Ever since early February, when former FDA Commissioner Robert Califf publicly bemoaned the progress of precision medicine, I wanted to find out whether other people felt similarly discouraged.
I got my chance in October when an MD Anderson Cancer Center public relations professional pitched an interview with a senior MD Anderson radiation oncologist, who just happened to be my dearly departed brother’s close friend. In fact, Dr. Prajnan Das, professor and department chair of Gastrointestinal (GI) Radiation Oncology, did not hesitate to open his home to us in June 2015, insisting that my brother be evaluated by a kidney cancer specialist at the storied institution. My brother had his cancerous right kidney removed in Houston thereafter, and it was a significant comfort to both him and me to be in Dr. Das’ home as we wrestled with the Stage 4 kidney cancer diagnosis and uncertain future ahead. My brother did have genomic testing done, but his cancer had spread too far. He lived another two years before succumbing to complications from the disease, something that Dr. Das had thankfully prepared us for right from the start.
But that was eight years ago. And sitting in Feb 2025, Califf, who also lost his brother to cancer — in January — was saying that the practical applications of precision medicine had been largely a failure. In October, Dr. Das spoke at HLTH on a panel on precision medicine I missed, and so in my Zoom interview with him after HLTH, that’s where I began.
RCM Outsourcing: Balancing Human Expertise & Smart Technology
Greenway Health blends smart technology with human expertise to elevate revenue cycle.
What follows is an edited Q&A about precision medicine and the potential of AI.
MedCity News: What Dr. Califf was saying is that his own brother had pancreatic cancer and passed away and he was very upset that he had to go shop around to try to find a protocol. So he thinks precision medicine has not lived up to its promise. Are you as harsh?
Dr. Prajnan Das: I think a lot of it depends on what you define by precision medicine. On the panel, I talked about the precision medicine aspects of radiation therapy because that’s my focus. And that’s also quite important because when we think about precision medicine, a lot of our focus tends to be on drugs. The FDA is all about that. But when you take a step back, 50-70% of cancer patients are going to get treated with radiation during the course of their treatment, and so innovations in radiation also have a huge ability to have an impact. So when I think about how I take care of patients today versus 10 years ago or 20 years ago, it’s completely different in every aspect. Ten, 15 years ago, we were using radiation primarily in patients whose cancer had not spread. If the cancer had spread, then we would use radiation to help palliate symptoms in critical parts of the body.
But now what we have learned in the last 10 years is that even in cancers that have spread to limited areas of the body, we have identified a distinct entity, which we call oligometastatic disease. And in those patients giving radiation or other forms of ablative treatment to those oligometastatic areas improves patients’ disease-free survival, their long-term survival, their ability to stay off chemotherapy for extended periods. So one of my colleagues in renal cell cancer has this study that if you treat areas of metastasis with radiation, those patients can stay away from the traditional systemic treatments for renal cell cancer for extended periods. And those treatments have side effects as you and I both know. So Chad Tang from MD Anderson published a really important study on that.
Enhanced Direct Enrollment: Georgia Access and the Power of Partnership
Across the nation, state-based health insurance exchanges are searching for ways to better serve their communities while making enrollment easier, more personal, and more accessible for everyone.
Even in pancreatic cancer, we used to think that, ‘Hey if the cancer has metastasized, the only option is chemo.’ My colleague Ethan Ludmir published this important randomized trial, which showed that even in pancreatic cancer, if you can treat the metastasis with radiation or surgery, those patients, their disease disease-free survival gets extended. We would never have known that a few years ago.
MedCity News: But is that knowledge coming from doing genomic testing? How is precision medicine related to this?
Dr. Prajnan Das: A lot of it is actually coming from genomic testing as well. So in addition to the clinical trials, where we are trying to figure out why patients are doing so much more better with local treatments in oligometastatic disease than we thought they would, there’s a lot of translational studies that are going in that are looking at things like immune signatures and specific molecular types that are also transforming our understanding of why these treatments are working. But going back to the molecular subtypes, so what we are also learning is that now we can use molecular subtypes to figure out which patients need radiation and which patients don’t. That is precision medicine too.
I’ll give you an example from one of the cancers I treat: rectal cancer. The standard treatment for rectal cancer used to be a combination of chemotherapy, radiation and surgery. Now we know that you can do a simple molecular, or even a simple immunohistochemistry test on the biopsy for rectal cancer to identify a subgroup that’s called microsatellite unstable rectal cancer. And in that subgroup, all you need to do is treat them with immunotherapy. So there have been studies that have shown a 100% response rate to immunotherapy in that specific patient population. You don’t need radiation, you don’t need chemo, you don’t need surgery.
So I think we’re getting much, much better now at figuring out this specific molecular subtypes and treating them and figuring out which treatments work for them, but also importantly, which treatments they don’t need.
MedCity News: But do you think that the vast knowledge that we have today on the human body that we didn’t have even 10, 15 years ago, do you think that has translated to great treatments? Or do you think there’s more to realize in that? I know you are optimistic about precision medicine. I get that. I’m trying to be skeptical, so quite the opposite.
Dr. Prajnan Das (smiles): That’s why we have different jobs.
MedCity News: Absolutely. So what more needs to be done? Do you think that not just in cancer, but also across diseases, molecular testing should be the first thing that people do? It’s still not something that I think is widely done.
Dr. Prajnan Das: Yeah, so I think it’s becoming much more widely done, as routine. I think the important thing to also appreciate is that ultimately it takes a number of years, sometimes a couple of decades, even between when a test becomes routinely used and when we really start understanding how to use that information.
With that microsatellite unstable rectal cancer example I gave you, we knew that there was this molecular subtype entity — we have known that for about 20 years, but only about 10 years ago did people start exploring, ‘hey, maybe immunotherapy will work on this patient.’ And only in the last two years have we had the prospective clinical trials that have actually rigorously studied that and proven the role of immunotherapy. So I feel like it takes 10 to 20 years almost, between that initial testing to coming up with specific targeted treatments. But I hope that going forward, that time lag is going to decrease, and we are going to get better and more efficient at bridging that gap between the test and the application of it.
MedCity News: On that hopeful note, let’s segue to AI. In the field that you’re in, and within the context of improved efficiency, where do you think you, in your practice, you at MD Anderson are using AI?
Dr. Prajnan Das: So our common friend has been threatening that I’m going to lose my job every year because of AI. He’s been telling me that for the last five to 10 years at least.
MedCity News: And you’re still there.
Dr. Prajnan Das: AI obviously is transformational and in many different ways, but part of it is also figuring out where we use it and where we don’t use it, right? So in some areas like diagnostic mammograms, or even in colonoscopies to help detection of polyps, AI has really augmented the clinician’s ability. In my field of radiation oncology, AI has the potential to be transformative too.
So the way radiation is designed today is like this: a person gets a scan for that simulation to the scan, and then a physician has to manually work on designing the radiation plan. And they do it by sitting on a computer, outlining what the target is and what the normal structures are, which takes several hours. And then a second professional, called the dosimetrist, uses that information to design the radiation beams, the beam angles, what’s blocked, what’s open, and then a third professional, a physicist, checks all of that and makes sure that the radiation machine is providing the output that we have designed this way.
Every step of that can be replaced by AI.
So I collaborated with Lawrence Court, a colleague who is a physicist, and we worked on this project with a PhD student, which we have now published, which shows that you can develop an end-to-end AI program for rectal cancer radiation planning. And that whole process can be done in just a few minutes, replacing all of this time and effort. So I feel like this can be truly transformative in our field, and what this will accomplish is that it’ll standardize care, it’ll make care more accessible, cheaper and better.
MedCity News: And reduce variability?
Dr. Prajnan Das: Absolutely reduce variability. And we’ve also done studies on that showing that when you start using these tools, even the variability between physicians can go down dramatically. That’s the exciting part.
MedCity News: I live in Silicon Valley, and one of the biggest complaints of all the tech people here is that it’s very hard to convince doctors to change the way they do what they do. Some of that is legitimate because you’re dealing with people’s lives here, so you are conservative as you should be. But do you think that when you have this, as you call it, a transformational technology like AI, do you think doctors are going to embrace it? What do you see amongst your own colleagues?
Dr. Prajnan Das: As long as you give the doctors the right tools, they will absolutely embrace it. So my group uses these tools on a routine basis, Not just that, Lawrence has taken this tool and has developed a platform where it’s being used globally in hospitals in Zambia and South Africa. Essentially, it’s a knowledge-based AI tool based on MD Anderson physicians’ knowledge, and now a physician in a low-resource setting can upload their patient’s information into this program and the program will [create] a radiation plan at MD Anderson standards based on that information.
MedCity News: What kind of AI is it? Did MD Anderson develop it, or was it an off-the-shelf tool?
Dr. Prajnan Das: It’s a deep learning tool. It’s our internal product developed by our brilliant physicist and his group of PhD students and postdocs and other physicists. The reason also I think that his project worked is that he works very collaboratively with physicians, so that we come up with tools that physicians actually will use.
MedCity News: As opposed to the EHR which was imposed on physicians.
Dr. Prajnan Das: Right, exactly. So I think that’s part of the problem. When physicians are thinking about electronic tools, the first thing that comes to mind is the EHR, and that has never met any of our needs or goals, right? If you can come up with the AI tools that will actually help physicians, they will embrace it. I mean, radiologists are using it, endoscopists are using it. We are using it.
MedCity News: Any closing thoughts? On the future of precision medicine?
Dr. Prajnan Das: I think where we really can be innovative is coming up with good hypotheses that are based on strong existing data that leverage molecular tools. And maybe AI will help us figure out what those questions are. Because right now, I’m sure there’s a lot of signal in our data that we are missing.
AI can help pick out those data that, ‘Hey, maybe that this can work for this patient.’ There is an amazing example of this. Recently, a paper came out in Nature by one of our residents in radiation oncology that got worldwide attention. He leveraged the data of patients who were getting treated during Covid on immunotherapy and what he found using some innovative data analysis was that patients who had a Covid vaccine had much higher responses to immunotherapy compared to patients who did not get a Covid vaccine.
So on the surface makes no sense. The Covid vaccine is targeting Covid, not the tumor, but he was able to use data tools to identify that signal. Then with that signal, he actually did rigorous analysis on clinical data sets to show that there is clearly this difference we are seeing. And then he went to the lab to find an explanation for why this was happening. And he did rigorous studies in mouse models, other areas.
I think that is the future of personalized medicine, where data tools help us find unique signals, and these then generate the next generation of studies that potentially lead to more treatments.
Photo: MD Anderson Cancer Center