Skip to main content
Premium Trial:

Request an Annual Quote

Q&A: Winship Cancer Institute's Dan Brat on the Challenges of Using MDx for Cancer


NEW YORK (360Dx) – Earlier this month, Winship Cancer Institute of Emory University was designated a comprehensive cancer center by the National Institutes of Health, placing it in the top 1 percent of cancer centers in the US, according to Winship. 

Each year, Winship treats about 15,000 patients who come from Georgia, the Southeast, and elsewhere. Among the research programs at the institute are those looking at the underlying cell biology of cancers, cancer genetics and epigenetics, and therapeutics development. With the advent of molecular technologies, such as next-generation sequencing, and now liquid biopsy, Winship has also embraced the newest tools to more quickly and accurately diagnose patients, and then direct them to the most appropriate treatment. 

Dan Brat, director of the cancer tissue and pathology shared resource at Winship, and whose work is primarily in the diagnosis of primary and metastatic brain tumors and other brain-related diseases, recently spoke with 360Dx about the impact of molecular methods on cancer diagnoses. 

While new technologies are ushering in a "period of renaissance" in diagnosing different forms of cancer, he said, reimbursement issues may be discouraging labs from leveraging them. 

Below is an edited version of the interview. 


Can you describe the kinds of technology development being done at Winship?

Our medical laboratories are in a constant state of evolution, bringing in new platforms, new tests for targeted sequencing, for whole-genome analysis by copy number alterations. It's really an evolution, cancer type by cancer type, from sarcomas to leukemias to brain tumors.

They're all kind of evolving on their own with their own subtype expertise kind of driving that.

What do you see as the biggest obstacles when it comes to cancer diagnosis?

I think, actually, cancer diagnoses are really in a period of renaissance, and new techniques are being applied to specific cancer types to create molecular diagnoses really for the first time in just about every single type of cancer.

That has been done for quite some time in the hematologic malignancies. There have been molecular subtypes known within those categories of leukemia and lymphomas, but now other organ sites, including brain tumors, lung cancers, breast cancers all have known molecular subsets that are clinically relevant and treated differently.

The obstacles really are not the diagnostic platforms themselves. Some of the barriers are the payment for them. In many instances, laboratories throughout the country are hesitant to develop the most cutting-edge molecular diagnostic platforms and techniques because the payors, the insurance companies and Medicare, are not reimbursing the testing at a level that makes financial sense for hospitals to implement them.

Is this a matter of the technology being too far ahead of the payment mechanisms?

I think so. I think there's also a question of what level of molecular testing is reasonable, what level of testing is appropriate, and what level of testing guides clinical care – and whether we should be doing the testing as single tests or platform tests, like panel tests.

I think our diagnostic capabilities would have advanced much further and faster if there was a reimbursement schedule that took into account the cost of the testing and the cost of the interpretation of the tests by pathologists.

Does Winship feel as if it's protected from some of these challenges because it's part of a large healthcare system?

Yes, I think the fact that we are a large tertiary center that attracts cancer patients statewide, that the expectations for us to perform more advanced diagnostic tests leads to us implementing them more so than community-based hospitals and cancer centers.

I think we do a pretty decent job of offering the appropriate diagnostic test. I think the testing that we do is clinically appropriate. I think that centers like St. Jude [Children's Research Hospital], Mass General, or UCSF, and Washington U., those that are the highest-ranking research institutes, have been willing to use these large multigene panel approaches.

We do a very good job at providing molecular diagnostics at a level that is clinically justifiable.

But those medical centers that have been at the leading edge of implementing panel-based, next-generation sequencing [tests] are losing money by providing those tests. I'm not sure when changes in reimbursement will be coming, but I think it's going to be necessary for us to really advance precision medicine and cancer across the board.

Is Winship losing money with its NGS testing for cancer?

That's not a question I can answer with any level of detail. But I would say that on the professional side, certainly that means with the physician interpretation of the testing, that it takes more for the interpretation than is being reimbursed.

On the technical side, I think the reimbursement is also not sufficient for these multigene panels for next-generation sequencing.

How do you decide when to do single-gene testing versus testing using large panels?

The pathologists who are making the diagnoses are aware of the different cancer subtypes and the necessity for doing specific molecular tests to make a given diagnosis. And based on that, we implement specific molecular testing in our laboratory. In the case of brain tumors, for example, there is a major distinction between those tumors that have isocitrate dehydrogenase mutations and those that don't.

We have all of the testing necessary to do that type of diagnostic, both at the immunohistochemical and at the sequencing level. And then there are additional layers of molecular testing for other biomarkers that might be targetable, like BRAF mutations, for example.

When you say molecular testing, are you talking specifically NGS or some other type of technology?

It can be PCR, targeted sequencing where it's just a specific gene region that's known to have hotspot mutations, or it can be more generic and with some next-generation sequencing. We use both approaches, and we try to implement tests as they're needed clinically, and that's sometimes diagnostically oriented, and sometimes that's therapeutically oriented.

In the case of breast cancers and lung cancers, there are specific molecular subtypes that direct therapies.

How is Winship using molecular technologies versus non-molecular technologies?

For example, in brain tumors, one specific type of brain tumor, oligodendroglioma, is now defined as having IDH mutations and whole chromosomal arm losses of 1p and 19q. In order to make that diagnosis, you need to establish that there is an IDH1 or IDH2 mutation, and we have multiple levels of tests to do that, one looking at the mutant protein by immunohistochemistry, and following that, doing sequencing on the less common IDH1 and IDH2 mutations.

Then we define whole chromosomes arm losses of 1p and 19q, based on cytogenetic microarray of the entire genome, and that shows copy number losses of 1p and 19q that are whole arm and definitional of the disease.

That's one approach where we absolutely have to have those molecular tests in order to establish a diagnosis. And as the World Health Organization classification of cancers evolves to include specific biomarkers or molecular markers, then diagnosticians like myself will need to keep pace with those because it becomes part of the definitions of diseases, or a part of the established mechanism of establishing the diagnosis that directs therapy.

That occurs within each cancer type, so that there are subspecialists that diagnose lung cancers, breast cancers, prostate cancers, renal cell carcinomas, and each of those has molecular subtypes that are becoming known to be more and more clinically relevant.

As a clinician are you finding it increasingly difficult to keep up with advances in technology?

I think it has been changing dramatically. I would say that each diagnostician within their subspecialty domain is very aware of the evolution of the molecular subtypes within those cancers. The platforms are also evolving and hard to keep up with, and the folks in the molecular pathology labs and the cytogenetics labs, those are the experts that really implement the platforms and develop the testing strategies that go with them, as well as the bioinformatics pipelines that are needed to assess the molecular alterations.

It's much more than just buying an instrument and plugging it in and having it do its testing. There needs to be a fairly sophisticated bioinformatics pipeline associated with these advanced platforms.

It would be especially important, then, that the clinician and the pathologist are in close communication with each other as these tests are being performed.

Oh yes, that is exactly why these molecular tumor boards have evolved. The traditional tumor board includes a surgeon, an oncologist, a radiation oncologist, a radiologist, and a pathologist, as well as the clinical care team that go along with them. And they discuss patients on a traditional manner with traditional diagnoses.

These days, molecular diagnostics have gotten so evolved and complex that there is a need to have molecular tumor boards where the findings of the molecular testing are discussed in depth, and each of the findings are interpreted as well, because it's not always a black and white result.

Is the next-generation sequencing you're doing very targeted in nature?

In general, individual tumor types have a limited number of mutations that are relevant to classifications for directing therapy. We do not sequence a huge number of genes for clinical care. It's more of a targeted approach, in general.

Is sequencing done on a patient as a regular part of the diagnosis, or does it depend on the symptoms?

It can be both. Some mutations like p53 are very common across all forms of cancer or many forms of cancer and wouldn't necessarily be diagnostic. But in some settings — for example, a medulloblastoma with sonic hedgehog activation — p53 mutation discriminates between different types of disease that occur at different ages and progress with different courses. A lot of it is context-dependent.

Is the technology being used at Winship to confirm a prior diagnosis performed elsewhere?

In some cases, yes. We also have a large population of people who come to us for their primary workup of disease. That's probably the larger fraction — it's patients that are having their biopsies performed at our clinics, or their cancers are being resected within our operating rooms, or having their bone marrows aspirated within our ambulatory care areas.

And the diagnostics material comes to our pathologists, and we work them up primarily and give an initial diagnosis. That's the more common route. And the molecular pathology is done within that context.

Are you using liquid biopsies?

We have not implemented that type of testing. I think that is going to be the next wave of diagnostic testing in the molecular era. It's a fairly noninvasive way in order to evaluate disease progression and to evaluate any response to therapies, especially for brain tumors where there are significant issues [around tissue access] at recurrence or following therapy. I think it would be very beneficial to access the DNA of the tumor, either through the blood or through the cerebrospinal fluid.

Both of those approaches are being attempted. For other types of cancers, the yield is much higher than it is for brain tumors in the blood, and there are platforms available for those types of analyses for lung cancer or breast cancer, and the other types of cancers, I think, will soon follow.

The clinical question to be addressed would need to be validated, as would the test accuracy and the clinical utility. Just because you have the ability to sequence specific genes from cancer in the bloodstream, doesn't necessarily imply that the tumor has progressed or altered or responded to therapy. All of that needs to be validated, and it is being validated.

But I think the technical ability to do sequencing needs to be translated into better clinical care. That's the ultimate goal.

I think the interpretation of the tests that are returned from liquid biopsies need to be put in a clinical context to know if the cancer is responding to therapy, for example, is progressing or regressing, for example. All of those endpoints need to be established.

A recent survey found that doctors think genomic testing for cancer is important, but that the tests are overhyped. What's your view of these tests?

I would say that molecular testing for many forms of cancer is state of the practice at this point, it's state of the care. I'm sure that there are some institutions that have implemented tests that provide results that may not be clinically useful, but certainly overall, I consider molecular testing the standard of care in 2017.

Another survey found that doctors prefer using immunotherapies over experimental drugs based on results from an NGS test. What's your opinion?

I think that depending on the type of cancer, immunotherapies are becoming either an alternative or sometimes a first-line therapy. Certainly, in melanoma, some forms of lung cancer, it has become relatively routine if the specific markers are positive. That's not true for all cancer types.

For brain tumors, we really hope that immunotherapies will become more relevant and more widely used, but we still need to find those drugs that work in brain tumors, and thus far, we haven't found them.