Letter to the Editor: Integrating immune-related metrics with personalize first-line immunochemotherapy in human epidermal growth factor receptor 2-negative advanced gastric cancer

Abstract

Yao et al recently published a study in the World Journal of Gastrointestinal Oncology, which presented a timely prediction model using nomogram combined with immune, molecular and host-inflammation metrics, including programmed death ligand 1, microsatellite instability status, tumor-node-metastasis stage, tumor differentiation, neutrophil-to-lymphocyte ratio and the C-reactive protein-albumin-lymphocyte index to forcast in 200 human epidermal growth factor receptor 2 negative advanced gastric cancer patients treated with firstline sintilimab plus chemotherapy. Through internal validation, the models have showed promising discrimination with C-index of 0.78-0.82 offering a potentially practical tool for individualized risk stratification. This article commends the authors for integrating tumor biology and host systemic status into a single clinical instrument as well as highlighting routinely measurable inflammationnutrition metrics. At the same time, we raise methodological and implementation issues that merit attention before routine clinical use, including the need for external and prospective validation, risks of selection and treatmentspecific bias, assay heterogeneity for programmed death ligand 1 and microsatellite instability, potential overfitting with relatively small sample size, and the absence of tumor microenvironment or dynamic biomarkers. We propose concrete next steps to accelerate responsible translation by sharing the model with external validation across different programmed death 1 agents and populations, comprehensively assessing added value versus standard staging with decision curve and costeffectiveness analyses, and exploring incorporation into adaptive clinical trials and longitudinal monitoring.

Key Words: Gastric cancer; Programmed death ligand 1 inhibitor; Predictive model; Neutrophil-to-lymphocyte ratio; C-reactive protein-albumin-lymphocyte index; Efficacy

Core Tip: This study developed and validated a novel survival prediction model for advanced human epidermal growth factor receptor 2 negative gastric cancer patients receiving immunotherapy combined with chemotherapy. A retrospective analysis of 200 patients identified programmed death ligand 1 expression, microsatellite instability, tumor-node-metastasis stage, tumor differentiation, neutrophil-to-lymphocyte ratio, and C-reactive protein-albumin-lymphocyte index as independent prognostic factors. Based on these factors, nomogram models for progression-free survival and overall survival were constructed and validated using the concordance index and area under the curve. The results showed good discrimination ability in both the training and validation sets, indicating that the model effectively predicts patient survival outcomes and provides strong support for personalized treatment decisions.

TO THE EDITOR

The recent study published in the World Journal of Gastrointestinal Oncology by Yao et al[1] has addressed an urgent clinical problem for individualized treatment about how to identify which advanced human epidermal growth factor receptor 2 (HER2) negative gastric cancer patients would derive meaningful benefit from programmed death 1 (PD-1) inhibitor plus chemotherapy. The authors have built nomograms with encouraging internal performance that C-index greater than 0.75 for both overall survival and progression-free survival by combining tumor markers like programmed death ligand 1 (PD-L1), microsatellite instability (MSI), classical clinicopathologic features like tumor-node-metastasis stage and differentiation, and host systemic inflammatory indices including neutrophil-to-lymphocyte ratio and C-reactive protein-albumin-lymphocyte index. This integration of tumor biology with inflammationnutrition status exemplifies a pragmatic approach by using routinely available data to inform individualized prognostication in the realworld setting.

Firstline immunochemotherapy has become standard for many patients with advanced gastric cancer[2,3]. It’s warranted for prediction tools to stratify expected benefit concerning personalize treatment intensity, weigh toxicity risk, and prognosis counseling[4]. By including routinely available blood and tumor biomarkers with representativeness of immunological and nutritional status, the construction of prediction tools has potential as readily usable instrument for bedside risk estimation[5,6]. Yao et al’s work[1] has showed their clinical significance that the chosen predictors were easy-to-reach from the real clinical practice, increasing feasibility for rapid adoption if validity would be demonstrated. Since the reported discrimination abilities have suggested the model’s promising prognostic power for both short-term and long-term outcomes, the inclusion of the C-reactive protein-albumin-lymphocyte index and neutrophil-to-lymphocyte ratio has reinforced the emerging evidence that systemic inflammation and nutritional status materially affect immunotherapy outcomes which should be considered in trial design and clinical management. The nomogram model has revealed often-underappreciated role of nutritioninflammation interplaying in immunotherapy outcomes as a clinically accessible biomarker for dynamic monitoring.

While Yao et al’s work[1] shows the value in individualized treatment of HER-2 negative advanced gastric cancer patients, deeper considerations with specific recommendations are intended to strengthen the model’s translational value and stimulate constructive dialogue for further clinical implementation. The generalizability with treatment-specific bias should be considered that concerning the cohort included was single-center and exclusively treated with sintilimab plus SOX/XELOX, it remained unclear whether the model’s prognostic associations held for other PD-1 inhibitors, such as nivolumab or pembrolizumab, or different chemotherapy backbones. The external validation in independent multicenter cohorts including other PD-1 agents and chemotherapy strategies might need to be prioritized that the treatmentspecific effects could bias coefficient estimates and limit transportability. Otherwise, the overfitting risks of a cohort retrospectively including comparatively small sample size of 200 patients with multiple predictors couldn’t be ignored with optimism in performance metrics even with reasonable internal discrimination. Moreover, full disclosure of the model’s coefficients and scoring algorithm would be essential by disclosing the calculator and underlying statistical code publicly available and depositing anonymized datasets or synthetic data for independent testing to foster data transparency and reproducibility. Beyond the prediction model’s discrimination and calibration, decision-curve and cost-effectiveness analyses are necessary for evaluation of net clinical benefit to demonstrate that modelguided strategies improve outcomes or resource allocation.

Yao et al[1] has taken easy-to-reach biomarker without full consideration of existing assay heterogeneity and reproducibility. PD-L1 and MSI testing has been found subject to platform, antibody clone, scoring system, and specimen source variability[7,8]. Therefore, the model inputs might perform differently across centers without standardization that detailed assay methods with harmonization strategies or sensitivity analyses should be provided using different PD-L1 combined positive score thresholds. While pragmatic, the evaluation of tumor microenvironment in gastric cancer consisted of various pathological biomarkers[9] that the present model omitted promising predictors such as tumorinfiltrating lymphocytes, immune phenotypes, lymphocyte activation gene-3/T cell immunoglobulin and mucin-domain-containing-3 expression, circulating tumor DNA dynamics, and serum cytokines like interleukin-6/interleukin-8. In future iterations, it’s needed to evaluate whether adding tissue and circulating immune biomarkers would improve net reclassification and clinical utility beyond the current model. The nomogram generated prognostic score was useful but was lack of prespecified thresholds that would alter therapy choices for intensifying combination or switching regimens when linked to management decisions. The risk categories with recommended clinical actions and test should be defined whether using the nomogram would alter treatment selection or patientcentered outcomes in prospective studies.

Yao et al[1] presented a pragmatic and promising prognostic tool built from accessible clinical and laboratory markers which revealed strengths in its interpretability and potential for near-term clinical translation. For short term, we recommend to coordinate external validation in retrospective cohorts treated with other PD-1 inhibitors and to assess model calibration. The further translation needs prospectively test of the model in a multicenter cohort, embedding it as a stratification or decision aid in observational registries and randomized pragmatic trials to test clinical impact. The integration of dynamic immune, molecular and host-inflammation metrics would assist in building a modular and updateable prognostic platform.

ACTIONABLE RECOMMENDATIONS TO ACCELERATE RESPONSIBLE TRANSLATION

Multicenter external validation and transportability testing

Prioritize external validation across independent, multi-center cohorts treated with different PD-1 agents and chemotherapy regimens to test transportability and recalibrate coefficients where necessary. Share model coefficients, scoring algorithm, and de-identified data and analytic code to enable reproducibility and independent benchmarking.

Prospective pragmatic evaluation and adaptive trial integration

Embed the model as a stratification tool or decision aid in pragmatic prospective registries or adaptive trials to evaluate both prognostic accuracy and clinical impact, for example, guided escalation vs standard care. Use platform/adaptive designs to test model-guided treatment strategies efficiently across multiple regimens and patient subgroups.

Define and validate clinically actionable thresholds

Predefine risk strata with explicit candidate management actions. Validate thresholds in retrospective external cohorts, then prospectively in pragmatic trials, using pre-specified endpoints such as overall survival, progression-free survival, toxicity, and quality of life. Use decision-curve analysis and net-benefit frameworks to select thresholds that maximize clinical utility and patientcentered outcomes.

Broaden and modularize biomarkers; adopt longitudinal modeling

Evaluate whether adding tumor microenvironment markers like tumor-infiltrating lymphocytes, immune checkpoints like lymphocyte activation gene-3/T cell immunoglobulin and mucin-domain-containing-3, ctDNA kinetics, and serum cytokines improves discrimination and reclassification over the current model. Develop modular, updateable prognostic platforms that permit incremental incorporation of tissue and circulating biomarkers and that support longitudinal time-varying prediction models reflecting dynamic treatment response.

Standardize assays and operationalize biomarker use

Report detailed PD-L1/MSI assay methods and perform sensitivity analyses across combined positive score thresholds and assay platforms. Advocate harmonization protocols using standard clones, scoring rules, and specimen handling before broad implementation. Consider assay turnaround time, cost, and laboratory capacity in implementation plans; incorporate health economic evaluation early to inform adoption.

Evaluate cost-effectiveness and implementation feasibility

Perform early health-economic modeling to estimate the incremental costs and benefits of model-guided strategies for treatment selection and monitoring intensity to inform policy and adoption decision.

CONCLUSION

Yao et al[1] have contributed a valuable starting point, a realworld, multiparameter nomogram that captured both tumor and host determinants of outcome for HER-2 negative advanced gastric cancer patients receiving immunochemotherapy. Our recommendations aim to strengthen the evidence base on Yao et al’s study[1] by improving generalizability, and guiding responsible clinical implementation. This model could become an influential tool after iterative external validation, prespecified clinical thresholds, and prospective testing of clinical utility to identify patients’ prognosis in order to optimize individualized therapeutic strategies and achieve precision immune-oncology in advanced gastric cancer.