NEW YORK (GenomeWeb) – A year after launching its cancer sequencing test for patients with solid tumors, MSK-IMPACT, investigators at Memorial Sloan Kettering Cancer Center are beginning to assess the test's clinical utility and recently started to report inherited cancer variants.
Launched last summer after being fully approved by the New York State Department of Health, MSK-IMPACT, which stands for Memorial Sloan Kettering-Integrated Mutation Profiling of Actionable Cancer Targets, is available to patients treated at MSKCC, primarily those with advanced metastatic disease. The hope is that test results will help identify patients eligible for clinical trials of targeted therapies.
Following an update earlier this year, the test now encompasses 410 cancer genes, including selected introns for some, as well as certain other non-coding regions. Later this year, the panel might be further expanded, according to MSKCC. All patient testing is performed by Memorial's Molecular Diagnostics Service in the Department of Pathology, and turnaround time has been less than three weeks for most cases.
The test relies on hybridization-based target capture followed by next-generation sequencing on an Illumina HiSeq 2500 in fast mode. It detects several classes of DNA alterations, including point mutations, small insertions and deletions, copy number variations, and structural rearrangements.
Unlike many other cancer sequencing tests, such as Foundation Medicine's FoundationOne assay, MSK-IMPACT involves sequencing of matched tumor and normal samples, which helps separate true somatic mutations from germline alterations and has recently allowed the center to return germline results. "Logistically, there are many challenges in the sequencing of normal [samples], [and] it does cost more, but we think the value exceeds the costs," said Michael Berger, associate director of the Marie-Josée and Henry R. Kravis Center for Molecular Oncology, which co-developed the test with the Department of Pathology.
This spring, Memorial Sloan-Kettering researchers published their validation study of an earlier version of MSK-IMPACT, which comprised 341 cancer genes, in the Journal of Molecular Diagnostics. That study showed that the test has a 2 percent detection limit for hotspot mutations and a 5 percent limit for other mutations.
So far, MSK-IMPACT results have been reported for more than 4,500 patients. Initially, samples from tumor types where actionable mutations are prevalent were prioritized, but more recently, the test has been offered for all types of solid tumors. A "very large percentage" of patients with advanced metastatic tumors at Memorial now receive the test, Berger said. So far, about 15 percent of samples have failed for technical reasons, mostly because the tissue blocks contained insufficient tumor material or because they did not produce enough DNA.
Much of the cost of testing is currently shouldered by Memorial Sloan Kettering through institutional and philanthropic funds, and insurance carriers are only billed for a small number of tumor types where molecular profiling is standard of care or where clinical guidelines recommend such testing.
"Ultimately, it's not sustainable indefinitely as we continue to expand, given the cost of the sequencing," Berger said, "but we would never expect full reimbursement until we generate the data demonstrating the clinical utility" of the test, which is why the center funds its use in "non-billable" tumor types, such as breast cancer and prostate cancer. He declined to disclose the internal cost of testing, which he said has been "a moving target."
Berger and his colleagues have recently begun gathering data on the clinical value of testing, looking, for example, at how many patients enrolled in molecularly targeted clinical trials after receiving the test.
MSK-IMPACT might be particularly useful for identifying candidates for so-called basket trials, which enroll patients based on actionable mutations rather than tumor type. As of the end of 2014, 200 patients were enrolled in basket studies at Sloan Kettering, though not all based on MSK-IMPACT testing.
Physicians who ordered the test also receive a survey six months later, asking them whether it revealed any clinically relevant mutations, and if so, whether they acted upon them. This will also allow the researchers to assess how far doctors agree on the clinical relevance of reported alterations.
While no firm results from the clinical utility studies are available yet, "certainly anecdotally, there are many cases where we have detected alterations, often unexpected alterations, in patients' tumors, and the treatment decisions were influenced [by that] and led to positive outcomes for patients," Berger said.
An anonymized analysis of the first 1,600 or so test results also yielded intriguing results about heritable cancer mutations: almost 7 percent of patients harbored pathogenic germline variants in one of 26 genes that are both part of MSK-IMPACT and of the American College of Medical Genetics and Genomics' recommended gene list for secondary findings.
While some of these patients may have had a known family history of cancer, many probably did not, Berger said, because they came from one of the oncology clinics and not the clinical genetics service. "One of the interesting and somewhat surprising findings was that for about half of the patients where we found a pathogenic [germline] alteration, it was in the context of a tumor type not normally associated with that variant," he said. For example, half of the germline variants in BRCA1 and BRCA2 were found in patients with tumor types other than breast, ovarian, prostate, or pancreatic cancer.
Based on these findings, Memorial Sloan Kettering decided to start reporting not only somatic tumor mutations but also germline results from MSK-IMPACT, and in March it obtained approval from New York State for doing so. No additional sequencing is required for the expanded test because it already analyzes normal controls along with the tumors.
Last month, the first patients started obtaining germline results, initially patients with breast, prostate, or gynecological cancers. Receiving germline variants is optional, and requires an additional consent. "Ultimately, it's up to the oncologist to decide whether to offer this to the patient, and up to the patient whether they want to opt into this analysis," Berger said. Patients who had the MSK-IMPACT test in the past can also sign the consent form and have their data reanalyzed for germline variants. Germline testing will be ramped up further later this year, he said.
Overall, MSK-IMPACT has been well received by doctors, and every oncologist who sees patients with solid tumors has ordered the test for at least some of them. In clinical meetings, "the MSK-IMPACT data is just one more aspect of the patient's clinical history to be considered, along with pathology and radiology and other clinical information," Berger said. "And because it's being used to match patients to clinical trials, for the [principal investigators] who run the trials, it's a very important part of what they are doing to find the right patients."
One way he and his colleagues have improved patient matching to clinical trials is an automated alert system that makes PIs aware of new patients with specific DNA alterations. "Just providing the information back to the primary treating oncologist may not be sufficient because they are unlikely to be aware of all the clinical trials that are available throughout the institution," he said, "but by making the PI of a specific trial aware of a new patient who has an alteration in a gene that's relevant for their study, they can then proactively reach out to the primary treating oncologist and suggest that they send their patient for this clinical trial."
At least some patients are aware of MSK-IMPACT, too, inquiring about it on MSKCC's blog or asking about it when visiting their oncologist.
One of the challenges with the test has been reporting the results in a manner that is easy to understand by all recipients, including the treating oncologist, the patient, and clinical researchers. "I think there is still room for improvement there," Berger said.
For example, in a list of mutations, it is important to see which ones are drivers and passengers, targetable and non-targetable. "We are in the process of expanding and enhancing our reports to make the information clearer, but we're also making use of the cBioPortal for Cancer Genomics," he said.
cBioPortal, an MSKCC-developed resource, includes both internal and publicly available de-identified cancer sequencing datasets and serves as a "tool for clinicians to interpret the significance of the alterations and the clinical opportunities that arise," he said. "As we go forward, knowing about whether patients with [a particular] mutation have had better or worse outcomes, or have responded favorably to particular therapies, is especially important information, and whether clinical trials are available for patients with specific alterations."
cBioPortal, which is only available to MSKCC investigators in its full version, can also be used to determine whether certain mutations occur frequently enough in patients to justify opening up new clinical trials, he said, or to assess how many patients could potentially participate in a clinical trial a pharmaceutical company might plan to open up.
A related internal database, OncoKB, provides annotations for oncogenomic mutations that have been curated by the center's disease teams.
Reporting and interpreting the test results will remain a focus going forward, Berger said. "We'll learn as we collect the data from the surveys and retrospectively review the cases whether important alterations ever were missed," he said. "I certainly hope they were not, but I would not be surprised if in some cases they were. That's something we are still continually trying to improve."
Over the coming months, Berger and his colleagues plan to publish on their clinical experience with MSK-IMPACT, and various disease teams at Memorial Sloan Kettering are preparing their own reports on specific tumor types.