NEW YORK – A study design like one used in a glioblastoma trial by a Dana-Farber Cancer Institute-led team may pave the way for faster, more efficient neuro-oncology drug trials.
Building on the emerging "adaptive platform trial" (APT) model, the team is testing multiple drugs simultaneously on biomarker-defined glioblastoma subtypes, adjusting specifics — even adding and dropping treatment arms as the trial progresses — in response to data as it emerges.
"The way we currently evaluate drugs in the conventional fashion would be to do trials for each one," said Rifaquat Rahman, a radiation oncologist at Dana-Farber and the study's first author. "That's a very slow pace, [and] what we're hoping to do is change that. The goal of this trial is to see how we can actually speed up and improve upon that process."
The Phase II Individualized Screening Trial of Innovative Glioblastoma Therapy (INSIGhT) study reported initial results on Monday in the Journal of Clinical Oncology. Researchers from Dana-Farber led the trial in collaboration with 10 brain tumor centers across the US.
They began the trial with three experimental arms and a shared control arm, examining whether Eli Lilly's Verzenio (abemaciclib), Puma Biotechnology's Nerlynx (neratinib), or CC-115, an investigational DNA-PK/mTOR inhibitor from Celgene, delivered a survival benefit over standard-of-care temozolomide in participants.
Patients in the study were newly diagnosed with O6-methylguanine-DNA methyltransferase-unmethylated glioblastoma, a hard-to-treat subtype of glioblastoma that carries a poor prognosis, and had undergone tumor genotyping via whole-genome copy-number analysis and whole-exome sequencing.
Based on that genotyping, the Dana-Farber team sorted participants into three biomarker-defined subgroups: EGFR-positive, PI3K-positive, and CDK-positive. These are the three most commonly altered pathways in glioblastoma, the team noted, adding that they have direct relevance to the drugs being evaluated.
Investigators randomly assigned 73 patients to the abemaciclib arm, 81 to the neratinib arm, 12 to the CC-115 arm, and 71 patients to the shared control arm.
The shared control arm, Rahman explained, is one of the study features that enables simultaneous testing. This contrasts with more traditional randomized trial designs, which generally compare each treatment to its own matched control cohort.
"Here, we're actually being much more efficient because essentially far fewer patients are getting the control arm compared to an experimental therapy," Rahman said. "But you're not losing anything by still being a randomized trial and still being a trial that's randomized to a control group."
During the course of the trial, study statisticians evaluate rolling progression-free survival (PFS) results on a monthly basis, Rahman explained, adding that the investigators use that data to determine the randomization probability for sorting new participants into any given arm.
In practice, this means that when analyzing each experimental treatment arm, the team used the "concurrent control group" of those patients accrued while each experimental arm was open during enrollment. In this way, all 71 control arm patients were used in analyzing neratinib because that was the last arm to complete accruing patients. The other experimental arms used subsets of the full shared control arm corresponding to patients accrued while the experimental arm was open. These comparisons then inform ongoing randomization of new patients.
"So it doesn't stay 1:1:1 over time," he said, "it actually changes, [so] if there is a drug that is beneficial, more patients are more likely to get that."
In the INSIGhT trial, for example, this led to fewer patients being randomized into the CC-115 arm, as this drug showed no improvement to PFS.
Though CDK-positive patients showed improved PFS with abemaciclib and EGFR-positive patients showed greater PFS with neratinib, none of the drugs evaluated in the current publication showed a benefit to overall survival.
While unfortunate from a patient perspective, Rahman pointed out that having a more efficient way to identify drugs that are not providing sought-after benefits can still be seen as a success, as it enables investigators to stop pursuing those drugs earlier in the trial process.
The adaptive design of the INSIGhT trial appears to be one of the first of its kind in neuro-oncology, though it was followed three years later by the larger GBM AGILE study, sponsored by the nonprofit Global Coalition for Adaptive Research.
Adaptive trials in general have been gaining traction as patients, regulators, and drugmakers begin to shift away from the "maximum-tolerated dose" model.
Motivated by experiences with severe treatment-related side effects, the patient advocacy organization Patient-Centered Dosing Initiative has called upon the US Food and Drug Administration to push trial sponsors toward more patient-centric models.
Merck, for instance, evaluated ongoing study results every three weeks in a Phase II trial of Keytruda (pembrolizumab), adjusting the trial doses in accordance with those analyses.
Getting adaptive trials approved can still be more challenging than for traditional trial designs, Rahman said, as these are still new and people are not used to them.
"There is a lot more education and time in terms of working through the steps you have to go through from that regulatory perspective," he said, "but what we've found is that it is feasible and doable, and so we were able to work through those kinks."
During the time the study was making its way to publication, the Dana-Farber team added two more experimental arms to the INSIGhT study. New participants can now be randomized to receive either a novel brain penetrant chemotherapy called QBS10070S, an immunotherapy regimen consisting of a tumor vaccine VBI-1901 and a PD1 antibody, or standard therapy.
Rahman and his colleagues expect to publish another batch of interim data as the trial progresses, but it is too soon to say when that might occur.
"We're excited to show the feasibility of this trial," Rahman said. "We think this can be part of the future for glioblastoma drug development."