Dual-layer spectral-detector CT for predicting microsatellite instability status and prognosis in locally advanced gastric cancer

Insights into Imaging

Original Article

Jun Jiang¹  Peng Wang¹  Ying Li¹  Liming Jiang¹  Yongjian Zhu¹  Zhichao Jiang²  Yuxin Zhong³  Bingzhi Wang⁴  Liyan Xue⁴ 

[1] Department of Diagnostic Radiology, National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, 100021, Beijing, China;Department of Medical Oncology, National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, 100021, Beijing, China;Department of Pancreatic and Gastric Surgery, National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, 100021, Beijing, China;Department of Pathology, National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, 100021, Beijing, China;

关键词: Gastric cancer;    Dual-layer spectral detector CT;    Microsatellite instability;    Quantitative parameters;    Nomogram;   

DOI  :  10.1186/s13244-023-01490-x

 received in 2023-05-31, accepted in 2023-07-31,  发布年份 2023

来源: Springer

【 摘 要 】

ObjectiveTo construct and validate a prediction model based on dual-layer detector spectral CT (DLCT) and clinico-radiologic features to predict the microsatellite instability (MSI) status of gastric cancer (GC) and to explore the relationship between the prediction results and patient prognosis.MethodsA total of 264 GC patients who underwent preoperative DLCT examination were randomly allocated into the training set (n = 187) and validation set (n = 80). Clinico-radiologic features and DLCT parameters were used to build the clinical and DLCT model through multivariate logistic regression analysis. A combined DLCT parameter (CDLCT) was constructed to predict MSI. A combined prediction model was constructed using multivariate logistic regression analysis by integrating the significant clinico-radiologic features and CDLCT. The Kaplan–Meier survival analysis was used to explore the prognostic significant of the prediction results of the combined model.ResultsIn this study, there were 70 (26.52%) MSI-high (MSI-H) GC patients. Tumor location and CT_N staging were independent risk factors for MSI-H. In the validation set, the area under the curve (AUC) of the clinical model and DLCT model for predicting MSI status was 0.721 and 0.837, respectively. The combined model achieved a high prediction efficacy in the validation set, with AUC, sensitivity, and specificity of 0.879, 78.95%, and 75.4%, respectively. Survival analysis demonstrated that the combined model could stratify GC patients according to recurrence-free survival (p = 0.010).ConclusionThe combined model provides an efficient tool for predicting the MSI status of GC noninvasively and tumor recurrence risk stratification after surgery.Critical relevance statementMSI is an important molecular subtype in gastric cancer (GC). But MSI can only be evaluated using biopsy or postoperative tumor tissues. Our study developed a combined model based on DLCT which could effectively predict MSI preoperatively. Our result also showed that the combined model could stratify patients according to recurrence-free survival. It may be valuable for clinicians in choosing appropriate treatment strategies to avoid tumor recurrence and predicting clinical prognosis in GC.Key points• Tumor location and CT_N staging were independent predictors for MSI-H in GC.• Quantitative DLCT parameters showed potential in predicting MSI status in GC.• The combined model integrating clinico-radiologic features and CDLCT could improve the predictive performance.• The prediction results could stratify the risk of tumor recurrence after surgery.Graphical Abstract

【 授权许可】

CC BY   
© European Society of Radiology (ESR) 2023

【 图 表 】

42004_2023_990_Article_IEq32.gif

13690_2023_1170_Article_IEq28.gif

13690_2023_1170_Article_IEq25.gif

Fig. 2

Fig. 1

Fig. 3