Expert: Raise the Bar for Diagnostic Tools in Oncology

Daniel Hayes, MD

Although the liquid biopsy test is showing potential in some areas of oncology, more needs to be known about this emerging diagnostic tool to understand its true clinical utility, and tighter regulation of the diagnostics market might aid in the search for answers, Daniel F. Hayes, MD, said in an interview with OncLive^®.

Hayes and his colleagues once coined the term prospective-retrospective trial to describe a higher standard for the approval of diagnostic tests, which are not subject to the same rigorous assessment the FDA requires of therapeutics.

Higher standards are not likely to be popular with some in the oncology community, said Hayes, the Stuart B. Padnos Professor of Breast Cancer Research, professor of internal medicine, and codirector of the Breast Oncology Program at the University of Michigan Rogel Cancer Center in Ann Arbor. However, a more rigorous approvals process is the most direct way to meaningful answers to important clinical questions, he noted.

Hayes, a veteran in the investigation and treatment of breast cancer, focuses on evaluation and clinical use of tumor markers and cochairs the American Society of Clinical Oncology’s Tumor Marker Guideline Expert Panel. Below, he discusses the potential of liquid biopsies in breast and other cancers and highlights some major challenges that keep these tests from wider use.

OncLive: What are the current applications of liquid biopsies?

Hayes: Let’s start out with a definition. The term liquid biopsy was coined about 10 years ago to mean any sort of test you could do in blood that might give insight into what tumor activity is going on at the tissue level. It was specifically coined, frankly, to refer to circulating tumor cells [CTCs]. It was then appropriated by people who started doing work with cell-free DNA [cfDNA]. In my opinion, the term liquid biopsy fundamentally refers to a liquid biopsy that can tell you anything about what’s going on with the cancer, and that could be protein, CTCs, DNA, etc.

So far, the main clinical utility of liquid biopsies is to monitor patients, across almost all cancers, who have established metastatic disease to determine if they are getting worse or getting better. I exclusively treat patients with breast cancer, and that’s where most of my work has been, but people have used CA [cancer antigen] 125 to do this [monitoring] in ovarian cancer, CEA [carcinoembryonic antigen] in colon cancer, PSA [prostate-specific enzyme] in prostate cancer, and so forth. If a marker is going up, it usually means the patient is getting worse and you should think about restaging or changing treatment. If the marker is getting better or staying the same, that might indicate that the treatment you have them on is doing what you hoped it would do and you’ll keep doing it.

Then the question is whether, instead of just as a marker that tells us if a patient is doing well, we can use liquid biopsies the way we use tissue biopsies to actually predict what therapies might work. For example, in breast cancer, could we do biopsies to determine estrogen receptor (ER) status for endocrine therapy and HER2 status for HER2- targeted therapy? I think that is probably the biggest promise and has panned out at least a little bit in other cancers, although tumortissue biopsies are still used for confirmation, because the sensitivity is not great but the specificity is high.

The 1 other place that people have tried to use circulating tumor DNA [ctDNA] is in clinical trials of precision medicine. The data to support that, in my opinion, are pretty weak. Although it is interesting and provocative, we really don’t have good evidence that it works.

How close are liquid biopsies to attaining clinical utility?

I’ve been involved in a number of studies where we’re trying to get there, but this gets really shaky. If you want to get a new drug approved by the FDA, you have to prove it is safe and effective, and they usually require a prospective, randomized trial. A lot of work goes into proving that a drug works and is safe. The reason the drug companies do that is because the payoff—if they can get such a drug to market—is pretty big.

Diagnostics are not on the therapeutic side of the FDA; they are regulated as devices. There are 2 issues with this. On the diagnostics side, the FDA doesn’t necessarily require that you show that a new diagnostic actually improves clinical outcomes; you have to prove that it has high analytical validity, that the assay can be run well and reproducibly, but you don’t have to show it actually improves some sort of meaningful end point. A lot of assays approved by the FDA, including liquid biopsies, may not be worth using.

Also of concern is that about 30 years ago, the FDA allowed the development of laboratory developed tests, so if you developed an assay in your laboratory that is approved by the CLIA [Clinical Laboratory Improvement Amendments]—which is regulated by Centers for Medicare & Medicaid Services, not the FDA—as long as you didn’t sell the assay to anyone else and you performed the assay in your laboratory, then you could provide that to patients, and doctors could bill for it. A lot of big pathology labs make and use assays and provide the information to patients without ever going to the FDA.

That means that there are assays being used that really haven’t been tested for any kind of clinical utility or even analytical validity. There are also assays approved by the FDA that aren’t worth using. On the other hand, there are both LDTs and FDA-approved assays that are terrific. The problem is that it is hard to figure out which is which.

If we treated diagnostics the way we treat therapeutics, in that you’d have to go to the FDA and prove that they are effective, this would raise the bar significantly. But the payoff for diagnostics is so low that most companies don’t have the resources to do that. A number of us in the field have been saying for a while that we need to value diagnostics the way we value drugs, and we need to have FDA regulation over all diagnostics that directly determine how a patient with cancer should be treated, so they should be both safe and effective—improve some kind of meaningful end point—and we need to improve the reward for that.

There would be 2 ways to do that: Run prospective trials that directly test the assay—a variety of trial designs have been suggested for that—and use specimens that have been collected in a prospective clinical trial retrospectively to see if they improve patient care. About 10 years ago, I coauthored a paper in which we coined the term prospective- retrospective studies, which would be a lot different than just grabbing samples out of the freezer and hoping to get lucky, which is how a lot of retrospective studies are done. Rather, this would be a rigorous prewritten protocol and would probably take a couple of such studies to see if a test is improving outcomes. That has gained a lot of traction, at least among the guidelines panels. The issue is, should we have diagnostics be the Wild West and just warn users to be careful, or should we regulate them the way we do drugs and make sure they are safe and effective? Obviously, I agree with the latter, but it’s messy right now.

Will liquid biopsies eventually replace tumor tissue biopsies?

I think these may be complementary. A tissue biopsy allows you to get a lot more cells to study, especially in terms of mutational analysis. But you’re testing just 1 site of that patient’s total tumor volume, so they may have metastasis in their liver or lungs, and you’ve just got 1 site. A blood test obviously is a single window in time of a very small volume of blood, in which there is an even smaller volume of CTCs, ctDNA, or protein, but it may also represent the patient’s entire tumor burden. This is total conjecture; it still needs to be shown definitively whether this will predict outcomes, and I think we’ll see more and more studies of this down the road.

In your opinion, what would be the most exciting development with respect to liquid biopsies?

I think the most exciting would be a demonstration, with good scientific rigor, that using a liquid biopsy, whether done with CTCs or ctDNA, actually gives you the same kind of information that you get from a tissue biopsy and results in better outcomes for patients. A lot of people are working hard on that.

What are the most significant challenges to the development of liquid biopsies?

I think the major challenges are funding and support for the kinds of studies that need to be done and a more consistent regulatory environment so that necessary studies are performed. Otherwise, we will see more and more assays on the market that are touted to do this or that but don’t have much accompanying evidence that they can. My soapbox position on the need for more regulation is not necessarily accepted by many people in the field, but part of the problem with limited regulation is that it actually takes much longer to figure out if something works. If we had tighter regulation, we would get answers much more quickly, because people would do what was needed right away instead of trying to get lucky. I’m hoping that we’ll see markers that you can get out of a simple blood draw that really do the same thing as a tissue biopsy, but we’re not there yet.

Disclosures: Daniel F. Hayes’ laboratory receives funding from Menarini Silicon Biosystems, which develops circulating tumor cell and other cancer diagnostic technology, and his institution has a patent licensed to Menarini on which Hayes is named.