构建肝癌大范围肝切除患者术后骨骼肌减少的风险因素模型，并验证其应用效果。

回顾性分析2019年3月至2021年3月山东省立第三医院收治的260例肝癌大范围切除患者临床资料，采用随机数字表法，将所有患者按照7∶3比例分别纳入训练组（n=182）、验证组（n=78）。使用CT影像学测量评估其骨骼肌减少情况，比较训练组（n=182）骨骼肌减少和未见减少患者临床资料差异。使用Logistic多因素回归分析，归纳肝癌大范围切除患者术后骨骼肌减少的风险因素，并将风险因素纳入Nomogram预测模型；绘制预测模型预测验证组患者肝癌大范围切除患者术后骨骼肌减少的受试者工作特征曲线（ROC），使用临床决策曲线分析（DCA）用于验证模型的临床实用价值。随访至术后3年，通过门诊复查及电话随访记录患者生存状态，采用Kaplan-Meier法评估骨骼肌减少对总生存期的影响，Cox比例风险模型校正手术时间等混杂因素。

训练组、验证组术后骨骼肌减少发生率分别为42.31%（33/78）、44.51%（81/182），差异无统计学意义（P>0.05）。训练组中术后骨骼肌指数变化率≤－3.6%的为骨骼肌肉减少组（n=81），其余101例为骨骼肌肉正常组。骨骼肌肉减少组、骨骼肌肉正常组术前肌肉减少症发生率、术后WBC水平、手术时间、失血量、微血管侵犯、并发切口感染发生率、并发器官/腔隙感染发生率、并发菌血症发生率比较，差异有统计学意义（P<0.05）。Logistic多因素分析显示，术前肌肉减少症、WBC水平、手术时间、失血量、微血管侵犯、并发切口感染、并发器官/腔隙感染、并发菌血症，均为影响肝癌大范围切除患者术后骨骼肌减少的独立危险因素（P<0.05）。模型在训练组和验证组的曲线下面积（AUC）分别为0.853（95% CI 0.802~0.904）和0.808（95% CI 0.721~0.895）。验证组的灵敏度、特异性分别为80.66%和81.30%。采用Hosmer-Lemeshow拟合优度检验评估校准度，训练组（χ²=6.32，P=0.612）和验证组（χ²=5.84，P=0.665）均显示良好的校准性能，DCA分析示模型临床净收益率较高。随访时，存活的患者中位随访时间为28个月，总队列的中位总生存期为35.7个月。单变量分析中，骨骼肌肉减少组的总生存期较短于骨骼肌肉正常组（P=0.014）。多因素分析显示骨骼肌减少是术后死亡的独立危险因素（P<0.001）。

肝癌大范围切除患者术后骨骼肌减少风险较高，且与患者术前肌肉减少症、炎症反应、手术时间、失血量、微血管侵犯、并发症等因素有关，基于上述因素建立的预测模型能够为患者肝癌大范围切除患者术后骨骼肌减少风险评估提供可靠参考。

To construct a risk factor model for postoperative skeletal muscle reduction in patients undergoing extensive hepatectomy for liver cancer and verify its application effect.

A retrospective analysis was conducted on the clinical data of 260 patients with extensive resection of liver cancer admitted to Shandong Provincial Third Hospital from March 2019 to March 2021, using the random number table method, all patients were included in the training group (n=182) and the validation group (n=78) in a ratio of 7∶3 respectively. CT imaging was used to evaluate the reduction of skeletal muscle, and the differences in clinical data between patients with and without skeletal muscle reduction in the training group (n=182) were compared. Logistic multivariate regression analysis was used to summarize the risk factors of postoperative skeletal muscle reduction in patients with extensive resection of liver cancer, and the risk factors were incorporated into the Nomogram prediction model. The receiver operating characteristic (ROC) curve of the predictive model for predicting postoperative skeletal muscle reduction in patients with extensive resection of liver cancer in the validation group was drawn, and the clinical decision curve analysis (DCA) was used to verify the clinical practical value of the model. The patients were followed up until 3 years after the operation. The survival status of the patients was recorded through outpatient reexamination and telephone follow-up, the Kaplan-Meier method was used to evaluate the impact of skeletal muscle reduction on overall survival, and the Cox proportional hazards model was used to correct for confounding factors such as operation time.

The incidences of postoperative skeletal muscle reduction in the training group and the validation group were 42.31% (33/78) and 44.51% (81/182), respectively, there was no statistically significant difference between the groups (P>0.05). In the training group, those with a postoperative skeletal muscle index change rate of ≤-3.6% were classified as the musculoskeletal reduction group (n=81), and the remaining 101 cases were classified as the normal musculoskeletal group. The incidences of preoperative muscle atrophy, postoperative WBC levels, operation time, blood loss volume, microvascular invasion, incidence of concurrent incision infection, incidence of concurrent organ/cavity infection, and incidence of concurrent bacteremia in the skeletal muscle reduction group and the skeletal muscle normal group were compared, with statistically significant differences (P<0.05). Logistic multivariate analysis showed that preoperative sarcopenia, WBC level, operation time, blood loss, microvascular invasion, concurrent incision infection, concurrent organ/cavity infection, and concurrent bacteremia were all independent risk factors affecting postoperative skeletal muscle reduction in patients with extensive resection of liver cancer (P<0.05). The area under the curve (AUC) of the model in the training group and the validation group was 0.853 (95% CI: 0.802-0.904) and 0.808 (95% CI: 0.721-0.895), respectively. The sensitivity and specificity of the validation group were 80.66% and 81.30%, respectively. The Hosmer-Lemeshow goodness-fit test was used to evaluate the calibration degree. Both the training group (χ²=6.32, P=0.612) and the validation group (χ²=5.84, P=0.665) showed good calibration performance, and DCA analysis indicated that the model had a high clinical net rate of return. At the follow-up, the median follow-up time for the surviving patients was 28 months, and the median overall survival of the total cohort was 35.7 months. In the univariate analysis, the overall survival period of the skeletal muscle reduction group was shorter than that of the normal skeletal muscle group (P=0.014). Multivariate analysis showed that skeletal muscle reduction was an independent risk factor for postoperative death (P<0.001).

Patients with extensive resection of liver cancer have a relatively high risk of postoperative skeletal muscle reduction, which is related to factors such as preoperative sarcopenia, inflammatory response, operation time, blood loss, microvascular invasion, and complications. The prediction model established based on the above factors can provide a reliable reference for the risk assessment of postoperative skeletal muscle reduction in patients with extensive resection of liver cancer.