20 In both trials, the HRQoL assessments were

comprehensive, providing a strong basis for evaluating the relationship between HRQoL and disease progression. Differences in the findings of the ANCOVA results between LUX-Lung 1 and LUX-Lung 3 may reflect the fact that clinically meaningful ATM inhibitor review changes in HRQoL may be harder to achieve in heavily pretreated patients such as those included in LUX-Lung 1. While the findings reported here indicate that disease progression is accompanied by a statistically significant worsening of HRQoL, it should also be considered whether the results represent a clinically meaningful change in HRQoL. There is continued debate as to what constitutes a meaningful change in oncology HRQoL scores, with data suggesting that patients are more responsive to improvement than decline,22 and that the thresholds for clinically significant improvement and decline are not always uniform.23 While a 10-point change in an individual patient’s EORTC QLQ-C30 item or domain is an accepted threshold for clinically meaningful improvement,24 different thresholds have been proposed for intergroup changes for individual QLQ-C30 QOL scales.25 For the QLQ-C30 Global health status/QoL scale, a mean difference of 0–4 points represents a trivial effect, 4–10 point difference represents a small but clinically

important effect and a 10–15 difference represents a moderate effect.25 These thresholds for QLQ-C30 Global health status/QoL imply that most of the findings reported here are clinically meaningful. For the EQ-5D UK Utility and EQ VAS scores, changes of 0.06–0.11 and 7–12 points, respectively, have been suggested to represent a minimally important difference,26 27 although there is no established consensus on how best to determine the minimally important difference in HRQoL measures.26 Using these values as a guide, some of the changes observed in our study should be considered

clinically meaningful. Brefeldin_A Limitations should be considered. As more HRQoL assessments were conducted up to the time of progression, and fewer at follow-up visit(s) following progression, limited data were available on the health state of patients with progression; this is a common limitation of this type of analysis.4 Further evaluation of HRQoL in these patients may have revealed more pronounced differences in HRQoL between patients with and without progression. Of the two trials, LUX-Lung 3 had more HRQoL data after progression than LUX-Lung 1, indicating that the results from analysis of LUX-Lung 3 data are potentially more robust. Accounting for, and minimising the impact of missing data (which are often not missing at random as assumed here) is an important factor in analyses such as ours.