NEW YORK – A cost effectiveness analysis published last month by a team of Canadian authors has added another factor to the ongoing competition and debate between various companies offering genomic breast cancer recurrence assays.
The study, which appeared in The Pharmacogenomics Journal at the end of May, was the subject of an announcement by Myriad Genetics this week — heralding investigators' finding that Myriad's EndoPredict assay was more cost effective in their model than Genomic Health's Oncotype DX. However, the paper itself actually compared three tests: Oncotype DX, EndoPredict, and NanoString Technologies' Prosigna, finding that both of the latter assays beat Oncotype in terms of cost effectiveness, though authors cautioned the conclusions should be taken with caution.
Study authors wrote that they believe their study represents the first "comprehensive and independent cost-effectiveness analysis" of this kind "using different available assays compared head-to-head." But it does join a host of other models applied to individual tests, often sponsored by the test manufacturers themselves, as well as previous inter-assay assessments conducted by regulatory and reimbursement bodies like the UK's NICE (National Institute for Health and Care Excellence), aimed at determining which tests meet clinical utility and health economic thresholds for adoption.
In their analysis last month, the Canadian investigators, led by a group at the Schulich School of Medicine and Dentistry at London Ontario's Western University, built a decision model that projected lifetime clinical and economic consequences of different adjuvant treatment-guiding strategies — either dependent on clinical factors alone, or on each of the three analyzed genomic tests.
According to the authors, Genomic Health's test has been most widely adopted and is most widely reimbursed in Canada. But with the emergence of other assays, the group hoped to try to explore "which GEP assay provides more value for money compared to other competing assays."
To inform their model, they used follow-up data from a secondary analysis of the Anastrozole or Tamoxifen Alone or Combined (ATAC) randomized trial along with Canada-specific cost data.
To model the impact of gene expression testing on patient outcomes, the group inputted risk distributions for the three assays alongside 10-year survival estimates from the ATAC data. They combined these factors with rates of chemotherapy administration from a recent Canadian field evaluation of adjuvant therapy decision-making with and without Oncotype Dx testing.
Based on their calculations, the researchers concluded that EndoPredict, Prosigna, and Oncotype DX had cost-effectiveness ratios of $36,274, $48,525, and $74,911 per quality-adjusted life year gained. This corresponded to total gains of 379, 284.3, and 189.5 QALYs/year and total budgets of $12.9, $14.2, and $16.6 million/year, respectively.
Considering these numbers, "GEP testing using any of these assays is likely clinically and economically attractive," the authors wrote. "The [incremental cost-effectiveness ratios] of the three GEP assays were within a range of $20,000 to $100,000 per QALY gained, a level which has been suggested in Canada to define 'moderate evidence for adoption and appropriate utilization'," they added.
That said, the data also suggested that the NanoString and Myriad tests could improve the cost-effectiveness of gene expression testing and offer higher value for the money, with Myriad taking the top spot in terms of cost-effectiveness.
Importantly, the authors stated that they believe prospective evidence is still needed to confirm the findings from their modeling using the ATAC data, and as such, "results should be interpreted with caution."
For one, the model treated all three tests similarly, assuming relative benefit from adjuvant chemotherapy to be the same across patient risk groups. But since the cutoff for the team's analysis, the field has now seen the finalization and publication of the prospective TAILORx trial, which confirmed that Genomic Health's Oncotype DX assay, unlike other tests, does directly predict benefit or lack of benefit from chemotherapy. Altering the model to reflect this could thus change the QALY results.
Genomic Health Chief Scientific Officer Steve Shak said in an email that "cost-effectiveness analyses are complex and highly sensitive to the assumptions underlying individual economic models."
He also stressed that Genomic Health has collated "more than 20 studies" showing that the Oncotype DX test is "cost-effective, and often cost-saving, even with varying assumptions, practice patterns and healthcare costs." Of note, the Canadian model did not find Oncotype DX to be non-cost-effective, but rather to be potentially less cost-effective than the two competing assays.
Ralf Kronenwett, Myriad Genetics' director of international medical affairs, said in an email that though the study did not include the most recent TAILORx data, "the transATAC dataset is widely accepted" and has been the preferred dataset for other analyses, like the NICE health economic model used for the UK's most recent diagnostic guidance of tumor-profiling tests to guide adjuvant chemotherapy decisions in early breast cancer.
Another potential issue, authors wrote, is that the risk group cutoffs for NanoString's test were originally defined in the ATAC trial, so using that data for the model could overly favor Prosigna over the other two tests.
They stressed that although prospective randomized test comparisons are unlikely to be practical and may also not be ethical, analysis of the performance of competing assays in real-world Canadian clinical practice would be valuable and could help update the current model and verify results.
Considering current willingness among Canadian payors to cover Oncotype DX, pursuing comparative field evaluation of the three available tests in real-world practice "may have a large societal benefit in Canada," and "appears to be the only means to verify their clinical utility and identify the most cost-effective assay," they concluded.