Validation of Eurolung risk models in a Japanese population: a retrospective single-centre analysis of 612 cases

Abstract

Abstract OBJECTIVES The objective of this study was to evaluate the validity of Eurolung risk models in a Japanese population and assess their utility as predictive indicators for the prognosis. METHODS Between 2007 and 2014, 612 anatomic lung resections were performed among 694 lung cancer patients in our institution. We analysed the cardiopulmonary morbidity and mortality and compared them with the predicted results. We also investigated the association between the Eurolung aggregate risk scores and the long-term outcomes using the Kaplan–Meier method and a multivariable analysis. RESULTS The percentage of cardiopulmonary complications was lower than that predicted by Eurolung 1 (22.4% vs 24.6%). The mortality rate was significantly lower than predicted by Eurolung 2 (0.7% vs 3.0%). The morbidity rate was stratified by Aggregate Eurolung 1. The stratification of the mortality rate by the Eurolung 2 aggregate score was also in line with the increase in score, although the observed number of deaths was quite small (4 cases). The 5-year overall survival was clearly separated according to the stratified Aggregate Eurolung 1 and 2 (P < 0.01 and P < 0.01, respectively). Besides pathological stage, both the Aggregate Eurolung 1 (score 0–7 vs 8–20) and 2 (score 0–8 vs 9–19) scores were shown to be independently associated with overall survival on multivariable. CONCLUSIONS Eurolung risk models cannot be directly applied to the patients in our institution. However, Eurolung aggregate risk scores were helpful not only for stratifying morbidity and mortality after anatomic lung resection but also for predicting the long-term outcomes.