增强CT影像组学与胃肠道间质瘤病理参数的相关性及对预后的预测价值OA

Objective:This study aimed to investigate the correlation between contrast-enhanced computed tomography(CT)radiomics features and pathological parameters(DOG1,CD117,CD34,Ki-67 index)in gastrointestinal stromal tumors(GIST),and to evaluate their predictive value for patient prognosis.Methods:A total of 220 GIST patients treated at the First Affiliated Hospital of Hebei North University between January 2016 and March 2024 were enrolled.Based on their status during follow-up,they were divided into a non-progression group(n=150)and a progression group(n=70).All patients underwent abdominal contrast-enhanced CT.Radiomics methods were employed to systematically extract texture,morphological,and histogram-based characteristic parameters of the tumors.Pathological parameters(DOG1,CD117,and CD34 expression status,Ki-67 index)and clinical data were collected.The bivariate Pearson correlation coefficient was used to analyze the correlation between radiomics features and pathological parameters.The Cox proportional hazards regression model was applied to screen for independent prognostic factors.The receiver operating characteristic(ROC)curve was utilized to assess the prognostic predictive efficacy of the radiomics features.Results:Among the radiomics features,texture entropy was positively correlated with CD117 expression(r=0.224,P<0.01),and positively correlated with DOG1 expression intensity(r=0.231,P<0.01)and CD34 expression(r=0.208,P<0.01).Gray-level variance was negatively correlated with CD117 expression(r=-0.142,P<0.05),while positively correlated with DOG1 expression intensity(r=0.159,P<0.05)and CD34 expression(r=0.172,P<0.05).Sphericity was negatively correlated with CD117 expression(r=-0.262,P<0.01),and and positively correlated with Ki-67 index(r=0.187,P<0.01)(all P<0.05).Multivariate Cox regression analysis showed that texture entropy(HR=4.374 95%CI:2.204～8.682,P<0.001),gray-level variance(HR=1.024,95%CI:1.013～1.036,P<0.001),sphericity(HR=0.034,95%CI:0.004～0.268,P<0.001)and Ki-67 index(HR=2.655,95%CI:1.408～5.003,P=0.003)were independent prognostic factors for GIST patients.ROC curve analysis demonstrated that texture entropy(AUC=0.705,95%CI:0.627～0.782),gray-level variance(AUC=0.733,95%CI:0.654～0.813),sphericity(AUC=0.704,95%CI:0.630～0.778)and Ki-67 index(AUC=0.703,95%CI:0.623～0.784)all had certain predictive value for poor prognosis in GIST patients(all P<0.001).Conclusion:Contrast-enhanced CT radiomics features are closely associated with pathological parameters in GIST and demonstrate high predictive value for patient prognosis.They can assist in clinical assessment of disease status and the formulation of personalized treatment strategies.