Chapters Transcript Video Step-by-Step Look at Online Adaptive Radiation Therapy Hi, I'm Doctor Kate Mittower. I am the technical lead of the MR guided online adaptive program at Miami Cancer Institute. And hi, I'm Doctor Michael Chong. I'm a GI radiation oncologist and the medical director of radiation oncology, uh, here at Miami Cancer Institute. Today we're excited to walk you through our online adaptive workflow which starts with image acquisition to treatment verification imaging and treatment delivery. This is with the Meridian MR LiNAX system that we will be demonstrating today. So this is a patient who was diagnosed with a locally advanced pancreatic cancer, which means she didn't have metastases, but her tumor was too locally extensive around the blood vessels for surgery to be feasible. So these patients typically get chemotherapy first, and if there's no evidence of any distant metastases, they will often referred for radiation therapy. Uh, the standard radiation approach in most centers is a non-ablative or moderate dose approach, um, that really does not effectively, um, prevent long-term, um, uh, or does not, does not achieve long-term, um, tumor control. So typically 2-year control rates after that type of standard radiation dose is about 50%, which obviously is not very good. Uh, the approach that we are, have been using here for the last 8 years since our MRLINAC program, uh, started in early April 2018 has been an ablative, uh, 5 fraction, uh, regimen with the MR Linux with the VA system. Uh, and it's using a simultaneous integrated, uh, boost or SIB approach where 50 05 fractions is prescribed to the visible tumor or the gross disease and there's an elective volume, uh, that is prescribed 33 05 fractions. Uh, this is delivered in consecutive days as opposed to every other day which might be done at other institutions, um, and it's done using an online adaptive, uh, workflow. So what's been really exciting is the outcomes that we published from this uh dose escalated MRI guided approach which, which really are among the best in the literature. So as opposed to standard um uh radiation outcomes that um that typically results in about 50% of patients having local failures within 2 years, uh, the incidence of local failure within the same period of time at our institution has been. Around 6%. So if you flip that around, it's, it's local control rates of over 90%, which really are unheard of. Um, so very excited to be a part of that. We also think that this is helping patients live longer. So if we're preventing death as a result of local growth of tumor, which can cause bleeding and infection and other, other bad things, uh, potentially we're helping patients live longer. So the standard radiation outcomes in terms of two year survival rate is typically about 20%. Uh, the, the, our institutional outcomes and others that are using this dose escalated, uh, MR guided approach, uh, are closer to 50% or even above 50% within two years. So that represents, uh, roughly doubling of 2-year overall survival. The online adapted workflow and MR guided radiation is a very, um, team, uh, I would say it's a very team-oriented as opposed to an individual type of treatment. Um, because there are many factors within the workflow including physician, physicists, and radiation therapists. Uh, I'm fortunate to be a part of, uh, a world-class team here that has been really together for a long period of time, uh, to the point where we know exactly what the other person is thinking and the other step, the steps that are gonna happen along the way without even, even talking. So, um, you know, it's a very well-oiled machine that helps, uh, drive efficiency but also quality. Welcome to our patient demonstration. We will start with a brief technical summary. The patient is being prescribed 50 grain 5 fractions to the pancreatic gross disease with a simultaneous integrated boost of 33 gray and 5 fractions to the elective nodal chain. The plan is a step and shoot IMRT 19 beam arrangement with nominal. Delivery time just under 10 minutes. No immobilization is being used. Immobilization has less of an importance in MRI guided adaptive radiotherapy because patient day to day interfraction reproducibility is not as relevant. And we are performing continuous Senate MRI for inter-fracture motion monitoring. At approval of patient alignment starts the parallel workflow, where the therapist, physician, and physicists are all contouring simultaneously on 3 separate workstations. While our institution has kept the parallel workflow as an in-person at the console due to IT restrictions, remote capability is possible with this user interface. Welcome to our adaptive console area, which is an L-shaped layout. On the left is a therapist meridian workstation for the therapist's adaptive driver, in addition to an Ario workstation for the second therapist. On the other side of the console area is a physics meridian workstation and a physician meridian workstation for a parallel in-person adaptive workflow. OK, we're starting the adaptive workflow with our image acquisition. Um, this is a trophy scan. It's a balanced, uh, MRI scan. It takes 25 seconds. The patient is in a breath hold. Uh, the patient's also in a very comfortable position, as you can see on the screen. The arms are down by side. There's no immobilization being used. The patient's just laying on a pad and has the coils, abdominal coils, uh, around them. The image you saw was 25 seconds, um, very fast for our breath hold purposes. You see on the screen, we have uh turned on the GTV contour in red, and then the um adaptive, uh, contouring ring, which is uh an expansion of 3 cm in the axial plane and 2 cm in the sid direction. So right now how our therapist is doing a GTB to GTB localization. The pink image is the SI image and the gray image is our image of the day. So, she's localizing there? Once she is complete, completed her localization, we're gonna do a check of the field of view and skin. The field of view is very important to check for online adaptive radiotherapy, um, because we are generating a new plan and any patient anatomy being cut off, um, can impact the dose calculation accuracy. So we are doing our skin check right now and our field of view check. So she turned on the ring for me. Yep, that's good. We're outside the ring now. Starting there is where we care about field of view and skin, and the skin and field of view are appropriate on every axial slice that the, um, ring is present on. OK. That looks good. You're good with the skin. Yeah, I'm good with the skin. Um, Doctor Chung, can you just check our GTV to GTV localization? Sure. That looks good. All right. So now we're going to move on to our contouring, um, Our therapist has begun contouring on the tracking structure and. And we have a parallel workflow where the 3 users can be contouring at the same time. So, I am contouring electron densities, Doctor Chung is going to be contouring. Um, the targets and the organs at risk that are relevant for this case. And then Paula is gonna be doing the tracking structures. So I'll just add a comment that I, um, the uh OARs are. Uh, deformed. Although I find it's a lot faster to just clear the contour if the, if there's going to be a lot of changes that are needed and then just interpolate. So that's what I just did here. If there are only minor tweaks, then I wouldn't do this. Uh, but I think it, I think it does save time this. If you start from scratch and just interpolate. Uh, the other strategy that we use when we contoured is not contouring all of the OARs, uh, but really it's just within, um, a certain distance of the PTB, which is the red structure that you see. Uh, and that really is because OARs outside of a certain distance are not going to influence whether or not the plan needs to be adapted. Uh, and that largely is because the OARs that drive the, um, decision making for adapting are, uh, the GI luminal organs. And, you know, those are going to be um Those are going to be a uh a volume receiving a certain high dose, and so that's going to be within a certain relatively close proximity of the targets. Uh, so that, I mean, that is all. All of those, um, that, that kind of mentality is all within. Uh, the theme or keeping within the theme of making the procedure and the process as efficient as possible, so we want to You know, focus on what is going to influence our decision making and the adaptive process. Uh, and it's, I think it takes a little bit of getting used to, if you're not, you know, if you're not familiar with the adaptive workflow and that. Uh, it's the goal here is not making everything look perfect, but it's really just focusing on what is important and what's irrelevant for ensuring that we accurately identify, uh, you know, what dose is going where, at least on the predicted plan, and then being able to make decision making in terms of, um. The adaptive process. Um, additionally, uh, some things that we do for our online adaptive program to make it more efficient and streamlined. As we actually do a bulk density MR only based approach for all of our abdominal uh patients, uh, particularly our pancreas patients. And so that's what you see here. Um, I am going over from the MRI scan of today to the electron density of today. So this is a bulk electron density assignment. You can see I started with the, uh, GI luminal air, and now I've tweaked the bone contour. Um, for the vertebral body. And lastly, um, I will update the lung, um, lung override cause we have a little bit of lung that is overlapped into the adaptive contouring, right. So just a quick comment in terms of contouring the GI organs at risk. So a very common. Um, mistake in a lot of ways is not contouring enough of the structure itself, and so specifically that is the entire thickness of the wall of that structure. So you can see here that I contoured all the way to the outside of the wall, common. Um, mistake in a lot of ways is, is doing something like this, and The trying to get the, the T2 bright contents within the lumen is not, is not, that's not what we care about. I mean, we care about the actual Wall of the organ itself. And so that's, it's, there's a a common tendency to under contour in, in, in a lot of situations, and I think just being cognizant of Uh, the fact that we need to contour the whole thickness of the wall is really important. Uh, and then you'll notice that I do include sometimes a little bit beyond exactly the wall of the structure. Uh, that is in part sometimes because of maybe some uncertainty if there was some motion artifact, uh, and I just want to make sure that we don't miss. And then two would be, you know, there are some small intrafraction changes that we see from time to time when we re-image. Uh, so in particular, the around the stomach is actually where I have, and this is just, um, a nuance in my practices, you know, I will actually end up more generously contouring around the stomach, um, especially here in the lesser curvature. Uh, and actually I could probably extend this out more, uh, because, you know, this is, this tends to be the area where, uh, we see the most infraction changes. Uh, as well as, uh, potential duodenum or potential ulceration. I mean, also here in the first part of the duodenum as well. I mean, if patients have GI toxicity, it almost always is in the pyloris of the stomach or in the first part of the duodenum. So I end up being a little bit more generous there just. Uh, just in case Um, I liked what Mike, Doctor Chung said earlier about, um, online adaptive contouring is figuring out how to contour accurately in the places where the dose is going to be impactful and clinically relevant, um, and that's something that, you know, we, as a hallmark of our program, is not having perfect contours, you know, particularly, um, away from the high dose gradient. Um And because it's important to go quick in online adaptive contouring because the patient's anatomy is constantly moving. So this GI anatomy with parasalsis is moving um every minute. And so if we spend a lot of time based on this scan, when we do our verification imaging, uh, we may see more uh potential of um movement, internal movement. Um, if we've taken more time for the online adoptive process. So in addition to um Doctor Chung is contouring the GI luminal organs at risk and the GTV and CTV, um, I do a couple of the um more uh standard straightforward organs at risk. So physics here at Miami Cancer Institute does the, uh, cord and kidneys. You can see that the deformation on the uh The canal did not, was not great. I just wanna make sure it included all of the canals. So it's OK if it's uh overgenerous, um, just wanna make sure that included everything. So I am done with my electron density portion. Um, I slip over to make sure the density assignments were appropriate, the lung, bone, and we have the skin, um, set to water, um, and this is all set from the initial plant. So I, you know, you'll, you'll see I'm, I'm not nitpicking the, I mean, there's a little bit of overlap here with some of the OARs. You know, it, it does not matter at all if I clean that up there. It just takes time. I don't want to take more time for that, so. You know, uh, you know, obviously we want to avoid huge volumes of overlap, but I mean things like Like what you see here is, it's really not. It's not worth taking the time. Although I'll say, you know, I think it is important, especially when you have See a transition between the stomach and the duodenum. I mean, there can be sometimes gaps in between those contours. I mean, I think that's something that You would want to actively avoid. I mean, I'd rather have some overlap as opposed to gaps, uh, especially in a higher risk area like where the stomach transitions to the duodenum, and as I mentioned before, that's where we typically see more of the toxicities, um, if patients are going to have ulcer or GI bleed, which are rare, but I mean that's. Most often where that happens. So I'll I'll show you, come back up here, I mean, you can see. There ends up being a little triangle in a way sometimes, and actually, if anything, I, I might actually. Just kind of fill this in OK. So, The, uh, the targets are regis rigidly registered. Um, so I will eyeball this first and then make adjustments as needed. Uh, you know, the volumes here are In a lot of ways different than what other centers may choose to use, especially with with respect to Uh, a CTV. And, uh, so. A good number of years ago we actually had moved to a routine anatomically derived CTV for these pancreas patients. This was based off a pattern of failure data from other institutions. And so this is specifically designed around expansions from vascular structures including the celiac artery and the portal vein, the SMA, and the aorta. Uh, and so this is a volume here that is prescribed to 330 5 fractions, and the gross tumor is prescribed 50,000 fractions using an SIV approach. Uh, The structure I'm editing here is essentially a, a baseline uncropped structure that then through rules is cropped out of the GIOARs. Uh, and, you know, having an asymmetric CTV like this, you know, we, we find there's a lot of time savings when we use this approach as opposed to starting with the crop structure because you can see how complex this structure is. I mean, it's hard to really replicate this from scratch each day, uh, if you don't have a baseline that's, um, a standard volume to start with. And it's especially difficult if you have coverage from different physicians because, you know, they may or may not be as familiar with what the CTV should look like. Um, so this really helps standardize that across the board. So there's a rule built in then for cropping. You'll see I'm turning off the OARs, and I actually I'm only looking at the Targets You know, I started doing this more recently, uh, because I think that There ends up being, uh, sometimes there can be an underappreciation or it's more difficult to visualize where there may need to be some further tweaks. So, for example, here, I actually I'm gonna tweak the stomach. A little bit more generously. For that reason, because the, the 33 great volume gets pretty close there. So this is, this is something that I, I think is pretty helpful and Probably started doing this with the OARs being turned off. Uh, probably within the last, um, 6 months or so, but I, I think it is something that And really be helpful. So I am I'm happy with this. Right, uh, Paula, we're ready. So Paula is now advancing us on to the next phase of the workflow. Um, the only, um, the next phase of the workflow can only be approved by the main console which the therapist is driving currently. Um, while we're waiting for the plan to optimize, which you can see down here, Uh, I just want to talk about our planning technique, um, at MCI for Pancreas. So these are the planning rules that Doctor Chung was referring to, um, and as he talked about, we have the uncropped structure that is propagated from SEM rigidly, and we don't, um, want any of the GI luminal structures to be overlapped with our final GTB and CTV so we crop. The GTV and CTV based on the luminal wall, um, that he accurately defined today. As previously explained, the GIOARs are individually added together to create the all OARs contour. The GIPRV is created from a 3 millimeter expansion of the all OARs contour to create a GIPRV. Then in our planning technique we take GTV uncropped and CTV uncropped and subtract the GIOARs to create the final GTV and CTV. Additionally, we create 3 optimization structures of the targets a GTV opt, a CTV opt, and PTV opt. And this is the GTV, CTV, and PTV minus the all PRVs. And in planning, we are, we are driving this to get full dose. We're also using it in our Plan RX to make sure that the part of the GTV, CTV, and PTV not overlapped with the organ at risk is getting full dose, so it's helpful from a statistical perspective when we're reviewing the adoptive plan. Uh, next we create the PTV 50 and PTV 33. The PTV 50 is the GTV expanded 3 millimeters isotropically. The PTV 33 is the CTV expanded 3 millimeters isotropically. Um, we for conformality with our planning technique, we have, uh, two ring structures we use. We use a ring 2CM, which is a 1 centimeter, uh, width, uh, ring that is PTV + 3 minus PTV + 2, and this is helpful to control the 50%, uh, isotose conformality as well as a low dose ring, which is the skin minus the PTV plus 3. So those are our planning technique structures that we're using for our pancreas planning. The reason why we use PRM is not, not for dosymmetric reasons, it's actually to create a PTV op and GTB op, and that is what's used um to drive the full dose and as well as for our statistics to review. While Doctor Zhong was speaking about contouring, I reviewed his contours, so that is an important part in the adaptive workflow, um, that we always do as a second check of the contours by physics. You are looking now at the plan generated by Bray. It is, it has been completed. It's the fully re-optimized plan. The predicted dose, just to make a quick comment, is the initial plan recalculated on today's anatomy based on today's contours. So you can see we're exceeding the duodenum and the small bowel, and the coverage is a little less than the original plan. Um, the first thing I'm gonna do as a physicist, um, doing this adaptive is I'm gonna check my modulation. The modulation of the adaptive plan is, uh, a little less than the original, so that is fine. I don't want to get it overly modulated. The next thing I'm going to do is normalize to the closest organ at risk or optimize the plan as needed. So for this plan, I see my closest organ at risk is looks like it will be the stomach. Um, so I'm gonna go ahead and norm to the stomach. All right, and it's actually was for us to actually large bowels, so I'm gonna delete that norm. And non for a large bowel. 0.5 cc is 38 gray is our large bowel constraint. I'm gonna rename this for our Institutional techniques. And I, I'll quickly jump in here and just mention the OAR constraints that we use have been the same ones since we started our MRLIC program in 2017 or 2018 rather. So these were published in, uh, our initial pancreas paper in pro, uh, and we've stuck with it since. So it's been patients have done really well from a toxicity standpoint, um, and so, you know, we, we haven't changed it at all. So next, after um renaming my plan, I just looked at the uh QA. This is the secondary calc. Um, we see that we're passing at 99.5 gamma at 2% 2 millimeter. And now I'm ready to present to Doctor Chung. OK, all right. Ready for you, Doctor Cheng. I normalize to large valve 0.5 cc 38 gray. Um, stomach is well under, it's at 34. Uh, canal is fine at 16. Small bowel is good at 33, uh, liver is fine, large bowel no, kidneys mean are 6, duoum is under 35. Uh, PTP 50 coverage today is 81 compared to the original plan of 66. CTV is 99, uh, GTV is 92, uh, PTP 33 fully covered, and then PTV 50 op is 88. So you can see our coverage is significantly better than the original plan. Uh maximum dose is 74 compared to 69. And I would just comment, you know, the mean dose here is also obviously higher than prescription. So, you know, for 50 grade prescriptions, it's common to see a mean dose of 10 grade higher, uh, and you can see the CTB which is prescribed 33 as a mean of 40, uh, so the low 40s is very common for the CTB. All right, and then I will scroll through the plan. So we're looking here for hotspots, we're also looking at conformality of the plan. So I just have the PTB turned on and the isotose turned on. Um, Doctor Cheng and I have memorized the isotose, um, lines, uh, color, uh, we have standardized that. Yeah, I think that's also helpful for cross coverage. So if everyone understands what the isodas color language is, I mean, it makes it much faster to review plans. So the blue is the 120, the 50% is the bright green line. Exactly. So, yeah, I'm looking for a streaking of, say, the 50% and looking at the hotspot to make sure it's not, you know, directly abutting or too close to any organ that I'm concerned about, especially anticipating that things may shift some during the treatment, and this all looks really good to me, so I'm good again, uh, and actually one thing I'll add is when we walked through, um, when Doctor Matara walked through the, uh, predicted and then the re-optimized, uh, plan, you know, from a quality standpoint, and we, we've. Made a decision for that to be done just to make sure, um, you know, we do comprehensively look at, at everything together and, you know, we've kept that process. This patient we are using um a single sagittal plane. It is a um a 8 frames per second, um, so it's 8 frames per second. Um, it's a radial acquisition, um, and we, this contour is the one that she drew when we were performing a parallel workflow and contouring. So it's just that single slice, yeah, that works great. Um, the boundary is a 3 millimeter expansion, um, of that tracking structure, and this is our initial verification imaging if the, uh, sine con if the sine imaging looks off the patient's internal anatomy has changed. Nice, yeah, that looks great. And so she's flipping back between that 3D scan and the current sitting, and you can see that the hold right there, right. You can see that not only the tumor still looks the same, but also the surrounding stomach and bowel is also the same, yep, OK. And then she is starting to treat. All right, remember you're in control of that threshold. You're gonna hold for as long as you can, OK? If we see anatomy moving during the treatment delivery, we will perform a 3D volumetric scan and, um, shift the patient or need to readapt, um, as needed, how depending on how much motion we do see. So. Yeah. Reading about what I eat at home. Hold right there, hold it right there. And this is a gated delivery that's the patient doing by themselves. Um, we do give them coaching the therapist. Um, right in the they have this right in the center, a mirror and monitor inside with a smiley face to, to let them know that they're in the right location, um, so it's self-guided. They hold as long as they can and they free breathe when they need a break. That was perfect. Breathe. We have them breathe intentionally between the gantry angles, and we have our planning technique is optimized so that it's typically 1 or 2 breath holds and 1 or 2 segments per beam, um, to make it. The beta delivery go faster. OK, and that concludes our online adaptive workflow. Created by
