NEW YORK (GenomeWeb) – Although oncologists and researchers recognize there is a dearth of evidence proving that genomically guided cancer treatments improve patient outcomes compared to standard care, they find it difficult to conduct traditional, randomized-controlled studies due to the very nature of precision medicine and patients' desire for the latest advances.
As a result, experts are exploring alternative approaches — basket trials such as NCI's MATCH, umbrella trials like Lung Cancer Master Protocol (Lung-MAP), meta-analysis, and retrospective matched cohort studies — to discern the benefits of precision oncology. But as cancer centers and community practices around the country increasingly implement genomic profiling and personalized medicine programs, some experts say oncologists should proceed with caution until those strategies have been vetted more rigorously —ideally, in randomized controlled trials.
While experts studying precision oncology approaches acknowledge the need for more evidence, they point to a confluence of factors — the pace of genomic advances in cancer, the advent of immunotherapy, and the rarity of molecular markers that might be targeted by treatments — that make it difficult to conduct prospective randomized studies, long considered the gold standard in clinical investigations.
"Conducting a prospective trial in this arena is really challenging because of the diversity of genetic alterations that exist, and trying to randomize on genetic alterations is really hard," said Lincoln Nadauld, director of cancer genomics at Intermountain Healthcare. It's more likely, in his view, that the evidence for precision medicine approaches will come incrementally from a variety of studies with different designs, instead of any one large randomized-controlled trial.
Intermountain, a healthcare provider in Utah and Idaho, launched a precision oncology program three years ago and has genomically profiled approximately 700 patients since, with the aim of getting them on personalized treatment options. Many community healthcare providers like Intermountain are also enrolling patients into studies, such as Lung-MAP, which experts say offer an opportunity to investigate the precision oncology hypothesis while providing patients access to novel treatments and genomic profiling. "What good is precision medicine if you don't treat patients using that information?" asked Roy Herbst, an expert in personalized lung cancer approaches at Yale Cancer Center and a co-principal investigator in Lung-Map.
The squamous-cell lung cancer study is enrolling patients at more than 700 medical centers and community hospitals around the country and employs a design that allows researchers to enroll patients with rare tumor markers across multiple substudies and gauge their responses to targeted treatments. "A lot of people who should get profiling in lung cancer and other diseases don't get it," said Herbst. "It's not available or affordable. It's tragic."
But in Lung-MAP, all patients are genomically profiled, and if they match to one of the biomarker arms, they receive one of three targeted treatments. If they don't have a biomarker of interest, they are enrolled in the immunotherapy portion of the study. "It may or may not work, but we'll answer the question either way," Herbst said. "If we answer these questions, then that helps make new drugs available in the future."
We saw that change in progression-free survival and we thought that was important [information] for community docs out there who are trying to understand how to implement and whether to implement precision medicine.
Managing bias
Recognizing these tensions in precision oncology, Nadauld and colleagues conducted a study comparing 36 patients who got targeted treatments within Intermountain's precision oncology program between July 2013 and January 2015 against 36 historical controls who received standard chemotherapy or best supportive care over five years. The researchers reported recently in the Journal of Oncology Practice, that patients who received targeted drugs guided by genomic test results had average progression-free survival of 22.9 weeks compared to 12 weeks in the control cohort.
The data, researchers concluded, suggest a survival advantage for patients who received precision medicine, although they acknowledged the study doesn't look at whether those patients' tumor features made them more likely to live longer anyway. In an effort to control for this bias, the investigators tried to match cases and controls tightly across different parameters, diagnosis, age, sex, and number of lines of treatment.
"We saw that change in progression-free survival and we thought that was important [information] for community docs out there who are trying to understand how to implement and whether to implement precision medicine," Nadauld said.
The researchers also tracked costs in the two arms and found that the charges per week for patients in the study were comparable — $4,665 per patient in the precision care arm versus $5,000 for controls. However, the average total cost per patient — including cost of treatment, associated radiology and lab work, toxicity, and sequencing — was around $92,000 in the precision care arm compared to $41,000 for those in the standard care arm. Given the higher costs of targeted therapy, Nadauld and colleagues were surprised that the per week costs weren't higher.
Although the per patient drug costs in the precision oncology arm were three times more than for the control group, Nadauld highlighted that the per-week total costs were comparable between arms because there were fewer toxicities, emergency room visits, and hospital admissions among those who received targeted treatment.
"So, we were able to save money in some areas, such as the ancillary care portions, but we lost money in the overall cost of drugs," Nadauld said, adding that precision oncology strategies will be hard to sustain at these prices. He is considering strategies to address this issue within Intermountain's precision oncology program, and has bandied around ideas with drugmakers. His group will also continue to follow the patients in the present study to assess the impact of precision oncology approaches on overall survival and overall healthcare costs from the time they entered the study until death.
Community healthcare providers trying to figure out whether they should implement precision oncology approaches for their patients and if it's a financially feasible strategy must consider the data from this and other studies alongside the viewpoint of critics who say the available evidence isn't enough. In a recent Nature perspective, entitled "The precision-oncology illusion," Vinay Prasad, a hematologist-oncologist at Oregon Health and Science University's Knight Cancer Institute, wrote that the enthusiasm for the approach doesn't square with the available data.
Prasad told GenomeWeb that despite the authors' attempt to control for bias in the Intermountain study, people in the precision medicine arm are likely to be fundamentally healthier than matched controls. "They're healthy enough that they're willing to seek things out like [precision oncology]," said Prasad.
He cited a 1982 American Journal of Medicine paper in which researchers compared 50 studies using randomized controls and 56 studies using historical controls for six drugs, and found that while 79 percent of historical control-design studies concluded the investigational drug was better than the comparator regimen, only 20 percent of randomized studies found the same. While at first glance the use of historical controls may seem to be a reasonable approach, in Prasad's view, the 1982 study shows that potential for bias within these types of studies are not to be taken lightly.
"Based on the available data, you cannot know that that amount of benefit is simply due to the bias in the way that these studies are done," he said.
Prasad expressed concern that cancer centers around the country are implementing genomic profiling and precision oncology programs in the absence of more evidence. In his Nature perspective piece, he highlighted that in the only published randomized-controlled trial, called SHIVA, patients who received drugs based on their tumor molecular profile didn't derive much benefit compared to patients who let their doctors pick the treatment. Although SHIVA had weaknesses and isn't the definitive test of precision oncology, Prasad wrote that more randomized-controlled trials are needed.
It's hard to tell someone that you're doing a precision medicine program, we found this new genetic mutation, and we're going to randomize you to this chemotherapy, which has a 5 percent chance of a response and a 100 percent chance of side effects.
The difficulty with randomized studies
In the Intermountain study, few patients had tumor markers, such as EGFR mutations in non-small cell lung cancer, where based on published literature and guidelines the standard practice is to treat patients with EGFR inhibitors. Instead, researchers saw many of the durable responses in patients who received targeted drugs based on alterations that were not common in their tumor type and so there was limited information in published studies as to how those subtypes might be treated.
For example, a melanoma patient had a c-Kit mutation, which is commonly seen in gastrointestinal stromal tumors. Researchers also noted patient responses in FGFR1-amplifed squamous cell lung cancer, FGFR2-mutant cholangiocarcinoma, and MEK1-activated non-small cell lung cancer, which are all rare subpopulations.
Intermountain's tumor board selected "actionable mutations" that had been "validated in peer-reviewed literature," and for which a targeted drug was available. The board then selected treatment options for these actionable mutations based on published clinical or preclinical evidence. But for some of the study participants with rare tumor markers, the support for precision oncology approaches came from published case series or n-of-1 reports.
Given the challenges of studying precision medicine approaches, Nadauld understands the calls for randomized-controlled trials. "A constant concern among docs and reviewers is whether outcomes are the result of drugs targeting specific alterations or are the alterations we're targeting simply markers of better biology, and that's why patients survive longer," he said. "And that's why people would love to see a big randomized-controlled trial, where patients are randomized based on alterations."
But he doesn't see that happening because of the very nature of precision medicine. For example, Intermountain tests advanced cancer patients on an NGS panel that assesses alterations in 96 cancer-related genes, and there are dozens of mutations that can occur in each one. "Waiting till you find another patient with the exact same alteration is really hard," he said. "Think of all the different combinations you have to randomize across."
Nadauld believes basket studies — which allow researchers to look for treatment response signals across different biomarker-defined tumors at once — will be the preferred method for investigating whether precision oncology approaches are benefitting patients. Indeed, a number of pharmaceutical companies are using basket trials to quickly figure out if their drugs are effective in genomically defined patient subgroups and flag for larger studies.
But these types of studies also have their challenges. For example, the NCI's MATCH trial is using a basket-type approach to explore the precision oncology hypothesis across 10 treatment arms. The trial has generated significant interest, with nearly 800 patients registering in the first three months. But as of this March, 56 (9 percent) genetically tested patients had matched to a treatment arm, and ultimately, 16 patients (2.5 percent) were enrolled. Although the match rate is close to what was expected in the study, the NCI has noted that actual mutation prevalence rates were "much lower" than reported in large-scale projects, such as The Cancer Genome Atlas.
In addition to the rarity of many cancer-associated markers, the enthusiasm — critics like Prasad use the word hype — around precision medicine is also what makes it challenging to recruit patients into randomized-controlled trials. "They all want the targeted therapy. They all want the new therapy," Nadauld said. "They're not that interested in possibly randomizing to a chemotherapy."
But Prasad maintained that it is nonetheless possible to do randomized studies in precision oncology and highlighted Lung-MAP as an example.
But even that umbrella trial, since it launched two years ago, has had to make amendments to its design due to lackluster accrual. Initially, Lung-MAP was designed so patients in the biomarker-based sub-studies would be randomized to either targeted treatments or standard chemotherapy. "As we were designing the trial, immunotherapy became approved in this setting," which pushed chemotherapy into a third-line option, said Herbst,
This made it harder to randomize patients to chemotherapy. "It's hard to tell someone that you're doing a precision medicine program, we found this new genetic mutation, and we're going to randomize you to this chemotherapy, which has a 5 percent chance of a response and a 100 percent chance of side effects," Herbst said.
So, after discussions with the US Food and Drug Administration, the experts at the helm of Lung-MAP decided to make the biomarker-based sub-studies into Phase II single-arm trials, and allowed those without targetable biomarkers to enroll in the immunotherapy portion of the arm. Enrollment has picked up "enormously" as a result of these changes, according to Herbst, and Lung-MAP enrolled it's 1,000th patient two weeks ago.
If there is activity seen in these single-arm studies, researchers have the option of randomizing patients to a control arm. "As a physician being involved with a lot of patients and having had family members as patients, it'd be hard to imagine how you could see some activity and then randomize [to a control arm]," he said, pointing out that the FDA, with the support of patient advocacy groups, has been increasingly willing to evaluate and approve breakthrough therapies based on data from single-arm studies. "The last thing we would want is to deny patients access to some of these life-saving treatments."