Development and validation of a composite C-reactive protein-to-albumin ratio–tumor–node–metastasis index for prognostic stratification in gastric cancer: A multicenter cohort study

ObjectiveSystemic inflammation and nutritional status influence gastric cancer outcomes; however, existing inflammation-based indices show inconsistent prognostic value and rarely reflect the tumor burden.MethodsWe conducted a retrospective multicenter cohort analysis of 990 patients with pathologically confirmed gastric cancer registered in the nationwide Investigation on Nutrition Status and Clinical Outcome of Common Cancers cohort (2012-2021). Patients were randomly split into training (n = 693) and internal validation (n = 297) cohorts. The prognostic performance of the C-reactive protein-to-albumin, neutrophil-to-lymphocyte, platelet-to-lymphocyte, and lymphocyte-to-C-reactive protein ratios as well as systemic immune-inflammation index was compared using time-dependent area under the curves and concordance indices. The C-reactive protein-to-albumin ratio cutoff was determined in the training cohort and rounded to 0.09 for clinical application. A four-category C-reactive protein-to-albumin ratio-tumor-node-metastasis index was constructed by combining the C-reactive protein-to-albumin ratio status with the tumor-node-metastasis stage group and was evaluated using the concordance index, Kaplan-Meier analysis, sequential Cox models, and calibration plots.ResultsThe C-reactive protein-to-albumin ratio was selected for model construction based on its training cohort performance and biological interpretability. The C-reactive protein-to-albumin ratio-tumor-node-metastasis index showed improved discrimination compared with the C-reactive protein-to-albumin ratio alone, with concordance index values of 0.629 (95% confidence interval: 0.599-0.660) in the training cohort and 0.640 (95% confidence interval: 0.595-0.686) in the validation cohort. The four C-reactive protein-to-albumin ratio-tumor-node-metastasis groups were defined as follows: group 1.1, low C-reactive protein-to-albumin ratio (≤0.09) plus tumor-node-metastasis I-II; group 1.2, low C-reactive protein-to-albumin ratio (≤0.09) plus tumor-node-metastasis III-IV; group 2.1, high C-reactive protein-to-albumin ratio (>0.09) plus tumor-node-metastasis I-II; and group 2.2, high C-reactive protein-to-albumin ratio (>0.09) plus tumor-node-metastasis III-IV. In the clinically adjusted Cox model, compared with group 1.1, the hazard ratios were 3.24 (95% confidence interval: 2.18-4.80) for group 1.2, 1.62 (95% confidence interval: 1.02-2.58) for group 2.1, and 5.59 (95% confidence interval: 3.83-8.16) for group 2.2. A nomogram incorporating the C-reactive protein-to-albumin ratio-tumor-node-metastasis index, age group, and body mass index category showed acceptable calibration for 1-, 3-, and 5-year overall survival prediction.ConclusionsThe C-reactive protein-to-albumin ratio-tumor-node-metastasis index may serve as a simple adjunctive tool for prognostic stratification in gastric cancer by combining tumor stage with systemic inflammatory and nutritional status. Further external validation and model refinement are needed before its adoption in routine clinical use.