The Uphill Climb of Precision Medicine
By Chase Doyle
While clinical oncology is witnessing the widespread adoption of comprehensive genomic profiling, particularly in patients with advanced cancer, this activity has not necessarily led to more knowledge. In fact, where these approaches have been studied more formally, the data are still a bit disappointing. That’s according to Richard L. Schilsky, MD, Chief Medical Officer of the American Society of Clinical Oncology, who presented at this year’s Cancer Center Business Summit.
Driven largely by anecdotes of cancer patients with remarkable responses to previously unthinkable treatment, the enthusiasm for precision medicine is undeniable. "However,” Dr. Schilsky said, “this enthusiam has been tempered by those who have actually conducted the trials, which have not been as promising as hoped."
Barriers to Change
In order to achieve the promise of precision medicine—providing highly specific, minimally toxic, and dramatically effective treatments for patients—several barriers need to be overcome. For one, lack of standardization of preanalytical variables, such as tissue collection and testing.
“It all begins with the biospecimen that is obtained from patients,” said Dr. Schilsky. Variable amounts of tumor in the specimen fluctuate in quality. In addition, there are no standard guidelines for the way in which the specimens are obtained, handled, shipped, and preserved. “This whole area needs a lot of standardization and then a great deal of education regarding how to do it well,” he said.
The increasing complexity of molecular testing is another barrier to change. While current personalized medicine involves single gene-based molecular tests to select specific drugs for an individual patient, evolving technology incorporates multiplexed molecular tests with increased sensitivity and outputs. Then, there’s the future of personalized medicine, which, as Dr. Schilsky sees it, will entail an integrated genomic profile from high-throughput, next-generation sequencing to tailor targeted treatment for an individual patient.
If that weren’t complicated enough, understanding the basic language of “omics” poses its own challenges. There is a nomenclature issue, according to Dr. Schilsky, “because there’s no standardized approach that laboratories use to report their results.”
While some labs report mutations by virtue of changes in the nucleotide sequence in the gene, others report mutations according to changes in the amino-acid sequence of the protein that’s encoded by that gene. Furthermore, the process by which variants are labeled by different bioinformatics platforms is neither standardized nor transparent, and is largely proprietary in many cases.
One lab may call certain changes in a DNA sequence a ‘pathogenic mutation’; while another may label these same changes a ‘variant of uncertain significance.’ “It’s very confusing for doctors to know how these things relate to each other,” he said.
Finally, there’s a lack of basic confidence in the test being ordered.
“Oftentimes, the doctors don’t even know what they’re looking for,” said Dr. Schilsky. “They only know that if they look, they might find something.”
What they’re really looking for is a test with fairly broad coverage across commonly occurring mutations that is seen in many forms of cancer, he said. However, distinguishing one test from another with respect to the extent of genome coverage, the depth of coverage in any particular gene, and whether all mutations are reported or not can be a daunting task for a general oncologist.
Demonstrating Value: Is precision medicine worth it?
According to Dr. Schilsky, as technology continues to push the boundaries of science, oncologists find themselves in a curious position—that is, clinical application of precision medicine is moving faster than the generation of evidence to support its use.
Most of the real proponents of the field have not slowed down enough to do the studies that are necessary to demonstrate the real value of these approaches, he observed. “We have to either pull back on the clinical use or do it in a context that allows us to generate the information that we need.”
Speaking generally, Dr. Schilsky said that in looking at what’s been reported in the literature for genetic profiling studies, about 40% of the time something turns up in the tumor that is considered to be “actionable,” i.e., a target for a drug that is potentially available. In only about half of those cases, does the patient actually end up receiving that drug.
“Now you’re down to 20% of your total patient population, and only about 10% of these patients actually seem to respond to the therapies so far.”
This means only 2% of the total patient population actually achieve some benefit. And even from those 2%, Dr. Schilsky added, the benefits are usually short-lived.
“When you look at it that way, you have to ask, ‘is it really worth doing genomic profiling on 100 patients just to benefit 2?,’” he asked.
This isn’t to suggest a lack of enthusiasm on Dr. Schilsky’s part. He’s completely enthusiastic about continuing to do the research, but where his enthusiasm is tempered is the extent to which this should be done outside of the research setting. “The information’s just not in yet.”
While hospitals and health systems race to set up next-generation sequencing labs “to be perceived as cutting edge,” Dr. Schilsky advocates a more measured approach.
One goal he thinks that should be strived for “is to work in a context that enables us to learn from what we’re doing – to learn about how precision medicine is being used in clinical practice and to learn about the outcomes of patients who are receiving a treatment based on a precision medicine approach.”
ASCO’s TAPUR trial (Targeted Agent and Profiling Utilization Registry) is one such mechanism, but there are other clinical trials, too, including NCI-Match. With literally millions of potential combinations that can be created among the currently available oncology drugs, it may be impossible to assess every combination in every cancer subtype through the usual clinical trial strategies.
Dr. Schilsky believes in the need to create an information collaborative that would allow researchers to learn from every encounter with every patient, and then share that information across the medical community. “Otherwise, you’ve got the Wild West of precision medicine where everybody’s doing what they want to do and nobody’s learning from anybody else. That’s not a very sustainable approach in a field that is so complicated and so rapidly changing.”
In Dr. Schilsky’s opinion, the large numbers of new tests constantly evolving are difficult to assess in randomized trials, particularly because subpopulations of patients with targetable alterations treated with targeted agents are too small to analyze.
He thinks it’s clear that the traditional paradigm of prospective randomized clinical trials in every population of cancer patients can no longer be relied upon. “When you begin to apply the genomics, every common cancer becomes hundreds or thousands of rare cancer subtypes under the same histological label… it’s impossible to do randomized trials in all of those subsets.”
“Stakeholders need to get comfortable with the idea of using different types of information, generated in different ways,” he added.
Who’s going to pay?
Whether in a clinical-trial setting or a registry setting, Dr. Schilsky would like to see insurance companies go in the direction of so-called data development: providing reimbursement for genomic testing in a setting where data on patient outcomes are meaningfully captured.
He noted how insurance companies are under the gun to pay for testing, “but they’re often left in the dark about whether or not it works. They’re apprehensive about the associated costs of precision medicine – not the cost of genetic testing, but being asked to pay for the downstream treatments that ensue.”
Many of the treatments in these circumstances are not well supported by evidence. He says, they’re almost always off-label or in some cases investigational, and the commercially available drugs are all expensive. “While the tests themselves may be somewhat costly, they’re nowhere near as expensive as a drug treatment that costs $150,000 per year.”
With this financial threat looming, Dr. Schilsky sees a great opportunity for insurance companies to undertake data-driven approaches, if they are willing.
This would allow the tests to be available to patients while, at the same time, generate the kind of information everybody’s waiting to get. “Over time, based on the information that they actually collect, [insurance companies] can make a judgment on whether to continue to cover that testing on a routine basis.”
While there are plenty of anecdotes to suggest that the precision medicine approach could ultimately be transformative in cancer care, Dr. Schilsky thinks the medical community still has much to learn.
He would encourage all of those involved in precision medicine to do it in a way that allows them to collect information, learn about patient outcomes, and report that information so that everyone can derive benefit from it.