Integrating objective sarcopenia assessment into geriatric screening as a predictor of treatment tolerance.

Abstract

1665 Background: Geriatric assessment (GA)–based screening tools, including the G8 screening tool, the Korean Cancer Study Group Geriatric Score (KG-7), and the Cancer and Aging Research Group (CARG) toxicity score, are widely used to guide treatment decisions in older patients with cancer. However, these tools incompletely capture biological reserve and show limited accuracy in predicting treatment tolerance, particularly in patients with indeterminate risk. We evaluated whether incorporating objective sarcopenia assessment could improve risk stratification. Methods: We retrospectively analyzed 97 elderly patients with cancer (age ≥70 years) who underwent GA screening prior to treatment decision-making at a single tertiary center between May 2024 and December 2025. In addition to standard GA tools (G8, KG-7, and CARG toxicity score), skeletal muscle index (SMI) and functional frailty were assessed. SMI was measured using the DYPHI body water analysis (BWA) system, which estimates appendicular lean mass through multi-frequency bioelectrical impedance analysis. Functional frailty was evaluated using the Short Physical Performance Battery (SPPB). Treatment intolerance was defined as grade 3–5 toxicity, unplanned hospitalization, emergency room visit, or treatment discontinuation. Multivariable logistic regression and subgroup analyses were performed. Results: The median age was 76 years (range, 61–87), and 77 patients (79.4%) were male. Most patients had advanced-stage disease (72.2%), and 55.7% received palliative treatment. Initial dose reduction (<100%) occurred in 63.9%. GA tools, sarcopenia, and frailty showed modest concordance. Among GA tools, G8 demonstrated the strongest association with treatment intolerance (odds ratio [OR] per point increase 0.86, 95% confidence interval [CI] 0.74–0.99; p=0.03; area under the curve [AUC] 0.64). Sarcopenia was independently associated with treatment intolerance (OR 2.45, 95% CI 1.02–5.87; p=0.045) and improved discrimination across G8, KG-7, and CARG tools (ΔAUC up to +0.12). The benefit of sarcopenia was greatest in borderline or intermediate GA risk groups (ΔAUC approximately +0.10–0.12), with minimal improvement in low- or high-risk groups (ΔAUC ≤0.02). A minimal model incorporating G8, sarcopenia, and Eastern Cooperative Oncology Group (ECOG) performance status achieved the best performance (cross-validated AUC 0.68). Conclusions: Incorporation of objective sarcopenia assessment using the DYPHI BWA system significantly improves prediction of treatment tolerance beyond established GA tools, particularly in intermediate-risk subgroups. A pragmatic model combining G8, sarcopenia, and ECOG performance status may refine risk stratification and personalize treatment decisions in geriatric oncology.