Genomic Era Trials Drive Discovery

OncologyLive, Vol. 20/No. 23, Volume 20, Issue 23

Innovative clinical trials designed to exploit the explosion of data on the drivers of cancer are demonstrating value as tools to explore potential signals of efficacy across tumor types, according to oncology experts. These studies have become a feature of the drug discovery landscape in oncology and are likely to remain part of the fabric of clinical research in the genomic era.

Richard L. Schilsky, MD

Innovative clinical trials designed to exploit the explosion of data on the drivers of cancer are demonstrating value as tools to explore potential signals of efficacy across tumor types, according to oncology experts. These studies have become a feature of the drug discovery landscape in oncology and are likely to remain part of the fabric of clinical research in the genomic era.

Findings from TAPUR, the American Society of Clinical Oncology’s (ASCO’s) study of off-label use of FDA-approved targeted therapies, offer the most recent example of the potential of novel trial designs. Recently reported results from several of the many arms of TAPUR indicated promising antitumor activity, and more publications showing positive signals are on the way.

In one substudy, investigators found that patients with non—small cell lung cancer (NSCLC) responded to palbociclib (Ibrance), a CDK4/6 inhibitor currently approved in breast cancer settings, according to findings reported at the 2019 ASCO Annual Meeting (2019 ASCO).1 In another study arm, patients with metastatic breast cancer with high tumor mutational burden (TMB) displayed responses to pembrolizumab (Keytruda), a PD-1 inhibitor that is not approved in the malignancy.2

Additionally, 3 cohorts involving patients with colorectal cancers (CRCs) also recently met the study’s criteria for positive signals, and about 35 additional cohorts have expanded enrollment because of preliminary signals of activity, said Richard L. Schilsky, MD, ASCO’s chief medical officer and an architect of the TAPUR study.

The 3-year-old study will continue at least through 2023, and Schilsky said ASCO is in discussions with pharmaceutical developers about adding more agents and extending the trial further.

“As long as the study is generating interesting results, and we and the cancer community and the companies believe that the information is useful, we don’t necessarily see an end to the TAPUR study,” said Schilsky, a 2018 Giants of Cancer Care® award winner. “Because it’s not exploring a particular hypothesis, like a randomized clinical trial would be exploring, there’s no fixed sample size and there’s no specific number of events we’re looking for. We’ll just keep it going as long as there’s continuing interest to keep it going.”

A New Generation of Trials


TAPUR is one of several ongoing, multicohort, phase II signal-finding studies. These are often called master protocols, as well as basket, umbrella, or platform trials, depending on their structure. They were set up to give patients across large geographies access to multiple targeted therapies, collect high-quality data on biomarkers and end points, and expand, close, or adjust cohorts as evidence on efficacy becomes available. Other such trials include NCI-MATCH, I-SPY 2, and the National Lung Matrix Trial (NLMT).

A basic basket trial groups patients with different tumor histologies but the same genetic alteration and treats them with a targeted therapy. An umbrella trial tests multiple agents, enrolling patients with a single tumor type and grouping them by molecular alteration.

Click to Enlarge

Although investigators have used elements of these trial designs in the past, in the era of genomic medicine, such studies are particularly well suited to serving patients who have rare alterations and sorting out which agents are active in which disease variants, experts say.

“We’re now slicing up the molecular pie into many thin slices because we’re using a precision medicine approach. Finding these molecularly selected patients can be tough, so we need a good number of centers to actually find these patients, and an efficient way of doing that would be to do molecularly driven basket and umbrella studies where each arm represents a molecular subtype,” said Timothy Yap, MBBS, PhD, MRCP, medical director of the Institute for Applied Cancer Science at The University of Texas MD Anderson Cancer Center in Houston.

“But more important than that, with many of these drugs and aberrations, we don’t actually know what the signal is. We don’t know whether it even works. With basket studies, you can actually define your threshold for a response and make go/no-go decisions in terms of that,” said Yap, who is an investigator on the NLMT.

In the TAPUR protocol, a drug is considered worthy of further study if at least 7 of 28 patients exhibit disease control (DC). DC is defined as objective response (OR) or stable disease at 16 weeks (SD16+), which includes complete response or partial response (PR).

As it stands now, 7 treatment arms are open and recruiting patients, while 6 other arms have been closed to enrollment. Of the 12 cohorts studied in the closed arms, results have been positive in 2 groups, negative in 5, and are pending in another 5.

In the substudy involving palbociclib, 28 patients with NSCLC, CDKN2A loss or mutation, and no RB mutation received palbociclib 125 mg daily on 21-day cycles, followed by 7 days off until disease progression. One PR and 7 SD16+ were observed, for a DC rate of 29% (90% CI, 15%-37%). Eleven patients had at least 1 grade 3 or 4 adverse event (AE) or serious adverse event (SAE) at least possibly related to the palbociclib regimen, with the most common being cytopenias.1

In the cohort testing pembrolizumab, 28 women with heavily pretreated metastatic breast disease and high TMB received the drug at 2 mg/kg over 30 minutes or 200 mg every 3 weeks until disease progression. The DC rate was 37% (90% CI, 24%-46%), with an OR rate of 21% (95% CI, 8%-41%). In terms of toxicity, 4 patients experienced AEs; these included SAEs of grade 3 colonic obstruction and hepatic failure and grade 2 urinary tract infection at least possibly related to the drug.2

Other recent TAPUR cohorts have tested trastuzumab (Herceptin) and pertuzumab (Perjeta) in HER2-amplified CRC, vemurafenib (Zelboraf) plus cobimetinib (Cotellic) in BRAF-mutated colon cancer, and pembrolizumab in CRC with TMB. The 3 cohorts met TAPUR’s criteria for a positive signal, and investigators have submitted the data for presentation at a conference next year, Schilsky said.

A number of arms have also had negative findings. For example, TAPUR investigators reported that single-agent palbociclib has no meaningful clinical activity in patients with CDKN2A-mutated or -deleted advanced pancreatic or biliary cancer.3 Schilsky noted that TAPUR was created specifically to test the efficacy of FDA-approved drugs in off-label indications and to aggregate data on such use that would otherwise be unavailable to the oncology community.

“As long as we generate information that is fairly definitive, whether positive or negative, we think we’re achieving some success. Even a negative result is informative if it discourages physicians from prescribing an off-label drug that’s not likely to work,” Schilsky said.

“On a positive signal, the ultimate success would be if the results are sufficiently persuasive that, for example, they are incorporated in clinical practice guidelines or compendia, which would then make it a lot easier for that off-label therapy to get reimbursed. The ultimate, of course, would be if the company that owns the drug could use the data to seek a label expansion from FDA. We’re hopeful that at some point those things might happen,” Schilsky said.


NCI-MATCH, developed by the National Cancer Institute (NCI), differs from TAPUR in that it tests both approved and investigational drugs. Schilsky said ASCO has had discussions with NCI-MATCH investigators about possibly sharing data to accelerate learning.

The study yielded a positive signal this year in its arm H (EAY131-H), which evaluated the combination of the BRAF inhibitor dabrafenib (Tafinlar) and the MEK inhibitor trametinib (Mekinist) in patients with BRAF V600E mutations, according to results presented at 2019 ASCO. The combination is approved for BRAF-mutated metastatic NSCLC, melanoma, and thyroid cancer; the trial excluded patients with those tumor types and with CRC, which investigators said has responded poorly to this approach in prior studies.4

Participants received dabrafenib 150 mg twice daily plus trametinib 2 mg daily. In all, 33 patients with 17 different cancer types and BRAF V600 mutations were available for evaluation. Overall, 11 patients had a PR (33%), which met the study’s primary end point, and 13 had SD (39%) as their best OR. All 6 patients with gynecologic cancers had either a PR or SD; 4 patients with cholangiocarcinoma had a PR, and 3 with central nervous system tumors had either SD or a PR.

Median duration of response was 12 months. The median progression-free survival (PFS) was 11.4 months, and median overall survival was 28.8 months.

In terms of tolerability, most AEs were grade 1 or 2, with the most common events including fatigue (60%), nausea (54%), fever (51%), and rash (40%). Grade 3 AEs included fatigue (11%), neutrophil count decrease (9%), and hyponatremia (6%).4

“Dabrafenib and trametinib had clinically meaningful activity in this very heterogeneous, heavily pretreated patient population with BRAF V600E mutations,” April K.S. Salama, MD, a medical oncologist at Duke Cancer Center in Durham, North Carolina, said in presenting the results. “And I think very importantly, no new safety signals were identified.”

Another NCI-MATCH arm tested palbociclib in patients with heavily pretreated advanced nonbreast solid tumors characterized by amplification in CCND1/2/3 and did not find a signal, according to findings presented at the 2019 American Association for Cancer Research Annual Meeting. Among the 36 patients in the analysis, none showed an OR and 14 (38.9%) had SD. Median PFS was 1.8 months, and estimated 6-month PFS was 13% (90% CI, 5%-29%). Investigators concluded that the alteration did not predict response to palbociclib in the patient cohort, which included participants with 23 different histologies.5

Other arms of NCI-MATCH have previously found that the investigational AKT inhibitor capivasertib (AZD5363) demonstrated clinical activity in patients with AKT-mutated cancers and that nivolumab (Opdivo) showed promising activity in mismatch repair—deficient non-CRCs.6 Meanwhile, ado-trastuzumab emtansine (T-DM1; Kadcyla) did not meet its primary end point for OR rate in heavily pretreated patients with HER2-overexpressing tumors, but investigators said the clinical activity observed in salivary gland tumors warrants further study.7

Investigators also unexpectedly found that 62.5% of the first 6000 patients initially screened for NCI-MATCH have tumors other than the 4 most common cancers (breast, CRC, NSCLC, and prostate), higher than the 25% original target for recruitment.6

Basket trial arms that do not find signals are inevitable, said Shivaani Kummar, MD, a professor of medicine (oncology) and radiology at the Stanford University Medical Center in Palo Alto, California.

Although these studies offer important opportunities for patients who lack therapy options, the work of matching drugs against a huge number of potentially actionable genetic alterations is inherently challenging, said Kummar, who previously headed early clinical trials development at the NCI and was principal investigator in NCI’s MPACT trial.

“Each patient that we sequence has multiple abnormalities. The question becomes: Which one do you target, by which drug? Do you give a combination of drugs, and if so, is there safety established for the combination? We’re really trying to get more information and more data to better match the therapies to what we are finding in the tumors. That’s why you see mixed results for some of these platform trials. Also, all of this genetic sequencing and matching is only as good as the drug you give,” said Kummar.

I-SPY 2 Success

A study that is starting to achieve the goal of influencing drug approvals is I-SPY 2, a pioneering, decade-old platform trial of neoadjuvant personalized agents for locally advanced breast cancer. I-SPY 2 is a randomized trial that tests both investigative and approved drugs, takes pathological complete response (pCR) rate as an early surrogate efficacy end point, and uses an adaptive method to preferentially randomize new patients to the arm with the highest probability of efficacy based on the results so far.

In 2013, I-SPY 2 “graduated” its first 2 drugs: neratinib (Nerlynx) and veliparib (ABT-888).8 The FDA approved neratinib in 2017 for the extended adjuvant treatment of patients with early-stage, HER2-positive breast cancer after trastuzumab-based therapy. Veliparib, a PARP inhibitor, remains under study in breast and ovarian cancers.

I-SPY 2 investigators reported in 2017 that adding pembrolizumab to standard paclitaxel therapy increased the estimated pCR rate nearly 3-fold compared with paclitaxel alone (80 mg2 weekly for 12 weeks) in patients with triple-negative breast cancer (TNBC; 60% vs 20%) and patients with hormone receptor—positive/HER2-negative breast cancer (34% vs 13%).9

The FDA granted the combination breakthrough therapy designation, and this year, findings from the phase III KEYNOTE-522 trial confirmed a statistically significant improvement in pCR rates as well as a favorable trend for event-free survival (EFS).10 Roger M. Perlmutter, MD, executive vice president and president of Merck Research Laboratories, said the decision to pursue the trial “rested heavily on data from I-SPY 2.”11

The experience with pembrolizumab for TNBC demonstrates that the platform trial model can accelerate the development of new treatments and target the patients who will benefit most, said Laura J. Esserman, MD, MBA, I-SPY 2 principal investigator and director of the Carol Franc Buck Breast Care Center at the University of California San Francisco.

“What it shows is this works. We’ve seen now that the signal we’ve generated is reproducible,” said Esserman, a 2018 Giants of Cancer Care® award winner.

I-SPY 2 investigators are scheduled to present results from several trials at the 2019 San Antonio Breast Cancer Symposium. These include studies of the combination of chemotherapy and the HER3-targing antibody patritumab (U3-1287); the impact of residual ductal carcinoma in situ on breast cancer recurrence; the use of HER2 signaling, estrogen receptor, and proliferation biomarkers to predict response to multiple drug combinations; and cancer subtype—specific association of pCR with tumor volume progression assessed with magnetic resonance imaging during neoadjuvant chemotherapy.

I-SPY 2 has focused on fine-tuning end points and use of biomarkers to evaluate early response and determine subsequent treatment. One subgroup analysis found that pCR was highly associated with 3-year EFS (P <.001). The 3-year risk of distant recurrence was 5% among those who reached pCR compared with 21% in patients who did not reach pCR. The HR for EFS was 0.20 (95% CI, 0.11-0.36) for those who reached pCR versus those who did not.12 Another study based on I-SPY 2 data assessed circulating tumor DNA (ctDNA) as a predictor of tumor response to treatment, finding that patients who were positive for ctDNA after neoadjuvant therapy correlated with lack of pCR.13

In addition to graduating therapies to confirmatory trials, I-SPY 2 has goals of integrating drug testing more fully into clinical practice and achieving earlier and less toxic treatments for women with breast cancer, Esserman said. It aims to identify combinations that achieve pCR quickly and possibly avoid the need for chemotherapy.

“People really want to try [to] figure out who needs anthracycline and who doesn’t. The platform itself can be modified; it’s an efficient structure for learning. We have our biomarkers, which include the imaging end point, and this whole engine for discovery. We now have 22 sites on board and a precompetitive collaborative set of contracts with all pharma, with all the sites. The platform means you can be efficient in adapting and changing, and pivoting and learning, as the field evolves, and you can make sure you can get patients to what they want,” Esserman said.

Global Trial for Lung Cancer

Serving patients is also central to the United Kingdom—based NLMT, Yap said. In a national healthcare system in which some centers have access to new drugs and others do not, the trial was launched to provide next-generation sequencing and matched therapies to patients with NSCLC across the country while picking up rare mutations that might not appear in a single-site study, he said.

The NLMT has 22 mutation-defined cohorts treated with 7 targeted drugs, according to a presentation at the 2019 International Association for the Study of Lung Cancer (IASLC) World Conference on Lung Cancer.14 The study, which the IASLC described as the largest global precision medicine trial in NSCLC, screened more than 4000 patients to recruit 315 participants.15

An interim analysis of 19 cohorts identified several agents with promising activity in 4 pathways of genomic alteration. Those pathways and the drugs under study are: cell cycle progression (palbociclib): RAS activation (palbociclib, vistusertib, and selumetinib/ docetaxel); PI3K/PTEN/AKT/mTOR (vistusertib, capivasertib); and RTK signaling (FGFR inhibitor AZD4547, crizotinib [Xalkori], and osimertinib [Tagrisso]).

The RTK signaling group had 4 subgroups with durable clinical benefit (DCB) rates that exceeded targets, and 3 of the 4 exceeded response targets. Crizotinib for tumors with ROS1 gene fusions and MET exon 14 skipping had DCB rates of 71% and 80%, respectively, and OR rates of 68%, both associated with predictive probability of success (PPoS) >0.99. (A treatment is considered a “go” if the PPoS is >0.50). Osimertinib for EGFR T790M mutation had a 94% DCB and an OR rate of 76%, with a PPoS >0.99. AZD4547 for FGFR mutation passed the DCB target (31%), which was associated with a PPoS of 0.54.13

In the RAS activation category, the combination of selumetinib and docetaxel had a 50% DCB rate and a PPoS of 0.89 for the subgroup with NF1 mutation. Palbociclib for tumors with KRAS mutation resulted in a PPoS >0.99 and a DCB rate of 44%.

At the same time, other subgroups appeared unlikely to meet predefined targets. In the cell cycle progression category, the DCB target had yet to be reached for any of 4 subgroups. And in the PI3K/PTEN/AKT/mTOR category, the DCB rates for the 5 subgroups ranged from 9% to 21%, associated with a PPoS of 0.06 to 0.34. The PPoS for the response target was <0.20 for all the subgroups.

Pros and Cons of Genomic Approach

Gary Middleton, MBBS, MD, FRCP, who is heading the NLMT study, said during his IASLC presentation that the mutationmatching approach does come with limitations. He noted some drugs that appear ineffective against an oncogenic target might prove beneficial in combination with other agents.

“One of the risks with the paradigm of precision medicine is that we focus on very high levels of single-agent activity in oncogene addiction,” said Middleton, a professor of medical oncology at the University of Birmingham in the United Kingdom. “We might be missing a trait, or we might be missing a useful target in our patients.”

He also noted that tobacco-associated lung cancer poses a multiplicity of obstacles for therapeutic development: genetic instability, ongoing evolution, oncogene predominance, and numerous pathways of resistance.

“That brings up the criticality of the preclinical models that we need,” said Middleton. “We need to make sure that the targets we are gunning for are tested in models that completely replicate the genomic complexity and chaos of tobaccoassociated non—small cell lung cancer.”

Yap and others noted that in addition to producing a steady stream of data and enhancing patient access to targeted drugs, platform and basket trials facilitate research by avoiding the usual struggle to recruit enough patients.

With small cohorts, many study sites, and an adaptive approach to end points, trials can wait for patients to emerge or adjust cohort parameters over time. The greater challenge is in creating a robust and durable trial infrastructure to manage patients as they enroll, with the inclusion of staff with expertise in adaptive methods in some cases, investigators said.

“It’s challenging at the very beginning in terms of setting it up. Obviously, there’s a lot more investment and many different parties to bring into play. You need a very robust central team of dedicated members to actually ensure that the trial is continually conducted well and that there is accountability from all parties involved, just because it’s done on such a large scale,” Yap said.

Once that structure is in place, it gives a platform trial its ability to add more sites, patients, and therapies and to evolve along with the field of oncology without having to launch a new project from scratch, Esserman said. She said her team is designing a trial arm called I-SPY 2.2 that will conduct phase III studies to generate regulatory evidence, after concluding that using its existing research network was more efficient than creating a wholly new I-SPY 3.

However, Schilsky said he expects basket trials will continue to be used primarily for signal finding and not as confirmatory trials.

“There’s no sign yet that they are going to be replacing prospective, randomized clinical trials, but they can inform the development of a drug to the point where a decision can be made as to whether or not a randomized trial is appropriate,” Schilsky said. “I just see them as another tool in the clinical trialist’s tool belt, if you will, that allows us to make the best use of the available patient resources.”

Trials like TAPUR are basket-umbrella hybrids, testing a number of therapies in multiple diseases and assigning patients to cohorts according to their genomic alterations. If a study is open ended and regularly adds new therapies and patients, it is often called a platform trial. Master protocol-type trials may be adaptive, using Bayesian statistical methods to optimally adjust dosing and enrollment as the trial proceeds.