NEW YORK (GenomeWeb) – After releasing a draft document last month updating its recommendations regarding the use of molecular tests to guide chemotherapy decisions for breast cancer patients, the UK's National Institute for Health and Care Excellence (NICE) is now evaluating comments from stakeholders ahead of an expected finalization in May.
Among other changes, the draft amends an earlier recommendation in favor of Genomic Health's Oncotype Dx as an option for guiding adjuvant chemotherapy decisions, reporting now that no evaluated tests — the committee looked at Oncotype Dx, Myriad Genetics' EndoPredict, Agendia's MammaPrint, NanoString Technologies' Prosigna, and a laboratory-developed test called IHC4 — offer a significant enough impact on health outcomes compared to current practice to make them cost effective.
Genomic Health said last month that it believes that the NICE draft, if finalized in its current form, "would represent a significant step back of five years in breast cancer care with more patients potentially receiving unnecessary chemotherapy."
The company also sent a letter to NICE calling into question the appropriateness of the methods and conclusions of the DAR, citing "major concerns" that the assessment procedure was "unbalanced" and "fails properly to take [into account] the purpose of tumor profiling tests." The letter also argued that an author of the report had major conflicts of interest, and that the assessment appeared to be biased in favor of certain technologies over others, though it didn't specify which technologies it believes were benefitting from this alleged bias.
Robert Stein, a professor of breast oncology at University College Hospitals, London, said in an interview that he believes that he is the author to whom Genomic Health refers to in its letter.
As a consulting author on the DAR, Stein disclosed as a conflict that he is chief Investigator of the OPTIMA (Optimal Personalized Treatment of breast cancer using Multi-parameter Analysis) trial and has published results from the OPTIMA prelim study.
In the OPTIMA prelim study — the feasibility phase of the ongoing OPTIMA trial — Stein and colleagues sought to recruit patients for randomization to two arms: standard of care consisting of chemotherapy followed by endocrine therapy, and the use of Oncotype DX to assign patients either to standard care or to endocrine therapy alone.
As part of the study, researchers also compared different tests — Oncotype, MammaPrint, Prosigna, IHC4, BioNtech's MammaTyper, and Genoptix's NexCourse Breast — in 302 patient samples.
Authors reported that all the tests categorized comparable numbers of tumors into low- and high-risk groups, but there was only moderate agreement between tests at an individual tumor level. And in the end, the group chose NanoString's Prosigna as the test that they would use going forward for the OPTIMA trial.
In seeking comments on the draft recommendation last month, NICE reported that it was interested in hearing about whether it missed any important evidence, opinions on the validity of the report's cost effectiveness interpretations, and thoughts on the general soundness of the resulting recommendations.
According to Stein, NICE is required to consider (and respond in detail) to all comments by registered stakeholders, which include certain individuals, physician groups, and the various companies whose tests were evaluated. After doing that, the committee prepares final recommendations and a guidance on the use of these tests in England's NHS.
The new revision will replace NICE’s diagnostics guidance 10, which in its first draft also recommended against use of any genomic tests, including Oncotype DX. But after negotiating a confidential price discount, Genomic Health was able to garner a change to the final guidance recommending Oncotype DX as an option for guiding adjuvant chemo in lymph node-negative women at intermediate clinical risk.
In 2015, the NHS then formally adopted the test, providing it to patients who meet the described eligibility criteria.
Since then, Genomic Health has held a monopoly in regard to NHS patients, Stein said, which it could now lose based on the final outcome of NICE's guidance revision.
Although the current draft once again recommends against all evaluated tests, the final recommendation could absolutely be updated if companies can present a special price for their testing that would shift the health economic results into an acceptable range — as Genomic Health did in 2013.
According to Stein, some health economics experts and "NICE watchers" have suggested that the new reevaluation is predominantly cost-motivated.
The fact that the draft reverses its position on Oncotype DX indicates that despite the fact that the confidential pricing for the test negotiated in 2013 was deemed sufficient at the time, it isn't anymore. In other words, regardless of the outcomes for Agendia, Myriad, and Nanostring, if Genomic Health is to maintain availability of its own test in the NHS, it may have to lower its price beyond what was agreed in 2013.
For the other test makers who were evaluated, the revision of the guidance offers them the possibility to join Genomic Health in the NHS market. Considering the stakes, all of the firms made lengthy comments on the diagnostics assessment report (DAR) that informed the draft, and it appears that the comments on the draft itself have also been ample — Stein said that the organization received so many comments from stakeholders that it had to postpone its planned Feb. 8th meeting.
Genomic Health and several of the other companies whose tests were evaluated declined to comment publicly on their reception of the draft and how they have responded to NICE during the comment period, which ended on Wednesday. But their comments made earlier on the diagnostic DAR are publicly available.
Authored by an external group at the University of Sheffield's School of Health and Related Research (ScHARR), with consultation by Rob Stein, the DAR weighed available evidence for the various tests, analyzing validity, utility, and cost-effectiveness, comparing to current decision-making for adjuvant chemotherapy prescribing, which the authors wrote include risk assessment tools like PREDICT, the Nottingham Prognostic Index (NPI) and Adjuvant! Online.
According to the assessment, all the tests evaluated appear to provide additional prognostic information over most commonly used clinicopathological factors in lymph node-negative patients, with more varied results in lymph-node positive patients.
But the overall calculations made about clinical utility and health economics were also based on the question of whether tests could predict patients' response to chemotherapy.
The authors wrote that there was some available data that different Oncotype DX risk groups receive different benefits from chemotherapy. However "interaction tests sometimes became non-significant when clinicopathological factors were adjusted for."
"Oncotype DX RSPC (Oncotype DX plus age, tumor size and grade) was prognostic but not statistically significantly predictive for chemotherapy benefit, indicating that the incorporation of CP factors to Oncotype DX may reduce prediction of chemotherapy benefit," they added.
For MammaPrint, which is the only other test to claim direct prediction of chemotherapy benefit, the DAR authors concluded that evidence overall "suggests no statistically significant difference in effect of chemotherapy between risk groups."
As a whole, the evaluation group concluded that none of the tests that were considered had strong enough evidence of a positive effect on patient outcomes and that their cost effectiveness compared with current practice was uncertain. Therefore, none of the assays should be recommended for routine use in the NHS.
Responses from the various companies to the DAR document vary, but mainly reflect their positions in regard to the potential commercial ramifications of the finalized NICE draft. Genomic Health attacked the methodology of the assessment on all sides, while other firms focused critiques on aspects of the analysis that might elevate or demote their own tests.
Among many responses to the DAR, Genomic Health expressed opposition to the authors' characterization of studies that assess Oncotype DX's direct prediction of chemotherapy benefit, and argued that because the DAR fails to recognize this ability for Oncotype DX, the subsequent health economic analyses are affected.
But the authors maintained in response to the company's objections that there is legitimate reason to be wary that the observed differences between risk groups observed in studies so far "could be due to confounding by clinicopathological variables, rather than due to a real effect."
According to Genomic Health, there exists other evidence that the DAR did not include in its analyses, but the authors disagreed, writing that they believe they appropriately evaluated all relevant data.
In one example, Genomic Health wrote that NICE postpone its assessment until the results can be incorporated from the company's TAILORx trial. However, ScHARR authors contended that there are a number of factors in the design of that trial that give pause to the potential impact of its results.
For one, the group wrote, TAILORx uses different cutoff points than are currently recommended, and it is unclear whether this trial will provide evidence relating to the prediction of chemotherapy benefit. In addition, the trial recruited only lymph node-negative patients who met the NCCN guidelines for chemotherapy, a population which may not be indicated for chemotherapy in the UK, according to usual clinical practice.
In Genomic Health's eyes, the DAR authors should also not have used data from TransATAC to inform its economic modeling, and the company complained that a bespoke analysis used by the DAR group was not made available for review by stakeholders.
But authors of the report argued that this criticism is unusual considering the fact that Genomic Health's own economic model also used the TransATAC data to characterize recurrence risk.
Authors from the evaluation committee reported that they also looked at an additional dataset provided in confidence to NICE by Genomic Health in response to the initial DAR, which comprised real-world evidence on the use of adjuvant chemotherapy in the NHS following testing with Oncotype DX, and that it was "an important piece of real-world evidence for use in the economic model," despite the fact that "more complete data could potentially have been collected and reported."
Among the firms and tests evaluated, only Agendia's MammaPrint has also directly claimed the ability to predict adjuvant chemotherapy benefit, and potentially stood in this evaluation by NICE to garner the same recognition that Oncotype Dx had in the previous guidance.
A significant proportion of the company's comments on the DAR focused on how the authors did or did not consider evidence from the MINDACT trial. For example, Agendia argued that MINDACT data "should be included in the prognostic evidence supporting MammaPrint," whereas the assessment authors described their methodology as requiring that MINDACT be considered under clinical utility, but not for prognostic performance.
Agendia also objected to assumptions made in the DAR's modeling of the long-term effects of chemotherapy, and, like Genomic Health, to its reliance on the TransATAC study.
Interestingly, the company brought up hypothetical impacts of the DAR as it stands, suggesting that if the conclusion of the report is that MammaPrint is not cost effective and not available for UK patients, then it is implying that high risk patients in the UK should all receive chemotherapy.
"This would result in knowingly treating 46 percent of … high-risk patients with a very toxic therapy for which they do not receive any significant or clinically meaningful benefit," Agendia wrote.
In reply, the DAR authors said that this view contrasts both with clinical opinions received by the group, and with the assumptions employed in its economic modeling.
Finally, Agendia also complained that the DAR seems structured to create an insurmountable barrier for adoption, in that the ability to predict chemotherapy benefit (which it concludes neither Oncotype DX nor Mammaprint can do) is necessary for a positive assessment by NICE.
But authors of the report objected to this, saying that their model includes prognostic benefit in the base case, as well as a sensitivity analysis in which Oncotype DX is assumed to be predictive of chemotherapy benefit, which addresses this question.
"We do not suggest at any point in the report that a tumor profiling test can only be valuable if it is predictive of chemotherapy benefit," the group wrote.
The assessment for the new NICE guidance did not address the question of chemotherapy benefit for Myriad Genetics' EndoPredict assay as it did for Oncotype DX and MammaPrint, which somewhat simplifies the question of its assessment.
Regardless, Myriad had a number of its own objections. For one, the company argued that the DAR model excludes relevant long-term adverse events associated with chemotherapy, namely chronic heart failure, with the implication that this might shift the cost-effectiveness calculations.
Myriad also jumped right in to pricing negotiations in its response to the DAR. The company wrote that the price for EndoPredict is assumed in the report to be £1,500 (about $2,100) per assay based on an assumption that all assays will be conducted at the centralized Myriad Genetics laboratory. But since Myriad expects the majority of NHS tests to be run locally, the company said that NICE should consider both central and local testing scenarios in its analyses.
According to Myriad, test kits have already been supplied directly to the NHS on this basis, "for small volumes of privately funded patients (40-50 per annum), with node-positive disease … and going forward, [the company] is prepared to commit to a confidential discounted price across the NHS … for such local testing." Based on a redacted price, Myriad estimated that its true costs would be less than what the DAR economic models used for NanoString's Prosigna.
For centralized testing, Myriad said it is also prepared to offer a discounted price. And in response, the DAR authors wrote that this would indeed reduce the incremental cost-effectiveness ratios it calculated.
However, the number that this reduction would produce is currently redacted, and it is unclear whether the draft conclusions by the NICE committee, which nevertheless recommend against all the assessed tests, reflect this change or not.
For its part, NanoString wrote that it objected to the fact that the cost effectiveness analyses used by the DAR do not take into account the emotional and psychological strain that breast cancer patients face when considering chemotherapy and the impact of gene expression profiling on this quality of life outcome.
"Gene expression profiling tests have the potential to reduce anxiety and improve health-related quality of life," the company wrote.
Authors of the report contended, though, that it's unclear why this type of quality of life metric should be any different for a patient using a genomic test as opposed to a clinical tool like PREDICT or Adjuvant! Online. "Contingent valuation is not part of [the] NICE methods guide," the group added.
NanoString also argued that time spent waiting for test results is a factor that should be have been considered. But the DAR authors disagreed.
Regarding chemotherapy benefit, the company made an interesting proposition — that the ability to predict chemotherapy of one test can or should be extended to all similar tests. In other words, although there is no direct evidence that Prosigna can predict benefit from chemotherapy, if the NICE model included sensitivity analyses that assume Oncotype DX can do so, it should do the same for Prosigna (and arguably the other tests under evaluation as well).
"The use of sensitivity analysis to explore this issue is especially important as the timescale required to generate direct evidence on the predictive benefit of Prosigna is very long, and denying access to patients in the meantime risks continued inappropriate targeting of chemotherapy and associated potential harms," the company argued.
"It is anticipated that incorporating a predictive benefit for Prosigna would increase the estimated [quality-adjusted life year] gain markedly, which would significantly reduce the corresponding ICERs compared to current practice," it added.
In the eyes of the DAR authors though, available evidence doesn't support the suggestion that if one gene test has the ability to predict benefit from chemotherapy then this is likely to extend to all such tests.
"The tests comprise different genes/markers to each other, and as such, the assertion that all the tests can predict chemotherapy benefit if one can does not necessarily follow. There are no empirical data to suggest Prosigna can predict benefit from chemotherapy, and without this, a sensitivity analysis would carry no importance."
With the comment period closed as of the end of last month, the NICE committee is now faced with managing what appears to be a larger volume of comments to the draft decision than it expected. The organization is required to respond to comments by registered stakeholders, though not to all comments it receives from other individuals or groups.
According to Stein, although there is no way to know what the ultimate outcome of the final NICE update will be, based on the DAR analyses, Genomic Health, NanoString, and Myriad seem like they may be in a position to adjust their prices to a point where they meet the NICE threshold.
For Agendia, Stein said that despite uncertainty in the economic models, the costs of MammaPrint may be too high to be brought down to a level that can sway the guidance. And for IHC, there are issues with reproducibility and consistency across labs that seem to preclude adoption.
The guidance update was initially slated for finalization in May, but the project website now says that the expected publication date is to be determined.