Pharmacotherapy biomarkers — those that predict drug response — and their related pharmacogenetic (PGx) tests are seen as the pathway to personalized medicine in cancer and other diseases, but their uptake has met resistance. Often they are supported by poorly designed, small, or biased clinical studies.
“There aren’t good ways for purchasers to assess analytic and clinical validity or clinical utility,” says Lee Newcomer, MD, senior vice president at UnitedHealthcare. “Here is the issue: Is the test dependable enough that a physician can rely on it to determine therapy? The only tests where we have solid information about those three measures are the biomarkers that have gone through the FDA approval process with a new drug.”
Newcomer refers to the three universally recognized criteria for PGx tests:
- Analytic validity — ability to accurately measure an analyte biomarker
- Clinical validity — the marker measured is linked to the target disease; and, most important
- Clinical utility — solid evidence that targeting the marker will improve patient outcomes
Health plans have been especially concerned about clinical utility.
“There is almost a gold rush out there to find biomarkers for cancer and other diseases,” says Newcomer. “People are not slowing down and doing the research that is necessary to demonstrate clinical utility. The lab companies and test developers are saying that it is expensive and it takes time. In response, the health plans are saying, We can’t afford for you not to do that. If we have resources being expended and bad treatment decisions being made, that is a lose-lose for the patient, health plan, and sponsor.”
“We need universally accepted criteria about how much testing is needed to get a test approved and covered,” says Lee Newcomer, MD, senior vice president at UnitedHealthcare.
Health plans cover the genetic tests mandated on drug labels and many plans will add a few additional tests that they have studied and determined to be beneficial. Generally, large health plans will have coverage policies for about 20 tests, experts say.
The prospects for credible biomarkers and wider coverage of PGx tests are improving. Some health plans and PBMs are experimenting with a pragmatic approach to covering testing. Advances in clinical studies will simplify marker and test development, and a new coding structure for identifying and billing genetic tests (See “Coding, billing process” on page 53) should help to control costs and utilization.
Despite limited coverage, genetic testing is expanding rapidly. “At UnitedHealthcare, we are now paying more for molecular diagnostics than we are for chemotherapy,” says Newcomer. That includes pharmacogenetic tests and other categories, such as genetic screening for heritable disease. National expenditures for genetic testing are expected to reach $6.2 billion in 2014, according to one source.
“Genetic tests cost from $80 to $1,400,” says David Lassen, PharmD, chief clinical officer at Prime Therapeutics. Some health plans contract with specific labs for certain tests and are able to negotiate prices. The list price can be much higher: The sticker price for the Oncotype DX multigene diagnostic assay from Genomic Health is $4,125 according to its Web site.
“We have created a framework for thinking about where and how to rationally apply pharmacogenetic testing in ways that ensure quality” and control costs, says David Lassen, PharmD, chief clinical officer at Prime Therapeutics.
The industry says a four-digit price is justified by the years of work it takes to discover and complete clinical trials needed to provide evidence of the test’s validity and utility.
Prime Therapeutics and certain of its Blue Cross owners are rolling out a pragmatic approach to evaluating and covering PGx tests in a cost-effective way.
“Genetic testing is gradually improving and we work to identify appropriate opportunities where we may want to facilitate testing,” says Lassen. “We have created a framework for thinking about where and how to rationally apply pharmacogenetic testing in ways that ensure quality and clearly improve outcomes and total cost of care.”
Prime’s approach includes requiring clinical utility and then goes beyond it.
“We start by digging deep into the evidence supporting a test and its ability to improve value and outcomes,” says Lassen. If reasonable evidence exists, Prime looks at other factors.
“The first question we ask is whether or not testing is the standard of care,” says Lassen. That question is important because in many cases there is controversy regarding the value of a test or its testing methods. Lassen cites HER2 gene testing for Herceptin in breast cancer. About 1 in 5 patients has the gene, and testing is complicated by the existence of two different tests. Immunohistochemistry measures gene over expression while the fluorescence in situ hybridization (FISH) test measures gene amplification.
The American Society of Clinical Oncology and the College of American Pathologists have recommended either test. So the incidence of the HER2 mutation and the possibility of expensive testing present a management challenge to health plans. “We have evaluated this situation and believe that health plans are already managing it well in terms of the appropriate standard of care,” says Lassen.
“The next question we ask is, Is the test administratively feasible? There are situations where testing can be administratively impractical — an example is testing for sensitivity to warfarin,” says Lassen. “The guidelines state that for new warfarin starts, sensitivity testing should be done within five days, but the chances are the results won’t be back in time to know how the patient is responding.” Simpler International Normalized Ratio (INR) lab tests are the most common approach to monitoring warfarin therapy.
“The third question is whether there is a potential for positive return on investment. If the answer is yes then we will move ahead,” he says.
Prime is testing its new framework. “In cardiology we are facilitating the testing for mutations in the gene CYP2C19 that identifies poor metabolizers of clopidogrel. The objective is to identity patients who are using brand-name antiplatelet therapy and if they are poor metabolizers of the medicine where there is limited value to therapy, then we can recommend to the prescriber to consider using generic clopidogrel.” This savings from the generic provides the return on investment that Prime is looking for.
“The other area we are looking at is oral oncology, specifically Tarceva and the EGFR test in lung cancer. We are looking for EGFR-positive patients who have been shown to respond to therapy,” says Lassen.
An important step in Prime’s program is rapid engagement with prescribers for new starts in therapy. “When we see a claim that indicates a new start, within 24 hours we reach out to inform the prescriber of the opportunity to test.
“Our lab partners will give us quick turnaround,” says Lassen. Prime has contracted with a few labs that it has evaluated to provide high quality tests quickly.
UnitedHealthcare uses a similar approach. “The FDA mandates testing for 17 or 18 agents,” says Newcomer. “We are looking at coverage of other biomarkers, primarily based upon how much activity we are seeing — for example, a specific test billed frequently by one lab. We are using that approach with the BRCA1 and BRCA2 tests for the risk of breast cancer.”
Both health plans are similar in another way. Both work with specific testing laboratories that they have evaluated for consistent quality in their work, capacity, turnaround time, and prices. For example, UnitedHealthcare has a longstanding relationship with Genomic Health for its Oncotype DX gene array expression test that identifies patients who are at risk for recurring breast cancer.
“Our relationship with Genomic is unique because we require the laboratory to prescreen the request for appropriate patients as part of the contract,” says Newcomer. “I believe that was a first for laboratory contracting.”
Proof of clinical utility
The strategies of Prime Therapeutics and UnitedHealthcare reflect a real-world approach to PGx testing, but they do not eliminate the need for stronger evidence in pharmacogenetics.
“We need universally accepted criteria about how much testing is needed to get a test approved and covered. The science that verifies there is a statistically significant correlation between a gene and a disease trait is not keeping up with the development of markers,” says Newcomer.
Retrospective studies are commonly used for marker validation, but they can have inherent biases, and there is no agreement about what makes a good retrospective study.
“It is not important whether the criteria are set by the FDA or another agency; we need standards for studies that attempt to demonstrate clinical validity and clinical utility,” Newcomer says.
The National Comprehensive Cancer Network (NCCN) has taken one step forward in evaluating the amount of evidence that underlies cancer biomarkers. In December 2012, it released a biomarker compendium that ranks the strength of evidence for more than 900 biomarker uses that appear in the NCCN clinical guidelines.
The NCCN describes the compendium as a tool to guide the appropriate use of biomarkers to screen, diagnose, monitor, and provide predictive and prognostic information for the treatment of patients.
The compendium was compiled by panels of experts in a process similar to the one that the NCCN uses for its treatment guidelines. The strength-of-evidence statement uses the NCCN categories of evidence and consensus.
- Category 1: Based on high-level evidence, there is uniform NCCN consensus that the intervention is appropriate.
- Category 2A: Based on lower-level evidence, there is uniform NCCN consensus that the intervention is appropriate.
- Category 2B: Based on lower-level evidence, there is NCCN consensus that the intervention is appropriate.
- Category 3: Based on any level of evidence, there is major NCCN disagreement that the intervention is appropriate.
There is another step forward in PGx testing. “The assessment and validation of biomarkers is becoming easier because of changes in the ways drug studies are conducted,” says Sumithra Mandrekar, PhD, professor of biostatistics at the Mayo Clinic.
“Phase 2 trials are becoming much larger. We are using phase 2 trials to get maximum answers about the drug marker interactions. The attitude is that it is worthwhile to spend a little extra time and enroll more patients and to use phase 2 data to really refine your question, the population, your treatment, and any marker,” says Mandrekar. She says the data from phase 2 trials are being examined more closely, such as questioning the clinical validity of a biomarker, and whether the drug works selectively in a certain subset of patients.
Phase 3 studies are also facilitating marker development.
“Researchers are starting to realize that there is value in obtaining tissue specimens in phase 3 trials for possible later use, even if those specimens are not studied as part of the original drug trial,” she says.
“For example, the original study may have assessed a drug in the overall population or within a marker-specific subgroup for late stage cancer. It is then possible to use this phase 3 trial to discover a new marker or a cutpoint and also to refine the marker and perhaps assess its use in a new setting, such as EGFR and ALK mutations in early stage lung cancer,” says Mandrekar.
There are other uses of these specimens. “They can be used to evaluate the assay methodology, reproducibility, or cutpoint for positive versus negative marker status. Specimens from phase 3 trials have associated clinical follow-up data, so they are a rich resource,” says Mandrekar.