In a recently published cost-effectiveness analysis by Yang et al,1 the authors compared Oncotype DX (Genomic Health, Redwood City, Calif) versus MammaPrint (Agendia Inc USA, Irvine, Calif) for guiding the use of chemotherapy in patients with early-stage, estrogen receptor-positive breast cancer. Such a head-to-head comparison is consistent with the spirit of comparative effectiveness research, which the Institute of Medicine defines as “the generation and synthesis of evidence that compares the effectiveness of alternative methods to prevent, diagnose, treat, monitor, and improve delivery of care for a clinical condition.”2 The analysts developed a decision-analytic model to evaluate expected costs and quality-adjusted survival over a 10-year period. Recognizing that treatment preferences are not driven entirely by results from gene expression profiling, the analysts first classified patients in the decision tree as being at low or high risk according to Adjuvant! Online, which uses patient and tumor characteristics for risk stratification. They then cross-stratified patients according to Oncotype DX or MammaPrint risk level.
Although this structure is appropriate, the analysts assumed that 47% of patients in the Oncotype DX group and 74% of patients in the MammaPrint group were at high risk according to Adjuvant! Online. This assumption made for an unfair comparison between genetic tests. For a fair comparison, patients in both groups should have the same risk according to Adjuvant ! Online. Because the majority of women with high-risk clinicopathologic characteristics (according to Adjuvant! Online) also have high-risk genetic markers (61% with Oncotype DX and 73% with MammaPrint), more women receiving MammaPrint would have been expected to receive chemotherapy and therefore incur higher costs. However, despite the higher costs associated with MammaPrint testing ($4200 vs $3975), Yang et al1 reported lower expected costs with MammaPrint over 10 years ($21,598 vs $27,882). The authors reported neither the percentage of women receiving chemotherapy nor the percentage of women with distant disease recurrence. Therefore, it is not clear how cost savings were achieved with MammaPrint compared with Oncotype DX.
For the expectations of comparative effectiveness research to be realized, the careful use of methodological tools such as decision analysis will be necessary to make fair head-to-head comparisons between technologies outside the context of randomized clinical trials. In the case of cost-effectiveness models, reporting should go beyond the presentation of expected costs and quality-adjusted life-years and include estimates of intermediate end points such as treatment rates and the incidence of disease recurrence.