Radiomics based on machine learning algorithms could predict prognosis and postoperative chemotherapy benefits of patients with gastric cancer: a retrospective cohort study.

Background

Traditional clinical characteristics have certain limitations in evaluating cancer prognosis. The radiomics features provide information on tumor morphology, tissue texture, and hemodynamics, which can accurately reflect personalized predictions. This study investigated the clinical value of radiomics features on contrast-enhanced computed tomography (CT) images in predicting prognosis and postoperative chemotherapy benefits for patients with gastric cancer (GC).

Methods

For this study, 171 GC patients who underwent radical gastrectomy and pathology confirmation of the malignancy at the First Affiliated Hospital of Wenzhou Medical University were retrospectively enrolled. The general information, pathological characteristics, and postoperative chemotherapy information were collected. Patients were also monitored through telephone interviews or outpatient treatment. GC patients were randomly divided into the developing cohort (n=120) and validation cohort (n=51). The intra-tumor areas of interest inside the tumors were delineated, and 1,218 radiomics features were extracted. The optimal radiomics risk score (RRS) was constructed using 8 machine learning algorithms and 29 algorithm combinations. Furthermore, a radiomics nomogram that included clinicopathological characteristics was constructed and validated through univariate and multivariate Cox analyses.

Results

Eleven prognosis-related features were selected, and an RRS was constructed. Kaplan-Meier curve analysis showed that the RRS had a high prognostic ability in the developing and validation cohorts (log-rank P<0.01). The RRS was higher in patients with a larger tumor size (≥3 cm), higher Charlson score (≥2), and higher clinical stage (Stages III and IV) (all P<0.001). Furthermore, GC patients with a higher RRS significantly benefited from postoperative chemotherapy. The results of univariate and multivariate Cox regression analyses demonstrated that the RRS was an independent risk factor for overall survival (OS) and disease-free survival (DFS) (P<0.001). A visual nomogram was established based on the significant factors in multivariate Cox analysis (P<0.05). The C-index was 0.835 (0.793-0.877) for OS and 0.733 (0.677-0.789) for DFS in the developing cohort. The calibration curve also showed that the nomogram had good agreement.

Conclusions

A nomogram that combines the RRS and clinicopathological characteristics could serve as a novel noninvasive preoperative prediction model with the potential to accurately predict the prognosis and chemotherapy benefits of GC patients.

Informations de copyright

2023 Journal of Gastrointestinal Oncology. All rights reserved.

Déclaration de conflit d'intérêts

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jgo.amegroups.com/article/view/10.21037/jgo-23-627/coif). ML reports funding support from the Natural Science Foundation of Zhejiang Province (Grant No. LY20H160015) and the Project of Zhejiang Provincial Department of Education (Grant No. Y202147047). EDN received financial support from Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) and Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP – 2014/26897-0). QD reports funding from the Zhejiang Provincial Health Department Medical Support Discipline-Nutrition (Grant No. 11-ZC24), the Special Fund of Zhejiang Upper Gastrointestinal Tumor Diagnosis and Treatment Technology Research Center (Grant No. jbzx-202006), the Fund of the Society of Parenteral and Enteral Nutrition of Chinese Medical Association (Grant No. Z-2017-24-2211), the National Key Clinical Specialty (General Surgery), the First Affiliated Hospital of Wenzhou Medical University. The other authors have no conflicts of interest to declare.