首页|期刊导航|肿瘤预防与治疗|基于肠屏障、免疫功能及术后疲劳指标构建腹腔镜结肠癌术后不良预后的列线图预测模型

基于肠屏障、免疫功能及术后疲劳指标构建腹腔镜结肠癌术后不良预后的列线图预测模型OA

Developing a Nomogram Prediction Model for Poor Prognosis after Laparo-scopic Colon Cancer Surgery Based on Gut Barrier Function,Immune Status,and Fatigue Indicators

中文摘要英文摘要

目的:腹腔镜结肠癌术后短期不良预后发生率偏高,现有评估体系仅依靠病理指标,缺乏肠屏障、细胞免疫、术后疲劳等机体功能维度的综合评价.本研究整合多维度客观指标筛选术后 6 个月不良预后独立危险因素,构建可视化预测列线图并完成内外验证,量化新增功能指标对模型预测效能的增益,为临床高危患者分层干预提供量化工具.方法:选取 2024 年 7 月至 2025 年 1 月于本院行腹腔镜结肠癌根治术患者 420 例为建模集,按术后 6 个月是否发生不良预后分为不良预后组 79 例、良好预后组341例;另选取同院2025年2月至2025年7月收治的 120 例患者为外部验证组.检测肠屏障、免疫功能、术后疲劳量表(Inventory of Cancer Fatigue Scale,ICFS)及临床病理指标,经单因素分析、Lasso 回归筛选关键变量,构建多因素 Logistic 回归模型并绘制列线图.以 ROC 曲线、Hosmer-Lemeshow 检验、决策曲线分析(decision curve analysis,DCA)评估模型区分度、校准度与临床净获益,外部验证验证模型稳定性.随访截止至2026年3月.结果:建模集不良预后 79 例(18.81%)、良好预后 341 例(81.19%);验证集不良预后 22 例(18.33%)、良好预后98例(81.67%).单因素分析显示:年龄、BMI、肿瘤直径等 20 项指标差异有统计学意义(均 P<0.001),验证集趋势一致.经 Lasso 回归与十折交叉验证(最优 λ=0.085),筛选出 pTNM 分期、分化程度、脉管侵犯、D-乳酸(D-lactic acid,D-Lac)、闭合蛋白(occludin,OCLN)、CD4+/CD8+、术后第3天ICFS 等7项关键变量.多因素Logistic回归显示:上述变量均为术后6个月不良预后独立危险因素(均P<0.001);其中临床分期Ⅲ~Ⅳ期(OR=6.512,95%CI:3.485~12.163)、低分化(OR=4.701,95%CI:2.623~8.437)、脉管侵犯阳性(OR=4.183,95%CI:2.391~7.312)、D-Lac升高(OR=1.244,95%CI:1.159~1.335)、术后第3天 ICFS 指标升高(OR=1.166,95%CI:1.114~1.221)为危险因素,OCLN 升高(OR=0.139,95%CI:0.072~0.267)、CD4 阳性/CD8 阳性 T 淋巴细胞比值(CD4+/CD8+T lymphocyte ratio,CD4+/CD8+)升高(OR=0.282,95%CI:0.172~0.461)为保护因素.模型性能验证:建模集 AUC=0.809(95%CI:0.733~0.884),灵敏度77.22%,特异度80.35%;验证集 AUC=0.738(95%CI:0.631~0.781),灵敏度 72.73%,特异度79.59%.Hosmer-Lemeshow 检验显示建模集 χ2=6.884、P=0.549,验证集 χ2=5.927、P=0.641,模型校准度良好.DCA证实模型在阈值概率0.05~0.85(建模集)、0.06~0.83(验证集)区间内净获益显著,临床实用性高.结论:基于肠屏障、免疫功能、术后疲劳与临床病理指标构建的列线图模型,对腹腔镜结肠癌术后 6 个月不良预后具有良好预测效能,可视化强、临床实用性高,可用于高危患者早期识别与精准干预.通过与传统临床病理模型对比,证实肠屏障、免疫及术后疲劳指标可显著提升模型预测效能,临床增益明确.

Objective:The short-term poor prognosis rate after laparoscopic colon cancer surgery is relatively high.The existing assessment system relies solely on pathological indicators and lacks comprehensive evaluations of body functions such as intestinal barrier,cellular immunity,and postoperative fatigue.This study integrates multiple objective indicators to screen independent risk factors for poor prognosis after 6 months of surgery,constructs a visual prediction nomogram and completes internal and external validations,quantifies the gain in model predictive efficacy brought by the new functional in-dicators,and provides a quantitative tool for stratified intervention of high-risk patients in clinical practice.Methods:A total of 420 patients who underwent laparoscopic radical resection for colon cancer in our hospital from July 2024 to January 2025 were selected as the modeling set.These patients were divided into the poor prognosis group(79 cases)and the good progno-sis group(341 cases)based on whether they developed adverse outcomes 6 months after the surgery.Another 120 patients admitted to our hospital from February 2025 to July 2025 were selected as the external validation group.The intestinal barri-er,immune function,postoperative fatigue(Inventory of Cancer Fatigue Scale,ICFS),and clinical pathological indicators were tested.Key variables were selected through univariate analysis and Lasso regression.A multivariate Logistic regression model was constructed and a nomogram was developed.The discrimination,calibration,and clinical net benefit of the model were evaluated using the ROC curve,Hosmer-Lemeshow test,and decision curve analysis(DCA).The stability of the model was verified through external validation.The follow-up was completed by March 2026.Results:The training set showed 79 cases with poor prognosis(18.81%)and 341 cases with good prognosis(81.19%);the validation set showed 22 cases with poor prognosis(18.33%)and 98 cases with good prognosis(81.67%).Univariate analysis revealed that 20 indicators,in-cluding age,BMI,and tumor diameter,showed statistically significant differences(all P<0.001),and the trend was con-sistent in the validation set.Through Lasso regression and 10-fold cross-validation(optimal λ=0.085),seven key varia-bles were selected:pTNM stage,differentiation degree,vascular invasion,D-lactic acid(D-Lac),occludin,CD4+/CD8+,and ICFS on the third day after surgery.Multivariate Logistic regression showed that all the above variables were independent risk factors for poor prognosis 6 months after surgery(all P<0.001);among them,clinical stage Ⅲ~Ⅳ(OR=6.512,95%CI:3.485~12.163),poor differentiation(OR=4.701,95%CI:2.623~8.437),positive vascular invasion(OR=4.183,95%CI:2.391~7.312),elevated D-Lac(OR=1.244,95%CI:1.159~1.335),and elevated ICFS on the third day after surgery(OR=1.166,95%CI:1.114~1.221)were risk factors,while elevated occludin(OR=0.139,95%CI:0.072~0.267)and elevated CD4+/CD8+T lymphocyte ratio(CD4+/CD8+)were protective factors.Model performance verification:the AUC of the training set was 0.809(95%CI:0.733~0.884),with a sensitivity of 77.22%and a specificity of 80.35%;the AUC of the validation set was 0.738(95%CI:0.631~0.781),with a sensitiv-ity of 72.73%and a specificity of 79.59%.The Hosmer-Lemeshow test showed that the χ2 value of the training set was 6.884,P=0.549,and the χ2 value of the validation set was 5.927,P=0.641,indicating good model calibration.DCA confirmed that the model had significant net benefit within the probability range of 0.05~0.85(training set)and 0.06~0.83(validation set),with high clinical practicability.Conclusion:The nomogram model constructed based on intestinal barrier,immune function,postoperative fatigue and clinicopathological indicators exhibits good predictive performance for poor prognosis at 6 months after laparoscopic colon cancer surgery.It features favorable visual presentation and good clinical utility,and can be used for early identification and precise intervention of high-risk patients.Compared with the traditional clinicopathological model,this study confirmed that indicators of intestinal barrier,immune function and postoperative fatigue could significantly improve the predictive performance of the model,with evident clinical benefit.

席俊峰;王宏;刘毅

030013 太原,山西省肿瘤医院/中国医学科学院肿瘤医院山西医院/山西医科大学附属肿瘤医院 麻醉科030013 太原,山西省肿瘤医院/中国医学科学院肿瘤医院山西医院/山西医科大学附属肿瘤医院 麻醉科030013 太原,山西省肿瘤医院/中国医学科学院肿瘤医院山西医院/山西医科大学附属肿瘤医院 麻醉科

医药卫生

腹腔镜结肠癌肠屏障功能细胞免疫术后癌因性疲劳不良预后列线图预测模型

LaparoscopeColon cancerIntestinal barrier functionCellular immunityPostoperative cancer-related fa-tiguePoor prognosisNomogramPrediction model

《肿瘤预防与治疗》 2026 (9)

734-744,11

山西省大健康产业高质量发展科研专项课题(编号:DJKZXKT2023280) This study was supported by grants from Shanxi Federation of Social Sciences and Shanxi Association of Ex-perts and Scholars(No.DJKZXKT2023280).

10.3969/j.issn.1674-0904.2026.09.002

评论