Abstract
Objective The objective of this study is to contrast the predictive ability of targeted muscle groups as radiographic proxies of sarcopenia on computerized tomography (CT) with body mass index (BMI) in head and neck cancer patients (H&NCP) undergoing radiation at a safety net hospital, and to evaluate sarcopenia with survival, local progression, toxicities and treatment delays. Methods A retrospective review included 52 H&NCP treated between 2017-2019. The posterior neck muscles (PN), sternocleidomastoids (SCM), and their summed volume (AM) were contoured at C3 on patients' pre-treatment CT scans, then normalized to obtain skeletal muscle index (MI) values. Pre-treatment BMI was also evaluated. Cutoffs for sarcopenia were determined by receiver operating characteristic curves. Overall survival and local recurrence-free survival were evaluated by Kaplan-Meier. Acute grade 3 or higher toxicities were evaluated by binomial logistic regression. Results Using all neck muscles (AM-MI) produced the best model for predicting outcomes, outperforming individual muscle groups and BMI. Local progression-free survival was worse in sarcopenic patients at 25.81 months versus 35.40 months (p=0.026). Acute grade 3 or higher toxicities were associated with sarcopenia (p=0.005). Conclusions In this small, retrospective single-institution experience at a safety net hospital, a single axial slice of the combined sternocleidomastoids and paravertebral muscles at C3 performed better than either muscle group alone or pre-treatment BMI at predicting oncologic outcomes.