NEW YORK (GenomeWeb) – After presenting a preliminary analysis last year, researchers have now published in full results from the Phase II SUMMIT trial, which demonstrated how genomically guided basket trials can both expand and contract the eligible population for a drug with important takeaways for larger ongoing studies.
SUMMIT was structured to assess the HER-targeted therapy neratinib (Puma Biotechnology's Nerlynx) in patients with different tumor types and different mutations. The basket study was designed based on evidence that HER2-mutated cancers in various tissues might respond to off-label use of HER2-targeting drugs, such as Herceptin (trastuzumab), which the US Food and Drug Administration has approved in breast and stomach cancer.
In the new publication in Nature yesterday, researchers led by Memorial Sloan Kettering's David Hyman reported that patient responses indicated that both response to HER kinase inhibition varies not only based on the mutation present, but also the tumor type, and the pattern of co-mutations present.
In 141 patients with cancers harboring 31 unique HER2 mutations and 11 unique HER3 mutations who received neratinib in the trial, there was no single dividing point between responders and non-responders. Rather several different features — both biological and genomic — defined the best outcomes: Patients with breast, cervical, and biliary cancers, and individuals with tumors containing kinase domain missense mutations did the best on the drug.
Importantly, the SUMMIT results contradict some things that were predicted by preclinical models. In terms of molecular findings, the trial appears to have ruled out mutations in HER3 as conferring sensitivity to neratinib, despite lab evidence suggesting that these mutations might be oncogenic drivers.
In some part, the results uphold the growing understanding that mutation status, regardless of tumor type or location, can confer a response to targeted therapy. Patients with HER-2 mutations in five different tumor types — breast, non-small-cell lung, cervical, biliary, and salivary cancers, showed response to neratinib, enough so that the trial pushed for expanded enrollment of these tumor types.
The concept of pan-cancer targeted treatments — drug approvals in which genomic characteristics determine whether a patient can be prescribed a drug, not where in their body a tumor occurs — has garnered increasing excitement at recent medical and research meetings.
2017 saw the first of a new type of tissue-agnostic US Food and Drug Administration approval — of Merck's Keytruda (pembrolizumab) in advanced cancer patients with microsatellite instability-high and mismatch repair deficient tumors.
Soon after, Loxo Oncology said it planned to seek FDA approval for its drug larotrectinib as a treatment for patients with TRK fusion-positive tumors, after reporting results from a study of 55 patients across 17 unique cancer types, in which 76 percent responded to the therapy.
But results from SUMMIT also reiterate that basket trials won't always support a broad expansion of the population eligible for a targeted therapy, defining not just who can be treated, but who can't.
Neratinib did not seem to be as effective in many other tumor types recruited to the trial. No responses were seen, for example, in either bladder or colorectal cancer — comprising 16 and 12 patients respectively.
And based on the results so far, only breast cancer has met the primary endpoint for efficacy in the trial, though study authors wrote that enrollment is continuing for other tumor-specific cohorts.
According to Hyman, both broadening eligible populations for targeted drugs, and ruling out patients who will not respond, are worthy goals, especially as more patients are tested in the clinic using broad genomic sequencing panels.
"We approach most of these studies hoping for homogeneity or consistency of response – but implicit in the design is the expectation that most of the time we are not going to achieve that," he said. "We were expecting a lot of diversity of response, and designed the study to capture that for the first time."
The FDA granted marketing authorization to four next-generation sequencing panel tests in 2017, most recently Memorial Sloan Kettering Cancer Centers' MSK-IMPACT, and Foundation Medicine's FoundationOne CDx.
Wider adoption of broad sequencing implies more frequent discovery of so-called targetable mutations in cancers that do not match those that were recruited for the drug trials that led to initial approval for these therapies.
Clinicians have debated how this may impact their prescribing moving forward, highlighting hypothetical cases, where a patient might demand, or a physician might consider off label prescribing in cases of a targetable mutation in a tumor type without efficacy data.
"There will be a [growing] need to disincentivize off-label prescribing, so we will need studies like this to guide regulators and to guide physician prescribing behaviors," Hyman said.
An important achievement for SUMMIT, Hyman and his coauthors wrote, is that the "permissive enrolment strategy" — which brought in patients with not only different tumor types, but also different types of variants with varying degrees of biological characterization — and allowed researchers to ask questions that might otherwise only be answered by preclinical lab testing.
"The challenge is not just [to investigate] the question 'if you have a mutation do you respond?' but also whether mutations are functional. There isn’t just imperfect knowledge on how to act on potential biomarkers, but really what their biological role is at all," Hyman explained.
"We were able to look at whether we can make additional decisions about how to triage patients even in the absence of biological characterization of these mutations, at the same time as looking at those that we felt were [well understood] and seeing how our results compare to what we thought we'd see."
A lot of these insights, Hyman argued, were only possible because of the size of SUMMIT, which raises some anxiety about how the results of larger umbrella-protocol trials like NCI-MATCH can be best analyzed and received.
"The concern I have is that we will see a wave of reports from studies that may not define at first flush the kind of activities we are hoping for," he said. "So, as we analyze this data, we have to ask if studies were powered to signal-seek for areas of greatest benefit."
For example, he said, Genentech's MyPathway trial recently published a report on 251 patients with 35 different tumor types in the Journal of Clinical Oncology.
Authors concluded that the four targeted therapy regimens in the study did produce meaningful responses when administered in some solid tumor types not currently labeled for those agents.
"But if you look at the top line you might be disappointed," Hyman argued. "So, a lot of this is going to be how we process the data that comes and out and what we do with it in follow up studies," he said.