Prognostic Value of a Serological-Based Clinical Model for Gastric Cancer Patients

TL;DR

This study develops a blood-based model to predict survival in gastric cancer patients, improving prognosis evaluation beyond traditional methods.

Contribution

A novel serology-based clinical scoring system (SerScore) is developed and validated for gastric cancer prognosis.

Findings

The SerScore model, combined with clinical factors, achieved a C-index of 0.745 in training and 0.750 in validation cohorts.

Patients with a SerScore below −1.73 had significantly better survival rates.

A nomogram incorporating SerScore showed superior predictive accuracy compared to SerScore alone.

Abstract

Background: Surgery remains the cornerstone of diagnosis and treatment for gastric cancer. This study aims to develop and validate a serology-based clinical scoring system to predict and evaluate the prognosis of gastric cancer patients. Methods: Clinicopathological data of primary gastric cancer patients who underwent surgical treatment from 2009 to 2018 were collected and divided into training and validation cohorts. Preoperative serological indicators were screened, and a serum risk score (SerScore) was developed using LASSO-Cox analysis. Prognosis prediction models incorporating the SerScore were established and validated. Results: A total of 5493 patients were screened, and 43 serological indicators were assessed. Twelve serological indicators were selected to construct the SerScore. Patients with a SerScore below the cut-off value of −1.73 had significantly better survival rates compared to those with higher scores. Multivariate Cox analysis identified SerScore, age, tumor location, T stage, and N stage as independent prognostic factors for overall survival in the training cohort. A multivariate nomogram was developed, achieving a C-index of 0.745 in the training cohort and 0.750 in the validation cohort. The nomogram demonstrated superior predictive accuracy compared to the SerScore alone, with AUC values of 0.783 versus 0.639 in the training cohort and 0.805 versus 0.657 in the validation cohort. Calibration curves closely aligned with ideal predictions in both cohorts. Conclusions: The SerScore model provides an effective tool for prognostic assessment in primary gastric cancer patients. This model not only enhances prognostic evaluation but also establishes a foundation for developing advanced prediction tools for gastric cancer.

Figures