Quality-adjusted survival (Q-TWiST) analysis of EORTC trial 30853: Comparing goserelin acetate and flutamide with bilateral orchiectomy in patients with metastatic prostate cancer
I. Rosendahl et al., Quality-adjusted survival (Q-TWiST) analysis of EORTC trial 30853: Comparing goserelin acetate and flutamide with bilateral orchiectomy in patients with metastatic prostate cancer, PROSTATE, 38(2), 1999, pp. 100-109
BACKGROUND. The first data analysis of the European Organization for Resear
ch and Treatment of Cancer (EORTC) 30853 trial indicated a significantly lo
nger time to progression and duration of survival for the maximal androgen
blockade (MAB) treatment arm. However, the MAB treatment arm had a higher f
requency of reported side effects.
METHODS. The quality-adjusted survival (Q-TWiST) method was applied to perf
orm a secondary analysis of the EORTC 30853 trial in order to obtain a qual
ity-adjusted survival (QAS) analysis. Two models with different definitions
of the progression health state were used for the analysis. In the first m
odel, progression was defined by both objective and subjective criteria, an
d in the second model only by increase in pain score. The approach was also
extended to include an analysis using actual utility scores (Q-tility) of
patients in the relevant health states.
RESULTS. Based on Q-tility scores obtained from a separate study of a cohor
t of prostate cancer patients, the QAS analysis resulted in a 5.2-month dif
ference (95% CI, -1.1; 11.5 months) in favor of zoladex and flutamide, equa
l in magnitude to the benefit found in the unadjusted survival analysis.
CONCLUSIONS. A QAS analysis such as the Q-TWiST method may be preferred ove
r the unadjusted approach in clinical trials where the health states are cl
early distinct, and differ significantly in either duration or quality of l
ife (QOL), or both. The second model, with progression defined as increase
in pain score, made no difference to the results in this study because of t
he small difference in duration of the pain-progression health state betwee
n treatment arms. However, Q-tility scores from the separate cross-sectiona
l study that was used in this Q-TWiST analysis showed that a subjective def
inition of health states better reflects differences in QOL between the hea
lth states that the patients experience during follow-up. (C) 1999 Wiley-Li
ss, Inc.