Background: In patients with metastatic colorectal cancer, chemotherapy may lead to changes in body composition, including skeletal muscle quantity and quality, and body fat area and distribution. Longitudinal follow-up data in a homogeneous population are required to understand these changes better.

Aim: To comprehensively evaluate changes in body composition and their prognostic value in patients with metastatic colorectal cancer undergoing palliative chemotherapy.

Methods: This retrospective study included patients with recurrent or metastatic colorectal cancer who received palliative chemotherapy between 2008 and 2017. Computed tomography scans were analyzed at multiple time points (before each new chemotherapy regimen and after discontinuing all chemotherapy). Body composition was analyzed from each scan using artificial intelligence software (AID-UTM, iAID Inc.), and its association with survival was evaluated through time-dependent Cox regression to adjust for time-varying effects.

Results: This analysis included 1805 patients, with a median age at diagnosis of 57 years, and 62% were male. At first-line chemotherapy initiation, 4.7%, 30.9%, 36.5%, and 37.1% of the patients had sarcopenia, myosteatosis, and visceral and subcutaneous obesity, respectively. During treatment, approximately 54.5% of the patients experienced significant changes in body composition, with 9.1% and 19.2% developing new sarcopenia and myosteatosis, respectively. Sarcopenia and myosteatosis were associated with poorer survival outcomes [hazard ratio (HR) for sarcopenia, 2.55 (95%CI: 2.06-3.16, P < 0.001; HR for myosteatosis, 2.37 (95%CI: 2.00-2.82), P < 0.001]. In contrast, visceral and subcutaneous obesity were associated with improved survival [HR for visceral obesity, 0.69 (95%CI: 0.57-0.82), P < 0.001; HR for subcutaneous obesity, 0.78 (95%CI: 0.64-0.95), P = 0.015], with no negative impacts observed at higher fat levels. These changes correlated with end-of-life survival time.

Conclusion: Abnormalities and body composition changes were frequently observed during palliative chemotherapy for advanced colorectal cancer; myosteatosis was common. Comprehensive body composition assessment offers valuable prognostic insights without requiring additional testing.

Keywords: Artificial intelligence; Body composition; Chemotherapy; Deep learning; Metastatic colorectal cancer; Myosteatosis; Obesity; Palliative systemic treatment; Sarcopenia.

©The Author(s) 2025. Published by Baishideng Publishing Group Inc. All rights reserved.

背景： 在转移性结直肠癌患者中，化疗可能导致身体组成的变化，包括骨骼肌的数量和质量以及体脂面积和分布。为了更好地理解这些变化，需要对同质人群进行纵向随访数据的收集。

目的： 全面评估接受姑息性化疗的转移性结直肠癌患者的身体组成变化及其预后价值。

方法： 这项回顾性研究包括了2008年至2017年间接受姑息性化疗的复发或转移性结直肠癌患者。在多个时间点（每次新化疗方案开始前和所有化疗结束后）分析计算机断层扫描图像。使用人工智能软件（AID-UTM，iAID Inc.）从每张图像中分析身体组成，并通过时依Cox回归来评估其与生存率的关系，以调整时间变化的影响。

结果： 该分析包括1805名患者，诊断时的中位年龄为57岁，62%是男性。在一线化疗开始时，分别有4.7%，30.9%，36.5%和37.1%的患者患有肌少症、肌脂变性以及内脏肥胖和皮下脂肪型肥胖。在接受治疗期间，大约54.5%的患者经历了显著的身体组成变化，其中9.1%和19.2%分别发展为新的肌少症和肌脂变性。肌少症和肌脂变性与较差的生存结果相关（肌少症的危险比HR：2.55 (95%CI: 2.06-3.16, P P P P = 0.015），并且在更高的体脂水平上没有观察到负面影响。这些变化与终末期生存时间相关。

结论： 在晚期结直肠癌的姑息性化疗期间，身体组成异常和变化经常被发现；肌脂变性很常见。全面的身体组成评估可以提供有价值的预后信息，无需额外测试。

关键词： 人工智能；身体组成；化疗；深度学习；转移性结直肠癌；肌脂变性；肥胖；姑息性全身治疗；肌少症。