-58%) and 84% (95% CI 80-88%), respectively. Body composition parameters at the L3 level and in CAP volume were strongly correlated (r = 0.87-0.98). Differences between short and long survival groups were significant for all CAP volume-based body composition biomarkers, while area-based IMAT and VAT height-squared values were not discriminative at the L3 and T4 levels, respectively (p = 0.068 and p = 0.169). Cox models with muscle-to-bone ratio and SM index yielded similar c-indexes (0.616 and 0.618). The best c-index was achieved integrating quantity and density biomarkers of all tissues (0.621).

TotalSegmentator v2.8.0-derived CT CAP volume-based parameters were associated with immunotherapy overall survival, enabling routine volumetric prognostication to optimize treatment.