Comparison of Glasgow prognostic score and prognostic index in patients with advanced non-small cell lung cancer.

OBJECTIVES

Previous studies have shown that Glasgow prognostic score (GPS) and prognostic index (PI) are also powerful prognostic tool for patients with advanced non-small cell lung cancer (NSCLC). The aim of this study was to compare the prognostic value between GPS and PI.

METHODS

We enrolled consecutive patients with advanced NSCLC in this prospective cohort. GPS and PI were calculated before the onset of chemotherapy. The prognosis outcomes included 1-, 3-, and 5-year progression-free survival and overall survival (OS). The performance of two scores in predicting prognosis was analyzed regarding discrimination and calibration.

RESULTS

138 patients were included in the study. The area under the receiver operating characteristic curve for GPS predicting 1-year DFS was 0.62 (95 % confidence interval (CI) 0.56-0.68, P < 0.05), and the area under curve for PI predicting 1-year DFS was 0.57 (95 % CI 0.52-0.63). Delong's test showed that GPS was more accurate than PI in predicting 1-year DFS (P < 0.05). Similar results of discriminatory power were found for predicting 3-year DFS, 1-year OS, and 3-year OS. The predicted 1-year DFS by GPS 0, GPS 1, and GPS 2 were 62.5, 42.1, and 23.1 %, respectively, while actual 1-year DFS by GPS 0, GPS 1, and GPS 2 were 61.1, 43.8, and 27.2 %, respectively. Calibration of the Hosmer and Lemeshow statistic showed good fit of the predicted 1-year DFS to the actual 1-year DFS by GPS (χ(2) = 4.326, P = 0.462), while no fit was found between the predicted 1-year DFS and the actual 1-year DFS by PI (χ(2) = 15.234, P = 0.091). Similar results of calibration power were found for predicting 3-year DFS, 5-year DFS, 1-year OS, 3-year OS, and 5-year OS by GPS and PI.

CONCLUSIONS

GPS is more accurate than PI in predicting prognosis for patients with advanced NSCLC. GPS can be used as a useful and simple tool for predicting prognosis in patients with NSCLC. However, GPS only can be used for preliminary assessment because of low predicting accuracy.