Important Points Highlighted by the Individual Speakers
- Payers use well-recognized guidance from professional societies, evidence-based consensus reports, and health care organizations to establish the clinical validity and utility of molecular diagnostic tests.
- It is suggested that Medicare will likely not be the leader for setting genomic testing reimbursement policies.
- Payers will be increasingly challenged with processing an enormous volume of coverage and reimbursement requests as a rapidly growing number of new genomic tests become available for clinical use.
There are several steps between determining the actionability of a genetic variant (see Chapter 2) and deciding to take clinical action, said Bruce Blumberg, institutional director of graduate medical education, Northern California Kaiser Permanente, Permanente Medical Group. One of the steps between actionability and action is the consideration of whether or not genomic testing costs should be reimbursed by public and private payers. To shed light on this issue, representatives from the commercial and government payer sectors explained their processes for evaluating genomic or multi-gene panel sequencing tests as well as their criteria for deciding whether to provide coverage of such testing.
Aetna is a large multi-payer company that offers health insurance plans and other types of insurance throughout the United States. Its process for handling genomic testing requests is the same process used to develop clinical policy, said Robert McDonough, senior director of clinical policy and research and development at Aetna. The company applies very similar criteria for determining the coverage of such testing as it does for other types of medical technologies (e.g., positron emission tomography, or PET, scans). Aetna policies cover only medically necessary tests and treatments and exclude coverage for experimental and investigational technologies, McDonough said. All plans must have a definition for coverage that address certain elements—specifically whether use in clinical practice is experimental, investigational, or medically necessary.
“The goal is to develop objective, clinically supported, and defensible coverage determinations,” McDonough said. To assess and to provide information about whether specific medical services are necessary or investigational, the Clinical Policy Unit has developed more than 700 clinical policy bulletins (CPBs), or medical policies, including bulletins about genomic and genetic tests, which are all posted on the company website.1 CPBs describe which tests and procedures Aetna considers to be medically necessary versus those that are for cosmetic, experimental, or unproven uses. These determinations are based on information from consultation of different sources, including peer-reviewed medical journal articles and reviews, evidence-based consensus statements and other expert opinions, and guidelines from nationally recognized health care organizations. The CPB for genetic testing provides a list of criteria that all must be met before a genetic test can be considered medically necessary for disease diagnosis.2 These criteria are
- The patient displays clinical features, or is at direct risk of inheriting the mutation in question (pre-symptomatic); and
- The result of the test will directly impact the treatment; and
- After history, physical examination, pedigree analysis, genetic counseling, and completion of conventional diagnostic studies, a
1For more information, see Clinical Policy Bulletins, http://www.aetna.com/healthcare-professionals/clinical-policy-bulletins.html (accessed May 15, 2014).
2Clinical Policy Bulletin for Genetic Testing, http://www.aetna.com/cpb.medical/data/100_199/0140.html (accessed May 15, 2014).
definitive diagnosis remains uncertain, and one disease diagnosis, as defined by Aetna, is suspected.
Additional criteria are also defined for specific diseases before a test is considered medically necessary.
Explanations and references are included within these documents to help the reader understand why a particular test is or is not covered. The CPBs help to support a policy that is “applied consistently and fairly and has a sound basis” as well as to provide transparency about the coverage, McDonough said. Policies are not created for every technology; they are prioritized based on questions that arise during the claims, precertification, and preauthorization processes. Revisions to Aetna’s policies are typically generated by new evidence, guidelines, consensus statements, and alterations in relevant regulations for a particular technology.
Coverage Criteria and Creating Policies
Aetna uses established criteria developed by the Blue Cross and Blue Shield Association’s Technology Evaluation Center for assessing whether a medical technology warrants clinical coverage, McDonough said. The criteria3 include the following:
- The test or treatment must have final approval from appropriate governmental regulatory bodies, where required;
- scientific evidence must permit conclusions about its effect on medical outcomes;
- technology must improve net health outcomes;
- the technology must provide as much health benefit as established alternatives; and
- the improvement in health must be attainable outside investigational settings.
The CPB drafting process entails a comprehensive search of the peer-reviewed medical literature and an assessment of the current regulatory status of the technology of interest. The Clinical Policy Unit then considers evidence-based guidelines, such as those from the American Society of Clinical Oncology and the National Comprehensive Cancer Network (NCCN) for cancer diagnostic testing. Aetna also performs
technology assessments and solicits expert opinions (e.g., through a liaison group that regularly seeks input from specialty medical societies). All of this information is synthesized into an initial CPB draft which is then subjected to review by the head of Aetna’s national medical policy and operations department and the legal department, and then it is finally reviewed for approval by the chief medical officer or a designee. Upon approval of a new policy, the Clinical Policy Unit helps facilitate implementation across coding and reimbursement areas.
Covered Genetic Testing
Examples of genomic technologies that are currently covered include noninvasive prenatal detection of chromosomal abnormalities, such as found in Down syndrome. Guidelines from the American College of Obstetrics and Gynecology4 that supported the test were important in the decision to cover it, McDonough said; the covered tests sequence cell-free fetal DNA in blood samples taken from expecting mothers. When it comes to genomic sequencing for diagnosing individuals with suspected genetic disorders (e.g., cystic fibrosis), the approach that Aetna has taken is to reimburse genetic testing for core cystic fibrosis mutations. Because next-generation sequencing may potentially be used to identify additional cystic fibrosis mutations not included in the recommended panel, the testing laboratory may offer a separate test to the individual member, but the member, rather than the payer, would pay for it directly, McDonough said.
Many people are looking to the Medicare program to lead the way in determining coverage of molecular diagnostics, said Louis Jacques, director of the Coverage and Analysis Group at the Centers for Medicare & Medicaid Services (CMS). However, while it is true that genomic testing of cancerous tumors is relevant for the aging population, if a predisposition for lung cancer could be identified and used to prevent someone from smoking, the target age for genetic testing would be in the teenage years, not at 65 years old. Thus from a practical standpoint, Jacques said,
4Noninvasive Prenatal Testing for Fetal Aneuploidy: Committee Opinion, http://www.acog.org/Resources_And_Publications/Committee_Opinions/Committee_on_Genetics/Noninvasive_Prenatal_Testing_for_Fetal_Aneuploidy (accessed April 29, 2014).
Medicare is “not going to be the major driver” on reimbursement for the genomic testing space.
Under Medicare regulations, a basic requirement for finding a diagnostic test to be medically reasonable and necessary is that the treating physician has to order it to help manage the patient, which automatically raises the question of how the physician is using the test for the patient, Jacques said. When considering coverage for such technology, CMS applies the same procedures it uses to evaluate any tests or treatments, Jacques said: the national coverage determination process (which takes 9 to 12 months), the local coverage determination process (which takes roughly 3 months), or a quicker, claim-by-claim adjudication process. Because of the variety of coverage determination mechanisms, the system has been criticized for not being transparent enough.
In 2009, the agency undertook a national coverage determination for pharmacogenomic testing to evaluate genetic tests predicting how a patient will respond to treatment with warfarin, a commonly prescribed anticoagulant.5 Aside from a fair number of meetings by the Medicare Evidence Development and Coverage Advisory Committee (MEDCAC) to examine various genetic tests, “that’s about the last you’ve heard from us on this topic,” Jacques said. Because of the way that the current coding system functions, a non-unique test code makes it possible that the default decision may be to reimburse for test or services without needing an individual to make a conscious decision for the claim to be paid. However, Jacques noted that the agency has started a “novel and innovative” pilot program called MolDx, which manages molecular diagnostic services and identifies and establishes unique identifiers and coverage and payment for such testing.6 The program is overseen by Palmetto GBA, an administrative contractor in South Carolina. Before potentially expanding it nationally, CMS would have to examine how the pilot program could evolve or grow and decide whether a nationwide program would be overseen by a single contractor or by a few regional contractors.
CMS can grant conditional coverage of testing or a procedure—a status called coverage with evidence development (CED)—only in selected patients enrolled in research studies that could provide further evi-
5For more information, see National Coverage Determination (NCD) for Pharmacogenomic Testing for Warfarin Response (90.1), http://www.cms.gov/medicarecoverage-database/details/ncd-details.aspx?NCDId=333&ncdver=1&bc=BAAAgAAAA AAA& (accessed May 15, 2014).
6For more details, visit MolDx, http://www.palmettogba.com/palmetto/MolDX.nsf/DocsCatHome/MolDx (accessed May 15, 2014).
dence of its utility. In some cases, Jacques said, therapeutic knowledge is so well developed that it is possible to provide sufficient evidence to successfully argue for CMS coverage of a diagnostic test. For example, when the agency reviewed FDG-PET imaging (positron emission tomography using F-18 fluorodeoxyglucose) for cervical cancer, it found that the evidence indicated that physicians could use scan results to make meaningful changes in a patient’s treatment plan; because anticancer therapies typically come with toxic side effects, it is meaningful for a patient if the physician determines that chemotherapy or invasive surgery is not appropriate. “Even though we did not have the perfect clinical trial from soup to nuts, we said, well, we’ve got enough [evidence],” Jacques said.
Medicare used a MEDCAC to review the evidence for the use of beta-amyloid PET imaging for diagnosis or treatment of dementia or neurodegenerative disease, Jacques said. While the review found that there was insufficient evidence for coverage, a decision was made to identify the status as CED for the purposes of excluding Alzheimer’s disease in specific diagnoses and for the purpose of enriching clinical trials that addressed disease treatment and prevention strategies.7
Another example of evidence consideration was in the case of pharmacogenetic testing to predict patient response to warfarin. Variations in the genes that encode CYP2C9 or VKORC1 enzymes (among others) affect disposition, response, and toxicity for individual patients receiving chronic warfarin therapy (Dean, 2013). While there was no doubt that this testing would provide clinically valid results, the real question was what difference the results would make in the medical management of typical Medicare patients, Jacques said. The FDA package label8 for warfarin specifies that prescribers should individualize the dosing regimen for each patient, adjusting it based on the international normalized ratio response, or a test to measure the clotting time of blood; additionally, Jacques said, knowledge of genotype can inform initial dose selection if a patient is taking drugs known to affect warfarin metabolism and either has an adverse genetic profile or is an older adult who has multiple illnesses. The latter describes most Medicare beneficiaries, he said, “so we were left with a clinical utility vacuum” for the genotyping. Given that
7Decision Memo for Beta Amyloid Positron Emission Tomography in Dementia and Neurodegenerative Disease (CAG-00431N), http://www.cms.gov/medicare-coveragedatabase/details/nca-decision-memo.aspx?NCAId=265 (accessed April 29, 2014).
8Coumadin® Package Label, Bristol-Myers Squibb Company, Princeton, New Jersey 08543 USA, revised October 2011.
physicians already know if a given patient is elderly and has comorbidities and which medications the patient takes, Jacques said, “Why would you go ahead and wait a few days or potentially a week to get a genetic test result? Even if results say the patient’s genetic profile is fine, the doctor is not going to ignore the other factors.” Thus CMS decided to grant CED of warfarin pharmacogenomic testing to assess utility, he said. Several clinical trials of genotype-guided warfarin dosing and similar drugs have found varying results (Kimmel et al., 2013; Pirmohamed et al., 2013; Verhoef et al., 2013). There is still uncertainty about the use of genetic testing for warfarin dosing even after randomized controlled trials were conducted.
In its coverage determination process, Medicare is required by law to consider public comments, Jacques said, and how persuasive those comments are depends on a variety of factors. For example, the clinical policy unit will not find thousands of form letters driven by interested stakeholders to be as convincing as a few well-considered public comments that point out, for instance, that an entire body of relevant evidence was omitted from the determination analysis. But public sentiment might sway other players who can influence agency decisions; in particular, Congress has the final say in decision making.
Commercial insurance companies are highly regulated, McDonough said. In certain circumstances, individual states may mandate coverage for certain treatments despite a lack of evidence for the value of such treatments, such as the use of bone marrow transplantation in treating breast cancer patients. Demand from members and plan sponsors can also have an impact on policy, McDonough said, but overall the primary coverage decision is based on “high-quality evidence and outcomes.”
Because decisions are based on a rigorous review of the evidence, it is unusual for coverage to be withdrawn based on new information, McDonough said. A disinvestment would need to go through a process similar to the one described earlier for coverage decisions. Jacques said that the removal of coverage is also rare for Medicare, as most requests are for expanded coverage by those who say there is now evidence to support the use of a test or service.
How will payers cope as a rapidly growing number of new genomic tests become available and patients and their providers ask insurers to pay for them? As one participant put it, administrative systems are going to be facing “a giant train wreck” where they are unable to process the volume of reimbursement and prior authorization requests. A particularly troublesome challenge for CMS, Jacques said, is that the volume of new tests is so huge that is it difficult to assess all of them for clinical utility. Nonetheless, he said, if diagnostic testing companies or researchers “come to us with an argument that is fundamentally grounded in clinical utility, that will be a persuasive argument” for coverage. Other workshop speakers described examples of difficulties with obtaining reimbursement for clinical genomic testing, such as in cases where a patient at high risk for a genetic disorder would benefit if sequencing could be done on a DNA specimen from a family member who died from the illness (see Chapter 2).
Jacques acknowledged the general challenge ahead at CMS. If several genomic tests come along that are all claimed to do the same thing but that are built on different platforms or look at different parts of the genome with some overlap, he asked, “how do we know whether those differences are significant or not? I don’t know that the best paradigm currently exists to handle that.” CMS would be open to hearing input from stakeholders on innovative or collaborative solutions, he added.
McDonough observed that Medicare and commercial payers have addressed similar challenges, such as handling different indications for cancer drugs by recognizing medications that are listed in certain published compendia. Similarly, he suggested, it is possible that a molecular diagnostics compendia could be created for which payers would recognize oncology markers identified in certain listings. The challenge with compendia, Jacques said, is that they are proprietary publications and not publicly accessible.
McDonough also suggested that concerns about the huge volume of different genomic tests on the horizon will be tempered because payers will only be aware of the most common tests—those with specific, identified codes—which will limit the number of tests that need to really be evaluated. Other, uncommon tests will be billed with generic codes, which generally get paid without review, he said.
No matter what the disease of focus, Jacques said, the type of evidence that is acceptable to public or private payers in their coverage decisions about genetic testing could be the subject of a vigorous, open debate held in a credible and bias-free public forum, where participants can discuss which types of evidence they find to be persuasive and which types they do not believe to be persuasive. As one potential starting point, Jacques suggested that interested stakeholder groups (such as professional societies or public interest groups) could go to Capitol Hill with a proposal for rule making to determine what is “reasonable and necessary” and to establish a particular process within Medicare.
The evaluation process, McDonough noted, is an important source of information for effectiveness guidance documents (EGDs), which discuss the type of research that is needed to provide reliable evidence for payers on the effectiveness and safety of medical technologies for specific diseases. For example, the nonprofit Center for Medical Technology Policy (CMTP) in Baltimore has completed an EGD on evidence standards for studies of the clinical validity and utility of molecular diagnostics for oncology (CMTP, 2013).9 The center is now planning an EGD on evidence standards for next-generation sequencing in oncology.
McDonough observed that the CMTP effectiveness guidance documents for evidence standards recommend convening a multidisciplinary group of stakeholders to determine the preferred evidence-based decision-making process for certain classes of technologies. But, he added, “I don’t think we can rely solely on expert opinion in making these decisions.” If the expert opinion “is not really bounded by reliable evidence, then one really needs to question the validity of the expert opinion.” What may be more important is the quality of the evidence; however, if stakeholders can agree on an evidentiary framework for assessing new genomic tests, that will make the process predictable, he noted.
The CED process needs to be more efficient and accessible, Jacques said, but that is a matter of resources, because the process is mostly done under a national coverage determination, which takes a total of 2 years of staff work. The analysts on his staff have been affected by the sequester and furloughs, which means that his unit can only manage 5 or 6 national coverage determinations a year, including areas beyond diagnostic testing. “If we want CED more broadly,” Jacques said—and he agreed it
makes sense to do that—“we need to find a better way to do CED.” One possibility, he suggested, would be to have a voluntary process in which CED could “essentially be a default position for medical technologies that meet certain criteria.” Test developers might be given, say, a 3-year grace period to establish sufficient evidence for full coverage approval. Such a system, he said, could motivate everybody to say, “Okay, we’ve got 3 years to get the answer on this. Let’s do it.”