Figure 1

OS experience of the total cohort ( n =539) according to pretreatment uric acid levels. Two different cut-offs were used for this analysis to investigate the sensitivity of our results to cut-off choice. In A, an empirical cut-off at the 75th percentile was employed, whereas in B we used a ROC-derived cut-off that maximised the number of patients that were correctly classified as alive or deceased. This cut-off corresponded to the 93rd percentile of the distribution. Abbreviation: UA=uric acid.