Application of computed tomography-based radiomics signature analysis in the prediction of the response of small cell lung cancer patients to first-line chemotherapy

The aim of the present study was to investigate the utility of a computed tomography (CT)-based radiomics signature for the early prediction of the tumor response of small cell lung cancer (SCLC) patients to chemotherapy. A dataset including 92 patients from a clinical trial was retrospectively assembled. All of the patients received the standard first-line regimen of etoposide and cisplatin. According to the Response Evaluation Criteria in Solid Tumors 1.1, the patients were divided into two groups: Response and no response groups. A total of 21 radiomics features were extracted from CT images prior to and after two cycles of chemotherapy and a radiomics signature was constructed via a binary logistic regression model. The area under the receiver operating characteristics curve (AUC) was determined to evaluate the performance of the radiomics signature to predict the response to chemotherapy. The clinicopathological factors associated with chemotherapy in patients with SCLC were also evaluated, and a predictive model was established using a binary logistic regression analysis. The 21 radiological features were used to establish a radiomics signature that was significantly associated with the efficacy of SCLC chemotherapy (P<0.05). The performance of the radiomics signature to predict the chemotherapy efficacy (AUC=0.797) was better than that of the model using clinicopathological parameters (AUC=0.670). Therefore, the present study demonstrated that radiomics features may be promising prognostic imaging biomarkers to predict the response of SCLC patients to chemotherapy and may thus be utilized to guide appropriate treatment planning.