Bioimpedance-derived body composition parameters predict mortality and dose-limiting toxicity: the multicenter ONCO-BIVA study

Background: In patients with cancer, lean body mass loss is frequent and associated with worse outcomes, including reduced treatment tolerance and survival. Bioelectrical impedance analysis (BIA) is a popular method for body composition assessment. We evaluated the value of BIA-derived body composition parameters in predicting mortality and, for the first time, dose-limiting toxicity (DLT). Patients and methods: We conducted a prospective multicenter (n = 12) observational study in adult patients with solid neoplastic disease and receiving primary systemic treatment. We collected information on BIA-derived parameters: phase angle (PhA) <5th percentile of age and gender-specific normative values; standardized PhA (SPA) <−1.65; Nutrigram® <660 mg/24 h/m and <510 mg/24 h/m for males and females, respectively. The primary outcome and the key secondary were 1-year mortality and DLT (any-type severe toxicity requiring a delay in systemic treatment administration or a reduction of its dosage), respectively. Results: In total, 640 patients were included. At 12 months, death occurred in 286 patients (47.6%). All BIA-derived body composition parameters were independently associated with death: SPA, hazard ratio (HR) = 1.59 [95% confidence interval (CI) 1.30-1.95] (P < 0.001); PhA, HR = 1.38 (95% CI 1.13-1.69) (P = 0.002); Nutrigram®, HR = 1.71 (95% CI 1.42-2.04) (P < 0.001). DLT occurred in 208 patients (32.5%) and body composition parameters were associated with this outcome, particularly SPA: odds ratio = 6.37 (95% CI 2.33-17.44) (P < 0.001). Conclusions: The study confirmed that BIA-derived body composition parameters are independently associated not only with survival but also with DLT. Although our findings were limited to patients receiving first-line systemic treatment, the evidence reported may have important practice implications for the improvement of the clinical work-up of cancer patients.