A lack of clinically useful diagnostics has prevented personalized medicine from living up to some of the eager projections made in the wake of the completion of the human genome project, a Tufts report has concluded.
Joshua Cohen, a senior research fellow at the Tufts Center for the Study of Drug Development, surveyed drug and diagnostic companies, payors, and pharmacogenomics experts, publishing his findings in a recently released Impact Report. From the survey and follow-up interviews he concluded that a lack of evidence linking diagnostic tests to clinical outcomes has made payors skeptical, hindering the growth of personalized medicine.
"I had noticed, particularly in Europe but also in the US, that payors were reluctant to pay for diagnostics and/or drugs, at least initially, and felt [this] needed to be further examined," Cohen told PGx Reporter this week. "So we … sent questionnaires to the drug sponsors, to diagnostic sponsors, to payors, as well as … people who have published widely in the [pharmacogenomics] field for a more bird's eye view on the field of personalized medicine."
From his analysis, two main factors emerged as the key obstacles to personalized medicine development: a lack of clinical outcomes evidence and payor hesitancy. According to the report, the two are often linked, though he also found that even with increased outcomes data, reluctance among payors can linger.
"You can see in the study, even when companion diagnostics are on the label, recommended by FDA, even then there are payors reluctant to pay for the test," he said.
For example, the US Food and Drug Administration has updated the labeling for warfarin to inform doctors and patients that people with certain polymorphisms in their CYP2C9 and VKORC1 genes are at heightened risk for adverse reactions. However, private payors do not reimburse for genetic testing to dose warfarin due to limited clinical utility data showing that testing improves patient outcomes and save money compared to standard INR monitoring. Moreover, the Centers for Medicare & Medicaid Services as agreed to pay for testing only in the context of a clinical trial that shows such an intervention improves outcomes for Medicare beneficiaries (PGx Reporter 03/09/2011).
To prepare the report, which will be published later in peer-reviewed form in the Pharmacogenomics Journal, Cohen said he surveyed 38 payors, of which about 25 responded in full, allowing them to be included in the study. In addition, 14 out of 22 pharmacogenomics experts responded fully, as well as six of eight drug developers and four diagnostic developers.
Cohen declined to reveal the specific questions in the survey, but said that he used a combination of questionnaires and interviews on regulatory trends in personalized medicine.
"I don't think [these results are] pathbreaking, but there are a couple of take-home messages I think are really important," Cohen said. "Payors are the linchpin here and they are skeptical of the usefulness of diagnostic tests, but even when they see the [clinical validity of a test] they're still skeptical because they don’t see [improvement in] outcomes."
Respondents to the survey generally expressed that even if they viewed a diagnostic as accurate in stratifying patients, there still might be a lack of conclusive evidence linking the test to patient outcomes.
Cohen said that surprised him, because he thought payors would be more "onboard" with personalized medicine. "They are not, generally," he said.
Cohen cited the lack of regulatory guidance as one stumbling block for the growth of companion diagnostics. The FDA issued a long-awaited draft guidance on in vitro companion diagnostic tests last week, reinforcing earlier advice from the agency that a drug and its companion test should ideally be developed simultaneously." (PGx Reporter 07/13/2011)
"In my view and also in the experts' views, not having that guidance in place perhaps has prevented some further development, in that the FDA was reluctant to come out with a document that would incentivize companies to co-develop as opposed to post hoc develop," he said.
The distinction between co-development and "post hoc" development, in which the diagnostic is developed after the drug, is important, he said. "I did not know this going in but there is actually very, very little successful co-development [happening], which is really one of the main take-home messages in the report."
This finding is particularly interesting in light of a report that Tufts published last year that found that 80 percent of pharmaceutical firms claimed to be engaged in personalized medicine projects. That report, however, did not address whether these firms were actively co-developing companion diagnostics to guide personalized therapies. Rather, all of the companies in that study said they were using biomarkers to help study targeted therapies in the discovery phase of the drug-development process — a fact that led Tufts to conclude that drug companies are primarily using biomarkers to learn more about the investigational agent being developed, rather than to inform "prescribing or monitoring requirements on the label." (PGx Reporter 11/21/2010).
Post hoc test development, Cohen argued, is often an iterative process — trying one idea and then another, while co-development's specificity encourages more ready acceptance. "You see that with the success stories," Cohen said. An oft-cited Rx/Dx success story is Roche's Herceptin. Representing the first simultaneous launch of a drug/diagnostic combination product, the FDA approved Herceptin for HER 2-positive breast cancer in September 1998, and at the same time, the agency also granted approval for Dako's HercepTest to gauge HER2 overexpression. The drug label also mentioned HercepTest by name.
Some post hoc developed tests have also been successful, Cohen noted, citing Novartis' Gleevec as an example. Gleevec was approved initially in 2001 by the FDA as a treatment for chronic myeloid leukemia. The drug's label was updated in 2002 to change Gleevec's indication as a treatment for patients with Philadelphia chromosome positive CML. Despite the lack of an FDA-cleared test, among oncologists it is already standard practice to molecularly monitor BCR-ABL transcript levels to confirm patients have Ph+ CML and to ensure patients aren't relapsing or having a refractory response to treatment. However, nearly eight years later, Novartis announced that it would develop a BCR-ABL gene transcript monitoring test (PGx Reporter 10/13/2010).
Regardless of the specific Rx/Dx development strategy, Cohen said there still seems to be a lag where payors are concerned, having noticed in other studies that co-insurance rates for drugs like Herceptin and Gleevec have been rising in spite of the fact that he considers them to have relatively good clinical and cost-effectiveness data.
"That has to change," said Cohen. "Every one of us talks about value-based insurance design, but I see very few payors in the US tackling this. Gleevec is known to be very cost effective, especially for patients with CML who have taken the test … Yet I see co-insurance going way up to 40, 45 percent for some," he said.
"If you really have a value-based insurance design, you would correlate somehow your co-insurance with the product that is being reimbursed."
Cohen said that ten percent of currently marketed drugs have pharmacogenomic information on the label while only one percent have a companion diagnostic. However, he is optimistic that this trend is slated for change, predicting that in the next decade between 5 or 10 percent of marketed drugs will be launched with a companion diagnostics strategy.
"If we have evidence now that post hoc is not the way to go, that will itself generate incentives for drug developers," Cohen said, citing crizotinib — an anaplastic lymphoma kinase inhibitor developed by Pfizer — as a good example of a co-development partnership. Pfizer has a deal with Abbott to develop a companion test that will select best responders to the investigational drug (PGx Reporter 5/25/2011).
"The regulatory authority is happy, patients and providers will be happy because of the regulation, and payors will be happy because all the evidence they have to sort out will already be there at the point of approval," he said. "If anything, that example of crizotinib, is probably a wave of the future," he said.
Overall, in releasing this report, Tufts hopes to temper some of the enthusiasm over personalized medicine, Cohen said.
"That we've come this far is amazing," he noted. "But moving forward we need all the key players involved. We have a way to go."
Have topics you'd like to see covered in Pharmacogenomics Reporter? Contact the editor at mashford [at] genomeweb [.] com.