You’ve probably heard about powerful new cancer medicines like Keytruda and Opdivo. As advertised on TV, these drugs release brakes on the immune system to make tumors disappear and extend survival in deadly diseases like lung cancer and melanoma. But these agents, called checkpoint inhibitors, work in only a fraction of patients. Scientists are searching for diagnostic tests to predict who will be helped, and who won't.
A promising solution comes from Foundation Medicine, a molecular diagnostics company headquartered in Cambridge, Mass. Foundation offers a cancer genome profiling test that evaluates mutations in DNA. With results from its standard FoundationOne panel, Foundation calculates a Tumor Mutation Burden (TMB) score. This quantitative readout―based on the number and type of DNA changes per megabase, a length of DNA―may prove helpful to patients considering immune treatment for many advanced forms of cancer.
Published reports suggest that Foundation’s TMB score can predict responses to immune cancer drugs in bladder cancer, lung cancer and melanoma. Other, preliminary analyses for other tumor types, but without survival or treatment response data, have been presented at major medical conferences. Daniel Lieber, a computational biologist who works at Foundation, presented an update on the TMB assay at last week’s American Association for Cancer Research (AACR) in Washington, D.C.
Lieber presented a summary of Foundation’s TMB results in over 88,150 clinical samples. As expected, a relatively high proportion of advanced skin, lung, bladder and a few other cancers demonstrated high TMB. What’s becoming evident is that for some malignancies―including brain, breast, colon and endometrial cancers―for which immune treatments are usually not considered, a small fraction of cases turn out to have high TMB. It is possible that TMB predicts if cancer―of any type―will respond to treatment with immune drugs, but this remains to be tested.
Just last month, Bristol-Myers Squibb (BMS) and Foundation announced a partnership to examine biomarkers including TMB―using Foundation Medicine’s methods for evaluating molecular pathology―in tumor specimens taken from patients receiving any of over a dozen investigational or approved immunotherapy drugs, in 35 types of cancer. Relatedly, the ASCO-orchestrated TAPUR trial is enrolling patients with all types of advanced cancer. In that basket-type study, those with “high mutation load” as determined by molecular profiling in a CLIA and CAP-certified laboratory, may be assigned to receive pembrolizumab (Merck’s Keytruda).
I spoke with Foundation’s chief medical officer, Dr. Vincent Miller. “TMB is a very promising predictive biomarker,” he told me. “We’ve already seen data supporting TMB as a predictor in bladder cancer, melanoma, non-small cell lung cancer and in the subset of colon cancers that are MSI-high.” (MSI refers to microsatellite instability). “In those tumor types, patients with high TMB do particularly well. And in general terms, patients with low TMB don’t necessarily do better with immune checkpoint inhibitors than they would with cytotoxic chemotherapy.”
The FoundationOne test was first developed to evaluate DNA in cancer samples, such as biopsies of solid tumors. The panel lists for $5,800; it currently provides detailed results for 315 genes. If a doctor orders PD-L1 testing, some material from the same sample is set on slides for protein analysis (by IHC). The lab is CLIA- and CAP-approved. The FoundationOne test has been approved for use by the New York State Department of Health. You can’t order the Foundation’s TMB score apart from FoundationOne, Miller clarified. TMB is "a push-the-button calculation.” But that’s only after sequencing results have been obtained using FoundationOne. This is a validated platform, he emphasized
Last year, the FDA and CMS agreed to evaluate the FoundationOne test by parallel review. The FDA also accepted the test for evaluation by its Expedited Access Pathway. “We anticipate approval later this year, in 2017,” Miller told me. (His statement is consistent with a Foundation press release, here.)
Roche holds a majority stake in Foundation Medicine. Yet Foundation functions autonomously, Miller told me. “We’re not constrained by that relationship. In fact, the converse is true. We’re working with so many pharmaceutical companies.” Indeed, the company boasts an astonishing number of collaborative projects with clinical and academic oncologists, drug manufacturers and other industry. Late last year, Foundation entered a partnership with Flatiron Health, an oncology IT company that holds anonymized cancer information from 20,000 patients, to examine how its molecular biomarkers relate to clinical outcomes. At the AACR meeting, Phil Stephens, of Foundation, spoke about this “marrying of clinical and genomic data” as it might apply to the expanding potential of immune cancer treatments.
Immune cancer medicines are a huge deal. Yes, they cost too much and can be toxic. But these drugs are nothing to scoff at. Checkpoint inhibitors help some patients to a degree that oncologists could not have fathomed until recently. Consider the case of Jimmy Carter, who received treatment with Keytruda for metastatic melanoma, tolerated the drug and entered a durable remission at the age of 91 years. Since 2010, when the FDA approved the first in this class of drugs, Yervoy—followed by Opdivo, Keytruda, Tecentriq and the latest, Bavencio—patients and oncologists have had a widening range of treatments to consider.
The problem is that most patients who try these expensive drugs don’t benefit, and there’s no reliable way of predicting who will respond. Days ago, England’s NHS rejected nivolumab (Opdivo) due to high costs. As I have considered previously, accurate companion diagnostic tests could reduce the costs of precision oncology, because patients with matching tumors might selectively choose those drugs. So far, however, there is no reliable test for response to checkpoint inhibitors. Tests for PD-L1 are imperfect, at best.
"That would be amazing, to know that the drugs you are taking will work for you," said April Kennedy Knowles, a 41-year-old woman with metastatic breast cancer who lives in Pensacola, Fla. Knowles is a leader of METUP, an organization that advocates for research to cure metastatic disease.
Recently, I wrote about an ongoing trial of atezolizumab (Tecentriq) in which only 10% of patients with metastatic breast cancer responded to the checkpoint inhibitor. But for those women who did respond, the results appear to be durable.
“This is an example of when you would want to have a biomarker,” said Dr. Suzanne Topalian. She moderated the AACR press session when those findings were discussed. Topalian is an associate director of the Bloomberg-Kimmel Institute for Cancer Immunotherapy and a professor of surgery at the Johns Hopkins School of Medicine in Baltimore. In her lab, she studies how checkpoint inhibitors and other immune treatments work in melanoma and other cancers. She receives research grant support from Bristol-Myers Squibb.
“The breast cancer study is promising,” Topalian told me in an interview. Most of the women in that study, all with triple-negative breast cancer, had received multiple prior therapies, she considered. They have limited treatment options, and a very poor prognosis. “For a small group of those patients, there appears to be a survival benefit, and it’s long-lasting.”
“It makes sense that that a higher mutational burden could be associated with a greater likelihood of responding to an immune agent,” she considered. That’s because genetic changes in the cancers can generate alterations in proteins, and those changes in proteins might elicit an immune reaction.
Foundation Medicine is not the only organization that is measuring the tumor mutation burden, Topalian noted. “At academic medical centers we do look at TMB, usually by whole-exome sequencing,” she said. “But it’s on a research basis.” She referred to separate findings, an exploratory analysis of the Checkpoint 026 study of anti-PD-1 therapy in advanced lung cancer, presented by Dr. Solange Peters and colleagues, also at the AACR meeting. “From that study of patients with lung cancer, it appears that high TMB predicts for response to nivolumab,” she said. But we don’t yet know if this will be true for breast cancer, or if the Foundation Medicine method for assessing TMB will be equally predictive.
Topalian expressed caution about the use of any biomarker to evaluate if patients will respond to immune-modulating cancer drugs. “We don’t want to overlook any patients who might benefit from these treatments,” she said. “The predictiveness of these biomarkers is not strict. It’s not like checking for a BRAF mutation in melanoma,” she said. “The tests we are talking about for immune checkpoint inhibitors, like PD-L1 assays and perhaps TMB, are not black-and-white. That’s why we have to be careful.”
At the AACR meeting, Lieber considered that the relationship between TMB and immune responses to cancer is not straightforward: A crude measure of DNA mutations in a cancer will not predict a patient's response or survival. That’s in part because some, but not all, genetic changes cause protein variants to pop up in a tumor. Another confounder in the TMB equation is that a “driver” mutation in cancer may be detected out of proportion to its immune-stimulating potential.
Lieber explained how the TMB score is calculated, in terms of mutations per megabase of DNA. The score can be stratified as high, intermediate or low. A panelist asked how TMB compares with other biomarkers. He said that that there’s overlap, and possibly an additive effect; these findings might complement information about PD-L1 expression or MSI (microsatellite instability). A subset of cancers with high TMB reveal high MSI; Foundation is exploring the interaction between those, particularly as it pertains to colorectal cancers.
“There’s not one global TMB score for each indication,” Lieber said. In melanoma, 40% of patients with intermediate TMB scores did well, he mentioned. The predictive value and thresholds for sensitivity may vary across tumors.
Previously published studies based on Foundation Medicine’s TMB score support its use in several cancer types. A 2015 paper from investigators at the Universidade de São Paulo, Brazil, found that gauging TMB using Foundation’s panel of 315 genes compared well with whole-genome sequencing of lung cancer, and predicted patients’ responses to treatment with PD-1-blocking drugs. Although the results for 31 patients (based on supplementary data, table S3) were good overall, they were imperfect: two patients with high or intermediate scores did not respond, and a few patients with low scores did benefit. Overall, 69% of lung cancer patients with high TMB scores, and only 20% of those with low TMB scores, experienced durable remissions.
Mutational load, assessed by Foundation Medicine TMB, is significantly associated with clinical... [+]
In 2015, the journal Science reported that mutational landscape determines sensitivity to PD-1 blockade in non-small cell lung cancer. In that study, investigators based at Memorial Sloan Kettering Cancer Center and collaborators carried out whole-exome sequencing of lung cancers in patients who received pembrolizumab (Keytruda), an antibody that targets PD-1. In that study, a high tumor mutation load predicted good responses to Keytruda. Of interest, the immune drug’s power correlated with a molecular smoking signature―a pattern of DNA mutations seen in lung cancers of people who smoked cigarettes. Keytruda responsiveness also correlated with the number of neoantigens―protein variants that pop up in tumors due to genetic changes that can elicit an immune response and sensitivity to checkpoint inhibitor drugs.
Dr. Louis Weiner moderated a press session at the AACR meeting which included several papers on precision oncology. He is a medical oncologist and researcher who leads the Georgetown Lombardi Cancer Center in Washington, D.C. and directs the regional MedStar Cancer Network. MedStar has a relationship with Caris Life Sciences for molecular testing of cancer specimens. Weiner has worked extensively in cancer immunology and precision medicine. He is the founder of Jounce Therapeutics, advises Celldex Pharmaceuticals and has been a consultant to Genentech.
I asked Weiner about the potential value of biomarkers such as Foundation Medicine’s tumor mutation assay. He responded with a story about a 49-year-old woman with a rare neuroendocrine tumor that was growing quickly. It appeared to have begun in her mouth, and spread to her liver and lungs. “There’s no playbook for how to treat that,” he said. “Ten years ago, I probably would have recommended a standard chemotherapy cocktail.”
Weiner sent the patient’s tumor sample to Caris Life Sciences (Dallas, Texas; lab in Phoenix, Ariz.) for molecular profiling. Caris looked at around 600 genes and also looked at some proteins in the tumor, he told me. “From Caris we learned that she had a high mutational load, and high levels of PD-L1,” he said. “Armed with that information, I recommended that the patient be treated with an immune checkpoint antibody.” She lives far away, and didn’t want to be on a clinical trial. “Because we had all that molecular information, we were able to get her insurance company to pay for it,” he said. “She did great and, so far, has continued to do well.”
“This is a woman with a rare cancer who if she presented with the same problems 10 years from ago, she’d be dead,” Weiner said. “So if you ask me if molecular profiling can be a useful adjunct in determining who should receive immune checkpoint antibody therapy, the answer is an emphatic yes.”
One thing I like about Foundation is its extreme focus on accuracy. When a sample is received in the Foundation Medicine lab, a pathologist reviews the specimen before DNA is extracted. The point is to assure that the cancer type is indeed what the submitting physician says it is; to check, for instance, that a sample labeled as neuroendocrine tumor is a neuroendocrine tumor.
Here is a video from Foundation Medicine (the first in a series) explaining how samples are processed in their lab.
Last June at the annual ASCO meeting in Chicago, I spoke at length with Foundation’s medical director, Dr. Jeffrey Ross, and its chief medical officer, Dr. Vincent Miller. Ross is a pathologist, and Miller is an oncologist. They are both are well-published physician-scientists with “real-world” clinical experience. Ross, a professor at Albany Medical College, chairs his department, runs a research lab and edits the Molecular Pathology section of the American Journal of Clinical Pathology. His previous work in industry includes having led Millennium Pharmaceuticals’ program in molecular pathology. Miller practiced oncology and performed research at Memorial Sloan Kettering Cancer Center in New York City (where we overlapped, briefly, during some of my rotations as a resident and fellow). If you spend over an hour chatting with this pair about precision medicine and cancer, as I did, it’s hard not to be impressed. They are deeply into medical science.
