Objectives: The modified Glasgow prognostic score (mGPS) is known to be useful in determining the prognosis of cancers. However, the utility of mGPS for pancreatic cancer (PC) has been examined based primarily on a surgical series of early-stage cancers. The purpose of this study was to examine the utility of mGPS for PC of all stages using a retrospective cohort design.
Methods: We conducted a retrospective cohort study using data from a computerized database. A total of 807 patients with pathologically confirmed PC were analyzed (mGPS-0, n = 620; mGPS-1, n = 153; mGPS-2, n = 34).
Results: Median overall survival (OS) was significantly worse for the mGPS-1 group than for the mGPS-0 group (5.8 vs 15.8 months, respectively) but was comparable between the mGPS-2 and mGPS-1 groups (4.8 vs 5.8 months, respectively). After adjustment, both mGPS-1 and mGPS-2 were independent predictive factors of OS (mGPS-1: hazard ratio, 1.772; 95% confidence interval, 1.417-2.215; mGPS-2: hazard ratio, 2.033; 95% confidence interval, 1.284-3.219). Subgroup analysis showed that OS was significantly worse in the mGPS-1 and mGPS-2 groups than in the mGPS-0 group for all except the following 2 subgroups: localized disease and curative resection.
Conclusions: The present results show that the mGPS is an independent prognostic factor in patients with PC, especially for advanced-stage disease.