Transcript
Hello, everybody. So, my name is Mariangela Zamburlini and I work in the University Hospital in Zurich. And today, I would like to talk to you about our experience with Elements 2.0, and why we decided to go in this direction to treat in the future. So, maybe, I can go. So, just we are on a collaboration with Brainlab. So, just to give you an idea of how many patients we treat, So you see that the number of stereo patient have been growing in the past years. And about 30% of our patients are treated with our stereotactic treatment, and about half of them, they are brain patients.
So the workflow that we use is as follows. So, we take MRI and CT, and then we fuse the image, we contour like the doctor contours the lesions, then we proceed with the treatment planning. And then, on the day of the treatment, the patient is positioned. We also use ExacTrac, and then the treatment delivery. The key findings that we would like really to keep this workflow as short as possible. And, at the moment, we do it in less than a week, but, as you will see my presentation, one of the problems we have, and the reason why we go to Elements is that we would like to shorten it further when we can do our treatment planning faster.
So, in this presentation, I will focus on the treatment planning and I would talk a little bit about the treatment delivery. So, this is our requirement for the perfect plan so, you know, what I'm talking. So, we prescribe the dose to the 80% or equivalent to the 80%. So we want that the 100% of the PTV is covered by 99% of the dose. The Dmax has to be around 125%. We also have a special constraint on the GTV. We want the dose of the 95% of the GTV to be higher than 113%, and the conformity index we would like ideally to be under 1.3. I will talk in the presentation about conformity index and gradient index just to see there are different ways to define it so just that we are all in the same page. So, for me, the conformity index would be the ratio between the volume covered by the 100% isodose divided by the PTV volume, and the gradient index is the volume covered by the 50% prescription isodose to the 100%.
So what was our problem? So, we used to plan our stereo patient using Varian Eclipse, and the plan quality was actually good so we didn't have any problem with that. But the problem is the time required to make such a plan, especially now that we treat more and more lesions. So you see here an example, what is the problem? When you have multiple lesions, eclipse want to treat them all simultaneously, and this creates gap where there are more than one lesion aligned to the MLC. You see here an example.
So what the [inaudible 00:03:40] has to do, has to create a plan then look how is the MLC motion try to change the colli, try to change the couch kick until it reduces as much as possible this gap. You can see here an example of a patient. The same patient planned with a poor colli settings on the right side, and with a better colli settings in the left side. So, it does make a difference. You can see also in the portal dosimetry, you see these on unwanted dose in between the two lesions.
So the problem for us was what can we do to speed up our process, and not to take hours to do a plan? We look into the Brainlab solution and we like it very much. First of all, it's a dynamic conformal arc. This is the best solution to spare the normal brain tissue. And also what we like is that the system does exactly what we want. So, for each angle, it selects a certain number of lesions, and it treats only these ones so there are not these unwanted gap in the MLC. You see here an example if I can use it. So, in this example, we have five couch kicks than arcs, and these pictures refer on this arch that is treated orange one. So, when there is this orange arc, only the orange lesions will be treated. And when the arc comes in the other direction, so same couch kick, but other direction, it will treat selectively the other one.
So, we like the solution, but unfortunately in version 1.5 of Elements it was not possible to prescribe in homogenous dose. So we did small study with 10 patients so we thought, "Okay, if we cannot do in homogenous, we just try to change the normalization to get the same equivalent coverage of our PTV and the GTV, but these didn't really work, because you see here the conformity index also went up by normalizing all of our plan. So, yeah, the results were promising, but we were really not there. So after working some months together with the Brainlab, they come out with a very nice solution. Now, in the version 2.0, so now you see there is these advanced editing options in which the user can choose to have any homogenous dose prescription, and also for each target, you can choose the dose, and also at which dose level you want to prescribe.
It's okay we repeat the same thing. This time we made a bigger study with 30 patients, and we divide them into three groups. So 10 patients had a single lesion, 10 patients had few lesions, so 2 to 5, and 10 patients had more than 5 lesions. All the plans were done or a Truebeam high-definition 6FFF full dose rate. And you can see here the difference between the number of arcs so you'll see that for the single lesion we use templates for Elements with 4 arcs and 3 couch kicks. And, therefore, multiple lesion, we use up a template with 10 arcs and 4 couch kicks. Sometimes we have to add an extra arc to get to the right plan.
For VMAT, we usually use two couch kicks, rarely, we use additional couch kicks, and we use a number of arcs which ranges from 4 to 9 in these study. So we were a bit worried, because you see now VMAT has much less couch kicks than Elements. And, you know, the number of arcs is also higher, and it has to be this way because since it selectively treat only a few METs at each arc, it cannot be that with the same amount of arcs it gets the same result. But, I will come back later on this point.
So, these are the results. So this is the comparison regarding the PTV and the GTV dose between the two systems. So on the horizontal line, you always have the dynamic conformal arc, therefore Elements, and in the vertical, you have the VMAT plans. This is the unity line so anything that is below here means that it is higher with Elements. So you see that the PTV mean dose is higher, the PTV D99 is higher. The GTV mean dose is also higher. The GTV D95 is more or less the same. I don't see a big difference.
So from the PTV-GTV coverage, I would say, yeah, Elements works. But the big difference I see really is the amount of healthy tissue that is irradiated. If we look at the conformity index and we see Elements versus the Clinical Plan for their one metastasis, this is basically no difference. For more metastases, Elements is not as good as the VMAT. And we will see in a moment why this is the case. But when we look at the gradient index, so how the dose fall off, in Elements, this is really more favorable. And if we look also at the volume cover by the 5-gray of normal brain, of healthy brain, and the volume of 12-gray, so here I show you the ratio between the percentage difference in the V5 between Elements and the Clinical Plans. So you see that in every case, Elements has a much less V5. And I would say by quiet a large amount because you see here, it goes from 20% to almost 80%. Why do I care about V5? I care, because these patients in our clinics come back so we often have to re-irradiate these patients so their V5 also becomes important. But if you look at the V12, it's actually the same trend, only on a couple of patients who have that Elements performed worse.
So, this is an example for you of the distribution looks like, and this is what I want to show you as a motivation of the problems with conformity index for multiple metastases. What we notice is that you see here we have a cluster of lesions, and when we have a cluster, the dynamic conformal arc is not as good in shaping the dose around. And we are still working with the Brainlab to find a solution to improve the case. But what is important is that the gradient index anyway fall off quiet fast.
Coming back now to these worries that we had at the beginning. So we have more arcs. Is it going to patient who lie on the bed for two hours? This is not the case. So, if we compare the number of MUs, you see that for single lesion, there is no difference. But, for multiple lesion, is actually Elements, we have a statistically significantly lower number of MUs. You have also to consider that the dose rate is always at the maximum. And, therefore, if we measure the treatment time between delivering, the VMAT plan, and the Elements plan is actually faster with Elements. But, of course, in this rise, I only considered the treatment time, so the time it takes to treat the plan. I don't take into account the time that you want to use your ExacTrac, for example, to check that the patient did not move. Of course, if you want to check that the patient is in the right position at every couch kick, and you have more couch kicks, these will increase your treatment time. But this is up to the clinic by how many times you want to check the patient positioning during the treatment.
So, in conclusions, for us, Elements 2.0 was really a step forward because now it allows more flexibility to have the prescription, the inhomogenous dose prescription. It produces treatment plans which are at least as good as our previous clinical plan. But what was important for us was really that the process becomes faster and it becomes more user-friendly. And Elements is really user-friendly, and in a couple of clicks the plan is ready. And, therefore, it can help us to squeeze this time between the imaging and the treatment delivery, which is also important in these type of patients.
I would like to thank my colleagues, especially, Dr. Krayenbühl and Professor Andratschke, and all the team of Brainlab to help us achieving these results. And thank you all for your attention. And questions are afterwards.
So the workflow that we use is as follows. So, we take MRI and CT, and then we fuse the image, we contour like the doctor contours the lesions, then we proceed with the treatment planning. And then, on the day of the treatment, the patient is positioned. We also use ExacTrac, and then the treatment delivery. The key findings that we would like really to keep this workflow as short as possible. And, at the moment, we do it in less than a week, but, as you will see my presentation, one of the problems we have, and the reason why we go to Elements is that we would like to shorten it further when we can do our treatment planning faster.
So, in this presentation, I will focus on the treatment planning and I would talk a little bit about the treatment delivery. So, this is our requirement for the perfect plan so, you know, what I'm talking. So, we prescribe the dose to the 80% or equivalent to the 80%. So we want that the 100% of the PTV is covered by 99% of the dose. The Dmax has to be around 125%. We also have a special constraint on the GTV. We want the dose of the 95% of the GTV to be higher than 113%, and the conformity index we would like ideally to be under 1.3. I will talk in the presentation about conformity index and gradient index just to see there are different ways to define it so just that we are all in the same page. So, for me, the conformity index would be the ratio between the volume covered by the 100% isodose divided by the PTV volume, and the gradient index is the volume covered by the 50% prescription isodose to the 100%.
So what was our problem? So, we used to plan our stereo patient using Varian Eclipse, and the plan quality was actually good so we didn't have any problem with that. But the problem is the time required to make such a plan, especially now that we treat more and more lesions. So you see here an example, what is the problem? When you have multiple lesions, eclipse want to treat them all simultaneously, and this creates gap where there are more than one lesion aligned to the MLC. You see here an example.
So what the [inaudible 00:03:40] has to do, has to create a plan then look how is the MLC motion try to change the colli, try to change the couch kick until it reduces as much as possible this gap. You can see here an example of a patient. The same patient planned with a poor colli settings on the right side, and with a better colli settings in the left side. So, it does make a difference. You can see also in the portal dosimetry, you see these on unwanted dose in between the two lesions.
So the problem for us was what can we do to speed up our process, and not to take hours to do a plan? We look into the Brainlab solution and we like it very much. First of all, it's a dynamic conformal arc. This is the best solution to spare the normal brain tissue. And also what we like is that the system does exactly what we want. So, for each angle, it selects a certain number of lesions, and it treats only these ones so there are not these unwanted gap in the MLC. You see here an example if I can use it. So, in this example, we have five couch kicks than arcs, and these pictures refer on this arch that is treated orange one. So, when there is this orange arc, only the orange lesions will be treated. And when the arc comes in the other direction, so same couch kick, but other direction, it will treat selectively the other one.
So, we like the solution, but unfortunately in version 1.5 of Elements it was not possible to prescribe in homogenous dose. So we did small study with 10 patients so we thought, "Okay, if we cannot do in homogenous, we just try to change the normalization to get the same equivalent coverage of our PTV and the GTV, but these didn't really work, because you see here the conformity index also went up by normalizing all of our plan. So, yeah, the results were promising, but we were really not there. So after working some months together with the Brainlab, they come out with a very nice solution. Now, in the version 2.0, so now you see there is these advanced editing options in which the user can choose to have any homogenous dose prescription, and also for each target, you can choose the dose, and also at which dose level you want to prescribe.
It's okay we repeat the same thing. This time we made a bigger study with 30 patients, and we divide them into three groups. So 10 patients had a single lesion, 10 patients had few lesions, so 2 to 5, and 10 patients had more than 5 lesions. All the plans were done or a Truebeam high-definition 6FFF full dose rate. And you can see here the difference between the number of arcs so you'll see that for the single lesion we use templates for Elements with 4 arcs and 3 couch kicks. And, therefore, multiple lesion, we use up a template with 10 arcs and 4 couch kicks. Sometimes we have to add an extra arc to get to the right plan.
For VMAT, we usually use two couch kicks, rarely, we use additional couch kicks, and we use a number of arcs which ranges from 4 to 9 in these study. So we were a bit worried, because you see now VMAT has much less couch kicks than Elements. And, you know, the number of arcs is also higher, and it has to be this way because since it selectively treat only a few METs at each arc, it cannot be that with the same amount of arcs it gets the same result. But, I will come back later on this point.
So, these are the results. So this is the comparison regarding the PTV and the GTV dose between the two systems. So on the horizontal line, you always have the dynamic conformal arc, therefore Elements, and in the vertical, you have the VMAT plans. This is the unity line so anything that is below here means that it is higher with Elements. So you see that the PTV mean dose is higher, the PTV D99 is higher. The GTV mean dose is also higher. The GTV D95 is more or less the same. I don't see a big difference.
So from the PTV-GTV coverage, I would say, yeah, Elements works. But the big difference I see really is the amount of healthy tissue that is irradiated. If we look at the conformity index and we see Elements versus the Clinical Plan for their one metastasis, this is basically no difference. For more metastases, Elements is not as good as the VMAT. And we will see in a moment why this is the case. But when we look at the gradient index, so how the dose fall off, in Elements, this is really more favorable. And if we look also at the volume cover by the 5-gray of normal brain, of healthy brain, and the volume of 12-gray, so here I show you the ratio between the percentage difference in the V5 between Elements and the Clinical Plans. So you see that in every case, Elements has a much less V5. And I would say by quiet a large amount because you see here, it goes from 20% to almost 80%. Why do I care about V5? I care, because these patients in our clinics come back so we often have to re-irradiate these patients so their V5 also becomes important. But if you look at the V12, it's actually the same trend, only on a couple of patients who have that Elements performed worse.
So, this is an example for you of the distribution looks like, and this is what I want to show you as a motivation of the problems with conformity index for multiple metastases. What we notice is that you see here we have a cluster of lesions, and when we have a cluster, the dynamic conformal arc is not as good in shaping the dose around. And we are still working with the Brainlab to find a solution to improve the case. But what is important is that the gradient index anyway fall off quiet fast.
Coming back now to these worries that we had at the beginning. So we have more arcs. Is it going to patient who lie on the bed for two hours? This is not the case. So, if we compare the number of MUs, you see that for single lesion, there is no difference. But, for multiple lesion, is actually Elements, we have a statistically significantly lower number of MUs. You have also to consider that the dose rate is always at the maximum. And, therefore, if we measure the treatment time between delivering, the VMAT plan, and the Elements plan is actually faster with Elements. But, of course, in this rise, I only considered the treatment time, so the time it takes to treat the plan. I don't take into account the time that you want to use your ExacTrac, for example, to check that the patient did not move. Of course, if you want to check that the patient is in the right position at every couch kick, and you have more couch kicks, these will increase your treatment time. But this is up to the clinic by how many times you want to check the patient positioning during the treatment.
So, in conclusions, for us, Elements 2.0 was really a step forward because now it allows more flexibility to have the prescription, the inhomogenous dose prescription. It produces treatment plans which are at least as good as our previous clinical plan. But what was important for us was really that the process becomes faster and it becomes more user-friendly. And Elements is really user-friendly, and in a couple of clicks the plan is ready. And, therefore, it can help us to squeeze this time between the imaging and the treatment delivery, which is also important in these type of patients.
I would like to thank my colleagues, especially, Dr. Krayenbühl and Professor Andratschke, and all the team of Brainlab to help us achieving these results. And thank you all for your attention. And questions are afterwards.