First, the projected survival advantage is reflected in all three, presumably related, measures of disease-free, relapse-free and overall survival (Sargent, 2004). Second, the correlation among these is consistent with previous studies of adjuvant 5-FU/LV in colon cancer. Third, the outcome in the control arm is similar to previous customer reviews studies using the Mayo Clinic regimen (Haller et al, 1998). Furthermore, the strength of the statistical trend was reinforced by the finding that covariate-adjusted survival was significantly superior with capecitabine vs 5-FU/LV (hazard ratio 0.788, P=0.0208; Twelves et al, 2005). The estimation of the quantitative impact on survival required extrapolation beyond the observed trial period. In similar studies, investigators have approached this in a variety of ways; there is no uniform methodology.
We used the approach of fitting a curve to the observed data and extrapolating to the end of life. Both the log-normal and Weibull survival curves are commonly used for this, so both were tried. The fit during the trial period was slightly better for the log-normal curve so this was used in the base case. However, the log-normal distribution yielded a gain of nine QALMs, while the Weibull distribution produced 10.9 QALMs, suggesting that the overall survival results were not sensitive to this choice and were possibly conservative. The improved survival rates observed with capecitabine, together with the cost savings identified in this and other analyses, render it a viable alternative to 5-FU/LV both as a single agent and in combination.
Preliminary phase III data have shown the combination of oxaliplatin and infusional 5-FU/LV (FOLFOX) to be effective in the adjuvant setting (Arkenau et al, 2005; Ducreux et al, 2005; Sastre et al, 2005). Replacing 5-FU/LV with capecitabine in this combination is promising not only clinically, but also economically, as additional infusion and time costs would be avoided. One limitation of this model is the lack of direct measures of utility in the stable (prerelapse) health state following treatment. Based on the literature, we imputed a relatively high utility value of 0.86 for this health state, which was assumed for both arms. Thus, any impact would be due to the duration of time in this health state, vs the time in the postrelapse health state. The postrelapse value was also imputed from the literature to be 0.59, which is similar to the values reported for patients on chronic renal dialysis. Treatment phase utility was assumed to be the same in both arms: 0.80. The use of a societal perspective to measure the GSK-3 time and travel costs associated with the treatments illustrates the advantage of oral over infusion treatment.
No related posts.