局所進行食道扁平上皮癌に対する術前化学免疫療法および手術を受けた患者の予後を予測する複合免疫栄養・腫瘍マーカースコア

Neoadjuvant chemoimmunotherapy (nCIT) has improved outcomes in patients with locally advanced esophageal squamous cell carcinoma (LA-ESCC). However, accessible and reliable biomarkers to predict survival and pathological response remain limited. This study assessed the prognostic and predictive value of the Global Immune-Nutrition Index (GINI), cytokeratin-19 fragment (CYFRA 21 - 1), fibrinogen-albumin ratio index (FARI), and a composite score combining these markers-the GINI-CYFRA 21 - 1-FARI (GCF) score.

A retrospective analysis was conducted on 138 LA-ESCC patients who received nCIT followed by radical esophagectomy between 2021 and 2023. Optimal cut-off values for GINI, CYFRA 21 - 1, and FARI were determined using X-tile software. The composite GCF score was developed accordingly. Associations with disease-free survival (DFS) were evaluated using Kaplan-Meier analysis and multivariable Cox regression. Predictive performance was assessed via time-dependent ROC curves. Tumor regression grade (TRG) was analyzed using logistic regression, and a nomogram was constructed to predict pathological response.

Elevated levels of GINI (median DFS: 20 vs. not estimable [NE] months; p = 0.001), CYFRA 21 - 1 (19 vs. NE months; p < 0.001), FARI (11 vs. 26 months; p < 0.001), and higher GCF scores (0 vs. 1 vs. ≥2: 18 vs. 22 vs. NE months; p < 0.001) were associated with worse DFS. The GCF score was independently predictive of DFS (low vs. high: HR = 0.264; p = 0.028) and demonstrated consistent discriminative capacity. For pathological response, the GCF score showed predictive value for TRG (AUC = 0.625; p = 0.004) and was independently associated with poor TRG (high vs. low: OR = 2.170; p = 0.048). A nomogram incorporating the GCF score outperformed models excluding it (AUC: 0.787 vs. 0.662; p < 0.05).

The GCF score-a composite of GINI, CYFRA 21 - 1, and FARI-independently predicts DFS and pathological response following nCIT and surgery in LA-ESCC, and may hold potential as a practical, blood-based biomarker. Its integration significantly enhances predictive model performance and may aid in individualized patient risk stratification.