Tamara Syrek Jensen, JD
Director, Coverage and Analysis Group
Centers for Medicare & Medicaid Services
7500 Security Blvd.
Baltimore, MD 21244
Re: NCD CAG-00450N
Dear Ms. Syrek Jensen,
We recently published a Policy Forum in Science encouraging more widespread insurance coverage, including coverage with evidence development (CED), for genomic testing of tumor DNA. [R Eisenberg & H Varmus (2017) Insurance Coverage for broad genomic tests in oncology. Science 358:133-34.] We therefore welcome the CMS Proposed Decision Memo for Next Generation Sequencing (NGS) for Medicare Beneficiaries with Advanced Cancer (CAG-00450N) issued on November 30, 2017 and are grateful for the opportunity to submit comments about it.
We see much to admire in the Proposed Decision Memo. We particularly commend CMS for proposing full coverage of FDA-approved, technically validated, NGS-based diagnostic tests for cancer patients. We admire the aspiration to issue a broad coverage determination that provides similar coverage for similar tests nationwide rather than a narrower decision limited to coverage of the F1CDx test, although we have concerns (detailed below) about whether the current proposal fulfills that aspiration. And we applaud the use of CED to support further clinical studies and to provide better data in patient registries to answer important questions about clinical utility of these tests in the Medicare population. We welcome these important steps towards broader coverage of NGS testing for the significant clinical benefits they offer to current and future Medicare patients, as well as for the promise they offer of improving the control of cancer more generally. We hope that this important step by CMS will inspire similar coverage decisions by private insurers.
As the Proposed Decision Memo recognizes, FDA has recently taken important steps towards streamlining the regulatory pathway for bringing new NGS tumor profiling tests to market. These steps include reclassifying such tests as “class II” devices. [FDA (Nov. 15, 2017) Reclassification Order DEN170058, available at https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfpmn/denovo.cfm?ID=DEN170058.] This allows these tests to be cleared through the less onerous premarket notification process, under section 510(k) of the Federal Food, Drug & Cosmetic Act, without having to go through the more elaborate premarket approval process necessary for approval of class III devices. FDA also accredited the New York State Department of Health (NYSDOH) as a third-party reviewer of such devices with authority to clear devices without the need for further FDA review, opening the door to similar arrangements with other reviewers. FDA’s classification of the MSK-IMPACT assay as a Class II device allows subsequent “next generation sequencing based tumor profiling tests” to take advantage of the 510(k) premarket notification process upon a showing of substantial equivalence to that predicate device. This classification reflects FDA’s determination “that class II (special) controls provide reasonable assurance of the safety and effectiveness of the device type.” [FDA (Nov. 15, 2017) letter from Reena Philip to Christina England, https://www.accessdata.fda.gov/cdrh_docs/pdf17/DEN170058.pdf, at p. 2.] These steps highlight a cost-effective and efficient pathway for regulatory clearance of similar devices in the future, while assuring physicians and patients that they meet FDA’s standards for safety and effectiveness. This is all good news for those who seek to bring advances in cancer research to patients with advanced cancers.
We are concerned, however, that the proposed CMS coverage policy could undermine the benefits of these important steps by FDA. In particular, we worry that the proposed policies will give future test developers an incentive to pursue the more costly and time-consuming route of FDA premarket approval of their tests as “companion diagnostics” in order to get the benefit of full CMS coverage. This is so because only “an FDA-approved companion in vitro diagnostic“ is immediately eligible for full coverage under part A of the proposed policy without the further requirements imposed on CED for “an FDA cleared or approved in vitro diagnostic” under part B of the Proposed Decision.
For a test that has merely been “cleared” for use as a “next generation sequencing based tumor profiling test,” like the MSK-IMPACT test, further requirements for CED include (i) registration of the test in the NIH Genetic Testing Registry; (ii) enrollment of the patient in a prospective registry designed to answer specified CED questions; (iii) “a written executable analysis plan to address the CED questions” with the registry and a plan to “make data available in a form and manner specified by CMS upon request;” and a registry that (iv) tracks certain information at baseline and (v) tracks further information after each intervention.
Such a registry would undoubtedly be a valuable resource for learning more about the significance of genetic data generated by the covered test and about effects of cancer therapies provided to patients with genetically diverse cancers, and that should be encouraged. But to require participation in such a registry as a condition for coverage adds to the costs and uncertainty facing test developers in deciding whether to pursue FDA clearance for a test for tumor profiling, rather than pursuing premarket approval for the same test as a companion diagnostic. By making it more lucrative to pursue the more burdensome premarket approval process at FDA, CMS may thus undermine the impact of recent efforts by FDA to streamline and accelerate review of NGS-based tumor DNA tests.
We recognize that FDA’s recent de novo classification decision covers “next generation sequencing based tumor profiling tests” and does not yet extend to companion diagnostics. But that was the only issue that FDA was called upon to decide in the context of the de novo request for classification of the MSK-IMPACT test. FDA has repeatedly recognized in guidance documents for industry and FDA staff that companion diagnostics might be classified as class II devices eligible for clearance without requiring premarket approval. As FDA explained in 2014,
FDA will apply a risk-based approach to determine the regulatory pathway for IVD companion diagnostic devices, as it does with all medical devices. This means that the regulatory pathway will depend on the level of risk to patients, based on the intended use of the IVD companion diagnostic device and the controls necessary to provide a reasonable assurance of safety and effectiveness. Thus, the level of risk together with available controls to mitigate risk will establish whether an IVD companion diagnostic device requires a premarket approval application (PMA) or a premarket notification submission (510(k)). [FDA (Aug. 6, 2014) In Vitro Companion Diagnostic Devices: Guidance for Industry and Food & Drug Administration Staff, at p. 10.]
FDA’s approach to device classification is flexible and permits change over time as initial concerns over safety and effectiveness for a new technology prove to be manageable with less burdensome regulatory controls. FDA explicitly anticipates this for in vitro companion diagnostics. Moreover, it has already expressed its satisfaction with the adequacy of “class II (special) controls” to ensure the safety and effectiveness of very similar NGS diagnostic tests for tumor profiling. Under these circumstances, we foresee future submissions to FDA seeking de novo classification of NGS tests for use as companion diagnostics, and we see no reason why CMS should make coverage of these products contingent on a more extensive regulatory approval process for these devices than FDA considers necessary.
A comparison between the MSK-IMPACT test (cleared as a tumor profiling test, and therefore eligible only for CED) and Foundation Medicine’s F1CDx test (approved as a companion diagnostic, and therefore eligible for full coverage) suggests that they differ more in the regulatory strategies pursued by their sponsors than in their technical features and clinical utility. Whether the test is called a “companion diagnostic” or a “tumor profiling” test, its clinical utility resides in its capacity to guide selection of appropriate care for cancer patients. Toward that end, the MSK-IMPACT test uses nearly identical methods to analyze the genes categorized as critical components of companion diagnostics in the F1CDx test. Both tests provide very similar or identical information about variants in very similar sets of genes, with similar implications for clinical care. The MSK-IMPACT test has some notable advantages, including more rigorous validation of the existence of somatic mutations because both tumor and germ line DNA are analyzed. Both tests have been used in tens of thousands of patients and validated by stringent criteria, such as those of the NYSDOH.
The different treatment of these similar tests is puzzling in light of CMS’s explicit choice in the Proposed Decision Memo to broaden the number of tests to be considered beyond the F1CDx test that is before CMS. The Proposed Memo states at p. 64:
Since our evidence question and analytic framework apply broadly to molecular diagnostic tests for advanced cancer we are considering the type of technology so that claims for similar tests will be covered in a similar manner nationwide under title XVIII.
As repeatedly stated in the Proposed Decision Memo, CMS seeks answers to the following question:
Is the evidence sufficient to conclude that next generation sequencing when used as a diagnostic test for patients with advanced cancer meaningfully improves health outcomes?
The Proposed Decision Memo answers this question with an unqualified “Yes” only for “FDA-approved companion diagnostics.” Other NGS diagnostic tests – including FDA cleared tumor profiling tests that provide similar information to physicians and patients who would use that information for the same purposes, and with the same degree of confidence that FDA considers the test to be safe and effective – do not qualify for full coverage under the Proposed Decision Memo. Instead, coverage of these tests is limited to CED for patients who agree to participate in further clinical studies and for service providers who comply with complex requirements. The result is to treat similar tests in a dissimilar manner depending on the regulatory pathway used to satisfy the FDA standards of safety and efficacy.
We recognize that allowing full coverage under part A to tests like MSK-IMPACT could reduce the use of CED under part B of the Proposed Decision, and thus slow the accumulation of data to guide further improvements in cancer care. We wholeheartedly support the use of CED as a mechanism to support further clinical studies in patients with advanced cancer, including observational studies conducted with well-curated patient registries. But the potential benefits of further research do not justify withholding coverage of diagnostic technologies with proven clinical utility. Moreover, use of CED to support a significant fraction of NGS testing would impose a significant administrative burden on test providers.
Rather than diverting patients and physicians who choose an FDA cleared or approved test other than F1CDx into CED, we suggest broadening the criteria for use of CED to test clinical utility beyond patients with “recurrent, metastatic, or advanced stage IV cancer” who have “not been previously tested” and “have decided to seek further cancer treatment.” These limitations omit the use of NGS-based tests in a number of situations in which it would be useful to assess their value, including: testing earlier in the disease process to decide about adjuvant therapies or to predict the course of the disease; repeated testing for appearance of mutations that confer therapeutic resistance; and testing of tumor DNA circulating in the blood (“liquid biopsy”), rather than DNA obtained through an invasive biopsy. We particularly welcome the availability of CED for diagnostic tests that have not yet been approved or cleared by FDA when used to identify candidates for participation in new clinical trials, although we would contend that a broader selection of trials, not just those conducted by NCI’s National Clinical Trials Network, should qualify for CED.
Summary and Recommendations
We applaud the actions being taken by the FDA and CMS to accelerate the widespread use of improved diagnostic tests that employ NGS-based methods to identify underlying drivers of disease and to guide patients toward appropriate therapies and novel clinical trials. Nevertheless, we are concerned that the Proposed Decision makes too much of two regulatory distinctions: first, the distinction between tests that have been “approved” rather than “cleared” at FDA, and second, the distinction between “companion diagnostics” and “tumor profiling” tests. These are very similar tests with very similar clinical utility, and they should be subject to the same requirements for reimbursement.
We recommend the following changes in part A of the proposed decision:
First, in Part A.2.a., we recommend changing the wording of the first criterion for coverage of a test from “FDA-approved” to “FDA cleared or approved.” The proposed modification would expand coverage to include diagnostic devices that FDA has determined meet the statutory standards for classification as Class II devices. Such devices would still require premarket notification to FDA, along with submission of such information as FDA requires to implement special controls to ensure safety and effectiveness for devices of that type. Most relevant to the proposed decision, should FDA decide in the future to reclassify NGS in vitro companion diagnostic tests as Class II devices, they would be eligible for full coverage, not limited to CED, without the need to go through a further regulatory review that FDA has determined is not necessary to meet its standards.
Second, again in Part A.2.a., we recommend further expanding coverage by deleting the word “companion” in the phrase “companion in vitro diagnostics,” to permit coverage of other types of FDA cleared or approved NGS in vitro diagnostic tests such as tumor profiling tests. As explained above, we believe that this distinction in the proposed decision ignores the substantial similarity in both technical features and clinical utility for tests that are used for the same purposes and should receive the same coverage.
In addition, we recommend the following changes in part B:
First, in Part B.1., we recommend broadening the criteria for CED to support the testing of patients at an earlier stage in the course of their disease and to support repeated testing of patients to detect the appearance of mutations that confer therapeutic resistance.
Second, in Part B.2.b.1, we recommend broadening the set of eligible trials to include trials outside of the NCI National Clinical Trial Network.
Harold Varmus, M.D.
Weill Cornell Medicine
New York Genome Center
Rebecca S. Eisenberg
University of Michigan Law School