NEW YORK (GenomeWeb) – Most of the approximately 300 public comments that the Centers for Medicare & Medicaid Services has gotten for its draft national coverage determination (NCD) for next-generation sequencing cancer panels criticize it for being too restrictive.
A perspective that hasn't gotten as much attention is that the CMS draft coverage decision is perhaps not restrictive enough, and that the policy could be bad for patients by increasing off-label drug use based on the results of NGS panels.
Howard West, medical director of the thoracic oncology program at the Swedish Cancer Institute, said that the draft NCD, if finalized as written, would make it far easier to get broad genetic testing for his patients. He believes NGS testing is valuable in certain settings, such as non-small cell lung cancer, where there are multiple molecularly informed treatment options showing they benefit patients, but procuring sufficient tissue for repeated genomic testing is challenging in advanced patients.
But NGS hasn't yet proven its value as the standard of care, he said, and the draft NCD risks implying that this technology actually has strong evidence to support broad use today.
"Despite the breathless hype and feel-good proclamations about molecular testing and precision medicine made by academic, industry, and political leaders, the fact remains that beyond the individual markers that have an established role and should definitely be tested for, there is no evidence that broad genomic testing of a population leads to better outcomes for the population," he said.
CMS put out the draft NCD last year after reviewing Foundation Medicine's FoundationOne CDx in parallel with the US Food and Drug Administration. After evaluating the data submitted by Foundation and surveying the published literature on the use of NGS testing in cancer, the agency last year proposed a policy that would apply not just to Foundation's test, but to all NGS panels when they're used for recurrent, metastatic, or advanced Stage IV cancer patients.
CMS has proposed three pathways through which NGS cancer panels may garner Medicare coverage. The agency said it will grant full coverage for genetic markers on the panel that have premarket approval (PMA) from the FDA as a companion diagnostic — where testing is required for the safe and effective use of a drug. For markers without CDx status, but that the FDA has cleared or approved for use in a patient's treatment plan with other information (such as Memorial Sloan Kettering Cancer Center's 468-gene oncopanel MSK-IMPACT), CMS is willing to grant coverage with evidence development (CED), when the lab collects patient outcomes in a prospective registry.
Lastly, for lab-developed tests without the FDA's blessing that are performed in a CLIA-certified lab — the majority of tests on the market — CMS is proposing CED when tests are included in a National Cancer Institute National Clinical Trial Network study. CMS accepted public comments on this proposal until Jan. 17, and is slated to issue a final determination on Wednesday, Feb. 28.
The majority of patients, oncologists, and representatives from genetic testing labs have told the agency that the coverage proposal as written will harm cancer patients by restricting access to NGS tumor profiling and have asked CMS to expand its coverage criteria. They want CMS to pay for NGS cancer profiling even when tests are not FDA approved/cleared, cover testing more than once in a patient's lifetime, ease the registry requirements, and expand the types of studies that would qualify under CED.
However, some worry that the NCD still puts the public health at risk because tests like Foundation's have not proven that testing advanced cancer patients for the majority of genes on these panel — and directing care based on the results — benefits them. "The question facing FoundationOne CDx is not whether it contains some useful tests (e.g. BRAF in melanoma)," Vinay Prasad, a hematologist-oncologist at the Oregon Health and Sciences University, wrote in an Annals of Oncology commentary this month criticizing the draft NCD. "But whether [the panel] offers benefit beyond current testing of a handful of actionable genes in specific cancers."
Although CMS has proposed full coverage for only a handful of FDA-approved genetic markers on FoundationOne CDx, the prevalence of these markers suggests that many cancer patients will get negative results for covered CDx indications. For example, among advanced NSCLC patients tested on the panel, it would be expected that between 10 percent and 20 percent would have EGFR mutations (and up to 50 percent of patients treated with EGFR inhibitors would develop T790M resistance mutations); around 5 percent would have ALK mutations; and 3 percent would have BRAF mutations and receive drugs the FDA has approved in those molecularly-defined indications.
For those testing negative for these markers, Foundation will still report other mutations that patients might have in more than 300 genes, as well as microsatellite instability and a tumor mutational burden score. Their doctors would be able to use that information to decide whether to prescribe standard chemotherapy, palliative care, an immunotherapy, or a targeted drug, including those that may not be FDA-approved for their specific biomarker subpopulation.
Vincent Miller, chief medical officer at Foundation, disagreed that the draft NCD as written will lead to harmful off-label prescribing based on FoundationOne CDx results. He reminded that the CMS has proposed to cover testing only for advanced cancer patients, many of whom are out of options. "Except in the rarest of cases, metastatic cancer is incurable, the treatments are God awful, and they're really expensive," Miller said. "Many disease states have no approved therapies … So, the background is critical."
Moreover, even if there are off-label treatment options available, CMS may not agree to pay for those treatments. "It's usually a payor who's holding the cards on that," Miller said. "It's not that easy to do that."
For some, the biggest disappointment of the draft NCD were its CED proposals, which they felt would be too narrow, costly, and fail to foster a learning healthcare system that collects patients' experiences on precision oncology approaches and improves understanding of what's working and what isn't. "I'm not prepared to say that precision medicine is either a booming success or an abject failure," said Richard Schilsky, chief medical officer at the American Society of Clinical Oncology. "It's a treatment strategy that is in development that has sufficient scientific rationale and preliminary clinical experience to continue to study it. Where I have a problem is with its widespread application outside of any mechanism to learn from what you're doing."
Hype and hope
The hype and hope around precision medicine is most palpable in oncology because the stakes are high, the drugs are expensive, and the patients are motivated to access the latest advances in care. When it comes to precision medicine, the FDA has approved the most targeted treatments for patients with rare molecular markers driving their tumors, and last year, approved the first pan-cancer indication for the immunotherapy Keytruda (pembrolizumab).
The FDA is approving more drugs that require a companion test to identify best responders, and organizations that create guidelines now support testing for multiple markers in several cancer indications. Because advanced cancer patients often lack sufficient tissue to endure multiple single-marker tests, Miller said, it makes sense that the field is moving toward more NGS profiling. "How are you going to do that without comprehensive genomic profiling?" he posited. The FDA’s approval of FoundationOne CDx, "addresses where we are today but also where the field is going," Miller added.
The growing adoption of NGS has been good for Foundation, which markets FoundationOne CDx and other NGS tests. In January, the company preliminarily reported physicians ordered 67,375 clinical tests last year, which would be up 54 percent from a year ago, and it anticipates clinical testing revenues of $53.1 million in 2017, a 39 percent increase from $38.1 million in 2016.
The majority of comments to CMS' draft NCD suggests there are plenty of oncologists who have ordered NGS testing for their patients and support its expanded use based on the impact they've seen in some patients. "Every patient with advanced cancer is entitled to a fully informed treatment plan," said Lincoln Nadauld, executive director of precision medicine and precision genomics at Utah-based hospital system Intermountain Healthcare. "And fully informed in this day and age includes a comprehensive genomic analysis."
This is what is offered at Intermountain, where patients have had remarkable responses to precision oncology, but there have also been patients who didn't respond to a treatment as Nadauld and his colleagues would have predicted based on a specific genomic alteration. "That's hard for patients to accept when that happens," he said. "And it's hard for providers as well."
Nadauld recalled a young metastatic breast cancer patient whose tumor harbored a genomic alteration suggesting she would benefit from a new immunotherapy. The patient, who had already failed other standard therapies, received the immunotherapy but she didn't respond as doctors predicted based on genomic tumor profiling. Her cancer progressed quickly and she passed away.
"That was a case where there was evidence in the literature and elsewhere that the patient had a good chance at responding, and [she] didn't, unfortunately," Nadauld said. "That's a caveat not only for precision medicine but for all of cancer care. We prescribe treatment based on best available evidence but we know that each individual patient is a separate case, and that just because previous clinical evidence has suggested that a treatment will be effective, it doesn't mean it will be for that specific patient." Nadauld highlighted that Intermountain, in partnership with several other healthcare organizations, is collecting and sharing data on how patients fare on molecularly informed treatments within the Oncology Precision Network, so doctors can use that information in the care of their patients.
There are now a number of observational and case control studies showing that patients who receive molecularly informed targeted treatment fare better than those who don't. Nadauld and others at Intermountain published a study last year comparing 36 patients who got targeted treatments within Intermountain's precision oncology program between July 2013 and January 2015 against 36 historical controls who received standard chemotherapy or best supportive care over five years. The researchers reported in the Journal of Oncology Practice that patients who received targeted drugs guided by genomic test results had average progression-free survival of 22.9 weeks compared to 12 weeks in the control cohort.
But studies like this haven't convinced the skeptics that NGS-guided treatment is necessarily better for cancer patients than the standard of care. In West's survey of the published literature, he sees a lot of retrospective analyses, studies that use endpoints manufactured to make targeted therapies look good, and cherry-picked anecdotal reports that invariably fail to mention how many patients needed to be genomically tested in order to find the patients who had an actionable result.
"Coverage and reimbursement are not n-of-one decisions; they are population-level decisions, which is why a collection of anecdotes, no matter how compelling and how numerous, does not guarantee either," said reimbursement and regulatory policy expert Girish Putcha. (Putcha is director of laboratory science at Medicare contractor Palmetto and chief medical officer at Freenome, a company developing a noninvasive screening technology, but his comments for this article aren't on behalf of these organizations.)
ClearView Healthcare Partners recently surveyed twelve medical directors at commercial plans that together cover 44 million lives (14 percent of the US population). Roughly half of these medical directors said that they do not currently cover NGS tumor profiling tests and while the other half said they only cover NGS testing on a case by case basis. While the surveyed payors said they would comply with the NCD for their Medicare managed care covered lives, medical directors were more cautious about how CMS' coverage determination might impact their commercial plans. Some expressed strong reservations about the process CMS used to craft the draft NCD and the lack of data showing NGS panels are clinically useful broadly in the cancer population.
Roughly half the respondents said that in the near term they would not align their commercial plans with the proposals in the draft NCD, and the other half expressed a willingness to cover FDA-approved NGS panels for use in supported biomarker applications. Only a few plans indicated a willingness to cover NGS testing more broadly than the CMS proposal, for example, lab-developed tests that are not associated with registries or clinical trials. Over the next year to two years, however, around 75 percent of medical directors surveyed by ClearView said they expect their commercial coverage of NGS testing in oncology will align with CMS policy, while the rest maintained they would still only cover specific indications where testing has shown clinical utility.
"In my opinion, coverage and reimbursement requires that on balance, the use of a drug, test, or whatever strategy in a specific population of patients is more effective and/or more cost effective than the existing standard of care," said Putcha.
Prasad believes CMS should have asked Foundation to demonstrate clinical utility for its NGS panel in a randomized-controlled study, and in his Annals of Oncology commentary he even suggested two study designs that can be conducted in a reasonable time frame without incurring exorbitant costs.
He is also quick to point out that the only randomized controlled study investigating the benefits of NGS-guided cancer treatment, SHIVA, showed that patients matched to a molecularly targeted therapy did not experience improved median progression-free survival compared to the control group, but had more serious adverse events. While experts have criticized SHIVA as a flawed test of the precision oncology approach ― because patients in the study received single targeted drugs instead of combination treatments that would interrogate multiple molecular tumor drivers ― the fact is, precision oncology programs have found it challenging to match patients to targeted treatments.
Patients often have insufficient tumor material for evaluation and the markers tested by the NGS panel often occur in a rare subset of cancers. Even when patients match to a drug, the drugs may be inaccessible, patients may not fit the enrollment criteria for a study, and sometimes advanced cancer patients pass away before the test results or drugs are available.
A retrospective analysis published last year in JCO Precision Oncology of patients treated at MD Anderson Cancer Center found that only around 27 percent of around 1,400 molecularly profiled patients received a matched targeted treatment. Matched patients in this study had better outcomes (including longer overall survival) than not-matched patients, though West pointed out that the benefit seen in these studies tend to be largely due to patients who matched to a targeted drug that's already the standard of care.
"This is like saying that patients with a winning lottery ticket made more money than patients with a losing lottery ticket," he said. "You didn’t show that playing the lottery leads to better outcomes overall."
Meanwhile, within the NCI's MATCH trial, nearly 6,000 have been genomically screened on NGS tests, but around 1,000 had a genomic abnormality that matched to a targeted treatment arm, and under 700 patients eventually enrolled in one of those arms. Matching to a targeted treatment, however, doesn't mean patients will benefit from treatment, as suggested by SHIVA and another study, called MOSCATO ― in which around 200 out of 1,000 patients received targeted treatment based on a genomic alteration, and only 22 saw their tumor shrink.
West notes that too often, along with mutation results that have an on-label targeted therapy or one with compelling early clinical evidence, he has seen test reports that suggest treatment options based on a drug's mechanism of action or preclinical data alone.
In the absence of clinical utility evidence demonstrating that testing all advanced cancer patients for these other markers would benefit them, West said it's quite possible that patients could be harmed if NGS testing leads oncologists to prescribe targeted drugs with significant toxicities over more established standard treatments. Another possible outcome, he said, is that CMS' proposed coverage policy could increase costs without improving patient outcomes because testing will promote overuse of poorly selected drugs that can cost more than $100,000 per year of treatment.
"We don’t have irrefutable proof that universal NGS testing is harmful and wasteful," West said. "But we also don’t have evidence that testing beyond the limited number of established markers should be adopted as a standard. Rather than just have the FDA and the people with a career incentive to promote it say we should take it on faith that it will be better, we should recognize that it is absolutely plausible that it will not improve clinical outcomes in any meaningful way, but will only increase costs and/or lead to net harm for patients."
Unintended consequences
In its risk/benefit assessment for FoundationOne CDx, FDA looked at its analytical performance in gauging four classes of genomic alterations (as well as microsatellite instability and the tumor mutational burden score) in thousands of samples from multiple tumor types. For premarket approval of the CDx markers, Foundation had to compare its test's ability to gauge those genetic alterations against other FDA-approved companion diagnostics.
Bruce Quinn, a nationally recognized Medicare policy expert, noted that FDA's review documents for FoundationOne CDx points to an extensive evaluation of analytical validity, but according to the agency's review documents, it didn’t consider the clinical impact of the hundreds of non-CDx genes on the panel in its risk/benefit assessment. Meanwhile, in the draft NCD, CMS also doesn't say if testing for these other genes is clinically useful.
"They write at length about a literature review on clinical utility, but when they finish, all CMS concludes is that the test is covered because certain genetic markers have premarket approval because they are linked to [treatment-related] outcomes," Quinn said. "Like the FDA, they don't make any assertions or claims to the utility of the other hundreds of genes on the panel."
"The central question we as a community are trying to answer is: When compared to the standard of care, does tumor mutation profiling in a population of patients, for example, those wtih recurrent, metastatic, or advanced Stage IV cancer, improve health outcomes, patient safety, and/or quality of life," posited Putcha. "A study that tries to answer this question could certainly also help [us] understand the economic impact of tumor mutation profiling versus the standard of care on the total cost of care."
In a recent white paper, Quinn provided rough estimates of what the coverage proposal might cost the government payor. Assuming that between 80,000 and 130,000 advanced cancer patients will meet CMS' coverage criteria for NGS testing for one of the CDx markers, and assuming CMS uses CPT code 81455 (describing sequencing panels assessing 51 or more genes, priced at $2,920), the agency will pay between $230 million to $375 million.
Meanwhile, assuming between 100,000 and 150,000 advance cancer patients are eligible for coverage with evidence development if they have mutations in non-CDx markers ― although Quinn doubts that CED infrastructure could be scaled up to this level ― the maximum cost for CMS is in the range of $290 million and $435 million.
Labs seeking coverage with evidence development by submitting to a registry would likely also incur significant costs associated with collecting the data CMS is asking for. One of the most contentious parts of CMS' registry proposal, in Quinn's view, is the gulf between the $2,920 Medicare reimbursement for a NGS test versus the $10,000 annual RECIST imaging bill that could result from tracking a patient's response to treatments based on NGS results.
This could deter NGS testing labs from pursuing this CED pathway, especially if they already have Medicare coverage for CDx biomarker indications in key cancer populations. That would mean that patients getting off-label drugs based on other markers on the panel may not have their data collected as part of a registry to track if they benefited from the strategy.
This may also create significant economic uncertainty for patients. Oncologists that GenomeWeb spoke to said that insurers do initially balk at paying for off-label drugs based on NGS tests, but many times, after a discussion with the physician, they change their minds. But as Quinn pointed out in his white paper, if off-label prescribing increases dramatically as an unintended consequence of this NCD, Medicare contractors might take a harder stance on covering those treatments. "Will beneficiaries with advanced cancer and their families understand that CMS has covered the test, with its broad genomic reporting, but CMS may decline to cover the drugs that the gene test points to?" Quinn wondered.
The greater risk
ASCO's Schilsky acknowledged that when it comes to precision oncology the field is awash in positive anecdotes. "The anecdotes are the proof of concept that if you get the right target and the right drug remarkable things can happen," he said. "The problem is that we don't have the right target and right drug too frequently. And therefore the remarkable things happen to very few patients."
However, Schilsky sees little risk of runaway off-label prescribing as a result of the proposals in the CMS' NCD. The limited way in which CMS covers lab-developed tests without FDA clearance/approval and payor reluctance to cover off-label treatments, "should put a lid on it," he said.
And if some patients do get off-label treatments based on FoundationOne CDx, there is some reassurance in that the test has been vetted by the FDA. "People can argue whether going through the FDA guarantees a high-quality test, but that process involves a rigorous review and sets a high standard," Schilsky said. "Foundation Medicine put its test through that process and they have earned the right to promote their tests and have it be paid for."
What Schilsky finds most problematic with CMS' draft NCD, is the agency's coverage with evidence development proposal to only pay for non-FDA cleared/approved tests that are part of NCI-sponsored trials. ASCO has written to CMS recommending it also include commercially sponsored trials conducted under an investigational new drug (IND) application, as well as IND or IND-exempt trials conducted by not-for-profit research institutions and foundations.
"That would allow a much larger swath of clinical trials to participate in the CED, which would be a good thing," Schilsky said. Such an expansion would include a non-randomized basket study launched by ASCO, called TAPUR, which is investigating the safety and efficacy of commercially available cancer drugs given off label, based on results from NGS panels.
Last November, ASCO announced that TAPUR, which has enrolled 500 patients and involves 16 targeted therapies, had closed an arm with pancreatic cancer patients with CDKN2A loss or mutation after they failed to respond to palbociclib monotherapy but expanded a cohort of lung cancer patients with the same marker based on their response to the treatment.
Studies like this are going on at institutions around the country, and although many wouldn't qualify under CMS' draft NCD criteria, the learnings within them could further inform the broad utility of precision oncology. "There is a risk that off-label prescribing will continue and may grow under this determination," acknowledged Nadauld, who noted that Intermountain is considering "all relevant avenues" to ensure that the NGS cancer panel it performs internally is broadly covered.
"But the greater risk," he said, "is that we may not benefit from understanding the response that these patients have on these off-label therapies."