METHODS: The area under survival curves for each treatment group was partitioned into distinct health states of varying utility: toxicity (TOX), time without toxicity or disease progression (TWiST), and the period following disease progression until death or end of follow-up (REL). The utility-weighted sum of the mean health state durations was derived for each group. The threshold utility analysis evaluates how varying utility values across the states affects Q-TWiST differences between groups, although the method is limited by not varying utilities within each health state.
RESULTS: The primary analysis population was the HER2+subgroup (n=219). There was no significant difference between treatments in mean duration of grade 3/4 adverse events prior to progression (L + Let = 1.95 weeks; Let = 2.14 weeks; P=0.90). Using utility weights of 0.5 for TOX and REL, L+ Let was favored for quality-adjusted survival by 8.8 weeks (P = 0.09). The Q-TWiST difference between treatment groups ranged from 8 to 9.5 weeks, favoring combination therapy for all hypothetical utility levels, but none of the comparisons were statistically significant at P = 0.05.
CONCLUSIONS: No significant differences were found between L + Let versus Let in mean duration of severe adverse events. Quality-adjusted survival was favored for the combination treatment arm for all utility levels examined when toxicity was defined by grade 3/4 AEs, but differences between groups were not statistically significant.
