Incorporating liquid biopsies into clinical trials

Share :
Published: 4 Jul 2017
Views: 1866
Dr Daniel Hayes - University of Michigan Comprehensive Cancer Center, Ann Arbor, USA

Dr Hayes speaks with ecancer at WIN 2017 about ease of use and wealth of information made possible with liquid biopsy to derive tumour genomics.

He describes the utility of serial biopsies in guiding patients selection of treatment or trial participation, and also how interrogation of this data may determine suitable biomarkers for further research, or reveal the development of resistance pathways in real time.

Dr Hayes considers the use of circulating biomarkers as surrogates for clinical endpoints, and contemplates the historic context of personalised therapy from the first vaccines to the current understandings of tumour heterogeneity.

The question is how can we use liquid biopsies to fit into cancer research and we’re all pretty excited about that. Frankly the advantage of the so-called liquid biopsy is that it’s a blood test, as opposed to tissue which requires a biopsy. So biopsies are more expensive, they usually have to be invasive because although there are a few people who have lumps and bumps that are easily accessible most of them are obviously in the liver, the lung or whatever. So while that’s not impossible to do it becomes logistically and also maybe dangerously difficult to do serial biopsies. The other advantage of the so-called liquid biopsy is that it may give you a snapshot of the entire patient’s cancer load as opposed to a biopsy which is just the site you stick the needle in. In fact, I think these are complementary. We haven’t proven it yet but I believe what we’ll see are things we can pick up with a real biopsy and one advantage, of course, is you get many more cells. You get a billion cells in a biopsy so you don’t have to sequence as deeply, the cancer cells are right there, they’re not rare events. So that’s an advantage there whereas the liquid biopsy either the cancer cells or the cell free DNA are really rare events so you’ve got to sequence pretty deeply to get to them so they’re less sensitive than a true biopsy.

So these are going to complement one another; the issue is how do we use them in research. There are a number of ways. One would be because you believe that patients who don’t have whatever the marker is are so unlikely to benefit or need it that they shouldn’t put them on trials. We’ve been doing that for years. For example, in breast cancer we don’t put oestrogen receptor negative patients onto trials of endocrine therapy, a good example. The second would be that you actually use it to identify a marker that you think the drug will work. So that’s called predicting it will work and so that would be eligibility, again I’ll use the same example. We put ER positive patients onto clinical trials of new endocrine therapy because we believe it’s the target for the endocrine therapy. In fact we don’t believe, we know it’s true. So those are two different ways you could use it just to make your trials more efficient.

If you do trials in patients who either don’t need it or are unlikely to benefit from it you’ll dilute out the true effect and get a false negative so you don’t want to do that. Also you’ll spend a lot of money in patients where it won’t work and you expose those patients to drugs that won’t help them and that’s not the way we want to do things.

The second thing would be to use a circulating biomarker as an endpoint itself, as a surrogate. So, let’s face it, the most objective and meaningful endpoint is survival. Survival is very easy to measure but it takes a long time. So if you want to do a relatively quick study you don’t want to wait and see if patients are alive or dead. The next surrogate in from that is progression free survival. Progression free survival is actually difficult to measure in terms of looking at radiographs and that sort of thing. It would be nice if you had a circulating marker that either went up or went down and reliably predicted what was going to happen to the patient for the other two endpoints down the road that could be easily collected and measured. Again, I’ve gone through why we think liquid biopsies are good. This is also not unprecedented, for example in prostate cancer we’ve used PSA as a measure of response or not. In ovarian cancer we’ve used CA125 and so on and so forth.

Now increasingly we’ve been involved with looking at circulating tumour cells. It’s pretty clear that if you’re given chemotherapy and we look at circulating tumour cells one cycle in, if you haven’t reduced those down below a certain level you’re almost certainly not going to benefit from that chemotherapy and you need something else, as an example.

Then the third thing is whether or not you could actually use liquid biopsies sequentially to begin to identify the biology of why this drug isn’t working. So you might begin to see emergence of mutations or phenotypes, for that matter. Likewise you might see that in patients who do do well, they don’t have emergence of new markers and that might give us a way to explore whether or not other therapies could be added in or substituted for the drugs you’re using. That’s in the future but all of us believe it’s going to happen.

The final way that I believe they could be helpful is as a pharmacodynamic monitor in early trials. So, for example, if I have a down-regulator of oestrogen receptor and I have circulating tumour cells that make oestrogen receptor perhaps maybe within two weeks or a month I can begin to see whether or not I’m at the right dose for that drug if the oestrogen receptor is decreasing or disappearing in the circulating tumour cells that are left behind. That hasn’t been proven yet but it’s one of the ways many of us believe it’s possible. People have tried to do this in white blood cells with other kinds of drugs that downregulate things. The flip-side of that would be whether or not you believe the drug causes some change in the pattern of genetic expression and you could measure that going down. There are a number of people who are trying to do these sorts of things.

All told we can use these things in a very smart way to make our trials more efficient. That way we can get to the answer that our patients want which is are these going to make them live longer or feel better faster than the way we do things now.

It’s interesting to me, one of my friends, Mark Lipman [?], often says if you wait for ten years you can do what somebody else has done and call it something new and pretend that you made it up. Doctors have been trying to deliver personalised medicine with precision since the time of Hippocrates, this is not new. What’s new is we have better toys and better tools to deliver our personalised care with more precision. So unless you’re a public health doctor where, for example, sewers aren’t personalised care but they and vaccines have had the most impact on health, more than any drug we probably have if you think about it. But as an individual doctor we are trying to personalise our care with as much precision as possible. This is one step forward to becoming even more personal with more precision.

Ultimately the issue is whether we will get down to only just treating one patient with one drug or one set of drugs or still have groups. I think we’ll still have groups of patients, we’re not going to be able to get down to the individual that much. This gets into heterogeneity which has been recognised for 150 years as a big issue in cancer and, for the last 50 or more years, as a big issue in resistance to therapy. There are two issues, one is inter-patient heterogeneity so that one patient is in breast cancer ER positive, the other patient is ER negative. We call it breast cancer and yet those are two different diseases and I can go on and on and on. The other, for an individual patient, is intra-patient heterogeneity and, again, this is not a new recognition but we have better tools to begin to recognise. In fact, we’ve ignored this in many ways. We’ve known about it, it’s been written about for decades but we’ve just gone ahead and treated people.

I think we can do better so, for example, if we go back to liquid biopsies we and others have seen that in a patient who is called “oestrogen receptor positive” that individual circulating tumour cells some will be oestrogen receptor positive, some will be in-between, some will be completely negative, the same patient. We’ve known that in tissue but we can begin to see that evolve. For example, we’ve seen patients who 99% of their primary cancer is ER positive and really hot ER positive whereas by the time they have metastasis their circulating tumour cells maybe 10% are very hot and the other 90% are not and so on and so forth. We don’t know actually that that predicts what’s going to happen to those patients, that’s what we’re starting to do now is saying, ‘OK, now that we can do this, let’s do trials and see whether or not it actually predicts resistance to therapy.’ Better yet, begin to see if you treat those patients with endocrine therapy do they begin to evolve even more into ER negative cancers or, for example, the oestrogen receptor gene, ESR1, begins to develop mutations that make it resistant to one kind of therapy and not another.

I’m using breast cancer as an example because that’s what I do but we’re going to be doing this with every disease now. We’re going to be able to start sorting out the biology and heterogeneity of that biology disease by disease and case by case. That’s pretty exciting because it will guide us then into using our so-called precision therapy with more precision which is what we’re trying to do.