Transcript
Bogdan: Hello, everyone, and welcome to a new Novalis Circle webinar. My name is Bogdan Valcu. I'm the director of Novalis Circle. And today, I have the privilege of welcoming Dr. John Gleason and Mike Taylor from Alliance Cancer Care in Huntsville, Alabama. Together they will present their experience with elements Multiple Brain Mets SRS, specifically focusing on new developments with version 2.0 and improvements for the Elekta configurations, and highlight the potential clinical impact of such improvements such as dynamic jaw tracking for the Agility MLC. A review of the clinical program would also be provided together with considerations for single fraction versus fractionated radiosurgical treatments. Dr. Gleason is a radiation oncologist and the medical director of the Alliance Radiosurgery Program. Mike Taylor is the chief of medical physics as well as the director of medical physics and dosimetry.
As with our previous webinars, we continue to provide CE credits. If you are in need of CAMPEP, MDCB, or ASRT credits, please follow up with us at info@novaliscircle.org upon successful completion of this webinar, and we'll be more than happy to help you with obtaining your credits. And also don't forget to sign up for our upcoming webinar with Dr. Prasad from Roswell Park in Buffalo, New York who will discuss response assessment technologies and the use of our contrast clearance analysis tool. This upcoming webinar would be held on July 9th. Lastly, please remember to use either Google Chrome or Safari to log into the webinar. Should you have any technical challenges, please refresh your window. Utilize the chat line to send us questions that we will answer upon completion of the two lectures. We will also utilize the chat line to pose some questions to you, and we will review your answers in the Q&A session as well. And if you'd like to follow us on social media, please use the hashtag provided. With that said, I'd like to turn it over to Mike Taylor.
Mike: Thanks, Bogdan, for that introduction. As you mentioned, my name is Mike tailor. I'm a physicist. I'm joined by Dr. Jack Gleason. We're from Alliance Cancer Care in Huntsville, Alabama. And today, we're here to talk about "Evaluation of Enhanced Solution for the Treatment of Multiple Metastases with a Single Isocenter," which is a long time for us to talk about our experience with Brainlab's multiple mets element as well as their VMAT cranial element solution. Just a quick disclosure, I'd like to thank Brainlab for inviting us to present this material, as well as they provided a speaker fee, and then we have a research agreement in place to evaluate new versions of the software. Some quick acknowledgements. Some folks and companies that have helped us with our program in terms of development, and then they've also provided some material for this presentation. I'd also like to thank the Alliance Cancer Care team, very fortunate to work with very talented group of therapists or cemeteries, physicists, and physicians. And, I think, they're, you know, really the core of what's made our program so successful.
So, this is our agenda for about the next 45 minutes, everything that we're gonna go ahead and cover, and I'll go ahead and get started. Just some background. We are a fairly busy community practice based in North Alabama. We treat about 200 to 225 patients a day on 6 LINACs. Huntsville has a population of just over 400,000. We're the primary radiotherapy providers in that area. Just looking back, you know, about three years ago in our historical SRS program, you know, we treated maybe one to two patients a month. We tended to treat things that were larger in size, and we avoided functional SRS targets. In terms of planning, you know, these things took hours to plan, and in terms of delivery, we were looking at about at least an hour per isocenter per target. So it could take hours for delivery as well.
So, you know, the physicians about three years ago decided to invest in a new SRS program. We began looking at SRS equipment, and we looked at, you know, Gamma Knife, CyberKnife, and LINAC-based options. And, for example, the CyberKnife, we really like the option to do non-coplanar imaging. The challenge for us in a busy practice is that it isn't as efficient as say a LINAC in terms of treating non-SRS, SBRT deliveries, which we need to do. We made the decision to go with an Elekta Versa HD that's equipped with an agility, and then with Brainlab's ExacTrac, you get that non-coplanar imaging ability to see what the patient is doing in six dimensions and as well as correct them with the HexaPod couch. This is a standard four energy machine, and then we use it for external beam when we're not doing SRS and SBRT. We also include a standard set of Aktina cones down to 5 millimeters.
One of the key components of this system is the Agility MLC. There's a 5-millimeter leaf width MLC, 160 leaves with just under 0.5% transmission. It's a very fast MLC, which I think is important when we're treating multiple targets with a single isocenter. Even though this MLC does...is a 5-millimeter leaf width, I just like to point out that in the leaf direction, these leaves can travel in 1-millimeter increments, and then the extras behind the leaves can also translate in 1-millimeter increments. So you can actually mimic a 1-millimeter pixel.
And just a quick snapshot of what our SRS program looks like today. With this system, we now treat, you know, more than 125 SRS and fractionated SRS cases per year, and that includes a full suite of functional indications such as acoustic neuroma, AVMs, and trigeminal neuralgias. We also have a very busy SBRT program and treating more than 100 SBRT cases annually. And this is just an example of one patient that we've done. This is a 9 target plan. And if, you know, we were to do something like this with our historic system, you know, it would have taken upwards of nine hours just for delivery for each of these targets. Now, this is something that we can do in an approximate, you know, 16-minute window with ExacTrac imaging.
Just real quick. I'd like to talk about accuracy and a little bit about QA. It's a very fundamental component when you talk about multi-target single isocenter treatment delivery. Historically, when you think about SRS and the single isocenter, you know, the Winston-Lutz is kind of the gold standard for verifying isocenter. The ExacTrac system does come with a daily isocenter verification tool, and that will verify the isocenter within 1 millimeter. On the days that we do a single fraction SRS, we will go ahead and do a full Winston-Lutz procedure and verify that, you know, the gantry and the couch are within about 0.65 millimeters.
In terms of commissioning, validation, and just ongoing QA, end-to-end test is a really good test to make sure your system's performing accordingly. This is a great paper by Tim Solberg where they took a bony anatomy, anthropomorphic phantom and inserted a 5-millimeter ball bearing and then basically created a plant-centered on that ball bearing. You then go ahead and set the patient up for treatment. You use ExacTrac for alignment. Then the symmetric amateur is centered on that ball bearing. You use that as a pseudo-Winston-Lutz to verify the accuracy of your treatment.
There are some other options for checking your delivery. MD Anderson provides an anthropomorphic phantom as well. The phantom in the center is kind of their long-term photon SRS phantom. And, I think, this is probably a good choice for like a frame-based system. But as you can see, it includes a cylinder that contains OSLDs [SP] and fill. The cylinder is inserted in the phantom, and then it's filled with water, which makes it really challenging when you have a frameless system that relies on bony anatomy for alignment. So this probably wouldn't be a good choice for something that uses bony anatomy image guidance for alignment. You know, looking over the MD Anderson website, I see that they offer a proton brain phantom, which looks to be almost a duplicate, but this does contain bony anatomy. So that might be a better option.
A company that we've more recently started working with is RTsafe, and they provide a really nice bony anatomy anthropomorphic phantom that you can insert TLDs film. It has cutouts for chambers. And then what's really unique is that they also have a 3D gel insert. So you can actually test multiple targets, and this is really good end-to-end test to test, you know...excuse me, a single isocenter multiple met delivery.
Some other tools we use in our clinic are, you know, simple geometric phantoms. One that we tend to favor is the StereoPHAN by Sun Nuclear. And you can see just a picture of it over here on the right, but it's milled for a small volume ion chamber that you can do point doses with. They also offer an SRS MapCHECK, which is essentially a high-resolution 2D detector to replace film to do some planar measurements. And then one of the more recent additions is this really nice multi-target Winston-Lutz Insert. So now you can actually, you know, get some accuracy about how your targets or multiple targets are being treated.
And then finally, you know, I think, follow-up imaging, as you're able to obtain it is a really good follow-through in terms of seeing where things are at with patient delivery. This is a patient that if you look at the left-hand image, we treated the right side of the patient, and then we were able to follow them up and get an image seven months later, and we have nice resolution in the target area.
So multiple met strategies. So for us, you know, I'll start with, you know, when do we do SRS versus fractionated SRS. The first topic that comes up is location. You know, is the target in close proximity to an organ at risk such as the brain stem or the chiasm, or are there multiple targets in close proximity to each other? They're perhaps gonna create a dose bridge and therefore increase, you know, some of the intervening normal brain dose. The next thing that we look at is size. If a single target is greater than 3 centimeters, or as, you know, you start increasing the number of metastases much beyond five, you substantially increase the number of volume.
And once you start adding margins to these larger volumes, well, then you're substantially increasing the amount of normal tissue. As we approach these greater sizes and these greater numbers of mets, we tend to go to fractionation. For single fraction, we'll look at the V12 parameter. We like that to be below 10ccs. For fractionation, when we do a 5-fraction plan, we tend to prefer the mean brain dose below 4 gray, and for 3 fraction, we like the mean dose below 3 gray. Kind of going on about margins. If you look at the image on the right, and you think about accuracy, it takes on kind of a different meaning in the single isocenter multi-target world. If you have a target that is centered at the isocenter, and you're just treating that single target, and it's displaced, a rotation probably isn't going to affect it anywhere as much as it will as if that target is at some distance. Targets that are further away from the isocenter are gonna be much more displaced.
And this is an extreme example of, you know, a target 7 centimeters away from the iso, but I've taken a 1-millimeter deviation and rotated it to 1-millimeter, and now we have almost, you know, a little over 2 millimeters of offset. Depending on your margin size, this could obviously, you know, be a geometric miss. So for us, our margins are determined by the distance targets are from the isocenter. We take into account the ExacTrac tolerance, meaning, you know, the residual tolerance that will allow after image-guided correction with ExacTrac. And then as a function of distance from the isocenter, we'll determine a margin for those targets. It's a really good paper by Chang that talks about incorporating the rotational setup and certainty in planning target volume margins.
You know, I borrowed this image from Brainlab. This just kind of shows you what happened as you gradually increase margins. In this case, you know, as you add 1 millimeter, you get upwards of a 50% increase of additional brain volume that gets radiated at the B12 increases pretty substantially. There's a paper from Kirkpatrick where they, you know, randomized 49 patients respectively, and they gave them a 1-millimeter or 3-millimeter margin or GTV-PTV expansion. What they found was is that, you know, 3-millimeter margin tended to increase radionecrosis and that they recommended a 1-millimeter expansion.
Going to the elements software suite. This is basically...again, it's a suite of tools, includes everything from, you know, commissioning and validation through contouring and planning, and they have site-specific modules such as the one that we're discussing today, multiple mets. There's a spine application, and then there's also a cranial VMAT that we'll discuss a little bit today. But the nice thing about these components is that the modules have a very linear flow that include fusion, distortion, correction for MR that takes you right into normal tissue structure mapping as well as target identification and planning. So now, you know, when I think back to our historic process of planning taking hours and days, now the order of magnitude is minutes in terms of planning.
Historically, you know, when you think about treating multiple targets, and you wanna do that in an SRS fashion, the first approach was a multi-isocenter or unique isocenter for each target. And this was a little challenging to plan. It was also cumbersome to deliver for the patient. So with the advent of RapidArc, you know, folks got the idea to go ahead and place an isocenter in the geometric mean of the targets and then deliver a non-coplanar delivery to all the targets with a single isocenter. One problem that comes up with the RapidArc delivery is something, you know, it's referred to as the island blocking problem, and this essentially happens when you have two targets that lie in the direction of MLC travel. What tends to happen is the optimizer will open the leaves in between those targets in an attempt to optimize dose to those targets. What this results in is dose spill into the intervening brain tissue. There are some, you know, ways to manually, you know, optimize this and get around it.
As I mentioned, you know, one of the key things for delivery is non-coplanar imaging. It's really important, as I mentioned, with a single isocenter and a multi-target approach that you limit the residual patient motion. And this is a really good paper from Dr. Todorovic presented at the 2019 ESTRO Novalis Circle, where he actually...you know, they actually went in, and they wanted to see how often they had to reposition their patients based on a 0.5-millimeter, 0.5-degree tolerance for ExacTrac. What they found is that they had to position the patient or reposition the patient about two-thirds or 67% of the time. And so, you know, one might say, "Well, that might be machine-related. It might be gantry sag. It might be table run out," things of that nature. And kind of what they found is they went back, and they looked really closely at the machine parameters, and they found that this wasn't machine-related. You know, patients move. This is a frameless system. You're looking at really tight residual tolerances. You know, if a patient coughs or, you know, there's some slight movement, you're gonna exceed that tight tolerance.
All right. So, the Multiple Brain Mets element that we use. This is a different approach from RapidArc. It is a non-coplanar delivery, but it's an inverse optimized dynamic conformal arc. And what's different when you think about the island blocking problem is that when you have two targets that lie in the direction of MLC travel, what it does is break this down into two arcs. It'll treat one target on the first arc pass. They'll treat the second target on the second arc pass. This provides really two benefits. One, it substantially decreases the intervening normal brain dose, but then it increases the conformity of each of those particular targets. The argument might be that you're adding some time to the patient treatment, but usually, these arcs are on the order of 120 degrees. And, I think, at most, you're possibly adding 20 to 25 seconds to a treatment delivery for a particular couch position.
So as mentioned, this is an inverse optimized dynamic conformal arc. You know, it's a reduced planning time. I think once you've validated the system compared to a fluence modulated technique, I think you could say that the QA is reduced, if not, substantially reduced. This package even though Multiple Brain Mets is in the title, we have used this successfully for quite a few single-target cases. The new version, version 2, does provide jaw tracking. So this will actually, you know, improve conformity of peripheral targets, and you have to keep in mind that for a given couch position, that peripheral target might change. The nice thing about the software suite is, again, it does include imaging in any couch position with ExacTrac assuming you have gantry clearance. Some of the tools and their suite. One is an organ at risk sparing interface. This is really nice because if you think about a target that's in close proximity to an organ at risk, you know, a traditional dynamic conformal arc would just look at the margin around a particular target. What's nice here is that we can actually go in and identify that organ at risk, and just by simply identifying the organ at risk in the optimizer, we're able to bring the brain stem back within our OAR constraints.
So another interesting feature of version 2. Brainlab refers to this as beam's eye view margin optimization. I tend to think of it as a gradient optimization, but if you look at the top, this is kind of a standard plan delivery where you have a homogeneous dose distribution within the target and just kind of a standard sharp falloff. What you can do is you can use this beam's eye view optimization or gradient optimization to enforce a higher dose and a steeper falloff. Essentially, you're prescribing to a lower icy dose line, and that, you know, has the potential to, you know, decrease some of this surrounding normal brain dust. Cranial elements is the other module that we use or have started using more frequently for cranial SRS, and this is really for cases that are...you know, targets that are more irregularly shaped. It's a VMAT solution, and it's for a single target, though it does support a simultaneous integrated boost if you wanted to take something within a particular target to a little bit higher dose. Again, it supports ExacTrac, so you get the imaging with the patient in any couch physician.
The nice thing about these tools is that even if over a course of time, you treated a patient with different modules, you can combine them, or if you've got a patient say with 11 metastases, one is a regularly shape that you'd wanna treat with the cranial VMAT solution, and then you've got 10 smaller ones you'd wanna treat with multiple met solution, you can actually create a composite plan and then go in and evaluate that. I borrowed this slide just to show a little bit more detail. This is a Brainlab slide, but this is a case where they did 10 smaller metastases with the Multiple Brain Mets element. Then they did a central larger irregular target with cranial VMAT SRS, and then they combine those. And this really is just to show you that the product is somewhat vendor agnostic, and it supports both Elekta and variant platforms.
So our elements experience. As I mentioned, you know, we started down this journey...this path about three years ago. Most recently, we just added cranial elements. So I'll talk about that real quick because we don't have as much experience with that. But we've done 30 cases or probably just over 30 cases through June. We've treated everything from post-op cavities and intact metastases to some functional things such as vestibular schwannomas and gliomas. Dr. Gleason will go over a few cases of the cranial element. The largest component of our program is multiple mets or metastases treatment. This program started in late 2017, and since that time, we've treated more than 300 cases. And as I mentioned, you know, this solution is not just for multiple mets. A little less than, you know, 50% of those cases were treated as a single target. A little more were treated...you know, were two or more targets. A little over 50%, 164 cases were single-fraction SRS, and then for fractionation, we tend to favor five-fraction over three-fraction for fractionated SRS. And just looking at the graph, you can see the majority of our multiple targets patients are, you know, 3 targets at a time, but we have treated as many as 12 and 14 targets in a session.
Just some statistics from all of those cases for the multiple mets. People often ask, you know, "What size targets are you treating?" On average, the target size that we treat is about 18 millimeters, but it rages up to, you know, 76 millimeters. The images shown here show probably one of the smaller targets we've treated. This is a 6-millimeter target that we've treated. And then as far as volumes go, we've treated things that are fairly small all the way up to 55 ccs. Moving on, you know, some more statistics. The average conformity index is about 1.38, and the average gradient index is 4.3. I just like to make a quick comment here. You know, usually, if you look at the situation with the image on the left, you have this nice isolated target. In that situation, the conformity index is gonna be around 1.1 to 1.2. You're probably gonna have a gradient index of around 3. If you look at the image on the right, you know, where you have two targets in close proximity to each other, you start observing dose bridging and dose pull. Your conformity indices tended to degrade at that point.
And so, if you look back at our stats, you know, about 50% of the time we're treating, you know, multiple metastases patients. And more times than not, you know, you're gonna have targets in close proximity, and this is gonna be a reality that you're gonna have to face. So treating small targets with the agility. This is another, you know, topic that comes up. And, I think, historically, this has been considered a hardware limitation. And, I think, you know, in recent years, this has been overcome by software solutions. Now that we have the ability to optimize on things like, you know, anatomy and gantry and MLC and couch position, we're able to create some very conformal dose distributions. I like to think of optimized dynamic conformal arc as an elegant solution as opposed to something that's fluence modulated, which is a complex solution.
So, you know, moving on, you know, looking at leaf with a little bit closer and looking at version 2 of the software, Brainlab provided us with a version 2 model, and what we decided to do was go back and take 20 patients that were treated with version 1.5 of the software and then evaluate those patients in version 2. And so in addition to that, we got to look at the new jaw tracking feature with Elekta. And then they also provided a Varian HD120 model, so we could make some determinations about leaf width effects. We went ahead and did those comparisons. Then we looked at conformity index, gradient index, V12, and mean brain dose.
And so when we did that, we did a Wilcoxon, you know, paired comparison. If you look on the far left-hand side in dark gray, you can see the conformity index for version 1.5, and then moving to the center columns in orange and gold, you can see an improvement with version 2.0 and with version 2.0 with jaw tracking. And then, you know, over on the right hand column in light gray, the HD 120 was the best. Just keeping in mind that while there is, you know, a P value here at point 0.1, the numerical difference was 1.22 versus 1.25. Again, that's 1.22 for HD 120 versus 1.25 with the Elekta Agility with Jaw Tracking. Going on to the gradient index similar as the conformity index. We did see an improvement with version 2.0 continuing improvement. With the Jaw Tracking, again, the HD 120 showed the greatest improvement, but again, these numbers are pretty small. We went from basically 3.33 with the agility and Jaw Tracking to 3.2. For V12, again, improvements shown in version 2.0 continued improvements with Jaw Tracking. The V20 or excuse me, the V12 for the HD 120 was just under 11ccs versus 12ccs for the agility with Jaw Tracking.
Mean brain dose. Again, improvements with version 2.0 of the software. You know, for the HD 120, yes, this is an improvement, but it's 0.2 gray. It's a 0.2 improvement over the HD or excuse me, the agility with Jaw Tracking. We saw improvements with version 2.0 of the software both for, you know, jaw tracking and HD 120. The issue, I think, comes down to is that I can...we can demonstrate a statistical improvement with those numbers. For the HD 120, I think, the challenge there when you look at 2.5-millimeter versus 5-millimeter leaf width is that you're gonna have a challenge proving that there's a clinical impact for those patients treated with a smaller MLC.
Bogdan: Thank you for that review, Mike, and let's turn it over to Dr. Gleason to provide a clinical review of Alliance's Cancer Care program.
Dr. Gleason: All right. Well, thank you very much for that introduction, Bogdan, and thanks, Mike, for doing the majority of the work on this presentation and getting everything together. Mike thanked a lot of people at the beginning of his talk. So I'm gonna thank him. None of this would be possible without Mike's leadership in our program, and certainly, I feel much more comfortable doing something like a trigeminal nerve with a cone right next to the brainstem without a frame knowing that Mike has done all his end-to-end testing. So, I just wanna briefly do some case presentation showing how we've used the software to address specific patient needs. And I've intentionally tried to go from simple to more complicated as we move through this.
So we'll start with the first case, which is just gonna be a single target case. So this was a patient 52 years old was found to have a widely metastatic malignancy from an unknown primary, and so as part of that workup, she had an MRI of her brain that showed the small lesion in the right cerebellum. This was ultimately found to be from an occult breast primary actually. So we treated this with single-fraction SRS 22 gray in a single fraction ended up being 4 arcs, and this is a good example of the fact that even though the name of the software is multiple mets, we use it very frequently to treat single mets because, you know, most of these mets are around spherical targets. So the dynamic conformal arc delivery can still give you excellent, you know, conformity, and you can see our V12, in this case, is just 1.1cc. And this patient would have been treated with only four arcs. They may have been on the table for 15 or 20 minutes total to get this addressed.
So the second case, just stepping forward to the next progression, we'll have just two sites of disease. In this case, it's a patient with a metastatic non-small-cell lung cancer, and at the time of their diagnosis, they were found to have a metastatic disease of a liver, and an MRI of the brain was also performed as part of this initial workup. And the patient had these two lesions, both asymptomatic. We have this small lesion here in the posterior right temporal lobe and then a slightly larger lesion here in the left parietal lobe. So, again, we treated this with a single isocenter, but in this case at the geometric mean between the two targets, the small lesion here received 20 gray, and the lesion here on the left received 18 gray.
And one of the things as a physician that I like about the software is that if my dosimetrist or physicist sends me the plan, I can change things on the fly without having to send it back to them. So, you know, in fact, they may have sent this to me with 22 gray to this target and 18 to this, and within the software, I can quickly just change the dose myself if for some reason I wanted to do 20 gray and hit calculate, and within, you know, a couple minutes, I've got a new plan that I can quickly evaluate and sign without all that back and forth between yourself and the planner. But in this case both, again, with excellent conformity 1.13 and 1.20 and actually, again, in this case with 4 arcs. You know, we don't always use four arcs, but I did find going through my lecture today that seems to be a common topic. Here's that same case just with a different graphic from the software, but again, it just shows you how you can nicely create these clouds of radiation dose with all this intervening normal brain that's spared.
Our third case is just another step forward and complexity. So in this case, it's, again, a non-small-cell lung cancer patient who at the time of the initial staging workup is found to have these five asymptomatic brain metastases that are scattered throughout the brain. And so in this case, we, again, used a single isocenter and delivered 20 gray in single fractions all 5 sites. It was, again, four arcs. And our conformity index. This is a volume average conformity index across all the targets was 1.40, and our V12 was only 5.1cc. So, I think, this is a great example of the fact that you can still treat 5 targets and have a low V12 in a single fraction scenario. But the key here is that our target volume was quite small. So it's really a good illustration also of the principle that the volume of metastatic disease in the brain is likely much more important than the actual number of brain metastases when figuring out whether or not SRS is an option. So certainly, I'd rather treat these five tiny mets with single fraction SRS than treat, you know, three very large...three to four centimeter metastases.
Jumping to the next level of complexity. This was an 11 site case. So I will acknowledge that most of the time with someone with this many metastases, we would strongly consider whole-brain radiation potentially with hippocampal sparing and memantine unless the patient already had a significant amount of prior treatment or some other factor there. In this case, I really wanted to avoid whole brain radiation for a couple reasons. One, the patient was a little bit older, and I feel like they're a little bit more prone to the neurocognitive sequelae of that treatment, and more importantly, this was a melanoma. So, you know, we know that melanoma tends to be more radioresistant in many cases, and so I hate using whole-brain radiation in that scenario knowing the potential side effects of the treatment and knowing that it's often not all that effective for some melanoma patients.
But in this gentleman's case, even though there were 11 sites, they were nicely spaced throughout the brain, and they were all small. They were all less than 2 centimeters. Going into it, we didn't even try to plan this with a single fraction. I could just kind of tell based on experience that we wouldn't have an acceptable V12 with this many lesions, and so we went with a sixth grade times five fractionated SRS approach, again, with just a single isocenter, multiple max dynamic conformal arc approach. And we have volume average conformity index of 1.39 and a mean brain dose of 5 gray. So, again, with 5 fraction treatment, we often aim for that 4 to 5 gray range for the mean brain dose. So, I think, that's quite good in the setting of treating 11 tumors.
I tried to play with the angle when I was creating this screen capture here to figure out where I could kind of best show you these dose clouds. But if you look at this 15 gray line, there's only 2 areas where even the 15 gray, the 50% line bridges at all despite there being 11 sites of disease, and there was nowhere where the 20 gray line even became close to bridging. And the reason I like to look at the 15 and 20 gray lines is just from a practical standpoint. We know we can do three grade per fraction to the whole brain for 10 treatments or even 4 gray per fraction times...you know, 5 treatments to the whole brain. So certainly, in a fractionated, you know, just 5 treatment course, if we can keep that 3 gray per fraction volume to a low area, we're sparing a lot of intervening normal brain tissue.
This next case here. This is a patient who had renal cell carcinoma and came back with a large symptomatic metastasis in the right occipital lobe and was seen by one of our neurosurgeons and had a resection. And when we targeted our cavity...you can see that in this case, the cavity happens to be quite spherical. It's not an irregular target. And so in that case, we...you know, rather than trying to use a VMAT type delivery, we started with our dynamic conformal arc approach, again, using the multiple mets software albeit in this case just with a single target. And impressively, the conformity here is 1.14, and the gradient index is 2.58. So, I think, you know, a very elegant treatment plan that can be done in a quick throughput as far as the planning, the quality assurance, and then the delivery on the machine.
So this next case is an example of how...you know, as we know many patients who develop brain metastases will develop more brain metastases, particularly when we're not doing whole-brain radiation and particularly in this era where because of targeted therapies or immunotherapy, we have people with metastatic disease who are living longer. So this was a patient in her late 60s who had a metastatic lung cancer. It happened to be an adenocarcinoma, and she had an EGFR mutation and had been on that TKI for a number of years, but ultimately, in February of 2019 started to develop some resistance and was found to have some small lesions in her brain despite no active disease outside of the brain.
And so when I first saw her in that first treatment course, there were six lesions that we addressed. Five were treated with a single fraction, and then one lesion in the pons was treated with a five-fraction course. And then she came back two to three months later for that initial post-treatment scan in April. Everything looked great, but by the time we saw her in July, there were now, again, two new small lesions that had not been treated with the first course. So we again treated her with a single isocenter multiple mets plan, delivered I believe 20 grain, a single fraction to both those sites. When she came back for her two-month post-treatment imaging, again, everything that had been treated prior appeared to be controlled, either resolved entirely or much smaller by then, but there was, again, one new site of disease, and we'll show the plan on that here on the next slide.
So that one new site was here. It was about a centimeter or just under that in diameter, and so we again set her up for single-fraction SRS. Again, it's a single site of disease, but we used our dynamic conformal arcs within the MME software. And in this case, with it being a small target, very peripheral far away from normal structures, we were actually able to generate just a three couch position plan, which Mike has captured this for me. Apparently, it was just 12 minutes to deliver this patient's treatment. With this small target and only 3 couch positions, we have this conformity of 1.27 and a GI of 3.35. What this final picture here shows is just resolution of that metastatic site. So this is six or seven months later, and that site's no longer present. And, in fact, when I saw her most recently, she doesn't have any active disease in the brain. Everything ultimately resolved with treatment, and fortunately, she's found a systemic therapy regimen, has stopped on a clinical trial and so far has stopped making new brain metastases and is doing quite well.
This final slide is kind of busy, but I just wanted to give some examples of when I feel like the cranial element can be a very useful tool beyond just using the dynamic conformal arcs type approach. So if you look across the bottom of the screen here, these are all three post-operative cavities that were irregularly shaped, and all these were treated with either three-fraction or five-fraction SRS. And with this cranial element despite these unusual shapes, we have conformity indices on the 1.11 to 1.17 range and gradient indices on the range of 2.60 to 3.0. So, these are situations where I found that the cranial element can be quite useful when we have these irregular shapes. You know, physics still loves the dynamic [inaudible 00:41:00] on the standpoint of the less need for QA and other resources. So, often they'll...Michael run a question, and they'll do a decat plan, and in some cases, we'll find that the dynamic conformal arc plan is equally good. But with these irregular targets, clearly, the cranial element was superior.
And then these two cases across the top, I've just shown as examples where you can use dose painting or simultaneous integrated boost within the software. Now, you can't use it if the targets are remote from each other, but in a case like this...so we'll start with this one here in the top left corner of your screen. This was a gentleman with a metastatic colorectal cancer who had this ongoing oligometastatic pattern of disease where at presentation, he had a liver metastasis resected and was NED for a couple years after chemotherapy and management of his primary. And then he developed a lung metastasis, and that was treated with SBRT, and then a year or two after that, he developed this brain metastasis, and one of our neurosurgeons removed it.
And on the post-operative scan, you can see this small residual area of tumor on the anterior aspect of the cavity adjacent to the corpus callosum where the neurosurgeon was likely concerned about causing some type of morbidity there. So rather than treating this entire target, nine gray times three fractionated, what I did was I treated the gross disease nine gray times three, but I treated the other part of the cavity just eight gray times three. And again, you can see in this case a conformity to an X under 1.1.
And then this case here in the middle, this was a gentleman who had had a lung cancer, non-small-cell lung cancer stage two that was resected three to four years prior received adjuvant chemotherapy and then came back with symptomatic CNS metastases and no other measurable disease. He had a resection of a large metastatic site, and the only other area was this small satellite nodule that I'm indicating here that was not removed. I chose this exact slice so you could see both the targets, but if you could scroll in the other direction, you'd see that this cavity was actually quite large. And so I was reluctant to give six gray times five to such a large target volume, and so I opted to give five gray times five to the cavity where I really felt like we were dealing more with microscopic disease. Both the surgeon and the post-op imaging favor, you know, radiographic, you know, gross total resection, but I still was able to dose paint and give 6 gray times 5 to this unresected nodule just posterior and inferior to that cavity site and with the treatment able to get a conformity of 1.17 and a gradient of 3.07.
So, again, for us, this has revolutionized what we've been able to do in our community, and we continue to build on what we've done so far. And I would again like to thank all the support we've received from Brainlab as we've developed the program, and again I'd like to thank my physics team including Mike Taylor, my dosimetrist particularly Christy Bradley who does the majority of these plans, and then all our therapists out on the machine who deliver these SRS treatments every day. And thanks again for everyone who actually tuned in to listen to us. We're certainly happy to answer any questions everyone has regarding our experience.
Bogdan: Thank you, Dr. Gleason, thank you, Mike, for the great lectures. We have a few questions that hopefully you can answer. Dr. Gleason, let's maybe start with you. There was a question regarding a five-fraction dose regiment not being as effective as, of course, single fraction or a three-fraction treatment at nine gray. So, could you provide maybe some further explanations on the five-fraction treatment?
Dr. Gleason: Yes. So being close to UAB...a lot of the group from UAB, Dr. Fiveash, they do a lot of sixth gray times five fractionated stuff that they published, and so that's kind of where we started doing fractionated SRS. Because of the literature from Italy, Dr. Monetti, I believe it is, the nine gray times three, we also use that regiment as well. So I tend to use nine gray times three if it's say a single lesion that's 3.1 or 3.2 centimeters that I don't feel like I can treat single fraction. I know in Italy, I think, they use a 2-centimeter cut-off in that Monetti data, but we still tend to treat anything less than three centimeters in a single fraction. But once I switch to fractionated, if it's just barely too large, I sometimes do use that three fraction regimen, but if I'm treating numerous targets, we felt more comfortable doing the five fraction regimen knowing that there's always a balance of toxicity versus efficacy.
Bogdan: Great. Maybe I'll ask you another question. We polled everyone for what is the smallest volume that they've treated with the Agility MLC, and Mike did a good job explaining why software solutions really improve the symmetry that's possible with the Agility MLC these days. But from your perspective, Dr. Gleason, what targets would you not treat with the Agility MLC? And maybe give us an idea of where you would use cones in your practice.
Dr. Gleason: So, we always use cone when we treat our trigeminal neuralgia cases. We never have used an MLC for those, although I know there's some interesting work being done at UAB using a virtual cone with their HD 120 MLC, but we've always used a cone in those. For the majority of our mets cases, we have been able to treat those with the MLC. When I do have a really small target, I'll have physics do a comparison for me, and sometimes I'm surprised, and the conformity is almost identical even though the target's small. But occasionally, we have brought out, you know, a 7-millimeter cone to treat a very small lesion, and Mike might have a more detailed answer as far as the size of things that we have treated without the cone.
Mike: Sure. You know, I don't know if I'd add too much to that from what Dr. Gleason said, but I agree. I think we kind of take it on a case-by-case basis, and we've been actually pleasantly surprised by some of the smaller things we've been able to treat. And obviously, you know, if you can get away with doing with an MLC, it's a much easier day for us.
Bogdan: Okay. Mike, we have a lot of technical questions for you. So, in no particular order, I'll try to go through them. So, first question, when you perform a Winston-Lutz test, do you use ExacTrac to correct its time, you have a new couch angle? And maybe I'll ask you something related to this for the Versa HD LINAC. What is the overall couch runouts on your device, mechanical runouts, and then, you know, how do you compensate for that clinically?
Mike: Yeah. I'm gonna kind of combine both of those questions back by then. So, yes, to answer the first question, we do use ExacTrac for the Winston-Lutz procedure specifically on the couch, and that's twofold, one, to correct for the couch run out, which probably is pushing, you know, close to about 0.6 millimeters from our observations. The second reason is to make sure that the system is performing like we intend it to perform for patients.
Bogdan: Okay. Thank you for that. Another question. What is the low dose percentage when treating multiple targets with arc therapy?
Mike: Not sure I followed the question. Yeah. We have some parameters for what we look at for V12, you know, and then for, you know, fractionated SRS, you know, three-fraction SRS three gray, you know, five-fraction SRS four to five gray.
Bogdan: Okay. Let's recap a little bit. The margins that you use, and there's a related question to these on do you only use ExacTrac, or do you also use cone beam CT for alignment?
Dr. Gleason: I can take that one. As far as the margins, Mike showed that table where we select our margin based on the distance of the individual target from the isocenter as well as our ExacTrac tolerance. So we have a separate table for when we're using 0.8.8 ExacTrac tolerance and a separate table for when we're using 0.5 tolerance. And those tables will go anywhere from 1 to 2 millimeters depending on the distance from iso and the ExacTrac tolerance. You know, for a little over a year now, we've always chose our ExacTrac tolerance so that we can keep the margins either at 1 or 1.5 millimeters. So, it would be rare that we would use any larger margin. Now, this is obviously for intact metastases. We use more margin on a post-operative cavity more for clinical target volume rather than any issue with the setup, but usually, we use 1 or 1.5-millimeter margins.
As far as whether or not we use cone-beam CT. So there are plenty of centers that use ExacTrac alone and don't use CT. Our practice has been for single-fraction SRS to do a confirmatory cone-beam CT after the initial ExacTrac just as a second check because it's a single fraction treatment, and so it gives us additional confidence. I'll say that we've gone back and looked at those, and we've never really found anywhere we weren't already on target, but it does give us some additional confidence. For fractionated SRS, we only do the cone-beam CT on the first fraction. So if it's a three-fraction case, we'll do ExacTrac and cone-beam on day one, but on days two and three, we'll only use ExacTrac.
Bogdan: So when we polled the audience today, it seems a large majority of them are using an isomorphic 1-millimeter PTV expansion. Maybe in your practice, is that what you also see for the majority of your patients, and how frequently do you use a variable margin? And linked to this, of course, the 3.0 version of Brainlab software that is coming out in October will actually automatically generate variable margins for you. How relevant do you think that would be?
Dr. Gleason: You know, I think it depends on how you wanna look at it. I mean, in all honesty, we use a 1 or a 1.5-millimeter margin. So, with 0.5-millimeter additional expansion, you know, I'm not sure how clinically meaningful it is one way or the other if you do or don't use it. I suspect if you look at your...if you give 22 gray with a 1-millimeter margin versus if you give 20 gray with a 1.5-millimeter margin, you probably have very similar plans and similar efficacy and safety. But we use 1 or 1.5, and maybe we're overly detailed to add that 0.5-millimeter. That's been our current practice if they're farther from the isocenter.
Bogdan: Mike, there are some questions regarding those calculations, and one had to do with correcting for artifacts in your CT. So maybe I'll ask you as well if you've ever used Monte Carlo for multiple brands SRS or just stick to pencil beam. And then another question is how is the OAR sparing accomplished in multiple brands SRS?
Mike: So for the first one, you know, our CT, we have a Siemens Somatom, and it does have a metal artifact correction. I think the question involved the artifact from teeth. I haven't seen something in our practice...in our experience, I haven't seen anything that substantially affected that. We have started using to some degree Monte Carlo in evaluating that in comparison to the pencil beam to look at differences. The last part of that question. The OAR sparing is...that's tied up in the optimizer. So, you know, is identifying that. Then, you know, you can actually modulate that, you know, around those OARs, you know, just to kind of pull the dose back off of those structures that you've identified.
Bogdan: Okay. When talking about delivery time, does it only include beam on time, or does it also include IGRT time?
Dr. Gleason: So, yeah. The nice thing about ExacTrac is it's really quick. The ExacTrac part as far as imaging goes is incredibly efficient. The slow part for imaging would be on the days that we do the cone beam CT. So, you know, that can take upwards between a minute to two minutes to acquire that actual image. ExacTrac can be image that...you know, we do it at the cardinal gantry angles when we have imager clearance, and we've even, you know, started looking at doing some of these with imaging during treatment delivery so that they don't actually stop the treatment but just do the quick snap verification with both imagers. That's very quick.
Bogdan: Okay. Some questions regarding QA. How do you QA for multiple mets especially when you have more than 10 tumors, or does your QA change at all when you have more targets? And also when we polled today, it seems to be kind of an equal split between secondary checks and gamma analysis with mostly electronic arrays. Maybe you can also touch on what you routinely do in the clinic and if there's any rationale after initial commissioning to still to do gamma.
Dr. Gleason: So, yeah, when we first started the program, we went through and did a specific QA on every single target. That included a point dose on each one in addition to a second MU check. And so the challenge there obviously is now you're not placing your chamber in the isocenter. You need to go ahead and make sure that that chamber is in each target relative to the isocenter. In one sense, you know, that's a good, you know, end-to-end evaluation of your system. Once we accumulated, you know, about 20 patients, you know, we felt pretty comfortable in the second MU check. The only time that we default back to say a point dose at this point is really when the second MU check were having some issues there or some concerns, we'll go back and reevaluate that. I agree. You know, film and the 3D gel from somebody like RTsafe is a good end-to-end test of looking at spatial and antisymmetric fidelity. The SRS map check is probably one of the more convenient things relative to film. The only, you know, caveat there, again, depending on how far apart your targets are, you know, that device, I think it's about 7 centimeters square. So, you know, if your targets are beyond that, again, you're kind of looking at QA each individual target. From a physics perspective moving around each target and then delivering the entire plan can be time-consuming.
Bogdan: I know you predominantly showed plans today with a four-arc setup, but we have a question of really what is your default arc geometry setup. And also for these kind of our treatments, how frequently are the patients shifting within the mass system outside of your treatment tolerance?
Dr. Gleason: I'll let Mike take the first question as far as what our starting off template is. It seems like we usually end up with four to six couch kicks depending on the case. But as far as how often the patients move, I hope to have an answer to that. We had a medical student evaluating a set of patients and then COVID hit, and we stopped letting other people inside the department. That project kind of came to an end a couple months ago before we got the answer to that. But we actually are interested to kind of quantify how often do we have to make corrections at these couch kicks. It's not infrequent that when you move to the next position, you do have to make a minor ExacTrac adjustment, and that's what makes us, you know, comfortable to do the treatment because we do have the ability to image at these non-coplanar angles and make 60 corrections. I'll let Mike answer about...
Bogdan: Dr. Gleason, just on that comment, are you imaging every time you rotate the couch?
Dr. Gleason: Yes. Yeah. We imagine every couch position.
Mike: And I'd say, you know, I don't think we've kept the same numbers that Dr. Dworek has. I think, you know, yeah, we do see at least 50% of the time we're correcting particularly for a smaller residual tolerance. You know, for 0.5 and 0.5, we'll correct that more often. Regarding the number of couch kicks, I'm gonna go ahead...it looks like my dosimetrist, Christy Campbell, chimed in and gave me the cheat there. It looks like we usually end up with about four arcs. As Dr. Gleason said, we've seen as many as six. It's nice. You know, obviously, the fewer couch kicks you do, the faster the treatment delivery.
Bogdan: Okay. There are two questions that seem to be more directed in my way. So maybe I'll answer those on Brainlab's behalf, and then I have one more question for you. So, there were some questions regarding combining elements Multiple Brain Mets SRS with cranial SRS. And like you've showed today, Mike, and Camille, thanks for your answer as well, you're right. These things can be combined today. It's a little bit of a serial way of combining them, and then you have to do a composite evaluation in those review. However, this is changing with version 3.0 of elements. We're slotted to release version 3.0, as I said earlier, in October with availability to all of you starting November. And with that version, the integration between Multiple Brain Mets SRS and Cranial SRS is more streamlined. You can start a plan for however many tumors you want in brain mets, and if you choose to do VMAT further dose painting in cranial SRS, you can just switch to that element, and it will automatically understand the dose that has been already delivered by Multiple Brain Mets SRS and will allow you then to have a much more streamlined composite view. And what Mike has showed today in terms of dynamic jaw tracking for the Agility MLC, that would also be available for all variant MLCs with version 3.0 as well. And I guess here's a question for you. Are you considering Brainlab's new product ExacTrac Dynamic?
Dr. Gleason: We have looked at it. We are certainly considering it. It's an excellent tool.
Mike: Yes, agreed.
Bogdan: Okay. I think that's it. There's a comment regarding the Winston-Lutz workflow. Mike, if you'd like to take that regarding why...
Mike: Yeah. I think that was Marcel. I think there was something about including the larger gantry angles, and I'm not sure if the path there is, you know, are these gantry angles that we're not necessarily treating. And I agree. The larger, you know, run-out happens at the extremes, and perhaps we shouldn't include those. And, you know, our procedure right now is pretty standardized, and our workflow is pretty standardized. As far as going to Winston-Lutz procedure, you know, prior to patient treatment, we usually, you know, carve out at least, you know, 20 minutes or so, and if things go well, you know, that's like I said, a pretty quick process to get through.
Bogdan: Gentlemen, again, thank you very much for your lectures today, and thank you all for your participation. Have a great rest of the week and hope to see on our next webinar.
Mike: Thanks, Bogdan.
Dr. Gleason: Thank you, Bogdan.
Bogdan: Yeah, bye.
As with our previous webinars, we continue to provide CE credits. If you are in need of CAMPEP, MDCB, or ASRT credits, please follow up with us at info@novaliscircle.org upon successful completion of this webinar, and we'll be more than happy to help you with obtaining your credits. And also don't forget to sign up for our upcoming webinar with Dr. Prasad from Roswell Park in Buffalo, New York who will discuss response assessment technologies and the use of our contrast clearance analysis tool. This upcoming webinar would be held on July 9th. Lastly, please remember to use either Google Chrome or Safari to log into the webinar. Should you have any technical challenges, please refresh your window. Utilize the chat line to send us questions that we will answer upon completion of the two lectures. We will also utilize the chat line to pose some questions to you, and we will review your answers in the Q&A session as well. And if you'd like to follow us on social media, please use the hashtag provided. With that said, I'd like to turn it over to Mike Taylor.
Mike: Thanks, Bogdan, for that introduction. As you mentioned, my name is Mike tailor. I'm a physicist. I'm joined by Dr. Jack Gleason. We're from Alliance Cancer Care in Huntsville, Alabama. And today, we're here to talk about "Evaluation of Enhanced Solution for the Treatment of Multiple Metastases with a Single Isocenter," which is a long time for us to talk about our experience with Brainlab's multiple mets element as well as their VMAT cranial element solution. Just a quick disclosure, I'd like to thank Brainlab for inviting us to present this material, as well as they provided a speaker fee, and then we have a research agreement in place to evaluate new versions of the software. Some quick acknowledgements. Some folks and companies that have helped us with our program in terms of development, and then they've also provided some material for this presentation. I'd also like to thank the Alliance Cancer Care team, very fortunate to work with very talented group of therapists or cemeteries, physicists, and physicians. And, I think, they're, you know, really the core of what's made our program so successful.
So, this is our agenda for about the next 45 minutes, everything that we're gonna go ahead and cover, and I'll go ahead and get started. Just some background. We are a fairly busy community practice based in North Alabama. We treat about 200 to 225 patients a day on 6 LINACs. Huntsville has a population of just over 400,000. We're the primary radiotherapy providers in that area. Just looking back, you know, about three years ago in our historical SRS program, you know, we treated maybe one to two patients a month. We tended to treat things that were larger in size, and we avoided functional SRS targets. In terms of planning, you know, these things took hours to plan, and in terms of delivery, we were looking at about at least an hour per isocenter per target. So it could take hours for delivery as well.
So, you know, the physicians about three years ago decided to invest in a new SRS program. We began looking at SRS equipment, and we looked at, you know, Gamma Knife, CyberKnife, and LINAC-based options. And, for example, the CyberKnife, we really like the option to do non-coplanar imaging. The challenge for us in a busy practice is that it isn't as efficient as say a LINAC in terms of treating non-SRS, SBRT deliveries, which we need to do. We made the decision to go with an Elekta Versa HD that's equipped with an agility, and then with Brainlab's ExacTrac, you get that non-coplanar imaging ability to see what the patient is doing in six dimensions and as well as correct them with the HexaPod couch. This is a standard four energy machine, and then we use it for external beam when we're not doing SRS and SBRT. We also include a standard set of Aktina cones down to 5 millimeters.
One of the key components of this system is the Agility MLC. There's a 5-millimeter leaf width MLC, 160 leaves with just under 0.5% transmission. It's a very fast MLC, which I think is important when we're treating multiple targets with a single isocenter. Even though this MLC does...is a 5-millimeter leaf width, I just like to point out that in the leaf direction, these leaves can travel in 1-millimeter increments, and then the extras behind the leaves can also translate in 1-millimeter increments. So you can actually mimic a 1-millimeter pixel.
And just a quick snapshot of what our SRS program looks like today. With this system, we now treat, you know, more than 125 SRS and fractionated SRS cases per year, and that includes a full suite of functional indications such as acoustic neuroma, AVMs, and trigeminal neuralgias. We also have a very busy SBRT program and treating more than 100 SBRT cases annually. And this is just an example of one patient that we've done. This is a 9 target plan. And if, you know, we were to do something like this with our historic system, you know, it would have taken upwards of nine hours just for delivery for each of these targets. Now, this is something that we can do in an approximate, you know, 16-minute window with ExacTrac imaging.
Just real quick. I'd like to talk about accuracy and a little bit about QA. It's a very fundamental component when you talk about multi-target single isocenter treatment delivery. Historically, when you think about SRS and the single isocenter, you know, the Winston-Lutz is kind of the gold standard for verifying isocenter. The ExacTrac system does come with a daily isocenter verification tool, and that will verify the isocenter within 1 millimeter. On the days that we do a single fraction SRS, we will go ahead and do a full Winston-Lutz procedure and verify that, you know, the gantry and the couch are within about 0.65 millimeters.
In terms of commissioning, validation, and just ongoing QA, end-to-end test is a really good test to make sure your system's performing accordingly. This is a great paper by Tim Solberg where they took a bony anatomy, anthropomorphic phantom and inserted a 5-millimeter ball bearing and then basically created a plant-centered on that ball bearing. You then go ahead and set the patient up for treatment. You use ExacTrac for alignment. Then the symmetric amateur is centered on that ball bearing. You use that as a pseudo-Winston-Lutz to verify the accuracy of your treatment.
There are some other options for checking your delivery. MD Anderson provides an anthropomorphic phantom as well. The phantom in the center is kind of their long-term photon SRS phantom. And, I think, this is probably a good choice for like a frame-based system. But as you can see, it includes a cylinder that contains OSLDs [SP] and fill. The cylinder is inserted in the phantom, and then it's filled with water, which makes it really challenging when you have a frameless system that relies on bony anatomy for alignment. So this probably wouldn't be a good choice for something that uses bony anatomy image guidance for alignment. You know, looking over the MD Anderson website, I see that they offer a proton brain phantom, which looks to be almost a duplicate, but this does contain bony anatomy. So that might be a better option.
A company that we've more recently started working with is RTsafe, and they provide a really nice bony anatomy anthropomorphic phantom that you can insert TLDs film. It has cutouts for chambers. And then what's really unique is that they also have a 3D gel insert. So you can actually test multiple targets, and this is really good end-to-end test to test, you know...excuse me, a single isocenter multiple met delivery.
Some other tools we use in our clinic are, you know, simple geometric phantoms. One that we tend to favor is the StereoPHAN by Sun Nuclear. And you can see just a picture of it over here on the right, but it's milled for a small volume ion chamber that you can do point doses with. They also offer an SRS MapCHECK, which is essentially a high-resolution 2D detector to replace film to do some planar measurements. And then one of the more recent additions is this really nice multi-target Winston-Lutz Insert. So now you can actually, you know, get some accuracy about how your targets or multiple targets are being treated.
And then finally, you know, I think, follow-up imaging, as you're able to obtain it is a really good follow-through in terms of seeing where things are at with patient delivery. This is a patient that if you look at the left-hand image, we treated the right side of the patient, and then we were able to follow them up and get an image seven months later, and we have nice resolution in the target area.
So multiple met strategies. So for us, you know, I'll start with, you know, when do we do SRS versus fractionated SRS. The first topic that comes up is location. You know, is the target in close proximity to an organ at risk such as the brain stem or the chiasm, or are there multiple targets in close proximity to each other? They're perhaps gonna create a dose bridge and therefore increase, you know, some of the intervening normal brain dose. The next thing that we look at is size. If a single target is greater than 3 centimeters, or as, you know, you start increasing the number of metastases much beyond five, you substantially increase the number of volume.
And once you start adding margins to these larger volumes, well, then you're substantially increasing the amount of normal tissue. As we approach these greater sizes and these greater numbers of mets, we tend to go to fractionation. For single fraction, we'll look at the V12 parameter. We like that to be below 10ccs. For fractionation, when we do a 5-fraction plan, we tend to prefer the mean brain dose below 4 gray, and for 3 fraction, we like the mean dose below 3 gray. Kind of going on about margins. If you look at the image on the right, and you think about accuracy, it takes on kind of a different meaning in the single isocenter multi-target world. If you have a target that is centered at the isocenter, and you're just treating that single target, and it's displaced, a rotation probably isn't going to affect it anywhere as much as it will as if that target is at some distance. Targets that are further away from the isocenter are gonna be much more displaced.
And this is an extreme example of, you know, a target 7 centimeters away from the iso, but I've taken a 1-millimeter deviation and rotated it to 1-millimeter, and now we have almost, you know, a little over 2 millimeters of offset. Depending on your margin size, this could obviously, you know, be a geometric miss. So for us, our margins are determined by the distance targets are from the isocenter. We take into account the ExacTrac tolerance, meaning, you know, the residual tolerance that will allow after image-guided correction with ExacTrac. And then as a function of distance from the isocenter, we'll determine a margin for those targets. It's a really good paper by Chang that talks about incorporating the rotational setup and certainty in planning target volume margins.
You know, I borrowed this image from Brainlab. This just kind of shows you what happened as you gradually increase margins. In this case, you know, as you add 1 millimeter, you get upwards of a 50% increase of additional brain volume that gets radiated at the B12 increases pretty substantially. There's a paper from Kirkpatrick where they, you know, randomized 49 patients respectively, and they gave them a 1-millimeter or 3-millimeter margin or GTV-PTV expansion. What they found was is that, you know, 3-millimeter margin tended to increase radionecrosis and that they recommended a 1-millimeter expansion.
Going to the elements software suite. This is basically...again, it's a suite of tools, includes everything from, you know, commissioning and validation through contouring and planning, and they have site-specific modules such as the one that we're discussing today, multiple mets. There's a spine application, and then there's also a cranial VMAT that we'll discuss a little bit today. But the nice thing about these components is that the modules have a very linear flow that include fusion, distortion, correction for MR that takes you right into normal tissue structure mapping as well as target identification and planning. So now, you know, when I think back to our historic process of planning taking hours and days, now the order of magnitude is minutes in terms of planning.
Historically, you know, when you think about treating multiple targets, and you wanna do that in an SRS fashion, the first approach was a multi-isocenter or unique isocenter for each target. And this was a little challenging to plan. It was also cumbersome to deliver for the patient. So with the advent of RapidArc, you know, folks got the idea to go ahead and place an isocenter in the geometric mean of the targets and then deliver a non-coplanar delivery to all the targets with a single isocenter. One problem that comes up with the RapidArc delivery is something, you know, it's referred to as the island blocking problem, and this essentially happens when you have two targets that lie in the direction of MLC travel. What tends to happen is the optimizer will open the leaves in between those targets in an attempt to optimize dose to those targets. What this results in is dose spill into the intervening brain tissue. There are some, you know, ways to manually, you know, optimize this and get around it.
As I mentioned, you know, one of the key things for delivery is non-coplanar imaging. It's really important, as I mentioned, with a single isocenter and a multi-target approach that you limit the residual patient motion. And this is a really good paper from Dr. Todorovic presented at the 2019 ESTRO Novalis Circle, where he actually...you know, they actually went in, and they wanted to see how often they had to reposition their patients based on a 0.5-millimeter, 0.5-degree tolerance for ExacTrac. What they found is that they had to position the patient or reposition the patient about two-thirds or 67% of the time. And so, you know, one might say, "Well, that might be machine-related. It might be gantry sag. It might be table run out," things of that nature. And kind of what they found is they went back, and they looked really closely at the machine parameters, and they found that this wasn't machine-related. You know, patients move. This is a frameless system. You're looking at really tight residual tolerances. You know, if a patient coughs or, you know, there's some slight movement, you're gonna exceed that tight tolerance.
All right. So, the Multiple Brain Mets element that we use. This is a different approach from RapidArc. It is a non-coplanar delivery, but it's an inverse optimized dynamic conformal arc. And what's different when you think about the island blocking problem is that when you have two targets that lie in the direction of MLC travel, what it does is break this down into two arcs. It'll treat one target on the first arc pass. They'll treat the second target on the second arc pass. This provides really two benefits. One, it substantially decreases the intervening normal brain dose, but then it increases the conformity of each of those particular targets. The argument might be that you're adding some time to the patient treatment, but usually, these arcs are on the order of 120 degrees. And, I think, at most, you're possibly adding 20 to 25 seconds to a treatment delivery for a particular couch position.
So as mentioned, this is an inverse optimized dynamic conformal arc. You know, it's a reduced planning time. I think once you've validated the system compared to a fluence modulated technique, I think you could say that the QA is reduced, if not, substantially reduced. This package even though Multiple Brain Mets is in the title, we have used this successfully for quite a few single-target cases. The new version, version 2, does provide jaw tracking. So this will actually, you know, improve conformity of peripheral targets, and you have to keep in mind that for a given couch position, that peripheral target might change. The nice thing about the software suite is, again, it does include imaging in any couch position with ExacTrac assuming you have gantry clearance. Some of the tools and their suite. One is an organ at risk sparing interface. This is really nice because if you think about a target that's in close proximity to an organ at risk, you know, a traditional dynamic conformal arc would just look at the margin around a particular target. What's nice here is that we can actually go in and identify that organ at risk, and just by simply identifying the organ at risk in the optimizer, we're able to bring the brain stem back within our OAR constraints.
So another interesting feature of version 2. Brainlab refers to this as beam's eye view margin optimization. I tend to think of it as a gradient optimization, but if you look at the top, this is kind of a standard plan delivery where you have a homogeneous dose distribution within the target and just kind of a standard sharp falloff. What you can do is you can use this beam's eye view optimization or gradient optimization to enforce a higher dose and a steeper falloff. Essentially, you're prescribing to a lower icy dose line, and that, you know, has the potential to, you know, decrease some of this surrounding normal brain dust. Cranial elements is the other module that we use or have started using more frequently for cranial SRS, and this is really for cases that are...you know, targets that are more irregularly shaped. It's a VMAT solution, and it's for a single target, though it does support a simultaneous integrated boost if you wanted to take something within a particular target to a little bit higher dose. Again, it supports ExacTrac, so you get the imaging with the patient in any couch physician.
The nice thing about these tools is that even if over a course of time, you treated a patient with different modules, you can combine them, or if you've got a patient say with 11 metastases, one is a regularly shape that you'd wanna treat with the cranial VMAT solution, and then you've got 10 smaller ones you'd wanna treat with multiple met solution, you can actually create a composite plan and then go in and evaluate that. I borrowed this slide just to show a little bit more detail. This is a Brainlab slide, but this is a case where they did 10 smaller metastases with the Multiple Brain Mets element. Then they did a central larger irregular target with cranial VMAT SRS, and then they combine those. And this really is just to show you that the product is somewhat vendor agnostic, and it supports both Elekta and variant platforms.
So our elements experience. As I mentioned, you know, we started down this journey...this path about three years ago. Most recently, we just added cranial elements. So I'll talk about that real quick because we don't have as much experience with that. But we've done 30 cases or probably just over 30 cases through June. We've treated everything from post-op cavities and intact metastases to some functional things such as vestibular schwannomas and gliomas. Dr. Gleason will go over a few cases of the cranial element. The largest component of our program is multiple mets or metastases treatment. This program started in late 2017, and since that time, we've treated more than 300 cases. And as I mentioned, you know, this solution is not just for multiple mets. A little less than, you know, 50% of those cases were treated as a single target. A little more were treated...you know, were two or more targets. A little over 50%, 164 cases were single-fraction SRS, and then for fractionation, we tend to favor five-fraction over three-fraction for fractionated SRS. And just looking at the graph, you can see the majority of our multiple targets patients are, you know, 3 targets at a time, but we have treated as many as 12 and 14 targets in a session.
Just some statistics from all of those cases for the multiple mets. People often ask, you know, "What size targets are you treating?" On average, the target size that we treat is about 18 millimeters, but it rages up to, you know, 76 millimeters. The images shown here show probably one of the smaller targets we've treated. This is a 6-millimeter target that we've treated. And then as far as volumes go, we've treated things that are fairly small all the way up to 55 ccs. Moving on, you know, some more statistics. The average conformity index is about 1.38, and the average gradient index is 4.3. I just like to make a quick comment here. You know, usually, if you look at the situation with the image on the left, you have this nice isolated target. In that situation, the conformity index is gonna be around 1.1 to 1.2. You're probably gonna have a gradient index of around 3. If you look at the image on the right, you know, where you have two targets in close proximity to each other, you start observing dose bridging and dose pull. Your conformity indices tended to degrade at that point.
And so, if you look back at our stats, you know, about 50% of the time we're treating, you know, multiple metastases patients. And more times than not, you know, you're gonna have targets in close proximity, and this is gonna be a reality that you're gonna have to face. So treating small targets with the agility. This is another, you know, topic that comes up. And, I think, historically, this has been considered a hardware limitation. And, I think, you know, in recent years, this has been overcome by software solutions. Now that we have the ability to optimize on things like, you know, anatomy and gantry and MLC and couch position, we're able to create some very conformal dose distributions. I like to think of optimized dynamic conformal arc as an elegant solution as opposed to something that's fluence modulated, which is a complex solution.
So, you know, moving on, you know, looking at leaf with a little bit closer and looking at version 2 of the software, Brainlab provided us with a version 2 model, and what we decided to do was go back and take 20 patients that were treated with version 1.5 of the software and then evaluate those patients in version 2. And so in addition to that, we got to look at the new jaw tracking feature with Elekta. And then they also provided a Varian HD120 model, so we could make some determinations about leaf width effects. We went ahead and did those comparisons. Then we looked at conformity index, gradient index, V12, and mean brain dose.
And so when we did that, we did a Wilcoxon, you know, paired comparison. If you look on the far left-hand side in dark gray, you can see the conformity index for version 1.5, and then moving to the center columns in orange and gold, you can see an improvement with version 2.0 and with version 2.0 with jaw tracking. And then, you know, over on the right hand column in light gray, the HD 120 was the best. Just keeping in mind that while there is, you know, a P value here at point 0.1, the numerical difference was 1.22 versus 1.25. Again, that's 1.22 for HD 120 versus 1.25 with the Elekta Agility with Jaw Tracking. Going on to the gradient index similar as the conformity index. We did see an improvement with version 2.0 continuing improvement. With the Jaw Tracking, again, the HD 120 showed the greatest improvement, but again, these numbers are pretty small. We went from basically 3.33 with the agility and Jaw Tracking to 3.2. For V12, again, improvements shown in version 2.0 continued improvements with Jaw Tracking. The V20 or excuse me, the V12 for the HD 120 was just under 11ccs versus 12ccs for the agility with Jaw Tracking.
Mean brain dose. Again, improvements with version 2.0 of the software. You know, for the HD 120, yes, this is an improvement, but it's 0.2 gray. It's a 0.2 improvement over the HD or excuse me, the agility with Jaw Tracking. We saw improvements with version 2.0 of the software both for, you know, jaw tracking and HD 120. The issue, I think, comes down to is that I can...we can demonstrate a statistical improvement with those numbers. For the HD 120, I think, the challenge there when you look at 2.5-millimeter versus 5-millimeter leaf width is that you're gonna have a challenge proving that there's a clinical impact for those patients treated with a smaller MLC.
Bogdan: Thank you for that review, Mike, and let's turn it over to Dr. Gleason to provide a clinical review of Alliance's Cancer Care program.
Dr. Gleason: All right. Well, thank you very much for that introduction, Bogdan, and thanks, Mike, for doing the majority of the work on this presentation and getting everything together. Mike thanked a lot of people at the beginning of his talk. So I'm gonna thank him. None of this would be possible without Mike's leadership in our program, and certainly, I feel much more comfortable doing something like a trigeminal nerve with a cone right next to the brainstem without a frame knowing that Mike has done all his end-to-end testing. So, I just wanna briefly do some case presentation showing how we've used the software to address specific patient needs. And I've intentionally tried to go from simple to more complicated as we move through this.
So we'll start with the first case, which is just gonna be a single target case. So this was a patient 52 years old was found to have a widely metastatic malignancy from an unknown primary, and so as part of that workup, she had an MRI of her brain that showed the small lesion in the right cerebellum. This was ultimately found to be from an occult breast primary actually. So we treated this with single-fraction SRS 22 gray in a single fraction ended up being 4 arcs, and this is a good example of the fact that even though the name of the software is multiple mets, we use it very frequently to treat single mets because, you know, most of these mets are around spherical targets. So the dynamic conformal arc delivery can still give you excellent, you know, conformity, and you can see our V12, in this case, is just 1.1cc. And this patient would have been treated with only four arcs. They may have been on the table for 15 or 20 minutes total to get this addressed.
So the second case, just stepping forward to the next progression, we'll have just two sites of disease. In this case, it's a patient with a metastatic non-small-cell lung cancer, and at the time of their diagnosis, they were found to have a metastatic disease of a liver, and an MRI of the brain was also performed as part of this initial workup. And the patient had these two lesions, both asymptomatic. We have this small lesion here in the posterior right temporal lobe and then a slightly larger lesion here in the left parietal lobe. So, again, we treated this with a single isocenter, but in this case at the geometric mean between the two targets, the small lesion here received 20 gray, and the lesion here on the left received 18 gray.
And one of the things as a physician that I like about the software is that if my dosimetrist or physicist sends me the plan, I can change things on the fly without having to send it back to them. So, you know, in fact, they may have sent this to me with 22 gray to this target and 18 to this, and within the software, I can quickly just change the dose myself if for some reason I wanted to do 20 gray and hit calculate, and within, you know, a couple minutes, I've got a new plan that I can quickly evaluate and sign without all that back and forth between yourself and the planner. But in this case both, again, with excellent conformity 1.13 and 1.20 and actually, again, in this case with 4 arcs. You know, we don't always use four arcs, but I did find going through my lecture today that seems to be a common topic. Here's that same case just with a different graphic from the software, but again, it just shows you how you can nicely create these clouds of radiation dose with all this intervening normal brain that's spared.
Our third case is just another step forward and complexity. So in this case, it's, again, a non-small-cell lung cancer patient who at the time of the initial staging workup is found to have these five asymptomatic brain metastases that are scattered throughout the brain. And so in this case, we, again, used a single isocenter and delivered 20 gray in single fractions all 5 sites. It was, again, four arcs. And our conformity index. This is a volume average conformity index across all the targets was 1.40, and our V12 was only 5.1cc. So, I think, this is a great example of the fact that you can still treat 5 targets and have a low V12 in a single fraction scenario. But the key here is that our target volume was quite small. So it's really a good illustration also of the principle that the volume of metastatic disease in the brain is likely much more important than the actual number of brain metastases when figuring out whether or not SRS is an option. So certainly, I'd rather treat these five tiny mets with single fraction SRS than treat, you know, three very large...three to four centimeter metastases.
Jumping to the next level of complexity. This was an 11 site case. So I will acknowledge that most of the time with someone with this many metastases, we would strongly consider whole-brain radiation potentially with hippocampal sparing and memantine unless the patient already had a significant amount of prior treatment or some other factor there. In this case, I really wanted to avoid whole brain radiation for a couple reasons. One, the patient was a little bit older, and I feel like they're a little bit more prone to the neurocognitive sequelae of that treatment, and more importantly, this was a melanoma. So, you know, we know that melanoma tends to be more radioresistant in many cases, and so I hate using whole-brain radiation in that scenario knowing the potential side effects of the treatment and knowing that it's often not all that effective for some melanoma patients.
But in this gentleman's case, even though there were 11 sites, they were nicely spaced throughout the brain, and they were all small. They were all less than 2 centimeters. Going into it, we didn't even try to plan this with a single fraction. I could just kind of tell based on experience that we wouldn't have an acceptable V12 with this many lesions, and so we went with a sixth grade times five fractionated SRS approach, again, with just a single isocenter, multiple max dynamic conformal arc approach. And we have volume average conformity index of 1.39 and a mean brain dose of 5 gray. So, again, with 5 fraction treatment, we often aim for that 4 to 5 gray range for the mean brain dose. So, I think, that's quite good in the setting of treating 11 tumors.
I tried to play with the angle when I was creating this screen capture here to figure out where I could kind of best show you these dose clouds. But if you look at this 15 gray line, there's only 2 areas where even the 15 gray, the 50% line bridges at all despite there being 11 sites of disease, and there was nowhere where the 20 gray line even became close to bridging. And the reason I like to look at the 15 and 20 gray lines is just from a practical standpoint. We know we can do three grade per fraction to the whole brain for 10 treatments or even 4 gray per fraction times...you know, 5 treatments to the whole brain. So certainly, in a fractionated, you know, just 5 treatment course, if we can keep that 3 gray per fraction volume to a low area, we're sparing a lot of intervening normal brain tissue.
This next case here. This is a patient who had renal cell carcinoma and came back with a large symptomatic metastasis in the right occipital lobe and was seen by one of our neurosurgeons and had a resection. And when we targeted our cavity...you can see that in this case, the cavity happens to be quite spherical. It's not an irregular target. And so in that case, we...you know, rather than trying to use a VMAT type delivery, we started with our dynamic conformal arc approach, again, using the multiple mets software albeit in this case just with a single target. And impressively, the conformity here is 1.14, and the gradient index is 2.58. So, I think, you know, a very elegant treatment plan that can be done in a quick throughput as far as the planning, the quality assurance, and then the delivery on the machine.
So this next case is an example of how...you know, as we know many patients who develop brain metastases will develop more brain metastases, particularly when we're not doing whole-brain radiation and particularly in this era where because of targeted therapies or immunotherapy, we have people with metastatic disease who are living longer. So this was a patient in her late 60s who had a metastatic lung cancer. It happened to be an adenocarcinoma, and she had an EGFR mutation and had been on that TKI for a number of years, but ultimately, in February of 2019 started to develop some resistance and was found to have some small lesions in her brain despite no active disease outside of the brain.
And so when I first saw her in that first treatment course, there were six lesions that we addressed. Five were treated with a single fraction, and then one lesion in the pons was treated with a five-fraction course. And then she came back two to three months later for that initial post-treatment scan in April. Everything looked great, but by the time we saw her in July, there were now, again, two new small lesions that had not been treated with the first course. So we again treated her with a single isocenter multiple mets plan, delivered I believe 20 grain, a single fraction to both those sites. When she came back for her two-month post-treatment imaging, again, everything that had been treated prior appeared to be controlled, either resolved entirely or much smaller by then, but there was, again, one new site of disease, and we'll show the plan on that here on the next slide.
So that one new site was here. It was about a centimeter or just under that in diameter, and so we again set her up for single-fraction SRS. Again, it's a single site of disease, but we used our dynamic conformal arcs within the MME software. And in this case, with it being a small target, very peripheral far away from normal structures, we were actually able to generate just a three couch position plan, which Mike has captured this for me. Apparently, it was just 12 minutes to deliver this patient's treatment. With this small target and only 3 couch positions, we have this conformity of 1.27 and a GI of 3.35. What this final picture here shows is just resolution of that metastatic site. So this is six or seven months later, and that site's no longer present. And, in fact, when I saw her most recently, she doesn't have any active disease in the brain. Everything ultimately resolved with treatment, and fortunately, she's found a systemic therapy regimen, has stopped on a clinical trial and so far has stopped making new brain metastases and is doing quite well.
This final slide is kind of busy, but I just wanted to give some examples of when I feel like the cranial element can be a very useful tool beyond just using the dynamic conformal arcs type approach. So if you look across the bottom of the screen here, these are all three post-operative cavities that were irregularly shaped, and all these were treated with either three-fraction or five-fraction SRS. And with this cranial element despite these unusual shapes, we have conformity indices on the 1.11 to 1.17 range and gradient indices on the range of 2.60 to 3.0. So, these are situations where I found that the cranial element can be quite useful when we have these irregular shapes. You know, physics still loves the dynamic [inaudible 00:41:00] on the standpoint of the less need for QA and other resources. So, often they'll...Michael run a question, and they'll do a decat plan, and in some cases, we'll find that the dynamic conformal arc plan is equally good. But with these irregular targets, clearly, the cranial element was superior.
And then these two cases across the top, I've just shown as examples where you can use dose painting or simultaneous integrated boost within the software. Now, you can't use it if the targets are remote from each other, but in a case like this...so we'll start with this one here in the top left corner of your screen. This was a gentleman with a metastatic colorectal cancer who had this ongoing oligometastatic pattern of disease where at presentation, he had a liver metastasis resected and was NED for a couple years after chemotherapy and management of his primary. And then he developed a lung metastasis, and that was treated with SBRT, and then a year or two after that, he developed this brain metastasis, and one of our neurosurgeons removed it.
And on the post-operative scan, you can see this small residual area of tumor on the anterior aspect of the cavity adjacent to the corpus callosum where the neurosurgeon was likely concerned about causing some type of morbidity there. So rather than treating this entire target, nine gray times three fractionated, what I did was I treated the gross disease nine gray times three, but I treated the other part of the cavity just eight gray times three. And again, you can see in this case a conformity to an X under 1.1.
And then this case here in the middle, this was a gentleman who had had a lung cancer, non-small-cell lung cancer stage two that was resected three to four years prior received adjuvant chemotherapy and then came back with symptomatic CNS metastases and no other measurable disease. He had a resection of a large metastatic site, and the only other area was this small satellite nodule that I'm indicating here that was not removed. I chose this exact slice so you could see both the targets, but if you could scroll in the other direction, you'd see that this cavity was actually quite large. And so I was reluctant to give six gray times five to such a large target volume, and so I opted to give five gray times five to the cavity where I really felt like we were dealing more with microscopic disease. Both the surgeon and the post-op imaging favor, you know, radiographic, you know, gross total resection, but I still was able to dose paint and give 6 gray times 5 to this unresected nodule just posterior and inferior to that cavity site and with the treatment able to get a conformity of 1.17 and a gradient of 3.07.
So, again, for us, this has revolutionized what we've been able to do in our community, and we continue to build on what we've done so far. And I would again like to thank all the support we've received from Brainlab as we've developed the program, and again I'd like to thank my physics team including Mike Taylor, my dosimetrist particularly Christy Bradley who does the majority of these plans, and then all our therapists out on the machine who deliver these SRS treatments every day. And thanks again for everyone who actually tuned in to listen to us. We're certainly happy to answer any questions everyone has regarding our experience.
Bogdan: Thank you, Dr. Gleason, thank you, Mike, for the great lectures. We have a few questions that hopefully you can answer. Dr. Gleason, let's maybe start with you. There was a question regarding a five-fraction dose regiment not being as effective as, of course, single fraction or a three-fraction treatment at nine gray. So, could you provide maybe some further explanations on the five-fraction treatment?
Dr. Gleason: Yes. So being close to UAB...a lot of the group from UAB, Dr. Fiveash, they do a lot of sixth gray times five fractionated stuff that they published, and so that's kind of where we started doing fractionated SRS. Because of the literature from Italy, Dr. Monetti, I believe it is, the nine gray times three, we also use that regiment as well. So I tend to use nine gray times three if it's say a single lesion that's 3.1 or 3.2 centimeters that I don't feel like I can treat single fraction. I know in Italy, I think, they use a 2-centimeter cut-off in that Monetti data, but we still tend to treat anything less than three centimeters in a single fraction. But once I switch to fractionated, if it's just barely too large, I sometimes do use that three fraction regimen, but if I'm treating numerous targets, we felt more comfortable doing the five fraction regimen knowing that there's always a balance of toxicity versus efficacy.
Bogdan: Great. Maybe I'll ask you another question. We polled everyone for what is the smallest volume that they've treated with the Agility MLC, and Mike did a good job explaining why software solutions really improve the symmetry that's possible with the Agility MLC these days. But from your perspective, Dr. Gleason, what targets would you not treat with the Agility MLC? And maybe give us an idea of where you would use cones in your practice.
Dr. Gleason: So, we always use cone when we treat our trigeminal neuralgia cases. We never have used an MLC for those, although I know there's some interesting work being done at UAB using a virtual cone with their HD 120 MLC, but we've always used a cone in those. For the majority of our mets cases, we have been able to treat those with the MLC. When I do have a really small target, I'll have physics do a comparison for me, and sometimes I'm surprised, and the conformity is almost identical even though the target's small. But occasionally, we have brought out, you know, a 7-millimeter cone to treat a very small lesion, and Mike might have a more detailed answer as far as the size of things that we have treated without the cone.
Mike: Sure. You know, I don't know if I'd add too much to that from what Dr. Gleason said, but I agree. I think we kind of take it on a case-by-case basis, and we've been actually pleasantly surprised by some of the smaller things we've been able to treat. And obviously, you know, if you can get away with doing with an MLC, it's a much easier day for us.
Bogdan: Okay. Mike, we have a lot of technical questions for you. So, in no particular order, I'll try to go through them. So, first question, when you perform a Winston-Lutz test, do you use ExacTrac to correct its time, you have a new couch angle? And maybe I'll ask you something related to this for the Versa HD LINAC. What is the overall couch runouts on your device, mechanical runouts, and then, you know, how do you compensate for that clinically?
Mike: Yeah. I'm gonna kind of combine both of those questions back by then. So, yes, to answer the first question, we do use ExacTrac for the Winston-Lutz procedure specifically on the couch, and that's twofold, one, to correct for the couch run out, which probably is pushing, you know, close to about 0.6 millimeters from our observations. The second reason is to make sure that the system is performing like we intend it to perform for patients.
Bogdan: Okay. Thank you for that. Another question. What is the low dose percentage when treating multiple targets with arc therapy?
Mike: Not sure I followed the question. Yeah. We have some parameters for what we look at for V12, you know, and then for, you know, fractionated SRS, you know, three-fraction SRS three gray, you know, five-fraction SRS four to five gray.
Bogdan: Okay. Let's recap a little bit. The margins that you use, and there's a related question to these on do you only use ExacTrac, or do you also use cone beam CT for alignment?
Dr. Gleason: I can take that one. As far as the margins, Mike showed that table where we select our margin based on the distance of the individual target from the isocenter as well as our ExacTrac tolerance. So we have a separate table for when we're using 0.8.8 ExacTrac tolerance and a separate table for when we're using 0.5 tolerance. And those tables will go anywhere from 1 to 2 millimeters depending on the distance from iso and the ExacTrac tolerance. You know, for a little over a year now, we've always chose our ExacTrac tolerance so that we can keep the margins either at 1 or 1.5 millimeters. So, it would be rare that we would use any larger margin. Now, this is obviously for intact metastases. We use more margin on a post-operative cavity more for clinical target volume rather than any issue with the setup, but usually, we use 1 or 1.5-millimeter margins.
As far as whether or not we use cone-beam CT. So there are plenty of centers that use ExacTrac alone and don't use CT. Our practice has been for single-fraction SRS to do a confirmatory cone-beam CT after the initial ExacTrac just as a second check because it's a single fraction treatment, and so it gives us additional confidence. I'll say that we've gone back and looked at those, and we've never really found anywhere we weren't already on target, but it does give us some additional confidence. For fractionated SRS, we only do the cone-beam CT on the first fraction. So if it's a three-fraction case, we'll do ExacTrac and cone-beam on day one, but on days two and three, we'll only use ExacTrac.
Bogdan: So when we polled the audience today, it seems a large majority of them are using an isomorphic 1-millimeter PTV expansion. Maybe in your practice, is that what you also see for the majority of your patients, and how frequently do you use a variable margin? And linked to this, of course, the 3.0 version of Brainlab software that is coming out in October will actually automatically generate variable margins for you. How relevant do you think that would be?
Dr. Gleason: You know, I think it depends on how you wanna look at it. I mean, in all honesty, we use a 1 or a 1.5-millimeter margin. So, with 0.5-millimeter additional expansion, you know, I'm not sure how clinically meaningful it is one way or the other if you do or don't use it. I suspect if you look at your...if you give 22 gray with a 1-millimeter margin versus if you give 20 gray with a 1.5-millimeter margin, you probably have very similar plans and similar efficacy and safety. But we use 1 or 1.5, and maybe we're overly detailed to add that 0.5-millimeter. That's been our current practice if they're farther from the isocenter.
Bogdan: Mike, there are some questions regarding those calculations, and one had to do with correcting for artifacts in your CT. So maybe I'll ask you as well if you've ever used Monte Carlo for multiple brands SRS or just stick to pencil beam. And then another question is how is the OAR sparing accomplished in multiple brands SRS?
Mike: So for the first one, you know, our CT, we have a Siemens Somatom, and it does have a metal artifact correction. I think the question involved the artifact from teeth. I haven't seen something in our practice...in our experience, I haven't seen anything that substantially affected that. We have started using to some degree Monte Carlo in evaluating that in comparison to the pencil beam to look at differences. The last part of that question. The OAR sparing is...that's tied up in the optimizer. So, you know, is identifying that. Then, you know, you can actually modulate that, you know, around those OARs, you know, just to kind of pull the dose back off of those structures that you've identified.
Bogdan: Okay. When talking about delivery time, does it only include beam on time, or does it also include IGRT time?
Dr. Gleason: So, yeah. The nice thing about ExacTrac is it's really quick. The ExacTrac part as far as imaging goes is incredibly efficient. The slow part for imaging would be on the days that we do the cone beam CT. So, you know, that can take upwards between a minute to two minutes to acquire that actual image. ExacTrac can be image that...you know, we do it at the cardinal gantry angles when we have imager clearance, and we've even, you know, started looking at doing some of these with imaging during treatment delivery so that they don't actually stop the treatment but just do the quick snap verification with both imagers. That's very quick.
Bogdan: Okay. Some questions regarding QA. How do you QA for multiple mets especially when you have more than 10 tumors, or does your QA change at all when you have more targets? And also when we polled today, it seems to be kind of an equal split between secondary checks and gamma analysis with mostly electronic arrays. Maybe you can also touch on what you routinely do in the clinic and if there's any rationale after initial commissioning to still to do gamma.
Dr. Gleason: So, yeah, when we first started the program, we went through and did a specific QA on every single target. That included a point dose on each one in addition to a second MU check. And so the challenge there obviously is now you're not placing your chamber in the isocenter. You need to go ahead and make sure that that chamber is in each target relative to the isocenter. In one sense, you know, that's a good, you know, end-to-end evaluation of your system. Once we accumulated, you know, about 20 patients, you know, we felt pretty comfortable in the second MU check. The only time that we default back to say a point dose at this point is really when the second MU check were having some issues there or some concerns, we'll go back and reevaluate that. I agree. You know, film and the 3D gel from somebody like RTsafe is a good end-to-end test of looking at spatial and antisymmetric fidelity. The SRS map check is probably one of the more convenient things relative to film. The only, you know, caveat there, again, depending on how far apart your targets are, you know, that device, I think it's about 7 centimeters square. So, you know, if your targets are beyond that, again, you're kind of looking at QA each individual target. From a physics perspective moving around each target and then delivering the entire plan can be time-consuming.
Bogdan: I know you predominantly showed plans today with a four-arc setup, but we have a question of really what is your default arc geometry setup. And also for these kind of our treatments, how frequently are the patients shifting within the mass system outside of your treatment tolerance?
Dr. Gleason: I'll let Mike take the first question as far as what our starting off template is. It seems like we usually end up with four to six couch kicks depending on the case. But as far as how often the patients move, I hope to have an answer to that. We had a medical student evaluating a set of patients and then COVID hit, and we stopped letting other people inside the department. That project kind of came to an end a couple months ago before we got the answer to that. But we actually are interested to kind of quantify how often do we have to make corrections at these couch kicks. It's not infrequent that when you move to the next position, you do have to make a minor ExacTrac adjustment, and that's what makes us, you know, comfortable to do the treatment because we do have the ability to image at these non-coplanar angles and make 60 corrections. I'll let Mike answer about...
Bogdan: Dr. Gleason, just on that comment, are you imaging every time you rotate the couch?
Dr. Gleason: Yes. Yeah. We imagine every couch position.
Mike: And I'd say, you know, I don't think we've kept the same numbers that Dr. Dworek has. I think, you know, yeah, we do see at least 50% of the time we're correcting particularly for a smaller residual tolerance. You know, for 0.5 and 0.5, we'll correct that more often. Regarding the number of couch kicks, I'm gonna go ahead...it looks like my dosimetrist, Christy Campbell, chimed in and gave me the cheat there. It looks like we usually end up with about four arcs. As Dr. Gleason said, we've seen as many as six. It's nice. You know, obviously, the fewer couch kicks you do, the faster the treatment delivery.
Bogdan: Okay. There are two questions that seem to be more directed in my way. So maybe I'll answer those on Brainlab's behalf, and then I have one more question for you. So, there were some questions regarding combining elements Multiple Brain Mets SRS with cranial SRS. And like you've showed today, Mike, and Camille, thanks for your answer as well, you're right. These things can be combined today. It's a little bit of a serial way of combining them, and then you have to do a composite evaluation in those review. However, this is changing with version 3.0 of elements. We're slotted to release version 3.0, as I said earlier, in October with availability to all of you starting November. And with that version, the integration between Multiple Brain Mets SRS and Cranial SRS is more streamlined. You can start a plan for however many tumors you want in brain mets, and if you choose to do VMAT further dose painting in cranial SRS, you can just switch to that element, and it will automatically understand the dose that has been already delivered by Multiple Brain Mets SRS and will allow you then to have a much more streamlined composite view. And what Mike has showed today in terms of dynamic jaw tracking for the Agility MLC, that would also be available for all variant MLCs with version 3.0 as well. And I guess here's a question for you. Are you considering Brainlab's new product ExacTrac Dynamic?
Dr. Gleason: We have looked at it. We are certainly considering it. It's an excellent tool.
Mike: Yes, agreed.
Bogdan: Okay. I think that's it. There's a comment regarding the Winston-Lutz workflow. Mike, if you'd like to take that regarding why...
Mike: Yeah. I think that was Marcel. I think there was something about including the larger gantry angles, and I'm not sure if the path there is, you know, are these gantry angles that we're not necessarily treating. And I agree. The larger, you know, run-out happens at the extremes, and perhaps we shouldn't include those. And, you know, our procedure right now is pretty standardized, and our workflow is pretty standardized. As far as going to Winston-Lutz procedure, you know, prior to patient treatment, we usually, you know, carve out at least, you know, 20 minutes or so, and if things go well, you know, that's like I said, a pretty quick process to get through.
Bogdan: Gentlemen, again, thank you very much for your lectures today, and thank you all for your participation. Have a great rest of the week and hope to see on our next webinar.
Mike: Thanks, Bogdan.
Dr. Gleason: Thank you, Bogdan.
Bogdan: Yeah, bye.