血清丙氨酸与谷氨酰胺水平对胰腺癌患者术后预后效果预测的列线图模型建立与验证

Establishment and validation of nomogram model based on preoperative alanine and glutamine levels to predict the prognosis of pancreatic cancer after surgery

目的分析丙氨酸、谷氨酰胺水平及其他胰腺癌患者术后预后的影响因素,并建立基于丙氨酸、谷氨酰胺的胰腺癌术后预后效果的列线图预测模型。方法回顾性分析2016年1月至2023年1月在天津市中西医结合医院(天津市南开医院)接受胰腺癌手术治疗的220例患者的临床资料,选取其中2016年1月至2021年1月收治的147例患者作为训练集,2021年2月至2023年1月收治的73例患者作为测试集。使用logistic回归模型归纳影响胰腺癌患者术后预后的相关因素,建立可视化的列线图预测模型,并用测试集验证模型预测效能。结果训练集147例胰腺癌术后患者中,18例患者存活(生存率为12.24%)。多因素logistic回归模型分析显示,术前血清丙氨酸和谷氨酰胺水平、TNM分期、分化程度和淋巴结阳性率(LNR)是胰腺癌术后预后的独立影响因素(P<0.05)。根据logistic回归分析结果建立的列线图预测模型的受试者操作特征(ROC)曲线下面积为0.97(95%CI 0.95~1.00);测试集验证结果显示,ROC曲线下面积为0.97(95%CI 0.93~1.00),校正曲线接近理想曲线,预测概率与实测值基本一致。结论术前血清丙氨酸、谷氨酰胺、TNM分期、分化程度和LNR是胰腺癌患者术后预后的独立影响因素,据此建立的列线图预测模型对胰腺癌患者术后预后具有良好的预测效能,可用于指导治疗决策。

Objective To analyze the levels of alanine and glutamine and other factors related to the prognosis of pancreatic cancer patients after surgery,and to establish a Nomogram prediction model based on alanine and glutamine for the prognosis of pancreatic cancer patients after surgery.Method 220 patients with pancreatic cancer who received surgical treatment in Tianjin Nankai Hospital from January 2016 to January 2023 were selected for retrospective analysis.147 patients admitted from January 2016 to January 2021 were selected as the training set.The test set consisted of 73 patients who were admitted from February 2021 to January 2023.Multivariate logistic regression model was used to summarize the relevant factors affecting the prognosis of patients with pancreatic cancer after surgery,Nomogram was used to establish a visual prediction model,and test sets was used to verify the prediction efficiency of the model.Result Among 147 patients with pancreatic cancer after operation in the training set,18 patients survived(survival rate was 12.24%).Multivariate logistic regression model analysis showed that preoperative serum alanine,glutamine,TNM staging,degree of differentiation,and positive lymph node ratio(LNR)were independent influencing factors on prognosis(P<0.05).Apply the Nomogram graph established based on independent influencing factors to the internal validation of the training set row,with an AUC of 0.97(95%CI 0.95~1.00).The validation results of the test set show that the AUC is 0.97(95%CI 0.93~1.00),and the calibration curves are close to the ideal curve.The predicted probability is basically consistent with the measured values.Conclusion The Nomogram prediction model of postoperative prognosis of pancreatic cancer is based on preoperative serum alanyl,glutamine,TNM stage,degree of differentiation,positive ratio of lymph nodes and other clinical data can accurately predict the postoperative prognosis of patients with pancreatic cancer,indicating that the nomogram model has good prediction accuracy and can be used to guide treatment decisions.