Sarcopenia and muscle loss are prevalent in patients with cervical cancer. This study aimed to evaluate the prognostic significance of early muscle changes during concurrent chemoradiotherapy (CCRT) in cervical cancer, utilizing an automated deep-learning tool applied on routine CT scans. In this retrospective cohort study, treatment-naïve cervical cancer patients (International Federation of Gynecology and Obstetrics staging IB-IVA) undergoing CCRT were analyzed. A two-stage deep learning pipeline (3D UNet for L3 localization; 2D DPNUNet for segmentation) was used to automatically quantify skeletal muscle, muscle density, visceral fat, and subcutaneous fat at baseline and after 4 weeks of CCRT. Model performance was validated (Dice = 0.92; radiologist audit). The relationship between body composition parameters and overall survival (OS) and disease-free survival (DFS) was assessed through Kaplan–Meier analysis and Cox regression. Among 505 patients (mean age 53.4 years), pretreatment sarcopenia (34.1%, n = 172) was an independent predictor of worse 5-year OS (85.5% vs. 91.3%, adjusted HR = 1.801, 95% CI 1.043–3.105, P = 0.029), whereas its association with DFS was significant only when using the 38.5cm2/m2 threshold (adjusted HR = 1.843, 95% CI 1.247–2.725, P = 0.001). Early muscle loss (≥ 5% SMI reduction at 4 weeks of CCRT) occurred in 25.7% of patients and independently predicted poorer OS (adjusted HR = 1.784, 95% CI 1.031–3.088, P = 0.041). Patients with early muscle loss also experienced higher rates of treatment interruption (11.5% vs. 6.1%, P = 0.043). Baseline sarcopenia and ≥ 5% SMI loss during 4 weeks of CCRT are prevalent and associated with inferior survival in cervical cancer patients. The feasibility of automated CT assessment provides a rationale for body composition monitoring, enabling timely intervention.
Wang et al. (Wed,) studied this question.