Loading summary

Transcript35 lines

[00:03]
A
Hi, I'm Derek Angus, a senior editor at JAMA and host of Healthy Dialogue, a new podcast from the JAMA Network. Join me as we go beyond the latest discoveries with nuanced in depth conversations with the world's leading experts to explore the most pressing issues in health and healthcare, from trends in autism diagnosis to private Equity acquisition to AI and much, much more. Visit JamaneTWORKaudio.com or search Healthy Dialogue wherever you get your podcast to subscribe.
[00:48]
B
Welcome to listeners from around the world and thank you for tuning in to this JAMA Clinical Reviews podcast. I am your host today, Dr. Mary McDermott, deputy editor of JAMA, and I'm here today with Dr. Ben Parks, who is a professor of medicine at Vanderbilt in the division of Hematology Oncology and a physician scientist focused on the study of breast cancer. Dr. Park is also director of the Ingram Cancer center at Vanderbilt University. And today I'll be talking to Dr. Park about his JAMA translational review on the topic of liquid biopsies. Welcome Ben. It's so nice to have you.
[01:32]
A
Thank you so much, Mary. It's great to be here.
[01:35]
B
Let's get started. Could you tell us how did you get started studying liquid biopsies? Can you give us a little of the historical background, please?
[01:43]
A
Yeah, so probably about 15 or more years ago, I was at Johns Hopkins at the time and I stayed there after doing my postdoc with Dr. Bert Vogelstein. And Bert had actually just proven that you could find cancer DNA molecules in the blood of patients with cancer. And that was such a huge thing and many people don't realize it, but that really set off this whole field. It even set off next gen sequencing as we know it today for looking at tumors, genetic alterations, et cetera. So we got in on the ground floor because my lab was literally on the ground floor above his lab. And we started working in this field to develop assays to pick out cancer mutations, mutations of DNA floating in the blood. So liquid biopsies can mean a lot of different things, but we in the circulating tumor DNA business tend to think of it as the DNA derived from cancer cells that we can now query find mistakes that we can often target with drugs. But then this has evolved into a newer paradigm of how do we actually use this to see if someone's cured or not. And that was something when we did the first study in metastatic breast cancer patients and saw how well this worked because it's so challenging. Most of the DNA that we call cell free is from the normal cells, so it's proverbial needles in the haystack that we're looking for. And I was really blown away at how accurate this was and how precise it was. And it started to get me thinking that if we could actually drill down to early stage patients with breast cancer and also just other solid tumor malignancies, that's a big conundrum right now in oncology. We could figure out who's actually cured versus who is not. And that affords an opportunity to not treat everyone with all the drugs, because we don't know who's cured and who's not. And it also affords the additional opportunity of giving people who aren't cured more drugs or put them on clinical trials in an effort to cure more patients. That's where the whole concept came in to really change the paradigm of how we give additional therapies, what we call adjuvant systemic therapies, in early stage patients with cancer. And that's kind of one of the biggest conundrums we have right now, especially in breast cancer. In my field, we know from historical studies 70% of patients after surgery, let's say, are in fact cured. We just don't know who they are. And so we treat everyone if they fell into clinical trials eligibility, knowing that 70% of the patients are being exposed to chemotherapies that they don't need, we just don't know who they. But we're also curing an additional 15 to 20% more patients. So the risk benefit ratio heavily favors. Let's treat everyone, recognizing that we're also doing some harm unintentionally. And now we finally are at a point 15 years later where we can start changing that paradigm. It's super exciting.
[04:39]
B
To summarize, liquid biopsies are, for example, blood samples where you're looking for small DNA fragments from cancer cells, correct?
[04:50]
A
Yep.
[04:51]
B
And how do they differ then, say, from one of those pan cancer screening tests?
[04:56]
A
Yeah, you know, they're very related. First, I should take a half step back. Probably for the last 10 years or so, we've had liquid biopsies, but mostly for metastatic disease, where the overall tumor burden is much higher. So there's a lot more circulating tumor DNA that gets shed into the circulation. Those are mostly used qualitatively, meaning that I said this earlier. If you have a particular gene mutation, you're looking for that, we have a drug, we can then say, aha, that person has that mutation. So they're eligible to get this drug, this targeted therapy. On the other hand, what we're doing now in the early breast cancer space that I just talked about is figuring out in a quantitative way, is there cancer still left behind? And if so, can we track it and monitor response to therapies, or are they recurring? And can we get enough lead time to prevent that recurrence from being true overt metastatic disease? Because if that happens, we can't cure the patient anymore. Ironically, we only cure metastatic disease when we can't see it. But now we can see it. And getting to your question, the difference is there's two kind of flavors, if you will, of these types of circulating tumor DNA assays. There's what we call tumor agnostic or tumor uninformed. Those are actually looking at DNA molecules again that are free floating in the plasma. And we end up looking at the whole kind of gamish, if you will, of DNA. And there are subtle changes in cancer DNA, both at the genetic level, meaning mutations and other types of genomic alterations. But even what we call epigenetic alterations, meaning that cancer DNA is methylated at certain residues that are very distinct than normal cell free DNA. So companies now, scientists, academics, we're working together to figure out can we use a more tumor agnostic approach to actually say who's got cancer and who doesn't? As a general screening tool, we're not quite there yet. Many people on this podcast probably have heard of the Grail Test Gallery. The results weren't quite as what we wanted to be for specificity sensitivity, but. But I do think with improved technologies, we will get there someday. On the other hand, what I was talking about earlier and what we have done in the breast cancer space recently and many others is actually looking at tumor informed circulating tumor DNA assays. What that means is we have to have a specimen of the patient's cancer, we sequence it and we find a lot of mutations. The company that we partnered with looks at 200 to 1800 mutations per tumor. And that affords the highest level of sensitivity and specificity so that we can detect literally one cancer DNA molecule out of a million normal wild types. And that's really been game changing. We're now calling it the ultra sensitive level of testing. But that has allowed us to really separate out, hey, are you going to recur or do you have like a 95% chance of being cured because you're negative? That's where we are right now. And I think that's the exciting part because we can really start changing how we're going to treat people and do studies in the future.
[08:06]
B
Can you give us an example of a liquid biopsy that is currently in clinical use and walk us through exactly how it's used? You mentioned earlier that it can be useful after a surgery for cancer. Your manuscript talks about using it after chemotherapy. Maybe just give us one example of a currently clinically used liquid biopsy.
[08:27]
A
Yeah, so again, I'm going to shift and just talk about early stage because there's been a lot of work already done in the field of metastatic disease where we match drugs with a mutation, but in early stage disease. Right now this is a little bit controversial in our field because we have not definitively proven clinical utility, meaning that if we get a result back, can we definitively say that acting on this is going to lead to a better outcome? We're designing those studies. What we've proven is validation. That said, because of the nature of these tumor informed assays being so specific and so sensitive, many of us, myself included, have been very thoughtful about patients who might benefit from this. And I have one example that I'd love to share is a young woman, mother of two. I've treated her for breast cancer that had a specific receptor called HER2 and it was positive. Now, there's a lot of drugs we give after surgery for ERBB2 positive disease, and sometimes we give them before, but the premise again is to get rid of microscopic cells if they exist, cells we can't see, that if left untreated, are going to come back as incurable metastatic disease. So in this particular instance, this patient completed all her therapies and we ordered the test and she still was positive. Very low level, 3 parts per million, as we call it. But that's again, because they're so highly specific. That's real. That meant she was not cured. And I had in the back of my mind, if I had found something like that first. We scanned to make absolutely sure there's no evidence of overt metastatic disease. And unfortunately there was not. But there's another drug. We don't use that much, but it is approved in this setting to cure more patients. It's got a lot of toxicity and that's why we don't use it that much anymore for HER2 positive disease. But it is there. So I put her on that. She did have some toxicities with it, but in six months she cleared and it's prescribed for one year as adjuvant systemic therapy. And then after another six months, still clear, she's at 18 months now, doesn't have any sign of CTNA positivity that to me is a very compelling story that we can recapitulate that paradigm for patients who otherwise would have recurred and unfortunately would have probably relapsed with metastatic disease. And I feel very excited that this is something that is going to really change how we think about and administer therapies for early stage cancers.
[10:56]
B
Yeah. And your review, in addition to talking about situations like that, where you might begin treatment when otherwise you wouldn't have known to begin treatment, you talked about how this could be used to minimize toxicity.
[11:10]
A
Yes, absolutely.
[11:11]
B
Can you give us an example of that?
[11:12]
A
Yeah. As I was saying earlier, right now our paradigm is to do standard of care because they've been vetted in big clinical trials. But you can imagine how those trials were developed. There were, let's say, 1,000 patients with cancer in one arm, another thousand patients with cancer, very similar types of cancers, H stage. And after surgery for both arms, then one group got placebo and the other group got chemotherapy. And, and we follow those patients out year after year, decade after decade. And the group that got chemotherapy does better. There's probably about again, 15 to 20% or more patients alive. And that's significant because for a disease like breast cancer example, 300,000 new cases diagnosed every year just in the United states. So even a 5% difference translates into thousands, if not tens of thousands of additional lives that are cured. But what about all those 70% of patients, 700 patients in the study that didn't need the chemotherapy? Heretofore, we've never been able to identify them. So we are now planning de escalation trials where we're going to see if the absence of circulating tumor DNA is a good enough indicator to say, you know what, now the odds weigh that chemotherapies not going to be in your favor. It'll probably do more harm than good. That's where we're headed now. That said, outside of a clinical trial, I don't think we should ever de escalate therapies. We can do it in a clinical trial context because as you know, we have interim safety analysis points. And if we see signal early on that this isn't the right thing to do, we stop it. That's the best ethical way to do a de escalation study. So right now, out of a clinical trial context, I like to say that de escalation would probably be considered bad medicine at best and maybe even malpractice at worst. So we just don't have that data yet. And that's where the controversy Is right now between where we are in this space for clinical validation versus clinical utility.
[13:11]
B
And do we have data on sensitivity? You gave the example of your patient with three parts per million.
[13:19]
A
Yeah, yeah.
[13:20]
B
And so is there no false positive?
[13:22]
A
There is almost none. And if there are, it's probably a technical issue. And that's because again, there's so many mutations that are keyed in to that patient's disease. I actually have to counsel patients before I order any of these tests. It's like a genetic counseling talk for about 15 to 20 minutes. Because that also has its downsides. One is that I have to remind patients because of that tight sensitivity specificity with the original tumor, if the patient ever got a different cancer or a second new breast cancer for my clinic, this test would never pick that up because it will only pick up the original tumor that it sequenced and has these unique mutations. Every cancer is different, just like every person is different. And so that's why it would never pick up a new cancer, breast or otherwise. The other caveat that a lot of people I don't think have necessarily thought through, these so called ultra sensitive tests are a second generation test. There was a first generation test that was one in 10,000, not one in a million in terms of technical sensitivity. But remember these tumor informed assays, you need tissue and that's an exhaustible resource. So I have had patients who wanted to get the tests. I went through my whole kind of counseling with them. We ordered the test but there was no tissue left and they had had a previous first gen test done. And I remind patients, you know, there's probably going to be a third gen test in the next five to 10 years. And if we order it today, you might not have tissue for that test in the future. It's just one of these things that there's a lot of caveats and unknowns. And I think unless you're really well versed in this field, I would counsel patients and providers to make sure that they speak with someone as a second or third opinion who actually knows this space and knows the limitations and the caveats.
[15:07]
B
Your review talks about some cancers these liquid biopsies work better for than others. What cancers does it not work well for? And why would that be?
[15:18]
A
Yeah, so we're learning a lot as we go along. But there are some cancers that we call low shedders, meaning that they're not usually as proliferative and therefore they're not breaking down there and having a lot of apoptosis. That's where we see a lot of the circulating tumor DNA go into the cell free space into the plasma. And so my disease subtypes of it. For example, the most common type of breast cancer is what we call hormone receptor positive, HERB2 negative. They tend to be very slow growing. They tend to be very sensitive to endocrine therapies and not so much chemotherapies. But they also tend not to shed as much circulating tumor DNA as a result. So those are patients where doing serial measurements can sometimes mitigate that effect. But we and others are working on strategies to try to mitigate those types of concerns because that's ultimately what limits the sensitivity. It's not necessarily the technical part per million, it's really how much DNA you can get safely from a patient's blood and assay that.
[16:17]
B
What are some of the challenges associated with use of these liquid biopsies?
[16:21]
A
I think there's always the challenge of misuse where again providers and patients may not understand their limitations. And that also worries me and concerns me that they could be used inappropriate, as I mentioned earlier, to de escalate therapies where we don't have good safety data, we have zero safety data really to actually be able to say and do that. I think there are other things that we still have challenges technically. As I mentioned, sometimes you just don't get enough DNA from patients blood. The variability from person to person, probably even from day to day from the same person can be orders of magnitude. And believe it or not, even though we've been working on this for well over two decades now, we don't really understand the biogenesis of how cells secrete out this DNA. We do know that if your cells are dying, that will liberate out some more. And there have been really nice studies showing that if you give chemotherapy, for example, and it bursts the cancer cells, your CTDNA will spike up. But beyond that, there hasn't been a lot of work done on that. And that's what I had mentioned earlier. We and others are really trying to make some discoveries and leverage those so that we can maybe even someday give drugs, believe it or not, that will elevate the amount of circulating tumor DNA to make all tests more sensitive. And that is a clinical problem right now, where even the best current 2nd gen tests still have a false negative rate.
[17:45]
B
Thinking ahead, say three decades from now, what's your vision of how liquid biopsies might be used to manage cancer?
[17:53]
A
Yeah, so again the qualitative aspects, we could probably find all the mutations and mistakes just using a tube of blood as a True. Liquid biopsy. But then I'm most excited about what I mentioned earlier. We could use this to individualize everyone's therapy. We know that adjuvant therapy, chemo, hormone therapy, antibody therapies work to cure more patients in early stage solid tumors. But we also know that comes at a price of over treating all these patients that heretofore we had no idea how to test whether they're cured or not. This gives us this precise lens into who is cured and who's not. And so for us, for a disease like breast cancer, I imagine you could actually take blood serially after each component of therapy and make a decision, do I need to get the next component? So as an example, do surgery. If that patient now clears the CTDNA that was positive before surgery, we won't have to do radiation, we don't have to do chemo, we don't have to do five to 10 years of endocrine therapy. It will be game changing. The other thing that I think it's really going to accelerate is clinical trials in early stage disease. Getting back to my original kind of analogy of how we do clinical trials in the early stage, we have to wait year after year, decade after decade, to show that in that example, the group that got chemo does better, significantly than the group that doesn't. Now imagine instead, and we enroll thousands of patients, sometimes even more than that, to show that statistical benefit. Now, if you imagine we only enroll patients who aren't cured, right. Then everyone is at risk. It creates a much higher statistically powered, but smaller trial, and then we could quickly find out, hey, does this drug really work in early stage disease? You know, removing the noise of all the patients that were already cured. And so this really is going to also, I think, accelerate the way we think about and plan clinical trials and hopefully make them faster, better and cheaper, which ultimately I think will benefit patients everywhere.
[19:50]
B
And presumably you can select which treatment to use based on these results. How does that work?
[19:55]
A
Yeah, so if we have the molecular profile of these cancers, then we know the genetic alterations that made that normal cell into a cancer. And again, now we have a lot of drugs that can target those alterations. We still have a long way to go. And this is my concern, is that not everyone's going to get cured just because we know who isn't. But I think as we learn more about the biology of why those patients didn't actually clear their circulating tumor DNA, even though they got the best standard of care drugs, and maybe even experimental ones, we have an opportunity to leverage that knowledge creating a platform that we can then figure out what are the best therapies. Why aren't standard therapies or even experimentals not working on this group of patients? That to me is really for me, something we can even try right now.
[20:44]
B
One of the interesting things you said in your manuscript was that all human cells shed DNA.
[20:50]
A
Yeah.
[20:51]
B
And so I'm wondering, have you thought about whether there may be applications of this type of technology to other non cancer diseases? Could we ever detect someone that's going to develop diabetes, for example?
[21:05]
A
Yeah, you know, it's a great question and the unequivocal answer is yes, because we've already done it in some other field. This field is 10 years before us is maternal fetal medicine, where women no longer have to get amniocentesis or chorionic villus sampling, which has a slight but real risk of spontane abortion. We can draw a tube of blood from mom's arm and sequence the entire fetal DNA. And again, this has been about 10 years ahead of circulating tumor DNA that's become standard of care across the country. Really at this point, the other area that's really emerging, but also equally important, we've never had great markers or early markers for who's going to reject their organ transplant. So if someone gets a liver transplant or a kidney transplant, we have to follow certain clinical parameters. But now, because the donor DNA DNA is distinct, usually unless they're identical twins from the recipient, companies and scientists have put together assays where you can now track ratios of donor derived DNA from the recipient DNA. And above a certain ratio shows that the organ is being rejected. And that's another use. And I would even argue because we're planning and working on this right now, there are even cancer indications that become non cancer indications. And what do I mean by that? Well, we're working right now on a study that cancer patients who are ostensibly cured, let's say if they need, let's say a heart transplant or a kidney transplant, they can't get on it by most guidelines unless they're five years from their disease or if there's mitigating circumstances like they need it, like now, otherwise they're going to die. So it depends on the ratio of acuity. But there are a lot of patients on there who can't even get on the transplant list until they're five years cleared. What if we could look at CTDNA after curative intent and say, hey, a patient has now had two negative results in a year, we're 95% certain that means they're cured, then they can get on the transplant list because there is a mortality per year waiting on the transplant list. So we can actually now accelerate their candidacy for getting an organ transplant. And we're doing lots of studies like that where clearance of ctdna, that is Cure, can afford different medical outcomes and opportunities that heretofore we didn't have that opportunity to give to those patients.
[23:26]
B
Ben, this has been really interesting and informative. Is there anything we didn't cover that you'd like to add?
[23:31]
A
Well, I think, and thank you, it's been really fun. But I think the number one thing is that this is still in its infancy. The message I always want to leave is one of caution. You can order these things right now. Patients want it, they're demanding it. But it can be just as dangerous as a negative tool, as a positive one, without proper understanding and utilization of these technologies.
[23:54]
B
I'm Dr. Mary McDermott and I've been speaking today with Dr. Ben park about his JAMA translational review on the topic of liquid biopsies. You can find a link to the article in this episode's description. This episode was produced by Shelley Steffens at the JAMA Network. To follow this and other JAMA Network podcasts and please visit us online at jamanetworkaudio. Com. Thanks for listening.
[24:23]
A
This content is protected by copyright by the American Medical association with all rights reserved, including those for text and data mining, AI training and similar technologies.

Summary7 min read

Podcast Summary: Liquid Biopsy Tests for Cancer

Podcast: JAMA Clinical Reviews
Host: Dr. Mary McDermott
Guest: Dr. Ben Park, Professor of Medicine, Vanderbilt University (Director, Ingram Cancer Center)
Date: August 10, 2026
Episode Theme: Exploring the current science, clinical applications, and future directions of liquid biopsy tests for cancer, with a focus on their potential to transform cancer care.

Episode Overview

This episode features an in-depth conversation with Dr. Ben Park, a leader in cancer research, focusing on the evolution, current use, and future promise of liquid biopsy technologies in oncology. Dr. Park discusses how these minimally invasive blood tests can detect cancer DNA fragments, their ability to personalize and potentially de-escalate treatment, their remarkable precision, and the evolving controversies and technical challenges. The discussion also touches on broader implications for other medical fields and the cautious approach clinicians should take while the technology matures.

Key Discussion Points & Insights

1. Historical Context and Development

Liquid biopsy technology began burgeoning about 15 years ago at Johns Hopkins after Bert Vogelstein’s group demonstrated cancer DNA in blood, catalyzing the field of next-generation sequencing (NGS) for tumors.
Dr. Park's own entry began with developing assays to identify cancer DNA mutations in blood despite the challenge that most cell-free DNA in blood is non-cancerous.
Quote (Dr. Park, 03:01):
“It’s proverbial needles in the haystack that we’re looking for. And I was really blown away at how accurate this was and how precise it was … It started to get me thinking that if we could actually drill down to early stage patients ... we could figure out who's actually cured versus who is not.”
The potential: avoid treating all patients with intensive therapies by pinpointing who’s truly at risk of recurrence after surgery or chemotherapy.

2. Liquid Biopsy Defined and Types (04:38)

Liquid biopsies: Blood samples analyzed for small fragments of DNA shed by cancer cells (circulating tumor DNA, or ctDNA).
Used differently in metastatic vs. early-stage disease:
- Metastatic: Mainly to match mutations with targeted therapies (qualitative).
- Early-stage: Now being applied quantitatively to assess residual disease and risk of recurrence.
Distinctions:
- Tumor agnostic/uninformed: Looks for general cancer-related DNA changes without knowledge of the patient's tumor.
- Tumor informed: Involves sequencing a patient’s tumor to track specific mutations for markedly higher sensitivity and specificity.
Quote (Dr. Park, 06:30):
“We can detect literally one cancer DNA molecule out of a million normal wild types. And that's really been game changing... we can really start changing how we’re going to treat people and do studies in the future.”

3. Clinical Use Case & Validation (08:05)

In current practice, especially for early breast cancer, high-sensitivity tumor-informed assays have validated accuracy but lack definitive proof that acting on results improves outcomes (“clinical utility”).
Case Example:
Young woman, HER2+ breast cancer, completed standard therapy. Liquid biopsy remained positive (3 parts per million, highly specific)—signifying residual disease. She received an additional approved but seldom-used therapy, cleared her ctDNA, and remained negative at 18 months, likely preventing metastatic relapse.
Quote (Dr. Park, 09:22):
“That to me is a very compelling story that we can recapitulate that paradigm for patients who otherwise would have recurred...”

4. Potential to Minimize Unnecessary Toxicity (11:10)

Current oncologic practice “overtreats” the majority to benefit the not-yet-cured minority. About 70% of post-surgery breast cancer patients are already cured but can’t be identified with certainty—leading to overtreatment.
Ongoing and planned “de-escalation trials” will test whether negative ctDNA post-treatment can safely predict cure, allowing many patients to avoid chemotherapy.
Warning (Dr. Park, 12:48):
“Outside of a clinical trial, I don't think we should ever de-escalate therapies … de-escalation would probably be considered bad medicine at best and maybe even malpractice at worst. So we just don't have that data yet.”

5. Sensitivity, Specificity, and Limitations (13:10)

Modern tumor-informed tests:
- High specificity—almost no false positives
- High sensitivity—but dependent on availability of tumor tissue and the cancer’s DNA shedding rate.
Caveats for patients:
- Test only detects recurrence of the original cancer—won’t pick up a new or unrelated malignancy.
- Tissue is an exhaustible resource; repeated/newer testing in future may be limited if tissue is depleted.
Quote (Dr. Park, 14:41):
“Unless you're really well versed in this field, I would counsel patients and providers to … make sure that they speak with someone ... who actually knows this space and knows the limitations and caveats.”

6. Cancer Subtypes and Test Efficacy (15:07)

Not all cancers shed DNA equally into the bloodstream (“low shedders”), affecting test sensitivity.
Hormone-receptor-positive, HER2-negative breast cancers (“slow-growing”) are notable examples.
Serial measurements and improved strategies may compensate.

7. Technical and Practical Challenges (16:16)

Misuse by uninformed providers could lead to inappropriate or unsafe de-escalation of therapy.
Variability in DNA shedding between patients, limited understanding of DNA release mechanisms.
Even current “ultrasensitive” assays (part per million) still have a false negative rate—especially for slow-shedding tumors.

8. Looking to the Future (17:44)

Liquid biopsies could personalize cancer therapy, de-escalating unnecessary treatment and focusing resources on those truly at risk.
Transformational effect on clinical trials:
- Could enroll only those with persistent disease (positive ctDNA), increasing trial efficiency.
- Faster, smaller, cheaper studies with clearer results.
Quote (Dr. Park, 18:50):
“Now if you imagine we only enroll patients who aren’t cured ... it creates a much higher statistically-powered but smaller trial, and then we could quickly find out, hey, does this drug really work in early stage disease?”

9. Beyond Cancer – Other Medical Applications (20:50)

Prenatal testing: Now standard for detecting fetal genetic abnormalities, a model ~10 years ahead of cancer liquid biopsy.
Organ transplantation: Detects early signs of organ rejection by distinguishing donor from recipient DNA in blood.
New applications: Could help recently “cured” cancer patients access transplantation sooner, reducing mortality while waiting for organs.
Quote (Dr. Park, 22:12):
“There are a lot of patients ... who can’t even get on the transplant list until they’re five years cleared. What if we could look at ctDNA ... and say…they’re cured, then they can get on the transplant list ...”

10. Caution Moving Forward (23:25)

The field is in its infancy—caution is crucial in clinical adoption.
Liquid biopsy is a powerful tool, but misuse or misunderstanding of results can harm patients.
Final Message (Dr. Park, 23:43):
“It can be just as dangerous as a negative tool as a positive one, without proper understanding and utilization of these technologies.”

Timestamps for Notable Segments

Early history and motivation for liquid biopsies – 01:43–04:38
Difference between current tests and pan-cancer screens – 04:50–08:05
Specific clinical example and implications for adjuvant therapy – 08:05–10:56
Potential for de-escalation of therapy and toxicity reduction – 10:56–13:10
Discussion of sensitivity, specificity, and limitations – 13:10–15:07
Cancers that are ‘low shedders’ and affect test efficacy – 15:07–16:16
Technical challenges and future technological improvements – 16:16–17:44
A vision for personalized cancer management and trials – 17:44–19:49
Non-cancer applications: organ transplant, prenatal testing – 20:50–23:25
Final thoughts and cautions about premature clinical use – 23:25–23:54

Memorable Quotes

“It's proverbial needles in the haystack that we're looking for. And I was really blown away at how accurate this was and how precise it was.”
— Dr. Park (03:01)
“We can detect literally one cancer DNA molecule out of a million normal wild types. And that's really been game changing.”
— Dr. Park (06:30)
“De-escalation would probably be considered bad medicine at best and maybe even malpractice at worst.”
— Dr. Park (12:48)
“Unless you're really well versed in this field, I would counsel patients and providers to … make sure that they speak with someone ... who actually knows this space and knows the limitations and caveats.”
— Dr. Park (14:41)
“What if we could look at ctDNA after curative intent and say ... they're cured, then they can get on the transplant list.”
— Dr. Park (22:12)
“It can be just as dangerous as a negative tool as a positive one, without proper understanding and utilization of these technologies.”
— Dr. Park (23:43)

Conclusion

Dr. Ben Park provides a comprehensive, nuanced look at liquid biopsy technologies: their promise to personalize cancer therapy, reduce unnecessary toxicities, and accelerate research, as well as their technical and ethical limitations. The field is rapidly moving but requires careful, informed clinical application. For those interested in leading-edge cancer diagnostics and the evolving potential of blood-based testing, this episode offers valuable insights directly from a field pioneer.

Loading summary

Transcript35 lines

[00:03]
A
Hi, I'm Derek Angus, a senior editor at JAMA and host of Healthy Dialogue, a new podcast from the JAMA Network. Join me as we go beyond the latest discoveries with nuanced in depth conversations with the world's leading experts to explore the most pressing issues in health and healthcare, from trends in autism diagnosis to private Equity acquisition to AI and much, much more. Visit JamaneTWORKaudio.com or search Healthy Dialogue wherever you get your podcast to subscribe.
[00:48]
B
Welcome to listeners from around the world and thank you for tuning in to this JAMA Clinical Reviews podcast. I am your host today, Dr. Mary McDermott, deputy editor of JAMA, and I'm here today with Dr. Ben Parks, who is a professor of medicine at Vanderbilt in the division of Hematology Oncology and a physician scientist focused on the study of breast cancer. Dr. Park is also director of the Ingram Cancer center at Vanderbilt University. And today I'll be talking to Dr. Park about his JAMA translational review on the topic of liquid biopsies. Welcome Ben. It's so nice to have you.
[01:32]
A
Thank you so much, Mary. It's great to be here.
[01:35]
B
Let's get started. Could you tell us how did you get started studying liquid biopsies? Can you give us a little of the historical background, please?
[01:43]
A
Yeah, so probably about 15 or more years ago, I was at Johns Hopkins at the time and I stayed there after doing my postdoc with Dr. Bert Vogelstein. And Bert had actually just proven that you could find cancer DNA molecules in the blood of patients with cancer. And that was such a huge thing and many people don't realize it, but that really set off this whole field. It even set off next gen sequencing as we know it today for looking at tumors, genetic alterations, et cetera. So we got in on the ground floor because my lab was literally on the ground floor above his lab. And we started working in this field to develop assays to pick out cancer mutations, mutations of DNA floating in the blood. So liquid biopsies can mean a lot of different things, but we in the circulating tumor DNA business tend to think of it as the DNA derived from cancer cells that we can now query find mistakes that we can often target with drugs. But then this has evolved into a newer paradigm of how do we actually use this to see if someone's cured or not. And that was something when we did the first study in metastatic breast cancer patients and saw how well this worked because it's so challenging. Most of the DNA that we call cell free is from the normal cells, so it's proverbial needles in the haystack that we're looking for. And I was really blown away at how accurate this was and how precise it was. And it started to get me thinking that if we could actually drill down to early stage patients with breast cancer and also just other solid tumor malignancies, that's a big conundrum right now in oncology. We could figure out who's actually cured versus who is not. And that affords an opportunity to not treat everyone with all the drugs, because we don't know who's cured and who's not. And it also affords the additional opportunity of giving people who aren't cured more drugs or put them on clinical trials in an effort to cure more patients. That's where the whole concept came in to really change the paradigm of how we give additional therapies, what we call adjuvant systemic therapies, in early stage patients with cancer. And that's kind of one of the biggest conundrums we have right now, especially in breast cancer. In my field, we know from historical studies 70% of patients after surgery, let's say, are in fact cured. We just don't know who they are. And so we treat everyone if they fell into clinical trials eligibility, knowing that 70% of the patients are being exposed to chemotherapies that they don't need, we just don't know who they. But we're also curing an additional 15 to 20% more patients. So the risk benefit ratio heavily favors. Let's treat everyone, recognizing that we're also doing some harm unintentionally. And now we finally are at a point 15 years later where we can start changing that paradigm. It's super exciting.
[04:39]
B
To summarize, liquid biopsies are, for example, blood samples where you're looking for small DNA fragments from cancer cells, correct?
[04:50]
A
Yep.
[04:51]
B
And how do they differ then, say, from one of those pan cancer screening tests?
[04:56]
A
Yeah, you know, they're very related. First, I should take a half step back. Probably for the last 10 years or so, we've had liquid biopsies, but mostly for metastatic disease, where the overall tumor burden is much higher. So there's a lot more circulating tumor DNA that gets shed into the circulation. Those are mostly used qualitatively, meaning that I said this earlier. If you have a particular gene mutation, you're looking for that, we have a drug, we can then say, aha, that person has that mutation. So they're eligible to get this drug, this targeted therapy. On the other hand, what we're doing now in the early breast cancer space that I just talked about is figuring out in a quantitative way, is there cancer still left behind? And if so, can we track it and monitor response to therapies, or are they recurring? And can we get enough lead time to prevent that recurrence from being true overt metastatic disease? Because if that happens, we can't cure the patient anymore. Ironically, we only cure metastatic disease when we can't see it. But now we can see it. And getting to your question, the difference is there's two kind of flavors, if you will, of these types of circulating tumor DNA assays. There's what we call tumor agnostic or tumor uninformed. Those are actually looking at DNA molecules again that are free floating in the plasma. And we end up looking at the whole kind of gamish, if you will, of DNA. And there are subtle changes in cancer DNA, both at the genetic level, meaning mutations and other types of genomic alterations. But even what we call epigenetic alterations, meaning that cancer DNA is methylated at certain residues that are very distinct than normal cell free DNA. So companies now, scientists, academics, we're working together to figure out can we use a more tumor agnostic approach to actually say who's got cancer and who doesn't? As a general screening tool, we're not quite there yet. Many people on this podcast probably have heard of the Grail Test Gallery. The results weren't quite as what we wanted to be for specificity sensitivity, but. But I do think with improved technologies, we will get there someday. On the other hand, what I was talking about earlier and what we have done in the breast cancer space recently and many others is actually looking at tumor informed circulating tumor DNA assays. What that means is we have to have a specimen of the patient's cancer, we sequence it and we find a lot of mutations. The company that we partnered with looks at 200 to 1800 mutations per tumor. And that affords the highest level of sensitivity and specificity so that we can detect literally one cancer DNA molecule out of a million normal wild types. And that's really been game changing. We're now calling it the ultra sensitive level of testing. But that has allowed us to really separate out, hey, are you going to recur or do you have like a 95% chance of being cured because you're negative? That's where we are right now. And I think that's the exciting part because we can really start changing how we're going to treat people and do studies in the future.
[08:06]
B
Can you give us an example of a liquid biopsy that is currently in clinical use and walk us through exactly how it's used? You mentioned earlier that it can be useful after a surgery for cancer. Your manuscript talks about using it after chemotherapy. Maybe just give us one example of a currently clinically used liquid biopsy.
[08:27]
A
Yeah, so again, I'm going to shift and just talk about early stage because there's been a lot of work already done in the field of metastatic disease where we match drugs with a mutation, but in early stage disease. Right now this is a little bit controversial in our field because we have not definitively proven clinical utility, meaning that if we get a result back, can we definitively say that acting on this is going to lead to a better outcome? We're designing those studies. What we've proven is validation. That said, because of the nature of these tumor informed assays being so specific and so sensitive, many of us, myself included, have been very thoughtful about patients who might benefit from this. And I have one example that I'd love to share is a young woman, mother of two. I've treated her for breast cancer that had a specific receptor called HER2 and it was positive. Now, there's a lot of drugs we give after surgery for ERBB2 positive disease, and sometimes we give them before, but the premise again is to get rid of microscopic cells if they exist, cells we can't see, that if left untreated, are going to come back as incurable metastatic disease. So in this particular instance, this patient completed all her therapies and we ordered the test and she still was positive. Very low level, 3 parts per million, as we call it. But that's again, because they're so highly specific. That's real. That meant she was not cured. And I had in the back of my mind, if I had found something like that first. We scanned to make absolutely sure there's no evidence of overt metastatic disease. And unfortunately there was not. But there's another drug. We don't use that much, but it is approved in this setting to cure more patients. It's got a lot of toxicity and that's why we don't use it that much anymore for HER2 positive disease. But it is there. So I put her on that. She did have some toxicities with it, but in six months she cleared and it's prescribed for one year as adjuvant systemic therapy. And then after another six months, still clear, she's at 18 months now, doesn't have any sign of CTNA positivity that to me is a very compelling story that we can recapitulate that paradigm for patients who otherwise would have recurred and unfortunately would have probably relapsed with metastatic disease. And I feel very excited that this is something that is going to really change how we think about and administer therapies for early stage cancers.
[10:56]
B
Yeah. And your review, in addition to talking about situations like that, where you might begin treatment when otherwise you wouldn't have known to begin treatment, you talked about how this could be used to minimize toxicity.
[11:10]
A
Yes, absolutely.
[11:11]
B
Can you give us an example of that?
[11:12]
A
Yeah. As I was saying earlier, right now our paradigm is to do standard of care because they've been vetted in big clinical trials. But you can imagine how those trials were developed. There were, let's say, 1,000 patients with cancer in one arm, another thousand patients with cancer, very similar types of cancers, H stage. And after surgery for both arms, then one group got placebo and the other group got chemotherapy. And, and we follow those patients out year after year, decade after decade. And the group that got chemotherapy does better. There's probably about again, 15 to 20% or more patients alive. And that's significant because for a disease like breast cancer example, 300,000 new cases diagnosed every year just in the United states. So even a 5% difference translates into thousands, if not tens of thousands of additional lives that are cured. But what about all those 70% of patients, 700 patients in the study that didn't need the chemotherapy? Heretofore, we've never been able to identify them. So we are now planning de escalation trials where we're going to see if the absence of circulating tumor DNA is a good enough indicator to say, you know what, now the odds weigh that chemotherapies not going to be in your favor. It'll probably do more harm than good. That's where we're headed now. That said, outside of a clinical trial, I don't think we should ever de escalate therapies. We can do it in a clinical trial context because as you know, we have interim safety analysis points. And if we see signal early on that this isn't the right thing to do, we stop it. That's the best ethical way to do a de escalation study. So right now, out of a clinical trial context, I like to say that de escalation would probably be considered bad medicine at best and maybe even malpractice at worst. So we just don't have that data yet. And that's where the controversy Is right now between where we are in this space for clinical validation versus clinical utility.
[13:11]
B
And do we have data on sensitivity? You gave the example of your patient with three parts per million.
[13:19]
A
Yeah, yeah.
[13:20]
B
And so is there no false positive?
[13:22]
A
There is almost none. And if there are, it's probably a technical issue. And that's because again, there's so many mutations that are keyed in to that patient's disease. I actually have to counsel patients before I order any of these tests. It's like a genetic counseling talk for about 15 to 20 minutes. Because that also has its downsides. One is that I have to remind patients because of that tight sensitivity specificity with the original tumor, if the patient ever got a different cancer or a second new breast cancer for my clinic, this test would never pick that up because it will only pick up the original tumor that it sequenced and has these unique mutations. Every cancer is different, just like every person is different. And so that's why it would never pick up a new cancer, breast or otherwise. The other caveat that a lot of people I don't think have necessarily thought through, these so called ultra sensitive tests are a second generation test. There was a first generation test that was one in 10,000, not one in a million in terms of technical sensitivity. But remember these tumor informed assays, you need tissue and that's an exhaustible resource. So I have had patients who wanted to get the tests. I went through my whole kind of counseling with them. We ordered the test but there was no tissue left and they had had a previous first gen test done. And I remind patients, you know, there's probably going to be a third gen test in the next five to 10 years. And if we order it today, you might not have tissue for that test in the future. It's just one of these things that there's a lot of caveats and unknowns. And I think unless you're really well versed in this field, I would counsel patients and providers to make sure that they speak with someone as a second or third opinion who actually knows this space and knows the limitations and the caveats.
[15:07]
B
Your review talks about some cancers these liquid biopsies work better for than others. What cancers does it not work well for? And why would that be?
[15:18]
A
Yeah, so we're learning a lot as we go along. But there are some cancers that we call low shedders, meaning that they're not usually as proliferative and therefore they're not breaking down there and having a lot of apoptosis. That's where we see a lot of the circulating tumor DNA go into the cell free space into the plasma. And so my disease subtypes of it. For example, the most common type of breast cancer is what we call hormone receptor positive, HERB2 negative. They tend to be very slow growing. They tend to be very sensitive to endocrine therapies and not so much chemotherapies. But they also tend not to shed as much circulating tumor DNA as a result. So those are patients where doing serial measurements can sometimes mitigate that effect. But we and others are working on strategies to try to mitigate those types of concerns because that's ultimately what limits the sensitivity. It's not necessarily the technical part per million, it's really how much DNA you can get safely from a patient's blood and assay that.
[16:17]
B
What are some of the challenges associated with use of these liquid biopsies?
[16:21]
A
I think there's always the challenge of misuse where again providers and patients may not understand their limitations. And that also worries me and concerns me that they could be used inappropriate, as I mentioned earlier, to de escalate therapies where we don't have good safety data, we have zero safety data really to actually be able to say and do that. I think there are other things that we still have challenges technically. As I mentioned, sometimes you just don't get enough DNA from patients blood. The variability from person to person, probably even from day to day from the same person can be orders of magnitude. And believe it or not, even though we've been working on this for well over two decades now, we don't really understand the biogenesis of how cells secrete out this DNA. We do know that if your cells are dying, that will liberate out some more. And there have been really nice studies showing that if you give chemotherapy, for example, and it bursts the cancer cells, your CTDNA will spike up. But beyond that, there hasn't been a lot of work done on that. And that's what I had mentioned earlier. We and others are really trying to make some discoveries and leverage those so that we can maybe even someday give drugs, believe it or not, that will elevate the amount of circulating tumor DNA to make all tests more sensitive. And that is a clinical problem right now, where even the best current 2nd gen tests still have a false negative rate.
[17:45]
B
Thinking ahead, say three decades from now, what's your vision of how liquid biopsies might be used to manage cancer?
[17:53]
A
Yeah, so again the qualitative aspects, we could probably find all the mutations and mistakes just using a tube of blood as a True. Liquid biopsy. But then I'm most excited about what I mentioned earlier. We could use this to individualize everyone's therapy. We know that adjuvant therapy, chemo, hormone therapy, antibody therapies work to cure more patients in early stage solid tumors. But we also know that comes at a price of over treating all these patients that heretofore we had no idea how to test whether they're cured or not. This gives us this precise lens into who is cured and who's not. And so for us, for a disease like breast cancer, I imagine you could actually take blood serially after each component of therapy and make a decision, do I need to get the next component? So as an example, do surgery. If that patient now clears the CTDNA that was positive before surgery, we won't have to do radiation, we don't have to do chemo, we don't have to do five to 10 years of endocrine therapy. It will be game changing. The other thing that I think it's really going to accelerate is clinical trials in early stage disease. Getting back to my original kind of analogy of how we do clinical trials in the early stage, we have to wait year after year, decade after decade, to show that in that example, the group that got chemo does better, significantly than the group that doesn't. Now imagine instead, and we enroll thousands of patients, sometimes even more than that, to show that statistical benefit. Now, if you imagine we only enroll patients who aren't cured, right. Then everyone is at risk. It creates a much higher statistically powered, but smaller trial, and then we could quickly find out, hey, does this drug really work in early stage disease? You know, removing the noise of all the patients that were already cured. And so this really is going to also, I think, accelerate the way we think about and plan clinical trials and hopefully make them faster, better and cheaper, which ultimately I think will benefit patients everywhere.
[19:50]
B
And presumably you can select which treatment to use based on these results. How does that work?
[19:55]
A
Yeah, so if we have the molecular profile of these cancers, then we know the genetic alterations that made that normal cell into a cancer. And again, now we have a lot of drugs that can target those alterations. We still have a long way to go. And this is my concern, is that not everyone's going to get cured just because we know who isn't. But I think as we learn more about the biology of why those patients didn't actually clear their circulating tumor DNA, even though they got the best standard of care drugs, and maybe even experimental ones, we have an opportunity to leverage that knowledge creating a platform that we can then figure out what are the best therapies. Why aren't standard therapies or even experimentals not working on this group of patients? That to me is really for me, something we can even try right now.
[20:44]
B
One of the interesting things you said in your manuscript was that all human cells shed DNA.
[20:50]
A
Yeah.
[20:51]
B
And so I'm wondering, have you thought about whether there may be applications of this type of technology to other non cancer diseases? Could we ever detect someone that's going to develop diabetes, for example?
[21:05]
A
Yeah, you know, it's a great question and the unequivocal answer is yes, because we've already done it in some other field. This field is 10 years before us is maternal fetal medicine, where women no longer have to get amniocentesis or chorionic villus sampling, which has a slight but real risk of spontane abortion. We can draw a tube of blood from mom's arm and sequence the entire fetal DNA. And again, this has been about 10 years ahead of circulating tumor DNA that's become standard of care across the country. Really at this point, the other area that's really emerging, but also equally important, we've never had great markers or early markers for who's going to reject their organ transplant. So if someone gets a liver transplant or a kidney transplant, we have to follow certain clinical parameters. But now, because the donor DNA DNA is distinct, usually unless they're identical twins from the recipient, companies and scientists have put together assays where you can now track ratios of donor derived DNA from the recipient DNA. And above a certain ratio shows that the organ is being rejected. And that's another use. And I would even argue because we're planning and working on this right now, there are even cancer indications that become non cancer indications. And what do I mean by that? Well, we're working right now on a study that cancer patients who are ostensibly cured, let's say if they need, let's say a heart transplant or a kidney transplant, they can't get on it by most guidelines unless they're five years from their disease or if there's mitigating circumstances like they need it, like now, otherwise they're going to die. So it depends on the ratio of acuity. But there are a lot of patients on there who can't even get on the transplant list until they're five years cleared. What if we could look at CTDNA after curative intent and say, hey, a patient has now had two negative results in a year, we're 95% certain that means they're cured, then they can get on the transplant list because there is a mortality per year waiting on the transplant list. So we can actually now accelerate their candidacy for getting an organ transplant. And we're doing lots of studies like that where clearance of ctdna, that is Cure, can afford different medical outcomes and opportunities that heretofore we didn't have that opportunity to give to those patients.
[23:26]
B
Ben, this has been really interesting and informative. Is there anything we didn't cover that you'd like to add?
[23:31]
A
Well, I think, and thank you, it's been really fun. But I think the number one thing is that this is still in its infancy. The message I always want to leave is one of caution. You can order these things right now. Patients want it, they're demanding it. But it can be just as dangerous as a negative tool, as a positive one, without proper understanding and utilization of these technologies.
[23:54]
B
I'm Dr. Mary McDermott and I've been speaking today with Dr. Ben park about his JAMA translational review on the topic of liquid biopsies. You can find a link to the article in this episode's description. This episode was produced by Shelley Steffens at the JAMA Network. To follow this and other JAMA Network podcasts and please visit us online at jamanetworkaudio. Com. Thanks for listening.