Search tips
Search criteria 


Logo of biotechhealthLink to Publisher's site
Biotechnol Healthc. 2007 December; 4(6): 46–53.
PMCID: PMC2651720

A Wish List for Molecular Tests

BOB CARLSON, MHA, Contributing Editor


Companion diagnostics improve health outcomes, say developers, who are willing to work with payers to develop medical policies for coverage. Payers and clinicians want more evidence of long-term benefit. The FDA, meanwhile, is jostling for regulatory control. How long will patients have to wait? Part 2 of 2.

Every six months, The Zitter Group, the managed care consultancy, asks payers which criteria are most important in deciding whether they will pay for a new molecular diagnostic test. “FDA approval” tops the list.

“Payers live in constant fear of being branded as the stereotypical mid-1990s HMO that denies care, so what they seem to ultimately want is some FDA guidance saying, ‘This is appropriate, you can use this,’” says Thomas Baker, MPA, The Zitter Group partner and senior vice president for strategy and analytics.

Payers may get their wish, depending on how an epic behind-the-scenes battle over regulatory oversight of molecular tests plays out. But ask these same payers if FDA approval means they’ll cover a new molecular test, and they’ll say, “Not necessarily.” FDA approval may make it easier to get payer coverage, but it’s no slam dunk.

So what does it take to get payers to cover molecular tests?


“Are you familiar with the TEC criteria?” asks Frank Dubeck, MD, vice president and chief medical officer of medical policy at Excellus BlueCross BlueShield, in upstate New York. Dubeck is referring to the five criteria developed by the Blue Cross Blue Shield Association Technology Evaluation Center (TEC) and also adapted by Excellus BCBS. “Those are the basic tenets of what we look for in evaluating new technology [when making decisions about] whether we are going to cover it.”

This approach also carries considerable weight outside the BCBS Association. TEC is one of 13 evidence-based practice centers for the U.S. Agency for Healthcare Research and Quality (AHRQ).

“We spend other people’s money, so before we make a decision to pay for a new technology, we want to be very sure that we’re getting value for that expenditure and not just a new Star Wars, whiz-bang, gee-whiz, neat gizmo,” says Dubeck.

Staying on top of such new technologies as molecular testing is a full-time job for Dubeck and seven MSN-level nurses. Coverage decisions are made by the medical policy committee, which consists of 27 network physicians and 18 physicians employed by Excellus BCBS. Those decisions are communicated to providers and members in the form of medical policies via the Excellus BCBS Web site.


That necessary-but-not-sufficient payer regard for regulatory approval is spelled out in the first Excellus BCBS Medical Policy Guideline, which cautions that “… governmental approval does not necessarily mean that [a service] is of proven benefit or appropriate or effective treatment for a particular diagnosis or for a particular condition.”

Right now, molecular tests such as Genomic Health’s Oncotype DX Breast Cancer Assay and RedPath Integrated Pathology’s Path-FinderTG cancer test may be either FDA cleared or approved as in vitro diagnostic devices, or come to market as a laboratory-developed test (LDT) regulated under the Clinical Laboratory Improvement Amendments (CLIA) of 1988. Both Genomic Health and RedPath IP have taken the CLIA regulatory path for their LDTs.

But that may change.

“If you look at the FDA’s draft guidance on in vitro diagnostic multivariate index assays,”1 says Emanuel F. Petricoin, PhD, codirector of the Center for Applied Proteomics and Molecular Medicine and professor of life sciences at George Mason University, “I think it’s pretty clear that the FDA’s concern is that when you have tests like Oncotype DX where the algorithm is proprietary so that no one can independently validate it, then there’s a ‘black box’ component to the assay. The FDA appears to want to regulate that to ensure that the assay provides for the safest and most effective treatment.”

Petricoin’s CV includes seven positions of increasing responsibility with the FDA, the last as senior investigator in CBER’s office of cell, tissue and gene therapies from 2003 to 2005. This gives him an insider’s perspective on the agency’s molecular testing policy initiatives.

An external file that holds a picture, illustration, etc.
Object name is f1-bth04_6p046.jpg

Persuading payers of a molecular test’s value is a process similar to that which must be done for pharmaceuticals, says Kimberly Popovits, president and chief operating officer of Genomic Health. It requires “a lot of up-front clinical validation and development, much like you would do with a drug,” she says.

“I think the multivariate assay issue is important and a dialog between the diagnostic community, CMS, and the FDA is necessary. We can’t have a situation where a decision as to whether someone should get a drug is made based on an assay that is so complex that it can’t be explained to patients, physicians don’t understand how the result was determined, and there’s no way to independently validate the result,” says Petricoin.


Paul W. Radensky, MD, JD, is an attorney based in the Miami office of McDermott, Will & Emery, an international law firm that is one of the counsels for the Coalition for 21st Century Medicine. According to its Web site, the Coalition —which came into being shortly after the FDA’s two guidances were issued — represents “some of the world’s most innovative diagnostic technology companies, clinical laboratories, researchers, physicians, venture capitalists, and patient advocacy groups — all linked by a common mission to develop advanced diagnostics that improve the quality of healthcare for patients.”

“The FDA maintained for years that it wouldn’t regulate these types of tests,” Radensky notes. “About a year and a half ago, it started issuing letters to a number of small laboratories saying, ‘Are you aware of the medical device rules? We want you to come in and talk with us.’ And then this [IVDMIA] guidance document was issued.”

Radensky, who focuses on the IVDMIA guidance, says that the assays are not defined in the Food, Drug, and Cosmetic Act or any other regulation. In effect, the guidance is being promulgated as if it were a rule, even though guidances are not supposed to be binding. ThE Health and Human Services secretary has reportedly stated that the FDA would not enforce the IVDMIA guidance before any final policy.

“That’s why a key concern raised by the coalition, and others who commented on the draft guidance, was the need to proceed under notice-and-comment rule making if the FDA wants to proceed in this area,” he says.

Petricoin thinks there’s more to these guidances than FDA concerns about so-called “black box” molecular tests. Many of the markers not only have an impact on diagnosis, they also have an impact on therapeutic tailoring, treatment monitoring, and discovery of relapse, he says. “This is not about assays that are a better ‘mousetrap’ for diagnosis, a better prostate specific antigen test. We’re talking about assays that can be used to guide or decide therapy. The assay becomes a multi variate test using a biomarker fingerprint, which, many believe, falls under the FDA, because the FDA regulates therapeutics.”

What could help remedy the situation, Petricoin says, would be an assay that measures a handful of markers and doesn’t require sophisticated algorithmic-based informatics. “Instead of a 70-gene fingerprint test for therapeutic guidance, a one-or two-gene or protein test that has the same predictive accuracy potentially could go a long way to assuaging concerns.” Petricoin points to cholesterol testing to guide treatment decisions as an example. “We need more biomarkers like that.”


Excellus BCBS policy guideline II.B states that the medical literature must provide conclusive evidence that the service has a positive effect on health outcomes. “Where we usually get into differences of opinion with a new technology is the positive effect on health outcomes,” says Dubeck.

Oncotype DX, also known as ODX, predicts which breast cancer patients are more or less likely to respond to cancer therapy based on a “recurrence score” of 0 to 100 generated by the test. That’s not the same as saying that the test saves lives or even extends survival.

“We still need studies where Oncotype DX is actually being used to make treatment decisions, and test subjects have agreed to the protocol and will abide by whatever arm they get put in,” says Dubeck. “So often, that’s what’s missing in some of these new technologies.”

In fact, in the TAILORx (Trial Assigning Individualized Options for Treatment [Rx]) breast cancer trial, sponsored by the National Cancer Institute and underway since May 2006, all subjects undergo Oncotype DX testing. Patients with low scores will receive adjuvant hormonal therapy, and high-score patients will receive adjuvant chemotherapy plus hormonal therapy. In other words, for patients with low or high recurrence scores, ODX is used to make treatment decisions.

According to Excellus BCBS policy guideline II.C, the benefits of a new technology also must outweigh any harmful effects. In other words, it must improve net health outcome. For example, new diagnostic tests are typically more sensitive than existing tests and may generate more false positives, which could result in needless procedures and complications.

“How bad is the disease we’re trying to detect, and how bad is the disease we create by using this more sensitive test?” Dubeck asks. “We’re on the cautious side in that we want to see more evidence over the long term that there really is a benefit —that it’s worth the money.”


Excellus BCBS policy guideline II.D requires that new technologies at least equal the effectiveness of established technologies, or offer an appropriate alternative when there are no other options.

“When we look at technology that essentially does an old trick in a new way, we have to ask where the value is,” says Dubeck. “Will more people get the appropriate test because they won’t be turned off by the noxious elements of the test?” For instance, he says, colonoscopy is more readily acceptable than sigmoidoscopy because, typically, people are asleep when it is done.

Finally, guideline II.E says that a new technology must prove itself in the real world, and that proof must be documented in the peer-reviewed medical literature. Dubeck cites computer-aided detection (CAD) for mammography as an example of a new technology that looked good on paper, but is now questioned based on real-world experience.

About five years ago, a large, well-controlled study published in a prestigious medical journal found that smaller tumors were detected earlier by CAD than by experienced radiologists. Expert consensus quickly coalesced, and Excellus BCBS decided to pay extra for CAD. But a multicenter study published this year found that CAD did worse than unaided mammography. Exactly why that is probably will be reported in yet another study.

“This technology looked good in a research setting where people were vested in the outcome, but you put it in day-to-day practice in community radiology departments over four states, and you find out that it’s not so good,” Dubeck says.


Other payers may use a less structured evaluation methodology or rely on independent technology evaluation organizations such as ECRI, Hayes, or McMaster University in Ontario. Moreover, payers don’t want to lead the parade in evaluating and covering a new technology such as molecular testing.

Launching any new product in this environment is complicated. But for a molecular test, the challenge is unprecedented, because molecular testing is a radically new way of doing diagnostics, with development costs in the hundreds of millions of dollars and long timeliness — but also with greater potential value for patient care.

The economics of molecular diagnostics work much like that of the pharmaceutical business, according to Kimberly Popovits, president and chief operating officer of Genomic Health in Redwood City, Calif., who led the 2004 launch of Oncotype DX, one of the first and perhaps most successful molecular test to date.

“We were definitely blazing a new trail, because it’s really a new business model in terms of value-based pricing in diagnostics,” says Popovits, who has been doing this kind of work for 20 years. Traditional diagnostics are relatively low cost, low margin, she says, “so you don’t do the type of extensive work that we did in the payer community. This product required a lot of upfront clinical validation and development, much like you would do with a drug.”

That work with payers began in 2002, two years prior to launch, and focused initially on big payers, like Cigna, Aetna, UnitedHealthcare, and Medicare, that could potentially influence others. Specifically, Genomic Health wanted to know what data medical directors and medical policy committees needed to cover a test like Oncotype DX. The answers were:

  • Clinical evidence published in respected, peer-reviewed journals;
  • Significant demand from oncologists;
  • “Evidence that the product would actually change clinical decisions, not just add cost to the system,” says Popovits.

In effect, these payer “must haves” became the “to do” list for Popovits’ team of national managed care account managers, a medical education group, marketing and sales staff, and R&D staff.

The New England Journal of Medicine published the Oncotype DX validation study in 2004. Follow-on studies were published in 2005 and 2006, each of which required CMO-level coordination with the NCI, Cancer Cooperative Groups, and Kaiser Permanente of Northern California. And two additional studies with practicing breast oncologists were conducted in 2006, which established that Oncotype DX results change treatment decisions 25 to 30 percent of the time.

Early dialogue with payers also paid off with an unexpected benefit. Today’s Oncotype DX quantifies the likelihood of breast cancer recurrence in women with newly diagnosed, early-stage breast cancer, and also assesses the likelihood of chemotherapy benefit, based on the expression of 21 genes in tumor tissue. The initial test validation study measured the likelihood of cancer recurrence, but a subsequent study showed that the test could predict the chemotherapy benefit.

“If our test had answered only the question of likelihood of a recurrence,” says Popovits, “I don’t think that payers would have felt that it brought $3,500 worth of clinical utility and value to the system.”2


RedPath IP’s PathFinderTG isn’t as far along in covered lives as Oncotype DX, but promises equally dramatic advances in resolving diagnostic dilemmas, says Rina Wolf, MHA, vice president of reimbursement and payer relations at RedPath Integrated Pathology in Pittsburgh. In up to 20 percent of cases, the cancer is not visually obvious, leaving pathologists and clinicians with biopsy specimens that are “indeterminate, atypical, suspicious, equivocal, and nondiagnostic,” Wolf says. “When front-line pathology fails, patients find themselves cycling through processes of second opinions, testing and retesting, delayed treatment, and, sometimes, unnecessary or inappropriate treatment. PathFinderTG resolves those dilemmas with unprecedented accuracy so that the best treatment can begin.”

PathFinderTG is the result of 17 years of research and development and, according to Wolf, is clinically validated at “an average 95 percent accuracy rate across multiple organs, specimen types, and diagnostic questions for cancer.”

An external file that holds a picture, illustration, etc.
Object name is f5-bth04_6p046.jpg

A technology might look good in a research setting, says Frank Dubeck, MD, of Excellus BCBS, but day-to-day practice may prove otherwise.

Using a broad panel of genetic markers, microdissection techniques, and proprietary DNA amplification techniques, PathFinder TG integrates quantitative genetic mutational analysis and traditional pathology to generate diagnoses for organ lesions and cancers.

Before PathFinderTG, for example, there were two options if pancreatic cancer was suspected — watchful waiting, which could result in a delayed diagnosis, or the Whipple procedure (pancreatoduodenectomy), which is complex and difficult on the patient, Wolf says. Recovery time is long and may force a patient into diabetes. A high percentage of patients that undergo Whipple procedures turn out not to have pancreatic cancer, says Wolf. “Now clinicians can order a PathFinderTG to find out definitively and treat accordingly.”

The brainchild of RedPath’s founder and chief scientific officer Sydney D. Finkelstein, MD, topographic genotyping (the TG in PathFinderTG) is the subject of over 500 papers, of which Finkelstein co-authored 140. Medicare coverage in June 2007 was a big step for PathFinderTG because 40 percent of RedPath’s patients are eligible for Medicare. Both Oncotype DX and PathFinderTG Medicare claims submitted by laboratories are paid by the regional Part B contractor where each laboratory is located. For RedPath, that’s Highmark Medicare Services in Pennsylvania.

Wolf declined to estimate market penetration for PathFinderTG, launched in 2004, but said that it is being used in major cancer centers around the country. Like Genomic Health, RedPath still fights for reimbursement of individual claims (the list prices for a PathFinderTG test are $4,000 and $4,500, depending on specimen type), but with plans for a bigger sales force, Wolf may soon spend more time on developing medical policies with payers.

An external file that holds a picture, illustration, etc.
Object name is f2-bth04_6p046.jpg
An external file that holds a picture, illustration, etc.
Object name is f3-bth04_6p046.jpg
An external file that holds a picture, illustration, etc.
Object name is f4-bth04_6p046.jpg

Excerpts from Excellus BlueCross BlueShield’s medical policy


To evaluate services like Path-FinderTG, payers have gone to the typical pharma model — clinical studies that follow patients for 10 years to see what happens, says Wolf. “But that just doesn’t make sense anymore, because while you’re waiting for those study results, millions of patients could have been helped, and after 5 or 10 years, the technology has changed again.” With formalin-fixed, paraffin-embedded tissue and outcomes data, she says, payers can get answers far more quickly as to whether something is appropriate.


In August, Excellus BCBS published its policy establishing when Oncotype DX would be considered medically indicated and payable. Previously, Excellus BCBS had considered the test not medically necessary. The policy followed publication in early July by the BCBS Association TEC of the conditions under which Oncotype DX meets TEC criteria. The BCBS Association medical policy panel is currently reviewing its position, Dubeck says.

The “black box” technology behind Oncotype DX doesn’t faze Dubeck, and Genomic Health may have already defused that criticism. In an e-mail to BIOTECHNOLOGY HEALTHCARE, Popovits writes, “We have published our algorithm and all the data behind Oncotype DX. It really isn’t a black box. We believe this approach, however it is ultimately regulated, is both viable and necessary for the future of the field.”

In addition to publishing the ODX algorithm in peer-reviewed journals, Popovits says the test was used by a laboratory to validate its own method for measuring gene expression, with the study results published. Also, Oncotype DX is included in American Society of Clinical Oncology guidelines as a new topic: Multiparameter Gene Expression Analysis for Breast Cancer.

Because so few claims have been received, Dubeck says Excellus BCBS hasn’t even begun to work on a policy for PathFinderTG. For its Medicare, commercial Medicare, and Medicare Advantage lines, Excellus BCBS uses Highmark’s coverage criteria.

“If we got a commercial Path -FinderTG case, the medical director would see we don’t have a policy, he’d reach out to our med policy staff, who would in all likelihood find the Highmark policy on the Web, and we would have that as a basis for making a decision,” Dubeck explains. “Because we’re not seeing a lot of volume, it would be a case-by-case review.”

Responding to the 500-plus papers published on PathFinderTG, Dubeck says, “Just going by the volume of articles doesn’t do a lot for me, especially if they are uncontrolled case studies. When the technology evaluation committee does its reviews, they pick these things apart in terms of study design, methodology, and how many patients were lost in follow-up, which clouds their reliability. If they aren’t judged to be good studies, they’re not even considered.”

He adds that the biggest hurdle for new technologies is demonstrating significant long-term benefit. Depending on the natural history of the disease, that could take as long as a decade.

“If TEC came out with a review today, my guess would be that PathFinderTG would fail on the long-term benefit issue,” says Dubeck. Wolf counters that RedPath is in the process of working with its cancer center clients to gather and evaluate their PathFinderTG outcomes data and the impact on patient management.


“The molecular profile that we use most in our practice is Oncotype DX,” says Paula Ryan, MD, PhD, a breast oncologist at Massachusetts General Hospital Cancer Center and director of the breast and ovarian cancer genetics and risk assessment program. “It helps me in deciding whether to give chemotherapy in addition to endocrine therapy, which would be the population you’re dealing with here.”

Ryan is less enthusiastic about Oncotype DX recurrence scores in the intermediate range, which still leaves the decision about chemotherapy up to the physician and patient, but she also is confident that more precise tests are in the pipeline.

Not coincidentally, the bigger question for Ryan also figures prominently for payers like Dubeck.

The reason payers have been hesitant about tests, Ryan says, is that it hasn’t yet been proven clinically that molecular markers translate to disease-free or longer overall survival. “I’d like to see those data, and that really has to come from the proof-of-principle studies, which are classically prospective, randomized trials like TAILORx. These trials are really going to help seal the deal, so to speak, in terms of whether these prognostic signatures will really improve outcomes.”


1The FDA published draft in vitro diagnostic multivariate index assay (IVDMIA) guidances and “Commercially Distributed Analyte Specific Reagents (ASRs): Frequently Asked Questions” in Sept. 2006.

2The Oncotype DX list price of $3,650 supports ongoing development of the assay, as well as the development of tests in the pipeline for colon and prostate cancer.

Contributing editor Bob Carlson writes exclusively about health-care. He lives near Zionsville, Ind.

Articles from Biotechnology Healthcare are provided here courtesy of MediMedia, USA