Evaluating treatment strategies in chronic lymphocytic leukemia: use of quality-adjusted survival analysis.

ABSTRACT

To assess comparatively, in terms of quality-adjusted survival, three front-line treatments in patients with stage B- or C-chronic lymphocytic leukemia (CLL). To describe better and compare the survival after randomization of patients from the CLL90 trial that randomly compared ChOP (cyclophosphamide, doxorubicin, oncovin, prednisone), CAP (cyclophosphamide, doxorubicin, prednisone) and fludarabine in advanced CLL, we performed a quality-adjusted survival analysis. This consisted of defining four clinical states (toxicity, treatment free of toxicity, no treatment nor symptoms, relapse), then summing up the average times spent in each state weighted by utility coefficients that reflect relative value according to quality of life. The resulting quality-adjusted time without symptoms or toxicity (Q-TWIST) was compared between randomized groups, and sensitivity (threshold) analyses to the choice of utility coefficients was performed. Over 73 months after randomization, the fludarabine group gained a mean of 45 days of toxicity-free survival at CAP, and 61 days over ChOP. The mean TWIST was 27.05 months with CAP, 31.5 months with ChOP and 32.95 months with fludarabine. The threshold analyses showed that, whatever the utility weights, the mean Q-TWIST was always greater with ChOP or fludarabine as compared to CAP. Fludarabine was consistently a better treatment than ChOP, except in the unlikely case of high utility weights attributed to toxicity and low utility weights attributed to treatment. Nevertheless, from a clinical point of view, differences between ChOP and fludarabine were moderate or event slight (mean difference in TWIST of 1.45 months). We conclude that patients with advanced CLL have a moderate benefit in terms of Q-TWIST when treated with fludarabine over ChOP. These two treatments are always superior to CAP.