Adaptive Radiation Therapy: How AI Is Personalizing Cancer Treatment

Dale Shepard, MD, PhD:

Cancer Advances, a Cleveland Clinic podcast for medical professionals exploring the latest innovative research and clinical advances in the field of oncology. Thank you for joining us for another episode of Cancer Advances. I'm your host, Dr. Dale Shepard, a medical oncologist and co-director of the Sarcoma Program at Cleveland Clinic.

Today, I'm happy to be joined by Dr. Ehsan Balagamwala, a radiation oncologist here at Cleveland Clinic. He was previously a guest on this podcast to discuss SBRT for oligometastatic disease, and that episode is still available for you to listen to. He's here today to talk about Ethos therapy for treating cancers with radiation. So welcome back.

Ehsan Balagamwala, MD:

Thank you, Dale. It's a pleasure to be back and congratulations to you and your team for such a successful podcast.

Dale Shepard, MD, PhD:

Oh my gosh, you're too kind. We're like six years in at this point. That's crazy, right?

Ehsan Balagamwala, MD:

Pretty incredible.

Dale Shepard, MD, PhD:

So now, the other crazy part is that we've only had you here twice now, so we're going to have to come up with some more topics for you.

Ehsan Balagamwala, MD:

For sure.

Dale Shepard, MD, PhD:

So give us a little bit of an overview, what do you do here at Cleveland Clinic?

Ehsan Balagamwala, MD:

Yeah, so my interests are around the gastrointestinal tumors. And so I'm co-director of the colorectal cancer program, treating GI tumors from esophagus down to the colorectal and everything in between, including liver, pancreas, stomach cancers. And I also have a strong interest in metastatic spinal and brain tumors.

And so focused a lot around the technologies, the advanced technologies in radiotherapy, including stereotactic body radiotherapy, SBRT, as well as radiosurgery, Gamma Knife radiosurgery, and now the adaptive Ethos planning that we have, which I'm really excited about.

Dale Shepard, MD, PhD:

Being interested in the technology part, you're the perfect guy to talk to. So when we talk about this Ethos therapy, adaptive therapy, for people that might not be familiar with all the sort of ins and outs of radiation therapy, what does that mean?

Ehsan Balagamwala, MD:

Yeah, so let's set the stage. Radiation for many decades, radiation was one of the first cancer therapies out there. And for many, many decades, the radiation oncologist developed a treatment plan based on a planning scan. And whether the treatment was two weeks long or nine weeks long, every day a little bit of radiation based on that original treatment plan was delivered.

But as we all very well know, bodies are not statues. Things move. Tumors grow and shrink. And so, over the last about decade, decade and a half, we've developed intensity-modulated radiotherapy as well as SBRT now, which allows us to deliver highly precise radiation to the tumor while avoiding normal structures. But there's an important limitation to SBRT.

It is still done the same way that radiation has been done for many years where you develop a treatment plan and all you can do at the time of treatment is adjust the patient's position. And so yes, it is better than what we used to do 20 years ago, much higher doses, but the next iteration is to be able to adapt that plan to that patient's anatomy on that particular day of treatment.

Because if there's a gas bubble, bladder filling changes to adapt to that is really important. And I think there's two aspects to adaption that are important. One is to adapt to the changes in tumor size and location, and the other aspect is adapting to the changes in normal anatomy and sensitive structures that are near the tumor. And so it's been pretty incredible to be a part of that transition from regular radiation all the way to adaption now.

Dale Shepard, MD, PhD:

I'm going to double back really quickly just to make sure people don't get confused because we're talking about this essentially moving things and it has to do with moving things. So you mentioned about how historically we could move the patient. And so there might be people listening like, "Well, but I remember grandpa had tattoos so that he was in the machine there or there was a mask to keep things fixed." So maybe just sort of double back on that so people don't get confused.

Ehsan Balagamwala, MD:

Yeah, that's a really good point. And so the tattoos and the masks keep patients within about five millimeters of what we planned, but that's too much if you ask most of us. And so today my goal is, especially with SBRT, to have patients within a 10th of a millimeter and a 10th of a degree rotation.

When I talk about moving patients, I'm really talking about utilizing what we call a 6D couch or a six-dimensional treatment couch, which can not only move the patient in X, Y, and Z axes, but also rotate around those axes to get that ideal positioning so that we can target the tumor and avoid the normal structures as accurately as possible. And these couches are pretty incredible. They move patients to less than a 10th of a millimeter.

Dale Shepard, MD, PhD:

Which is really pretty impressive. Look, I mean, who am I? I give pills and IV therapies that treat everywhere the blood supply goes. And on the flip side, you're looking at millimeter and saying, "That's too much."

Ehsan Balagamwala, MD:

That's too much, right? Right. Yeah. And it's really been revolutionary because when I was in training, this was just being developed and within a millimeter it was okay. And now we're looking at a 10th of a millimeter.

Dale Shepard, MD, PhD:

Does this play more into importance, that being very, very precise, does that play more into avoiding normal structures, like if you have something close to a heart or an esophagus, or is this really across the board something we're trying to achieve in any tumor?

Ehsan Balagamwala, MD:

Yeah, so that's a really good question. So patient selection is really important. To me, when we think about adaptive, or at least in the last five years when adaption has really picked up, a lot of the focus in the community was normal tissue toxicity. And truth be told, our toxicities are not terrible. The grade three, grade four toxicities are actually very low because radiation oncologists in general are pretty conservative when it comes to that.

And so I do think adaptation will potentially reduce toxicity, but I think the biggest gain that we're going to get is going to be, and this is hypothetical, but the biggest gain that I think we're going to get is increasing the radiation dose to the target, better covering the target and achieving better control so that systemic therapies, they're really important in as much as their whole body, they're really important. And adaption will truly help us leverage the excellent advances in systemic therapy to better local control with radiation.

Dale Shepard, MD, PhD:

I guess I was just, again, being somewhat of a simple thinker. Surgeons do en bloc resections and have wide margins. And I guess the question is is a little over spray a bad thing? And it seems like that's a question always with radiation. How do you think about those margins?

Ehsan Balagamwala, MD:

Yeah, so those margins are really important. When the surgeon's taking out margins, they're looking for microscopic disease. And typically in radiation oncology when we're talking about what we... The jargon is PTV or the planning target volume. Those margins are actually setup margins, meaning day-to-day variations in setup. Whereas the microscopic margins won't change with radiation.

The CTV or the clinical target volume that covers the microscopic margins, those stay the same, except that every millimeter of PTV that you do because of the spherical pi r cubed, every millimeter leads to a cubic amplification of normal tissue that we're unnecessarily treating potentially because of our inability to be as exact. And so the adaption allows us to shrink those margins, not really impact the microscopic margins that we treat with radiation and the surgeons treat with an en bloc resection.

Dale Shepard, MD, PhD:

What's the technology that has enabled us to make the plateau change to more machine-based adaptive therapies?

Ehsan Balagamwala, MD:

Yeah, so several advances. And so what led from conventional radiation to SBRT has been better stereotactic immobilization of patients and the tumors, whether it's breath hold or knowing how much the tumor's moving. But the real big iteration with adaption has been the game changing of the workflow. Just to backtrack a little bit to describe the radiation oncology, typical workflow is after the planning scan is done, the physician contours the target and the normal tissues.

And so for a simple case, it could be as little as 20 to 30 minutes. But for a more complicated head/neck or a pancreas case, it could be three to four hours of physician time. And then we give it off to our planners who plan. And so for a simple case, it could be done in two to three hours. For a complicated case, it could take about a day or two. And that's a lot of computing power that they're using, hundreds of radiation beams all being optimized for that patient.

And so the goal has been to reduce that time as much as we can because you can't have the patient laying on the table for four hours while we're doing our work. And so there's been a couple advances. One has been what I would call as help that we're getting from AI or artificial intelligence. And so for the physician part of things, the contouring, the AI software can actually utilize my contours on the treatment planning scan and adapt them on the newest cone beam CT that the patient got, which with the HyperSight technology on the Ethos, that cone beam CT happens in six seconds.

And so the AI uses what it sees as my contour on the treatment planning scan, compares it to the current cone beam CT and adapts it to that. And so what typically would take me three hours to do, the AI can do that in about 30 seconds to a minute. And of course, it takes me to verify that. Because notoriously, especially in the upper GI tract with the bowel moving and the liver moving, the AI makes mistakes.

And so for me, typically on the Ethos machine for a complicated case, it takes me about five to 10 minutes to adapt those AI contours because most of the work is already done both prior to treatment as well as with the AI, which has done it within 30 seconds.

Dale Shepard, MD, PhD:

So, I have a three and a half hour savings in time.

Ehsan Balagamwala, MD:

In savings in time, right? Because I've put in the time upfront and AI is using what I put in upfront and adapting to it. And the second part has been that dose calculation after the contours are done. And so what used to take two hours per iteration can actually be done in about five to 10 minutes, depending on the complexity of the plan. So if it's traditional IMRT, it's about five minutes.

If it's volumetric modulated arc therapy or VMAT, which uses hundreds of beams rather than nine or 10 static beams, that's about 10 minutes. And what's really interesting is I know we're in the midst of a chip shortage nationwide and worldwide, but as these chips are getting better and the algorithms are getting better, that dosimetric calculation time is getting shorter and shorter.

So over the last even couple of years with the Ethos, the time to do the dosimetric calculations has gone down by about a third. And so over the next year or two, I anticipate about two and a half to five minutes and we have a good treatment plan.

Dale Shepard, MD, PhD:

That's impressive. So certainly we have a lot of savings on the front end in terms of physician time and time for planning, things like that. You can have more throughput through the machine, I'm guessing, because of that.

Ehsan Balagamwala, MD:

Potentially. So the time after the simulation has not changed much, but when you're using a set of contours that you already put three, four hours planning and the AI is just using that as a backbone, it can adapt to the new contours much faster. So at the treatment, if I were to adapt without the AI, I think it would be prohibitively long.

Dale Shepard, MD, PhD:

When we look at what has happened since AI has become more involved with these adaptive planning, has there been anything different clinically, less toxicity, outcomes from an oncologic standpoint? What has been identified at this point?

Ehsan Balagamwala, MD:

So the outcomes is very difficult to gauge. I think we're still learning about how AI is going to impact outcomes. But what AI, and I've incorporated AI both in my personal life, but also professionally, it's been pretty incredible. A lot of skepticism out there. And healthy skepticism is a good thing. But the way I look at AI is it's helping me do my job better. And what I mean by that is when I think of AI as part of Ethos, I think of it as one of my best residents ever who never gets tired, but still needs supervision.

And so I can't just let it do its thing because it does make mistakes. It does still need me. But when I leverage AI, what it lets me do is do the most important things that an oncologist needs to do is make good multidisciplinary oncologic recommendations and treatment decisions, but most importantly, take care of the person in front of them and take care of that patient. Because that human touch, no matter how good AI gets, it's not going to hold my patient's hands through their treatment.

Dale Shepard, MD, PhD:

The physicians that are skeptical, what's their biggest area of concern? Is it the fact that they see an error and they still have to correct what's the point in this? What's the skepticism?

Ehsan Balagamwala, MD:

It's hard. The sense I get with AI in general adoption in our society is people are worried that it's going to take their jobs away or make them so efficient that they have to do more during that. But I look at AI slightly differently. I look at AI that say if I have to write a paper, for instance, or a letter to the insurance company, if it took me 30 minutes to do that, I could have AI do it in about two minutes.

Or I could still take that 30 minutes and leverage AI to make that letter so good or that paper so good that it's perfect. So I think AI, the biggest advantage it gives us is to help us do our jobs better rather than to replace us. It adds to efficiency, but I think it's making our work more meaningful and better.

Dale Shepard, MD, PhD:

How about the patient side? Do you get much feedback from patients about how they love the fact that you're getting the help or they're a little, again, leery of the fact that the computer's planning their treatment?

Ehsan Balagamwala, MD:

Everyone's slightly different. There's some people who are like, "Hey, don't use any AI." I use AI for taking patient notes and stuff. And I've really noticed that the way I connect with patients now with AI notes is different. I don't have to jot notes anymore while I'm talking to them. I'm just talking to them like we are right now.

And so nobody's ever been like, "Hey, don't use the AI scribing tool." I do tell them and I show them the treatment plan that, "Hey, I had help with AI." And so some people are more skeptical, others aren't. But I think at the end, once they realize that, hey, I was able to do my job better as a result of that, I think they're all very satisfied and happy.

Dale Shepard, MD, PhD:

So we have recently gotten this Ethos adaptive therapy capability. If we think about types of tumors, things like that, what do we have at our disposal here that kind of sets us apart from some other places?

Ehsan Balagamwala, MD:

Yeah, so adaptive radiotherapy is being adopted worldwide. There's both CT-based and MRI-based adaption ability. So the Ethos system is CT-based. I think the best way to answer your question is perhaps with an analogy. If you were overly generous and you bought me an F1 race car, just because I'm driving an F1 race car does not make me better than Lewis Hamilton, right? He could still beat me in a Corolla. It's a tool in our toolbox. It really comes down to the team.

Oncology and radiation oncology are a team sport. And so what differentiates our program from other programs, really I look at it as threefold. One is the scale and standardization. So we treat a lot of folks. We have a lot of experience. And I know my colleagues have been on your podcast before talking about the workflow standardization that we've taken on in the department, is that every patient with that particular disease is treated in a very standardized manner, whether they see me or one of my partners.

And thereby if I'm out, I can be comfortable that, hey, when one of my partners is taking care of them, they're going to do an excellent job the same way I wanted it. The patient's going to get excellent care. The second is all of these decisions regarding whether to use adaptive radiotherapy or SBRT or conventional radiotherapy are made in a very multidisciplinary manner.

And multidisciplinary classically in oncology has been the medical oncologist, the surgical oncologist, the radiologist together make that decision. But for adaptive, it goes one step further. It's multidisciplinary within radiation oncology where we involve the physicists, we involve the dosimetrists. And we look at the scans together and be like, "Hey, would adaptive be appropriate for this patient?"

Just because we have the tool doesn't mean every patient needs to be on this machine. It's not appropriate for everyone, but having that tool allows us to offer it to the right patient. And the third is the significant integration of physics and dosimetry in our program where we've been planning this for a year before we got the machine to look at practice scans, do the workflow and see where the pitfalls are and correct them before ever treating our first patient.

And so we've actually been able to roll out Ethos adaptive radiotherapy from day one at a very high level, treating some of the most complicated liver and pancreas tumors essentially from the first day that we had it available.

Dale Shepard, MD, PhD:

Nice. When we think about those tumor types, are there certain things that you've encountered that seem to be oftentimes best with Ethos instead of the other approaches?

Ehsan Balagamwala, MD:

Yeah. So when it comes to adaptive radiotherapy, there's two big picture scenarios where I find that it's helpful. So one is where normal structures are moving. So, whether it's peristalsis in the bowels, or if the tumor is close to the bladder and you can have bladder filling while the patient's on the table for 30 minutes, or by the rectum where there might be some rectal gas that can abut the target, like for prostate cancer or bladder tumors. The other is if the tumor's moving.

So I have a patient with a high rectal tumor that's not only abutting the bladder, but it's also in a part of the sigmoid by the peritoneal reflection that any sort of peristalsis above in the small bowel or lower can make the tumor move. And so he's perfect for adaption. And in as much as I'll be treating him with palliative radiation, using adaption has allowed me to make it a smaller radiation field, thereby preventing more side effects, but also dose escalate to achieve a good local control in a person that has a seven, eight centimeter tumor that needs more radiation dose.

A lot of the questions I get is how do we decide whether are we going to use adoption in this patient or conventional approaches, IMRT, SBRT? It's really the abutment of that tumor close to normal structures as well as some tumors like the head/neck tumors, for instance, that can change size. So HPV-positive head/neck tumors respond very nicely to chemo and radiation, and so they can have a drastic change week to week.

And so those patients are also really great candidates for Ethos adoption. And one more thing that I'll mention about the Ethos adoption is probably an under discussed topic within adoption is simulation-free palliative treatments. So as we talked about every patient needs a simulation so that we can plan, but in patients that have bone pain and significant palliative needs, if you can minimize the number of visits and help them quickly with their pain, that's a huge advantage in terms of quality of life for the patient.

And so with the Ethos, you can do the cone beam CT in six seconds, plan right there and deliver treatment within about 30 minutes. And so you've saved the patient an hour of department time, not the travel time, but an hour from a simulation perspective and time until we can bring them back for treatment. And so there's a day or two for the planning and then they come back, whereas we jump straight right to the treatment plan. And so faster pain relief, more convenient for the patients, so overall a huge win in the palliative setting as well.

Dale Shepard, MD, PhD:

What's going to be the next big break?

Ehsan Balagamwala, MD:

I think we're going to see a lot of advances. In the short term, in the next year or two, I see adaption becoming much faster on the machine like we talked about. The processing power is going to get better. The AI is going to learn and get better at giving me a contour product that I'm more likely to like quicker and fewer edits. And so that will shrink the time from about 40 to 50 minutes down to hopefully 30 to 40 minutes.

10 minutes is a big deal when the patient's in the treatment position. So that'll be in the short term. I think in the long term, we're going to start seeing changes in how we do radiation. And what I mean is changes in fractionation. So we used to do radiation for prostate cancer eight and a half, nine weeks. We've come down to five treatments every other day.

But I can see things with combination systemic therapy that we're doing radiation once a week for three or four weeks and leveraging that with adaption and watch that tumor shrink, those normal structures go away from the tumor and you're able to dose escalate. And so a lot of exciting things that I think are coming in the next five to 10 years.

Dale Shepard, MD, PhD:

Very nice. Well, sounds like it's been a great addition as a tool to help our patients and appreciate you coming and giving us your insights.

Ehsan Balagamwala, MD:

Absolutely. Thank you, Dale, for having me.

Dale Shepard, MD, PhD:

To make a direct online referral to our Cancer Institute, complete our online cancer patient referral form by visiting clevelandclinic.org/cancerpatientreferrals. You will receive confirmation once the appointment is scheduled.

This concludes this episode of Cancer Advances. For more podcast episodes, visit our website, clevelandclinic.org/canceradvancespodcast. Subscribe on Apple Podcasts, Spotify, or wherever you listen to podcasts.

Thank you for listening. Please join us again soon.