David Raubach: I want to thank you all for joining us on today’s episode of the Cancer Project podcast. We’re really privileged to be joined by Mark Malin, the US president of RaySearch Laboratories, which is one of the most important companies in the world of radiation oncology. And we’re going to talk about why that is here. And then we’re also going to talk about all things treatment planning. Mark is a dosimetrist by training, but he’s also very good at translating what is a very complicated topic into a way to understand it for the average patient. And so that’s what we’re going to do today. So Mark, thank you so much for joining us.

Mark Malin: Thanks for inviting me. I’m glad to be here.

David: Yeah. So just to kick off, tell us about RaySearch since we’re sitting there at the RaySearch Laboratories headquarters in the Empire State Building there in New York City. What is RaySearch? What do they do? Talk a little bit about the history of the company.

Mark: Yeah, sure. So RaySearch Laboratories is actually based in Stockholm, Sweden. That’s where the development for our software is done, and they were formed — we were formed — in 2000 as a company, and then in 2011 we opened a US branch called RaySearch Americas. That’s the organization I’m the president of, and that’s based in New York and we also have an office in California. So RaySearch Laboratories is primarily a company that creates software for radiation therapy planning. There are a couple of different products that we do — both oncology information systems and treatment planning, which I’m sure we’ll get into here. And there’s about 450 employees total. I have an organization here in the US of about 40, probably going up to about 50 people this year. It’s been a great run and it’s a great company to work for.

David: And talk about the scale of the services that they provide — their customer base both here in the US and worldwide.

Mark: Right. So we have about 1,300 facilities around the world that provide cancer treatment, and most of those are photon treatments — your conventional linear accelerator that you find in many hospitals around the world. On the proton side, which of course you guys are doing protons and I assume that a lot of the patients that’ll be watching this are under treatment for protons as well — we have really most of the centers around the world. We have a very powerful product — just for example, in the United States there’s 47 or so operating proton centers and there’s others that are in development, and we have RayStation in all of those centers except about three. The product was released in the United States in 2011 and for protons in 2014, and it’s got a lot of great features that really make it safer and better for patients to be treated with.

David: So talk about your background, and I’d actually love to start with your growing up because you’re from New York. You had a little bit of a challenging upbringing. Talk a little bit about just growing up in the Bronx and then ultimately that path that led you to becoming a dosimetrist.

Mark: Yeah, it’s a fun story, but I’ll keep it simple. So I did grow up in the Bronx. My father died when I was quite young — about seven years old. My mother was disabled. One of those life stories that could kind of go any which way, so to speak. And ultimately I made the right life choices. I started taking martial arts, as a kid in the Bronx that had to survive. Went to the right schools with the help of my grandparents and then eventually made the choice to go to university at Stonybrook in Long Island. It wasn’t that far from the Bronx, but it felt very far. And I wanted to be a physician. Actually, my original intention was to become a medical doctor.

When I went to college, I just learned about myself as a leader. I was the president of an ambulance company. I was a resident assistant. I did all the things that leaders kind of do. And then I started doing research in medical physics, which is what brought me into radiation oncology — because to become a physician you need to have some sort of background in research. But I just loved radiation oncology. It was a mix of medicine, of computers, of dealing with patients — and it was just a great place to be. That was in 1989. So things were quite different then, but I just stayed in radiation oncology and I trained, became a medical dosimetrist. I was certified in 1994.

For those folks that don’t know, a CMD — or medical dosimetrist — is the person that actually operates the computers and the software that do the dose calculations and helps the physicians figure out what your treatment is going to be. So that’s what I did for quite a long time until 1997, when I was hired by a company to work as an application specialist.

David: That company was ADAC, and as the story goes you were working at a hospital and requesting service from ADAC, and they were a little bit slow to respond, and you said “what’s the problem?” And they said “well, we need more service people.” And then within a few days you had been hired by that company. Talk a little bit about ADAC because they were really one of the early pioneers in treatment planning.

Mark: Yeah, it’s so strange how there are certain times within the development of software and medicine that really make a difference. At about the 1997 time frame, a new algorithm came out — it was called the superposition convolution algorithm — and it was just a way to calculate the radiation dose in a much more correct way than had ever been done before. The original dose calculations assumed that everyone was a square box of water. That’s just one example of how things were different back then.

ADAC purchased a company called Geometrics, and there were quite a few famous names in that development. Their product was called Pinnacle. ADAC was then purchased by Philips in 2000, but Philips really didn’t buy ADAC for radiation therapy — they bought them for nuclear medicine. So I was with Philips from 1997 until 2007. And your story is exactly right — I was working in a hospital, I needed some help, I called them up. Took them a while, and I said “what’s going on?” And they said “well, we’re looking for people.” And I said “what kind of people?” “People like you.” Literally 48 hours later I was in California.

I realized then I loved working with patients as a dosimetrist, but when a company said “hey, you could come here and make a much bigger difference” — it’s not just the patients in your center you’re helping anymore. By making the product better and by helping hospitals use the product, now you’re helping an exponential number of patients. I was really happy to make the move.

At its high point when I left in 2007, we had 1,400 centers just in the United States. So it was an amazing experience. I built an organization of support, I was in charge of physics training — it was very hands-on and very involved.

David: I’ve read that there was a partnership between Pinnacle and RaySearch — they were working on IMRT, or intensity modulated radiation therapy planning. Talk about that relationship.

Mark: There was a gentleman named Johan Löf — he’s the founder and CEO of RaySearch Laboratories. At the time, RaySearch created algorithms for other companies. They didn’t have their own treatment planning system. They were primarily a laboratory that created algorithms and then sold them to companies like Philips, Varian, IBA, and a few others. And their algorithms were incorporated into the Pinnacle treatment planning system to help with a new technique at the time called IMRT.

And just in 60 seconds or less — in the old days we used to aim at a tumor with a square or rectangular field and use lead blocks to protect healthy tissue. Then with the advent of what’s called multileaf collimators — these are basically leaves that interdigitate — you had a lot more control over what area of the body you’d expose to radiation. By taking a whole bunch of these little leaf motions, you could build the perfect dose distributions. You no longer needed the blocks, and you were treating just the tumor and not the healthy tissue. RaySearch Laboratories was just key to that whole algorithmic development.

David: And you were actually a block cutter at one point in your career.

Mark: Yes, so in 1989 I started in medical physics research, and then in about 1991 I started training in dosimetry, and one of my jobs was to actually cut the blocks. The doctor would draw the shape on the X-ray film, and then I would use a piece of styrofoam with a hot wire, cut the block out, pour lead into it, and then attach that to a tray. It was a whole physical process that was — looking back — pretty crazy. But it was the best that we could do at the time. So yes, there was a lot of history, and knowing where we came from makes it amazing to see where the technologies have gone now: protons, IMPT, intensity modulated proton therapy. It’s just an amazing time.

David: So I’m a patient, and my radiation therapist is saying I’m probably getting two fields. But then my buddy I’m sitting next to in the lobby every day is getting three fields. And this lady down the hall is getting one field. Why are we getting different numbers of fields?

Mark: It really depends on the disease and where the tumor is, what’s around it, how big you are. There are also energies of the radiation to consider — higher energy goes deeper. Protons are really interesting because you have this delivery of spots and different energies you can use with intensity modulated proton therapy. It’s mostly about the same concepts as on the photon side: based on where the tumor is and what’s the best way to treat it without impacting healthy tissue — that’s really the determinant.

A good dosimetrist is going to sit there and think out of the box. And this is really where AI is going to help out — not the scary kind of AI you hear about in the news, but training the planning system to do things that we know were successful in that facility, with that physician, in that environment, at least as a starting point. The future of AI is going to take everything into account — blood tests, medical history, medications, family history, whether prior treatment was successful. There are so many variables that impact whether treatment is going to be super successful, and we just don’t know all of them yet.

David: So talk about the progression from 2D planning to 3D planning to IMRT — what each of those mean.

Mark: So 2D planning — the original assumptions were that everyone was a block of water. When we talk about 2D it just means one single plane. Let’s say someone was being treated for prostate cancer. We would do a tracing of their external contour and basically on a piece of paper we’d have a drawing of a person in cross-section and then we’d use a computer to calculate what the radiation dose would be within that drawing. But it didn’t take into account curvatures further up. A really good example is breast — you would take a tracing right through the middle of the breast, but the breast is conical tissue, so above and below that middle part there’s different amounts of tissue, and radiation scatters around in ways that the 2D calculation couldn’t capture.

3D planning was where we took a CT scan of the whole patient or at least the area being treated, and now the treatment planning systems were able to calculate dose much more correctly. When you’re trying to treat a tumor you’re trying to kill the tumor but protect everything healthy around it — all the organs at risk. In the early days you just couldn’t go as high of a dose because you were so concerned about what you’d do to healthy tissue around the tumor. But once 3D treatment planning systems came out, you had a much better idea of what was going on.

Then IMRT happened. With the MLC leaves, now you had intensity modulation. You could have anywhere from five to nine beams on average, and in each of those, you created that perfect dose distribution because the MLC leaves were able to move and really come up with just an amazing plan that was going to be good for the patient’s specific problem. And translating to protons — IMPT, intensity modulated proton therapy — you shoot these little directional small beams almost like spots and then you build the perfect dose distribution through optimization. That’s where computers are so important. That’s RaySearch’s strength — the optimization and dose calculations.

David: So one of the questions we get is: how do we know the right amount of radiation is getting to the tumor? I’m a patient — I don’t see anything, I don’t feel anything, I don’t hear anything. How do I know?

Mark: That starts way at the beginning. When a treatment planning system like RayStation is going to be implemented in a hospital, the physicists do a great job measuring all the radiation in phantoms, which are then put into the treatment planning system. There’s tons of research proving that all the treatment planning systems out there are safe and that the dose calculations are right. Then you start doing tests — there are phantoms from MD Anderson that are sent out, you expose the phantom to prove the dose calculations were right. There are crystals sent out annually to prove that the amount of radiation you think is coming out of your machine is actually coming out.

So there’s an ongoing daily, monthly, quarterly, annual quality assurance. When it comes to delivery, there’s always secondary dose calculations — other products will take the plan information, calculate it outside the treatment planning system, and say “yes, that’s the right amount of dose.” And on the delivery side, there’s EPID-based dosimetry where you look at the dose that exits from the patient and analyze whether that’s what you expect. There are log files that record exactly what happened and when.

With protons, you know, it’s like measuring water coming out of a faucet. There are even devices built into the actual delivery snout — ion chambers that are literally counting the number of protons coming out of the snout going into the patient.

David: One of the things that’s also a challenge is reproducibility. I’m a person and I breathe, and even when I’m laying on the table getting treatment, I’m breathing. There’s this concept of motion management.

Mark: Yes, and that’s another whole conversation in science. So for protons we use something called robustness — when we identify the target, we look at what kind of motion is expected there. And then we tell the system: “this thing might move by a half a centimeter or a centimeter,” and we do a plan that if the tumor does move that expected amount, it’s still going to be treated.

From a therapist perspective, there’s a lot of interaction between the patient and the immobilization equipment. Sometimes we just start and we’re like “this is not going to work” — we may have to remake a mask, especially if you gain or lose weight. Sometimes there are breathing holds we ask patients to do if they are able. There’s surface imaging cameras in the treatment room that track the patient’s motion over time, and if it gets out of tolerance, we can shut the beam off and get the patient set back up.

As imaging gets better in the rooms, even while the beam is on, if the patient moves, the beam could be interrupted and say “stop, let’s give the rest when the patient’s back in the right place.” These are all technologies still developing, but that’s definitely where we’re headed.

David: Talk about online adaptive therapy — what does that mean?

Mark: So you know, when a patient came in for treatment back in the ’90s, they would lay on the table, you would get tattoos, you would line up the patient exactly in the room using lasers, and then take an image to verify position. But the reality is that the original treatment plan done three weeks ago — when the patient first went in for their simulation CT — that tumor may not be in exactly the same place anymore. There’s always been that assumption: we have to replicate exactly how the patient was set up at simulation, every single day.

Online adaptive therapy is where we don’t have to do that anymore. We’ll actually just do a new treatment plan each day, on the table. So we don’t have to completely reproduce what that patient looked like the day before — we’ll just plan that day based on a new CT taken right in the treatment room. Now there’s things that make that complicated — when you’re changing things up every day, you have to keep track of what that means over 20 or 30 fractions. The computer has to say “here’s what that CT scan looked like on day one, here’s the cone beam CT from day eight, compare the two, overlay the two, add the doses together, and let’s make sure that as we go out to the prescription we’re going to do what we thought we were going to do.”

I want to make sure patients realize there’s a lot that goes on behind the scenes. That’s why it’s taken so long to get here — the computers just needed to be able to do all this stuff, and now we’re there.

David: So AI — how is that going to change everything about radiation oncology and the way it’s delivered going forward?

Mark: Online adaptive is such an important thing, and AI is really enabling it. AI is going to enable much faster response: when it comes time to look at where the tumor is that day, rather than a person sitting there going “okay, yeah, that kind of looks like it,” AI is going to take care of a lot of that. The physicist and dosimetrist are going to be less concerned about the little menial tasks that a computer can handle, and they’re going to be more involved in case management and deciding what’s the right way to treat the patient with the physician.

And then in the future, when we have other information — histology, medications, all of these variables — I can’t wait for the day when AI is just going to take all that information from your oncology information system and now you have a feedback loop that’s going to make every treatment better and every next patient’s treatment better.

Each person is like their own little large language model — there’s all these inputs: what drugs are they on that can impact how radiation interacts with cells, what other comorbidities do they have, have they had prior radiation, how old are they, do they have specific genetic mutations? There’s just so much data, and you need incredibly powerful, well-trained software systems to process all of that and then spit out a practical output.

Right now, the physicians are the natural intelligence. The physician is the one using her experience based on what she knows, what she’s studied, her previous patients — she’s the model of decision making. In the future, AI is going to take in a lot more variables, and she and AI are going to make those decisions together. That’s really going to be great because it’s very difficult to take all of these variables into account. Why not take all that experience, especially the good experience, and then use that to train the systems? And that’ll give physicians more time to just interact with patients — because that’s what a computer can’t do. It’s that relational element.

I think people are afraid of AI — dosimetrists in particular. “Oh my god, the computer’s going to do all this, I’m going to lose my job.” But I was with Philips in 2000 when IMRT was being introduced, and the arguments were exactly the same. And what we learned was our jobs became more difficult — but more sophisticated. I think in the future, everyone’s job is unfortunately still safe, and we’re going to be needed to work with the patients and work with the medical experts on what’s the right approach. Because we’re all still learning. There’s going to be new developments about what treatment techniques are needed, and you’re still going to need humans to implement those.

David: Well, this has been really good. I think this is going to be extremely helpful for patients. I can’t wait to get this out. I do have one last question for you — you’re the president of RaySearch Americas, and you’re also the unofficial president of the radiation oncology karaoke association. How did that come about?

Mark: I look at karaoke as the ultimate human exposure of weakness — none of us are expert musicians, or at least most of us aren’t. And there’s a real human side to it. One thing I really love about RaySearch and about working with people like you, David, and people that take care of patients — all of our academic partners, all of our hospital partners — is that everyone really cares for their patients. And I think the relationship between companies and clinicians should be strong. So when there’s a problem, you can call me up and say “Hey Mark, I know we sang Bohemian Rhapsody last night, but I have a problem I need help with today.” It became a thing because we have to keep the human side going. I really think some of my favorite relationships are based on singing in front of each other and feeling comfortable doing that.

David: If you had to sing a song today right now, what’s your go-to?

Mark: I’m a big Billy Joel guy. He’s got this great way of looking at life in so many different ways and he’s so human. Any of the Billy Joel songs, I would sing for you.

David: Well, thank you so much, Mark. I really appreciate it — and thank you for everything that RaySearch continues to do for us at the Proton Center and for the industry.

Mark: I appreciate the opportunity. I love sharing knowledge, and by all means if there’s anything in the future that we can help with to educate patients, we’re always there for them.