Numerous studies have been published in recent years on the use of AIs in the adjuvant setting, but none of these has taken the potential generic prices for the agents into account. A total of 20 publications were evaluated in a recent meta-analysis (8 on anastrozole, 8 on letrozole, and 4 on exemestane) [24
]. Median ICERs of $ 21,113 for letrozole, $ 21,428 for exemestane, and $ 24,932 for anastrozole were found. As all of the values were below the accepted threshold value of $ 25,000, the authors concluded that administration of AIs is always a cost-effective option in comparison with tamoxifen.
Another published meta-analysis of the available economic analyses of the 3 AIs in comparison with tamoxifen was carried out by the Health Technology Assessment (HTA) program in 2007 to provide data for decisions taken by the National Institute for Health and Clinical Excellence (NICE) [25
]. The AIs were evaluated as a cost-effective form of treatment. However, it was emphasized, qualifying this, that further research results are needed to confirm the usual extrapolation of the benefits in terms of disease-free survival and recurrence-free survival to the overall survival as well. These data are now available.
In the 100-month data for the ATAC study, a significantly longer disease-free survival (HR 0.85; 95% CI 0.76–0.94; p = 0.003) and a significantly longer period to recurrence (HR 0.76; 95% CI 0.67–0.87; p = 0.0001) was observed for anastrozole in comparison with tamoxifen [11
]. Although fewer deaths occurred in the anastrozole arm (472 versus 477), the overall survival was not significantly longer (HR 0.97; 95% CI 0.86–1.11; p = 0.7). In the BIG 1–98 study, an intention-to-treat (ITT) analysis of the 2 single-treatment arms (5 years of letrozole versus 5 years of tamoxifen) after a median follow-up period of 76 months showed a significant advantage for letrozole, both in relation to recurrence-free survival (HR 0.88; 95% CI 0.79–0.99) and also distant metastasis-free survival (HR 0.85; 95% CI 0.7–1.00). With regard to the overall survival, the ITT analysis showed a trend in favor of the AI (HR 0.87; 95% CI 0.7–1.02) [9
]. The problem with these data is that, due to the better effectiveness that was emerging for letrozole, unblinding of the study was carried out for ethical reasons, as a result of which 25% of the women switched to the letrozole arm. This selective cross-over made subsequent analysis more difficult, as these women continued to be assigned to the control arm in the ITT analysis, reducing the potential difference between the 2 treatment arms. This led to a censored analysis, in which a significant difference was found in the overall survival (HR 0.81; 95% CI 0.69–0.94) (fig. ) [10
]. To compensate for underestimation in the ITT analysis and overestimation in the censored analysis, an IPCW analysis of the BIG 1–98 study was published [10
], which provides an opportunity to see how the study might have fared without selective cross-over. It was shown once again that letrozole significantly improved the overall survival by 17% in comparison with tamoxifen (HR 0.83; 95% CI 071–0.97; p < 0.05) [10
]. As expected, the results were intermediate between the figures for the ITT and censored analyses.
The present cost-benefit analysis incorporates the survival data mentioned above; taking into account expected price reductions due to the expiry of the patent for anastrozole to 75%, 50% and 25% of the original price and the future expiry of the patent for letrozole, with the same estimated price reductions, the study compares the cost-benefit values for letrozole and anastrozole with tamoxifen in postmenopausal patients with hormone-receptive breast carcinoma. For the basic assumption with letrozole (100% of the price), the cost-benefit figure is € 29,375.15 per QALY. This is also the first cost-benefit analysis for letrozole in comparison with tamoxifen in the context of the German health care system that takes health utilities in Germany into account.
Comparable international data are available [26
]. Delea et al. [26
] investigated the cost-benefit ratio for letrozole in comparison with tamoxifen on the basis of the American health care system and data from the BIG 1–98 study, assuming a carry-over effect of 5 years. A gain of 0.409 QALYs in comparison with tamoxifen and additional costs of $ 9705 with letrozole treatment were observed. This led to costs of $ 23,743 per QALY (95% CI, $ 14,087–51,129).
Numerous published cost-benefit analyses are also available for anastrozole in comparison with tamoxifen, although most of these are based on the 60-month data from the ATAC study [28
]. Karnon [32
] previously carried out a meta-analysis of the published data for the AIs, showing that anastrozole is cost effective. An analysis from the point of view of the American health care system estimated an ICER of $ 20,246 per QALY for anastrozole in comparison with tamoxifen [31
By contrast, the present analysis shows a cost-benefit value of € 94,648.03 (95% CI, € 25,997–1,010,381) for up-front therapy with anastrozole at 75% of the original price in comparison with treatment with tamoxifen. At 100% of the original price, the figure would be as high as € 134,580.89 per QALY (table ). This result clearly diverges from the previously published data, including those reported in a study based on the health care system in Germany [34
]. Using the ATAC 100-month data, the latter study indicated costs of € 21,069 per QALY gained (95% CI, € 12,567–46,604). The differences are based on the problem that the ATAC study did not show a significant benefit in relation to overall survival, but only for disease-free survival. However, numerous cost-benefit analyses for AIs have extrapolated the disease-free survival to the overall survival. This procedure is a matter of controversy [35
]. But Cuzick argues in his paper that disease-free survival is a widely accepted end point in numerous clinical studies [35
], so that overall survival in a group with an initially good prognosis cannot be used, as the patients die due to other diseases and the overall survival resulting from a therapeutic innovation is no longer measurable. In the publication by Lux et al. [34
] mentioned above, a longer overall survival was estimated on the basis of a significantly prolonged time to recurrence due to treatment with anastrozole. Although this assumption is accepted from the health economics point of view, it was dispensed with in the present analysis, and the pure published survival data from the studies were used without extrapolation in order to avoid additional uncertainties in the model.
A comparable analysis using IPCW data from the BIG 1–98 study in comparison with data from the ATAC study has also been published [37
]. Over a period of 20 years, the discounted additional treatment costs for letrozole were ≤ £ 3618. Taking all adverse side effects and avoidance of recurrences into account, the cost difference from tamoxifen was ≤ £ 2964. With a gain of 0.297 QALYs, an ICER of < £ 9999 for letrozole was estimated. For anastrozole, this value was ≤ £ 48,829. The authors concluded that letrozole is a substantially better strategy in comparison with anastrozole. These data from England confirm both the present results from the German study and also the validity of the model based on survival data.
Although the results of this work are confirmed by several international studies, some limitations have to be mentioned. The results are based on a model. The validation with real data would be helpful and should be part of health services research. Moreover, the model is based on the IPCW analysis, which can only weight known patient characteristics and tumor parameters. An adjustment of unknown prognostic factors is not possible.
One further statistical limitation is the fact that the mortality tables for the 76–80-year-old women are based on the general population. A bias is possible, because mortality rates 15–20 years post early breast cancer will be low compared to all-cause mortality rates. This may slightly underestimate mortality. Increasing the mortality probabilities in this age group to 5% increases the ICER of letrozole by € 200; so the effect on the cost-effectiveness would be marginal.
Last but not least, compliance has to be mentioned, which could influence the results of the model. It is known that compliance in real life is different to clinical trials. This is especially relevant for endocrine therapies regarding the long duration of intake (at least 5 years, up to 10 years possible) und the side effects. Non-adherence and non-persistence due to non-compliance can decrease overall survival (HR = 1.26 and HR = 1.49) [38
]. Moreover, it can cause costs by additional relapses and metastasis, which need cost-intensive therapies. On the other side, results of cost-effectiveness models can have a positive influence on the perception of endocrine therapies and optimize compliance.