Q-TWiST analysis of cyclophosphamide, epirubicin, fluorouracil versus cyclophosphamide, methotrexate, fluorouracil treatment for premenopausal women with node-positive breast cancer

Abstract

Objective: To estimate the Q-TWiST (Quality-Adjusted Time Without Symptoms and Toxicity) for premenopausal women with axillary node-positive breast cancer receiving either cyclophosphamide, epirubicin and fluorouracil (CEF) or cyclophosphamide, methotrexate, and fluorouracil (CMF), using data from a phase III clinical trial (National Cancer Institute of Canada Clinical Trials Group [NCIC-CGT.MA5]).

Methods: Restricted mean times by treatment group were computed at 3, 4 and 5 years as the area under the Kaplan-Meier survival curves for the period of treatment-related symptomatic toxicities (TOX), disease-free survival (DFS) and overall survival (OS). Restricted REL (the period of relapse) and TWiST (the period of best possible QOL after breast cancer diagnosis) mean times were indirectly computed as OS minus DFS and DFS minus TOX, respectively. It was assumed that TWiST has a utility coefficient equal to 1. Patients' utility values were arbitrarily set to 0.5 for the toxicity and relapse health states. It was assumed that the utility scores were constant over time. A threshold utility sensitivity analysis was conducted allowing the trade-off between the two treatments to be assessed for the entire range of the utility coefficients. A bootstrap empirical test distribution of the Q-TWiST difference between CEF and CMF was constructed in order to test the statistical significance of the Q-TWiST difference point estimate.

Results: The Q-TWiST mean differences (CEF minus CMF) were -0.37, 1.20 and 2.62 months at 3, 4 and 5 years, respectively. Positive differences are in favour of the CEF arm but none of the above differences were statistically significant. The threshold-utility analysis confirmed the indifference of treatment choice with respect to the utility values, at least at 5 years. There were no combinations of TOX and REL utility values such that the Q-TWiST for the standard therapy (CMF) was greater than the Q-TWiST of the experimental treatment (CEF).

Conclusion: The computed Q-TWiST difference of 2.62 months in favour of CEF (p = 0.492 vs CMF), while not statistically significant, is an indicator that CEF treatment could be a better choice from the patient's perspective than CMF after 5 years of treatment. These results should be considered within the context of the methodological limitations posed by the assumptions in the study.