KEY POINTS
- This prospective multicentre REQUITE sub-analysis included 279 patients with lung cancer and 848 with prostate cancer, with clinician- and patient-reported outcomes assessed for up to 24 months.
- Skeletal muscle index and intramuscular adipose tissue ratio were extracted from radiotherapy planning computed tomography scans at T12 for lung cancer and L3 for prostate cancer.
- Body composition variables ranked among the five most important predictors for 9 of 12 endpoints and ranked above body mass index for 9 of 12 endpoints.
- Adding body composition variables reduced apparent, in-sample root mean squared error across all 12 models, but the absolute improvements were small: median −0.005 for lung models and −0.003 for prostate models.
- Baseline toxicity or quality-of-life score remained the most important predictor across all endpoints. Model performance was not assessed in an independent or held-out validation cohort.
CLINICAL TAKEAWAY
Computed tomography-derived muscle quantity and quality may capture clinically relevant information not reflected by body mass index alone. However, the incremental predictive benefit was small and assessed only in the data used to fit the models, so these measures are not ready for routine toxicity risk stratification.