目的

探讨不可切除肝细胞癌（uHCC）患者介入联合靶向免疫转化治疗后手术的影响因素，并构建列线图预测模型。

方法

回顾性分析2022年1月至2024年6月空军军医大学第二附属医院收治的190例初诊为uHCC患者的临床资料。患者均签署知情同意书，符合医学伦理规定。其中男163例，女27例；年龄31～75岁，中位年龄55岁。患者均采用介入联合靶向免疫治疗。Lasso回归筛选uHCC转化手术的预测因素，按照7∶3比例将全组患者分入训练集（133例）和测试集（57例），采用Logistic回归方法基于预测因素在训练集中构建列线图预测模型。分别采用ROC曲线、校准曲线和临床决策曲线分析模型的区分度、校准度和临床适用性。

结果

共有51例患者转化成功手术治疗，总体转化手术率为26.8%。Lasso和Logistic回归分析最终共筛选出4个因素，分别为美国东部肿瘤协作组-体能状态（ECOG-PS）评分、肝硬化、C-反应蛋白与白蛋白比值（CAR）和中性粒细胞与淋巴细胞比值（NLR）。基于以上4个因素在训练集中建立预测模型并绘制列线图。列线图在训练集和测试集中ROC曲线下面积（AUC）分别为0.784（95%CI：0.699～0.869）、0.806（95%CI：0.693～0.920），提示该模型有良好的区分度。校准曲线和Hosmer-Lemeshow 拟合优度检验一致表明，该模型有较高的校准度（χ²=7.410，P=0.493）。分别在训练集和测试集中绘制临床决策曲线，结果显示该预测模型具有良好的临床净收益率。

结论

ECOG-PS评分、肝硬化、CAR、NLR是初治uHCC患者接受介入联合靶向免疫治疗后转化手术的影响因素，基于上述影响因素建立的列线图具有较好的预测能力。

Objective

To investigate the influencing factors for surgery in patients with unresectable hepatocellular carcinoma （uHCC） after interventional therapy combined with targeted immunotherapy， and to construct a nomogram prediction model.

Methods

Clinical data of 190 patients with newly-diagnosed uHCC admitted to the Second Affiliated Hospital of Air Force Medical University from January 2022 to June 2024 were retrospectively analyzed. The informed consents of all patients were obtained and the local ethical committee approval was received. Among them， 163 patients were male and 27 female，aged from 31 to 75 years， with a median age of 55 years. The patients were received interventional therapy combined with targeted immunotherapy. The predictive factors for surgical conversion uHCC were screened by Lasso regression. According to the ratio of 7:3， all patients were divided into the training set （n=133） and test set （n=57）. A nomogram prediction model was constructed in the training set based on the predictive factors by using Logistic regression analysis. ROC curve， calibration curve and clinical decision curve were drawn to assess the degree of discrimination， calibration and clinical applicability of this model.

Results

A total of 51 patients were successfully converted to resection， and the overall conversion rate was 26.8%.Finally， 4 factors were identified by Lasso and Logistic regression analyses， including the Eastern Cooperative Oncology Group-Performance Status （ECOG-PS） score， liver cirrhosis， C-reactive protein-to-albumin （CAR）ratio and neutrophil-to-lymphocyte ratio （NLR）. Based on these 4 factors， a prediction model was constructed and a nomogram was drawn in the training set. The area under the ROC curve （AUC） of this nomogram in the training and testing sets was 0.784 （95%CI: 0.699-0.869） and 0.806 （95%CI: 0.693-0.920）， respectively indicating that a high degree of discrimination. Both the calibration curve and Hosmer-Lemeshow goodnessof-fit test showed that the model had a high degree of calibration （χ²=7.410， P=0.493）. The clinical decision curves were delineated in the training and test sets， revealing that the prediction model yielded high net clinical benefit.

Conclusions

ECOG-PS score， liver cirrhosis， CAR and NLR are the influencing factors of newly-diagnosed uHCC patients receiving interventional therapy combined with targeted immunotherapy before conversion to surgery. The nomogram based on these influencing factors has high predictive capability.