The use of next-generation sequencing to diagnose cancers and guide therapy decisions is expanding as more therapies are linked to genomic abnormalities.
David S. Klimstra, MD
The use of next-generation sequencing (NGS) to diagnose cancers and guide therapy decisions is expanding as more therapies are linked to genomic abnormalities. But advances in molecular testing face headwinds in the near term as clinicians struggle to manage the wealth of data produced by tests and payers balk at covering expensive assays backed by sparse evidence for improved outcomes.
The debate over the future of testing centers on broad-spectrum panels that evaluate tumor tissue for dozens or hundreds of genes at once, as opposed to standard in vitro diagnostics that test for a single aberration such as HER2, to predict response to trastuzumab (Herceptin) in breast cancer, and EGFR, to guide the use of tyrosine kinase inhibitors for lung cancer. Spurred by a 2013 US Supreme Court decision that genes are not patentable, along with the plunging cost of sequencing and the convenience of using 1 tissue sample to check for multiple abnormalities, private labs, academic hospitals, and cancer centers have developed panels that examine as many as 500 genes at once in the last 5 years.
The growing significance of NGS was illustrated in December 2017 by the first FDA approval of a large panel, Foundation Medicine’s FoundationOne CDx (F1CDx), which detects mutations in 324 genes and 2 genomic signatures in any tumor type. CMS conducted a parallel review that led to a proposal for limited Medicare coverage of the test.
Last year, the FDA also granted authorization for MSK-IMPACT, a 468-gene assay developed by Memorial Sloan Kettering Cancer Center (MSK) in New York, New York. The decision created a new pathway for regulatory analysis of large gene testing panels (FIGURE). The authorization could promote test adoption and cooperation with pharmaceutical companies, possibly payer coverage as well, while potentially paving the way for future approvals of other similar products.
However, the federal review process also underscores the hurdles that laboratories face as they seek broader acceptance of their products. Although testing companies and oncologists have applauded CMS for addressing the question of NGS coverage, many have also expressed dismay at the accompanying restrictions. Under the draft policy, CMS would only pay for tests that are FDA approved or cleared or are being used in National Cancer Institute-sponsored trials and would not cover retesting using the same test, among other limitations.1 A decision is due February 28, 2018.
The policy specifically mentions 5 tests that the FDA has approved so far (TABLE). Additionally, the draft identifies the patient population elgible for test coverage as those who have recurrent, metastatic, or advanced stage IV cancer.1
“What it would basically do is deny payment for the vast majority of assays currently being offered, unless they follow our lead and obtain FDA clearance. Even then, it’s a rather restrictive subset of cancer types that would be covered,” said David S. Klimstra, MD, chair of the Department of Pathology at MSK. He said he is also concerned that CMS’ proposed policy would essentially expand FDA authority over the assays without the usual congressional oversight. Such assays are currently considered laboratory-developed tests (LDTs) that do not need FDA approval.
“The draft policy is also only directed toward next-generation sequencing assays,” Klimstra noted. “CMS would continue to support existing payment structures for single-gene assays, which are less broad-spectrum, less tissueefficient, and less sensitive, but it would refuse to pay for the best assays out there. In a sense, that is a clear step backward.”The dominant technologies for molecular marker testing include immunohistochemistry, fluorescence in situ hybridization, and NGS methods. Use of all of these types of assays is increasing due to the expansion of targeted therapies, but the number of multigene NGS tests ordered annually appears to be rising particularly quickly. NGS assays represented 27% of molecular tests in the National Institutes of Health’s Genetic Testing Registry in February 2016, up from 10% 2 years earlier.2 Panels of at least 5 genes accounted for 16% of NGS tests. In addition, the number of oncology genetic tests of all types in the registry had jumped 153% over 12 months.
Multigene panels arose in response to the discovery of various molecular pathways that may be active in a tumor site. At Moffitt Cancer Center in Tampa, Florida, pathologists began developing larger panels when it became necessary to test for multiple abnormalities in non—smallcell lung cancer (NSCLC), including EGFR, KRAS, ALK, and ROS1, said Anthony M. Magliocco, MD, FRCPC, FCAP, chair of the Department of Anatomic Pathology at Moffitt.
“You have to test several different genes simultaneously to be able to determine the correct treatment for these patients. The test gets used very routinely for patients with advanced lung cancer.”
Thermo Fisher’s Oncomine Dx Target Test, which evaluates 23 genes associated with NSCLC and serves as a companion diagnostic for 1 combination regimen and 2 single-agent therapies, became the first NGS-based panel to win FDA approval last year. The company says it is working with more drugmakers to add indications for the test.
Labs are constantly assembling new panels or expanding existing ones. Moffitt has a 55-gene myeloid malignancy panel and a 170-gene solid tumor panel; the University of California San Francisco Medical Center (UCSF) has a 500-gene test, one of the larger panels in regular use. Nationally, tens of thousands of such tests are conducted every year, said Richard L. Schilsky, MD, senior vice president and chief medical officer of the American Society of Clinical Oncology (ASCO).
Broad-spectrum tests not only encompass the best-known mutations that are commonly evaluated through single-gene or small-panel tests, but they also allow for the detection of rare, actionable abnormalities. “BRAF V600E, which is a mutation common in melanoma, shows up in a couple percent of a lot of different kinds of cancer. You would never sequence every different cancer for BRAF, but it’s included in these large panels, so you get the information anyway and it allows you to catch those uncommon patients who could have a dramatic benefit. Those rare mutations would never be studied by individual gene assays,” Klimstra said. Findings from a recent study support the utility of broad-spectrum gene screening. The ProfiLER study in France sequenced 69 genes in patients with advanced cancer and found at least 1 clinically actionable alteration in 52% of samples tested.3
The 5-year overall survival (OS) rate was higher among 143 patients who received recommended treatments than it was among 533 who did not (34.8% vs 28.1%, respectively). The PI3K/ mTOR pathway contained the most common actionable mutations. Investigators are planning a follow-up trial to compare their test’s outcomes with those of a commercial 315-gene test.
Researchers are hoping for further insights from TAPUR, a nonrandomized, phase II basket trial sponsored by ASCO that provides patients with targeted drugs matched to the genomic profiles of their tumors. The study has enrolled 510 patients at 83 clinical sites in 20 states and provides 19 drugs for 16 therapy options. Results have not yet been released.
The advent of tumor-agnostic targeted therapies could help make large NGS-based assays more routine, at least for patients with advanced cancers or who otherwise lack satisfactory treatment options. Last year, the FDA approved pembrolizumab (Keytruda) for the treatment of patients with unresectable or metastatic microsatellite instability—high or mismatch repair deficient–solid tumors, marking the first time a drug gained an indication based on its efficacy against a molecular target rather than on the primary tumor site.
“Because the drug is now FDA approved to use in any tumor that has high microsatellite instability, and because many tumors have that but at relatively low frequency, almost every patient at some point will likely get their tumor tested to see if they could get that drug on label,” Schilsky said. “The more we see these kinds of things coming along, the more the nature of the testing is going to change. I’m sure we’ll begin to see testing for these specific alterations incorporated into the large-panel tests, because otherwise you just end up having to continue to test for multiple one-offs and that starts to get very inefficient after a period of time.”
Schilsky said he also expects the shift in testing practice to gain steam from the anticipated approval this year of larotrectinib (LOXO-101) for TRK fusion—positive cancers, which would make the drug the first targeted therapy developed in a tumor-agnostic manner. The oral drug has shown high response rates in both adult and pediatric patients with TRK fusions, with an objective response rate (ORR) of 76% (95% CI, 62%-87%) and consistent results across tumor types.4
Other TRK fusion drugs in development include Ignyta’s entrectinib, which has demonstrated efficacy against cancers with NTRK1/2/3, ROS1, and ALK rearrangements. In patients with ROS1-positive NSCLC, the ORR with entrectinib was 68.8%, which included 2 complete responses (6.3%).5
The FDA granted a breakthrough therapy designation last year for entrectinib as a treatment for adult and pediatric patients with TRK fusion—positive, locally advanced, or metastatic solid tumors who have either progressed following prior therapies or have no acceptable standard therapies.
Additionally, Loxo Oncology, the maker of larotrectinib, is developing LOXO-195 for patients with cancers that have acquired resistance to initial TRK therapy, which may emerge due to TRK kinase point mutations.
At a 2017 OncLive® State of the Science SummitTM on Gastrointestinal Cancers which was held in January, Luis Diaz, MD, head of the Division of Solid Tumor Oncology at MSK, said these new targeted therapies should demonstrate to community oncologists the importance of genetic testing.
“What I would like them to start realizing is that genomics is not just ‘nice’ to have. It is soon to become a ‘must have.’ Understanding some of the underlying principles, such as the advantages and disadvantages, will be important for the community oncologist,” Diaz said.A major obstacle to wider adoption of large-panel gene tests is their cost, which can be 10 times higher than for a single-gene assay. Magliocco said a commercial test of 1 gene costs $300 to $400 while big panels come in at $3000 to $6000. FoundationOne has a $5800 list price and FoundationOne Heme costs $7200, although actual payments are often smaller, according to the company. The 500-gene tumor-profiling panel at UCSF runs about $4000, according to Eric Collisson, MD, an associate professor at UCSF School of Medicine.
The costs are covered through various means. Schilsky said some labs cap payments or agree not to bill patients. At MSK, philanthrophic support pays for MSK-IMPACT testing, Klimstra said. In general, Medicare and private payers sometimes pay for smaller panels but do not cover the broad-spectrum tests, with some exceptions. Palmetto GBA, a regional Medicare contractor based in South Carolina, last year began approving payments for FoundationOne testing for patients with advanced NSCLC. Magliocco said Moffitt can sometimes obtain partial payment by focusing on genes with billing codes.
“There are frequent denials on the larger panels. Payers view them as being research rather than diagnostic assays—even though they’re clearly diagnostic—so we certainly have challenges there. For some tumors, we can break down the billing for the specific genes. We may break it out and bill for BRAF, EGFR, KRAS and so on separately, and those tend to be reimbursed. But it’s very challenging,” Magliocco said.
Schilsky said labs have rationales for including many nonactionable genes in their panels, based on data showing they are important in molecular pathways known to be activated in cancer cells. However, much of the resulting data is not useful to most physicians. “The broader that the panel goes, the more and more information you get that you frankly don’t know what to do with. That sets up this whole conundrum that is facing oncology these days. How do I interpret this information? How do I act on it? How do I get a therapy that potentially benefits my patient based on these findings? This potentially leads to lots of off-label prescribing without really, in many cases, a sound basis, and no mechanism by which to collect data as to whether patients benefit from this approach,” he said.
Off-label prescribing based on the molecular profiling of tumors has not been shown to improve outcomes. In the phase II SHIVA trial conducted in France, patients with metastatic solid tumors refractory to standard therapies were randomly assigned either to molecularly targeted therapies or treatment of physician’s choice. After a median followup of 11.3 months, median progression-free survival was similar in the experimental and control groups (2.3 vs 2 months, respectively).6 Grade 3/4 toxicities were more common among those who received targeted therapies (43% vs 35%). Findings from an updated analysis of the SHIVA population showed that patients with targetable mutations who were randomized had better OS outcomes than those with no druggable mutation, but the benefit did not reach statistical significance (HR, 0.85).7
Against concerns about the cost and utility of broad-spectrum NGS tests, proponents argue that they will inevitably be used much more frequently because of their greater efficiency and comprehensiveness. The true cost difference between differently sized panels is not significant enough to limit access to the larger ones, they argue. “We know that we never need to know 300 genes. That’s absolutely a true statement,” Collisson said. “But at the same time, that’s not really the issue. The issue is, you’re probably going to need to know more than 2, and once you get into low doubledigits, it’s way more economical to just do 300 and not have to guess which the right 200 are.”
Collisson also argued that it does not make sense to suppress easily available gene data because it is not prospectively actionable, as some labs do. “We don’t hold anything else in medicine to that criterion. When I order a chemistry panel, I get 12 things back that I mostly don’t want, but that’s because they all come together. We just have to get past this ‘I’m going to pay $1000 but not $4000.’ They should just argue about price rather than what the content is, and the market will probably figure out the sweetest spot of serving the most people for the least amount of money.”
Klimstra and others said large-panel testing is also an important tool for recruiting patients for clinical trials. MSK has a system that tracks genomic abnormalities as they are detected and notifies oncologists who are studying relevant novel therapies so they can enroll the patients in the trial, he said.
Schilsky said the use of NGS testing in oncology will continue to grow as the technology continues to advance and costs keep falling. At the same time, he said oncologists and other participants in the cancer ecosystem have a responsibility to make progress in key areas in order improve testing.
More basic research is required to understand the functional consequences of the abundance of genetic alterations that assays uncover, he said. Doctors need better test reporting and clinical decision support tools so they can quickly determine whether and how to act on test data.
In addition, precision medicine must proceed through a learning-systems approach that captures test data, interventions, and outcomes to help refine testing and ensure that patients benefit, Schilsky said. A number of data-collection and data-sharing efforts are under way, including programs sponsored by ASCO, the National Cancer Institute, and the American Association for Cancer Research.