Abstract Background The modified nutritional geriatric risk index (mGNRI) was developed as a novel index and provides a more appropriate prognostic index than the original GNRI, which was reported to be a useful index for predicting prognoses for various malignancies. This study investigated the prognostic significance of the mGNRI compared with that of the GNRI in patients with pancreatic cancer and the association with psoas muscle volume (PMV) for survival outcomes. Methods This retrospective study included 137 patients who had undergone pancreatectomy for pancreatic cancer. The enrolled patients were grouped as high mGNRI (≥ 85.3) or low mGNRI (< 85.3), and high GNRI (≥ 92) or low GNRI (< 92) for prognostic analysis based on cutoff values. A propensity-matched analysis was performed in this study. Results The 5-year overall survival of patients in the high mGNRI group or high GNRI group was significantly longer than those in the low mGNRI group or low GNRI group. Statistically significant differences for the 5-year OS were observed in the three groups with respect to the combination of mGNRI and PMV. Patients with low mGNRI/low PMV had a worse 5-year OS rate compared with patients with high GNRI/high PMV or those with high GNRI or high PMV, but not both. The concordance index of the mGNRI to predict the 5-year overall survival was greater than that of the GNRI or the combination of the GNRI and PMV, but lower than that of the combination of the mGNRI and PMV. Multivariate analysis revealed that the mGNRI was an independent prognostic factor for patients with pancreatic cancer (P = 0.005). Conclusions The mGNRI might be a more useful prognostic factor than the GNRI for patients with pancreatic cancer, and might predict prognostic outcomes more accurately when combined with PMV.
