基于CT特征的列线图模型对GIST核分裂象数的预测价值

The Value of the Nomogram Based on CT Feature for Predicting the Mitotic Index of Gastrointestinal Stromal Tumors

目的 开发和验证一种基于CT征象的列线图,用于术前预测胃肠道间质瘤(GIST)的核分裂象数(MI)。方法 回顾性分析100例GIST患者的临床及影像资料,根据病理资料将患者分为低MI组(63例)和高MI组(37例),利用单因素和多因素Logistic回归分析CT征象与GIST MI之间的关系,构建列线图预测模型。通过受试者工作特征曲线(ROC)曲线下面积(AUC)、校准曲线及Hosmer-Lemeshow检验来验证列线图性能。结果 单因素分析得出肿瘤形态、最大径、溃疡、坏死、坏死百分比、肿瘤供血或引流血管增粗(EVFDM)、强化程度、强化特征、转移和外周侵犯差异具有统计学意义(P均<0.05)。在多因素Logistic回归中,只有肿瘤最大径、坏死、溃疡和强化特征与高MI显著相关。将这4个CT征象纳入列线图模型中,该模型的AUC值为0.892(95%CI:0.829~0.956),校准曲线及Hosmer-Lemeshow检验(P=0.516)均说明该列线图模型的预测结果和实际结果之间的一致性良好。结论由肿瘤最大径、溃疡、坏死和强化特征组成的列线图模型在预测GIST患者的MI中有较高价值,有助于临床医师更全面评估GIST复发和转移潜能,从而作出更精准的临床决策。

Objective To develop and validate a nomogram based on CT features for predicting the mitotic index of gas⁃trointestinal stromal tumors(GIST)preoperatively.Methods The clinical and imaging data of 100 patients with GIST were retrospectively analyzed.According to the pathological data,the patients were divided into a low mitotic index group(63 cases)and a high mitotic index group(37 cases).The association of CT characteristics with mitotic index was analyzed using univariate and multivariate logistic regression analysis.A nomogram was developed for predicting the mitotic index.The predictive accuracy of the nomogram was determined by the area under the ROC curve(AUC),calibration curves and Hosmer⁃Lemeshow test.Results CT imaging features including tumor shape,the maximum diameter of tumor,ulceration,necrosis,the percentage of necrosis,enlarged vessels feeding or draining the mass(EVFDM),enhancement degree,en⁃hancement characteristic,metastasis and peripheral invasion showed significant differences between groups in univariate a⁃nalysis(all P<0.05).Only the maximum diameter of tumor,necrosis,ulceration and enhancement characteristic were in⁃dependent risk factors for high mitotic index of GIST.The nomogram model including these four independent factors a⁃chieved good AUC of 0.892(95%CI:0.829~0.956).The calibration curve and the results of Hosmer⁃Lemeshow test(P=0.516)demonstrated good agreement between the predictions by the nomogram and the actual results.Conclusion The nomogram model consisting of the maximum diameter of tumor,ulceration,necrosis and enhancement characteristichas high value in predicting the mitotic index of GIST patientsand is helpful for clinicians to evaluate the recurrence and metastasis potential of GIST more comprehensively so as to make more accurate clinical decisions.