The Festival of Genomics and Biodata is just around the corner, and we’ve had the opportunity to sit down with some of our expert speakers to get a sneak peek into what they’ll be discussing, and why they think you should come along to the event.
In today’s interview, we speak to Aadel Chaudhuri (Vice Chair for Translational Research, Mayo Clinic) about the promise of liquid biopsy in cancer and beyond, the important role of AI in biomedical science and the need for rigorous validation of multi-omics approaches.
Register for the Festival of Genomics and Biodata here.
Please note transcript has been edited for brevity and clarity.
FLG: Hi everybody. Today, we’re joined by Aadel Chaudhuri from the Mayo Clinic, who’ll be joining us next month in Boston for The Festival of Genomics and Biodata. Ahead of his talk, we have some time today to have a chat about his work, and I’m really excited to hear all about it. So Aadel, over to you, could you introduce yourself and tell us a bit about your background and your work?
Aadel: Yeah, thank you so much. I’m Aadel Chaudhuri. I’m here at the Mayo Clinic. I’m the Vice Chair for Translational Research in the Department of Radiation Oncology, and also co-direct our Cancer Precision Medicine programme across the Mayo Clinic enterprise, across the Comprehensive Cancer Centre.
FLG: And at the Festival next month, you’re going to be giving a talk on a multi-omic approach to liquid biopsy for cancer section and profiling. Could you elaborate on why a multi-omic approach offers an advantage over a single omic approach for cancer detection?
Aadel: Yeah, absolutely. One of the issues with different liquid biopsy approaches is the sensitivity. So, while with ctDNA we can achieve superb specificity, the sensitivity for certain applications such as minimal residual disease detection and early cancer detection are still lacking. One way to potentially overcome that is through multi-omics, by approaching detection from different orthogonal standpoints, by multi-omically analysing the same or different bioanalytes.
FLG: It sounds really useful, but I would imagine that it comes with its own challenges. And what are some of the challenges that you’ve encountered when integrating those different omics layers into a cohesive pipeline. How do you overcome those challenges?
Aadel: There are significant challenges. One is just, how do you get different forms of data into the same space for integration? Another is that you need to be cognizant that when we’re layering on different data sets that we’re not overfitting a biomarker to the outcome of interest. So, in the setting of multi-omics, it’s even more important to have completely held out validation data sets to ensure that our biomarker predictions are robust and not merely showing what we expected to show in the test set.
FLG: It does sound like it could be really revolutionary, though. And are there any specific cancer types or any other clinical applications where you’ve observed particularly strong results from that approach?
Aadel: We’ve looked at a number of different malignancies. We had a paper recently looking at metastatic prostate cancer, where we were able to layer on different cell-free DNA epigenomic features in order to identify patients with significantly worse prognosis. Patients with metastatic castration resistant prostate cancer that were resistant to AR-directed therapy went on to develop poor outcomes. We were able to identify this by inferring stemness from their cell-free DNA epigenome.
FLG: And how accurate is the approach when it comes to predicting treatment responses compared to traditional methods?
Aadel: We found it to be quite strikingly associated with outcomes. I will say we didn’t take the next step to establish it as a predictive biomarker, so it will be important in future work to show that, in addition to risk-stratifying these patients, that we can additionally recommend certain treatments in order to improve a poor outcome, to give a patient who otherwise might have a poor outcome a better outcome.
FLG: That’s really interesting, because I was wondering about the clinical implications for patients. For example, in a scenario where we could detect precancerous lesions much earlier using a method like this, what would the clinical implications be? Would we treat patients before disease develops, or would we offer increased monitoring? What would be the best pathway for those patients?
Aadel: It’s a great question. I mean, there is this concept of trying to pre-cure a patient, if there’s a patient who has a pre-cancer developing. We’re not there with the technology. But in the future, let’s say we have the technology where we can detect pre-malignancies reliably with high sensitivity and specificity, then one might have treatment paradigms for pre-cure, and can we do something to that patient in order to mitigate their risk of ever developing the malignancy? There’s some literature with canakinumab, in a study focusing on heart disease, they found that canakinumab also lowered the lung cancer risk. So, there’s presumably a set of patients who were at risk for lung cancer, and that risk was mitigated by pharmacological intervention. In that case, it was quite surreptitious. But if this could be done by design, it’d be very interesting, using liquid biopsy as the framework.
FLG: It could be amazing for patients if the evidence was there. It could be really, really great. But of course, you’ve mentioned there’s still work to be done, there’s still evidence to be seen, but what do you see as the main challenges that would hinder the widespread adoption of this approach in the clinic?
Aadel: Clinical performance is important. Right now, these assays, in the setting of early detection, even in the setting of minimal residual disease detection, we’re still optimising their sensitivities, their specificities, their reliability. And, ultimately, we don’t have FDA-approved assays in these spaces yet for that reason. So, I think once we have FDA-approved, reliable assays that a clinician can act upon, then it’ll truly be game-changing. Now I will say, when it comes to actionable mutation detection by non-invasive genotyping from plasma cell-free DNA in metastatic patients, there are FDA-approved clinically actionable assays, and those are in widespread use in the clinic.
FLG: And on the topic of FDA approval, what are your thoughts on the regulatory landscape for liquid biopsy assays, and how might it need to evolve to accommodate these multi-omic approaches?
Aadel: A great question. There’s still a favourable landscape for laboratory developed tests. They don’t undergo the same level of FDA scrutiny as, say, drugs. So, in that sense, there is more room to develop a test and have it utilised in patients. But ultimately, the landscape is changing, so we’ll see. There could be greater regulation in the future. It’s possible that liquid biopsy tests will have to undergo FDA approval prior to clinical use. That that all remains to be seen.
FLG: And do you anticipate that there would be any challenges with patient accessibility, for example, or cost effectiveness?
Aadel: Yeah, that’s important. Ultimately, these tests will have greater use when they’re Medicare reimbursed, or generally insurance reimbursed. Liquid biopsy assays for non-invasive genotyping, when we’re looking for actionable mutations in metastatic patients include approved tests that are insurance-reimbursed. But in more exploratory settings, such as for multi-cancer early detection, they’re not reimbursed, so those tests are pay-out-of-pocket. Now, the landscapes are changing, but ultimately, for a test to really have widespread approval, it needs to be covered by insurance.
FLG: Sounds like it’s quite a complex landscape to navigate in that sense. Onto what I what I imagine is everybody’s favourite question at the moment, what’s the role of AI and machine learning in this approach, if any? And if you’re not using it, do you have any plans to adopt those kinds of tools in any of this work going forward?
Aadel: With cell-free DNA or cell-free nucleic acid multi-omics more generally, AI is critical. In order to integrate these different data sets, we use advanced machine learning, deep learning, AI techniques. It is important, when using AI approaches, to ensure that the result is reliable, that the result is not over-fit, and ideally, we prefer to use models that are explainable, such that the AI integrative model is not a black box and that one can understand precisely what inputs are being used by the model in order to develop the result.
FLG: And do you expect that one day, AI could potentially be able to effectively make similar predictions on treatment response and disease risk? And how might that complement multi-omics, liquid biopsy approaches?
Aadel: There’s a whole field focusing on using clinical informatics, using data from the clinical chart, using data from pathology slides and imaging studies, and using AI to integrate those data sets in order to predict outcomes. I think all of these push towards a precision medicine future of early anticipation, early detection, early intervention. And I do think that these fields and technologies complement one another. Liquid biopsy and the data revolution; they go hand in hand. They are really two sides of the same coin.
FLG: That’s really interesting. And I know that cancer is the primary application for something like liquid biopsy, but beyond that, do you see any potential applications for a multi-omic liquid biopsy approach in any other disease areas?
Aadel: Yeah, liquid biopsy can really be applied across the entire spectrum of human health and disease. For example, we have a project in our lab focusing on sepsis, where we’re using some of these same liquid biopsy technologies that we’re applying to cancer, and we’re also applying them to sepsis, with the goal of early sepsis detection, early sepsis intervention. So, in many ways, I do think of many of these concepts as portable.
FLG: That’s really, really interesting. It sounds like it could be life changing for patients suffering from or at risk of something like sepsis. So, I’ll be keeping an eye out for that work! And that leads me on quite nicely to my next question; what are the next key research questions that your team is focused on addressing?
Aadel: We’re very focused on solving the big questions in the field – how can we detect disease early, how can we intercept disease early? How can we risk-stratify patients better? And ultimately, how can we improve outcomes? We’re mostly focused on cancer, but as I just gave you an example, we are looking at other etiologies as well. But those are our main questions, and we’re looking to do it all revolving around liquid biopsy, integrative omics, with AI baked in as well.
FLG: And are there any other breakthroughs in the field that are exciting you at the moment?
Aadel: There are a number of exciting areas. The recently FDA-approved test for early colorectal cancer detection from plasma cell free DNA, from Guardant Health, that’s interesting. I think that’s a very practical approach, using liquid biopsy to detect cancer early in a setting where you have a cancer type that’s rapidly increasing, especially in individuals who are relatively on the younger end of the spectrum. And now, in addition to the standard methods that we have, like colonoscopy and stool analysis, now we also have plasma cell-free DNA as another analyte for early detection.
FLG: Amazing. That does sound really exciting. Now, as I mentioned earlier, you’re joining us next month for The Festival of Genomics and Biodata. What are you most looking forward to about the event, and why would you encourage people to come along?
Aadel: What I love about the event is that it’s a really nice marriage between academia and industry. I’m looking forward to seeing what my colleagues have to say, and other leaders in the field, both on the academic side and on the industry side. Ultimately, for us to really move the needle and improve patient care, we need this sort of interaction, this sort of collaboration, across the entire spectrum, going all the way from academia all the way into medicine, all the way to the biotechnology industry.
FLG: Well, thank you for the kind words, and we’re really excited to have you there as well. Thank you so much for taking the time to talk to me today, and we’ll see you next month.
Aadel: You’re most welcome. Take care. Bye.
Register for the Festival of Genomics and Biodata here.
