NEW YORK – The NCI's Molecular Analysis for Therapy Choice (NCI-MATCH) trial is winding down, but researchers are still analyzing data from its many arms and taking notes for the design of future precision oncology trials based on what this pioneering initiative got right and the challenges it faced.
Launched in 2015, the multi-cohort trial is one of the first to nationally enroll patients with advanced cancers based on the molecular profile, not the site, of their tumors. Over the six years this trial has been running, drugmakers have embraced tissue-agnostic multi-cohort studies as part of their development strategies — and the US Food and Drug Administration has approved multiple tissue-agnostic biomarker-guided treatments.
Based on knowledge gained from NCI-MATCH, researchers are moving toward exploring biomarker-driven combination trials as the next logical step in developing master protocols exploring precision oncology treatment strategies.
Even as the study winds down, NCI-MATCH is evolving based on lessons learned. The ECOG-ACRIN Cancer Research Group, which is co-leading the study with the NCI, said last month that it is expanding two treatment arms based on emerging data.
Arm Z1M in NCI-MATCH was initially studying Bristol Myers Squibb's Opdivo (nivolumab) in patients with mismatch repair deficient (MMRd) solid tumors and LAG-3 expression who had progressed on anti-PD-1/PD-L1 treatments.
But based on data from other studies showing that not all patients respond to single-agent checkpoint inhibitors, and that LAG-3 inhibition may restore checkpoint inhibitor sensitivity in patients who have progressed on anti-PD-1/PD-L1 therapy, Z1M will be amended to add BMS's LAG-3 blocking antibody relatlimab to Opdivo. Patients must still have MMRd and LAG-3-expressing tumors to qualify for the trial. Melanoma patients will not be able to enroll in this arm, however, because the relatlimab-Opdivo combination, packaged in a new offering called Opdualag, was FDA-approved last month for patients with unresectable or metastatic melanoma.
In yet another NCI-MATCH cohort, arm H, which demonstrated the activity of Novartis' Tafinlar (dabrafenib) and Mekinist (trametinib) in patients with BRAF mutations, researchers are now seeking additional patients. Between the two expanded arms, the trial hopes to enroll 85 additional patients.
Greater than the sum of its parts
Arms Z1M and H are the last two enrolling trials in NCI-MATCH, after which it will close. Since it launched as part of Barack Obama's $215 million Precision Medicine Initiative with just 10 biomarker-defined treatment arms, more than 1,200 advanced cancer patients have received treatment within 39 arms. Researchers have published or presented the findings from 17 arms, and many more publications are underway.
The study is designed to enroll 35 patients in each arm. In order to be eligible to receive a targeted therapy based on a genetic aberration in their tumors, patients must be unresponsive to standard treatment. A positive signal in an arm means that five patients out of 31, or 16 percent, have had an objective response to the treatment, indicating that the agent is worthy of further study.
In the six years since it was launched, NCI-MATCH has racked up an impressive list of wins. Significantly, eight of the 39 arms met the primary endpoint.
And while a greater number of arms did not meet that endpoint, there were responses in nearly every arm, according to Peter O'Dwyer, a medical oncologist with Penn Medicine and co-chair of NCI-MATCH. "Not all of them get to that five out of 31 criterion, but we have responses," said O'Dwyer.
In fact, O'Dwyer said, the success or failure of each arm was never the point of NCI-MATCH. "How come there weren't more home runs in this study? That wasn't the design," he said. "The study didn't have as its underlying hypothesis, 'We're going to find a whole bunch of home runs.'"
Some study arms that seemingly turned up negative results have nonetheless yielded intriguing information. For example, study arm S2, focusing on tumors with GNAQ or GNA11 mutations only enrolled four participants. That's because, O'Dwyer said, GNAQ mutations tend to occur in ocular melanoma, but it turned out to be an extremely uncommon mutation outside of that cancer.
Still, two out of the four patients responded to Mekinist. The drug-biomarker strategy "was probably effective and didn't meet criteria for a positive study," said O'Dwyer.
Genomic sequencing laboratories at City of Hope and the Molecular Characterization Laboratory (MoCha) within NCI's Frederick Laboratory for Cancer are analyzing activated pathways and co-mutations in study specimens to understand the biology of response and resistance.
The experience with study arm S2 reflected one of NCI-MATCH's major early challenges — patient accrual. While there was great enthusiasm for the trial when it opened in 2016, the number of matches following genomic sequencing was disappointingly low, and many arms struggled to accrue the 35-patient goal.
Although enrollment was affected by early issues that needed to be addressed related to standardizing sample collection and other logistics of genomically profiling a large number of patients, O'Dwyer noted another factor was that certain gene variants researchers chose to include in the study based on their frequency in The Cancer Genome Atlas turned out to be far rarer among study participants. One possible reason O'Dwyer noted is that TCGA was built from analyzing primary tumors, but the patients enrolling in NCI-MATCH had metastatic disease, in whom some of these genomic alterations may show up in varying rates.
These details, unknowable in the beginning, will contribute to improvements in subsequent trial design, and elucidate the biological underpinnings of cancer.
An example of a biological insight from the trial was learning how BRAF inhibitors would perform in a tissue-agnostic setting. Nearly half of all melanomas harbor a mutation in BRAF at codon 600, which constitutively activates the MAPK pathway. Targeting that pathway with BRAF or MEK inhibitors has shown significant clinical benefit in melanoma patients. Disappointingly, however, BRAF- and MEK-targeted therapies had limited activity in colorectal cancers harboring BRAF mutations.
Arm H of NCI-MATCH studied the efficacy of BRAF/MEK inhibitors Tafinlar and Mekinist in cancers with a BRAF mutation that have progressed on standard therapy. "The question was, are most diseases like melanoma, where you're going to get a great effect, or are they like colon cancer, where you may only get an occasional hit from applying a targeted therapy," said O'Dwyer.
The data from arm H showed a more melanoma-like response pattern for most tumors. Data from arm H, combined with data from Novartis on Tafinlar-Mekinist in different tumors, may support tissue-agnostic approval for the combination, according to O'Dwyer. The agency approved this combo in 2018 for melanoma patients with BRAF v600E or V600K mutations.
'Combination trials are the future'
An equally important legacy of NCI-MATCH will be the knowledge base and infrastructure built by the trial. Unprecedented in size and scope, NCI-MATCH required the coordination of hundreds of researchers at over 1,100 clinical sites. In order to screen and enroll patients, techniques for sample handling and sequencing had to be established and standardized for patients at all trial sites.
Initially, four laboratories — MD Anderson Cancer Center, Massachusetts General Hospital, Yale University, and NCI Frederick Laboratory — sequenced patients' samples using an assay developed for the trial. Within the study, it was extremely important to ensure that the labs testing patients and matching to a study arm were performing comparably.
Over time the lab network expanded to include 40 labs that were screening patients. NCI-MATCH ensured the proficiency of labs in its network to detect genomic alterations by having them test tumors with known mutational profiles. Only labs that returned results consistent with the known mutational profiles of the tumors were allowed to screen patients for the study.
O'Dwyer said that the processes guiding NCI-MATCH were created by close to 20 committees that tackled various challenges, including design and implementation of the study, logistics, and patient advocacy, among others.
The knowledge base created by the organization will inform future precision oncology research. Among other things, NCI-MATCH demonstrated that a network of laboratories could be used to identify patients with target genomic alterations for a clinical trial; developed central assay and analytic platforms; and raised the profile of precision oncology, generating interest among oncologists. Most importantly, it showed the value of multi-cohort platform trials (also called umbrella or basket trials) for testing targeted therapies for tumors with actionable genomic alterations.
NCI-MATCH investigators led by Keith Flaherty from Massachusetts General Hospital reported in 2020 that 38 percent of 5,540 patients screened had an actionable marker as defined by the study, 26 percent of patients were eligible for enrollment in a study arm, and about 18 percent were assigned to a study arm.
Razelle Kurzrock, associate director for clinical research at the Medical College of Wisconsin Cancer Center, said that lessons learned from NCI-MATCH will carry forward to future studies in precision oncology. That especially applies to the relatively low number of study arms meeting their endpoint. "The important lesson learned is that targeting one alteration at a time is suboptimal in tumors that have multiple alterations that may be co-drivers, and that you have to target some of those co-drivers."
NCI's Lyndsay Harris concurred. "One observation from the published results of a limited number (11 of 39) of treatments arms so far is that single-agent targeted therapy may not be the optimal treatment approach for specific genomic alterations in advanced solid tumors and that other treatment strategies, for example, combination therapy, might be necessary to overcome the resistance to therapy in heavily pretreated patients," Harris said over email. "It also demonstrates the frequency of co-mutations in heavily pretreated advanced cancer patients that might be at least partly responsible for resistance to therapy."
The I-PREDICT trial headed by Kurzrock is an early prototype for a platform study that tested out combination therapy approaches. In that study, 73 patients with previously treated, refractory metastatic cancers were matched with one or more personalized, precision therapies after molecular profiling. That was out of 149 initially enrolled patients, yielding a match rate of 49 percent, which is much higher than the match rate achieved by other similarly designed trials.
Moreover, the higher the patients' match score — indicating that they received combination therapies targeting more of the molecular aberrations driving their tumors — the better the patients did. Three-fourths of patients with a match score above 50 percent had significantly improved progression-free survival compared to their last therapy, while 36.6 percent of patients with a match score below 50 percent saw a similar level of benefit.
"Combination trials are the future," said Kurzrock. "If you look at the genomic portfolios of these complicated tumors, they have more than one driver, and it just makes sense that a combination may do much better than a single agent targeting the most important drivers. And that's certainly what we found on a smaller scale in the I-PREDICT studies."
ECOG-ACRIN's follow-up trial, NCI-ComboMATCH, will pick up where NCI-MATCH left off, exploring targeted drug combinations to overcome drug resistance to single-agent therapy. ComboMATCH will test combinations of targeted drugs supported by in vivo preclinical evidence, particularly patient-derived xenograft and cell-line-derived xenograft data. The study will include both tumor-specific and tumor-agnostic studies.
Kurzrock noted that there have been six instances when the FDA has approved drugs for multiple cancer types based on a common biomarker, starting with Merck's Keytruda (pembrolizumab) in 2017. "This was kind of unimaginable, if you go back a decade, that the FDA would approve drugs for all solid tumors based on genomic alterations."
She also pointed out that NCI-MATCH's discovery that every gene abnormality being studied was less common than expected in no way invalidated the overall premise. "Just because something is rare and difficult to accrue, doesn't mean it's not important," said Kurzrock, who is principal investigator of TRACK, a platform trial that is studying how patients with rare tumors respond to molecularly targeted treatments. "When you start to add up all the rare alterations, having an alteration actually becomes pretty common."
According to Kurzrock, in the future, many more tumors will have therapeutically actionable alterations due to advances in technology. Already, many genomic alterations, for example KRAS mutations, that were previously considered undruggable are now druggable.
In addition to ComboMATCH, two other successor trials are being planned. MyeloMATCH is an umbrella trial led by the NCI and the SWOG Cancer Research Network that will study treatments for acute myeloid leukemia and myelodysplastic syndromes. The master protocol for the study is still under development, but generally, patients will be enrolled in the trial at the time of diagnosis and will be assigned to an initial sub-study based on next-generation sequencing results.
SWOG is also coordinating a trial called I-MATCH that will analyze immune characteristics that relate to response or resistance to tumors. And a third trial, according to O'Dwyer, will focus on lymphoma. While lymphoma patients were eligible for NCI-MATCH, none were enrolled and only one myeloma patient joined, which O'Dwyer said indicates that there is almost no overlap of genomic drivers between those cancers and the solid tumors included in the trial. The planned lymphoma trial will, in part, attempt to fill that knowledge gap.