Nomogram for predicting primary resistance to anti-PD-1 therapy in advanced gastric cancer based on dynamic inflammatory markers: a multicenter retrospective study.

BACKGROUND

While PD-1 inhibitors have improved clinical outcomes in advanced gastric cancer (AGC), primary resistance remains a significant therapeutic challenge. Although inflammatory markers have shown prognostic potential in immunotherapy, their role in predicting primary resistance to PD-1 inhibitors remains underexplored. We developed and validated an inflammatory marker-based nomogram for predicting primary resistance to anti-PD-1 therapy in AGC patients.

METHODS

This multicenter retrospective study enrolled 314 AGC patients separated into training (N = 191), internal validation (N = 82), and external validation (N = 41) cohorts. Dynamic changes in inflammatory markers were assessed at baseline, after two courses of treatment, and at primary resistance via Wilcoxon signed-rank tests. Multivariate logistic regression was performed to identify independent predictors of primary resistance, which were then included in the nomogram. Receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis (DCA) were used to assess the model's performance thoroughly.

RESULTS

Overall, primary resistance occurred in 143 patients (45.5 %). After anti-PD-1 therapy, patients who achieved a complete/partial response presented significant decreases in the neutrophil-lymphocyte ratio (NLR), platelet-lymphocyte ratio (PLR), and systemic immune-inflammatory index (SII) level (all P < 0.001). In contrast, progressive disease was associated with an elevated NLR (P < 0.001), PLR (P < 0.010), and SII (P < 0.001) and a reduced prognostic nutritional index (PNI, P < 0.001). At primary resistance onset, NLR levels were markedly greater and PNI levels were lower than those at baseline (all P < 0.001). Furthermore, treatment lines (P = 0.002), HER-2 negativity (P = 0.049), high SII (P = 0.001) and low PNI after 2 courses of treatment (P = 0.001) were independent predictors of primary resistance. The nomogram integrating these factors demonstrated robust discrimination (training AUC: 0.767, internal/external validation AUC: 0.756/0.729) with optimal calibration and clinical utility in DCA.

CONCLUSION

Longitudinal inflammatory marker profiling enables dynamic monitoring of anti-PD-1 therapeutic response. The primary resistance prediction nomogram developed on the basis of inflammatory markers provides an essential reference for risk stratification and therapeutic decision-making in AGC patients receiving anti-PD-1 treatment.