Transcript
Well, thank you very much for the invitation to talk about iterative planning for the treatment of multiple brain metastases. My disclosure is I'm a radiation oncologist, not a neurosurgeon. This is our Novalis STx, which we have a joint business venture with the Department of Radiation Oncology, and the Department of Neurosurgery, professor Van Dino [SP], Dr. [inaudible 00:00:24] Schwitzer [SP].
And today, I'm going to present our first clinical experience and also some patient outcomes of treatment with the Elements Multimets software version 1.5. And also, we've had recently the opportunity to perform plan comparisons between the current version 1.5 and a prototype of version 2.0, due for release. So, I don't need to mention this paper again, but since 2014, I think we've all been more comfortable with offering radiosurgery for five or more brain metastases. And actually, we had a Novalis STx after this time point, so we went straight in. And this is a patient of ours, with 9 brain metastases, a 50-year-old lady wild type, lung cancer. Total cubic of 7.9, who retreated with 1.20 gray to each metastasis.
The software basis for Elements 1.5 we've heard is the single isocenter dynamic conformal non-modulated arc technique. Some metastases were treated in one pass of the arc, the others on the way back. And so far, we've treated 47 patients with 106 brain metastases with this software. I must say that 24 patients had a single metastasis, and 23 had multiple metastases ranging between 2 and 9, a total of 49 brain metastases. We have clinical follow-up data so far on 23 patients, 14 clinical, and 9 with an MRI, at least 3 months after treatment.
And we've been able to show 89% local control rates, so 8 out of 9 patients, low numbers, but the sort of control rates that we would expect with radiosurgery for single metastases, so we're confident that we are actually meeting our target. The demographics, I think, are very standard in terms of age, gender, diagnosis. And as has been alluded to previously, the most frequent outcome is local control of the treated metastases, but then with new brain metastases at three months. And a lot of patients then come back for a further round of radiosurgery.
One patient had questionable radiation necrosis or progression, and it was a superficial temporal lesion that was then resected. And due to leptomeningeal disease on the follow-up MRI, did receive whole-brain radiotherapy. But we also had a complete remission in a patient with gastric cancer. You'll see that we don't always use just one isocenter, and I'll go into the reasons for this subsequently, but some patients have two or three isocenters. We've also combined version 1.5 with iPlan in a couple of cases. The reasons for these, in 8 out of 14 patients, we clustered the metastases into 2 or 3 isocenters. Because at present, we're unable to combine hypofractionation with single prescription doses in the software.
And hypofractionation in our center, as indicated in the post-operative setting, and for particularly large or cystic lesions or those in very eloquent regions, or if we were unable to meet our brain constraints, or V10 of 10 cubic, or to avoid organs at risk, such as the chiasm, or unable to consider previously treated metastases with version 1.5. And then there's always the geometric constraint of the distance between metastases regardless of the software.
Just to discuss a little further, this distance of the metastasis from the isocenter, I think a very useful rule of thumb from the paper from [inaudible 00:04:05] is that with every centimeter from the isocenter, there's a 0.1 millimeter per centimeter rotational uncertainty. So, with our ExacTrac system and the 6 degrees of freedom table, a distance of up to 4 centimeters in the isocenter can be confidently corrected with our 0.5-millimeter and 0.5-degree tolerances. But over 4 centimeters, our choice so far has been to then use multiple isocenters. A different approach is also to increase the margin for smaller lesions more peripherally, and we will be exploring that as a next step. Oh, and just to mention in version 2.0 now, when you look at the DVH, it does specifically give you the distance of the metastasis from the isocenter which is extremely helpful. It's otherwise very difficult to extrapolate in 3D.
I'm only going to show one case, but it's a very typical case. This was a patient with non-mutated lung cancer, who had declined all systemic disease treatment, who had no chemotherapy for the primary and then presented with neurological symptoms, raised intracranial pressure symptoms for three brain metastases, wanted an effective treatment that wouldn't have any side effects. And this was quite early on in our history with the Novalis STx, and we actually only gave 1.15 gray in this case.
Two months later, she had a repeat MRI which showed some partial remission of these lesions. As is the way three months later, three further lesions, again, treated with a single isocenter and another round of 1 times 15 gray because we turn this to be clinically effective in her case so far. And then she succumbed to extracranial disease six months later, at which 0.6 lesions were controlled to be no new lesions. And I say 1.1 times 20 gray is our standard, this was exceptional.
Moving on to the second part. The new features in version 2.0 of the software include the option to optimize coverage of the individual metastases. There's a slider to increase sparing of the normal tissues. There's the option to avoid organs at risk. There's a workaround that you can use previously treated metastases as organs at risk, and you can also add additional arcs to individual metastases. And these are the features that are then iterative planning or dose iteration. The current 1.5 is extremely user-friendly, as we've already heard. One click, you get a plan, but then you need to accept it or reject it. Whereas now you can actually modify any features that aren't to your liking. And these are aspects that we've evaluated so far.
We've also done a trial plan with up to 20 metastases, which is an increase from the 15 that can allegedly be treated with version 1.5. There's also now an option to prescribe to the isodose. I must say in our center, our interpretation of iso 91 is more that we apply 100% of the dose to at least 99% of the volume with a Dmax of 125%. So roughly equivalent to 80% prescription isodose. And in this software, there's also an option for the Monte Carlo dose calculation, but we have not explored that yet. And in our department, this would be future work.
So, to summarize our findings, we have planned in 1.5, with a single isocenter at 1 times 20, and replanned in version 2.0, also with the single isocenter with 1 times 20. Twenty patients with 2 to 9 brain metastases, totaling 85 brain metastases. And as the maximum dose relation or prescription isodose decreases from 83 to 75, you see, as expected, a parallel increase in mean PTV dose. The conformity was relatively unchanged, but the gradient index also improved. And this is with a slider for normal brains varying between low standard and high.
I think very interesting was this finding. I mean, we know that normal brain dose, so V10, increases with PTV size. We've also seen charts along this line this morning. But then as you use your slider to iterate the planning to spare normal brain between low, medium, and high settings, there is then less V10 dose. So, you can spare more normal brain through this dose iteration process. And probably if we did this further, we'd be able to reduce it yet again.
This was just a trial of a patient with 19 brain metastases. A melanoma patient was not treated, had a total volume of 20 cubic. I'm not sure that I would treat, but really just to confirm that a single click or with minimal input other than a little iteration does produce acceptable plans if we think that we have the dose metrics for normal brain for 19 metastases. And I think a really big improvement with version 2.0 compared to version 1.5 is the option to spare organs at risk. This was a reason why we rejected a few plans with version 1.5. You'll see here in the top left that when the arc covers the target but doesn't pass through an organ at risk, the MLC is conformed to the target. However, when the arc then passes over the chiasm, the MLCs will then conform to shield the chiasm.
And you might be able to see bottom left this was the one-click fully automated plan with version 1.5 giving 13 grays the chiasm, for us that was too high. And then in version 2.0, you have the slider, and if you set it too high, it's possible to reduce the dose to the chiasm by compressing these isodoses in fairly and reducing the dose to 6 gray. And, again, that could probably be reduced further should you want to in a different situation by further iteration.
So, in summary, the Multimets version 1.5 is routine in our clinic. Even for one metastasis, we really like the fusion algorithm. The auto contouring is very reliable, and the speed of plan generation and then the whole process, the plan checking, and the reporting is also very popular with the Medical Physics Department. With version 2.0, we found that we were able to produce single isocenter plans that we were not able to achieve with version 1.5. And that's due now to the ability to manipulate the conformity/gradient index trade-off for individual metastases, and also for the organs at risk. And we were able to show that you can reduce the dose to normal brain further by increasing normal brain sparing by improving the gradient index most probably, and this is likely to enable the treatment of a greater volume of brain metastases in the future.
Thank you for your attention. Thank you to our department back in Aarau, Switzerland. And a special thank you to Niki Lomax sitting in the audience who's done all the work for the planning and the replanning, without whom we wouldn't have a talk today. Thank you.
And today, I'm going to present our first clinical experience and also some patient outcomes of treatment with the Elements Multimets software version 1.5. And also, we've had recently the opportunity to perform plan comparisons between the current version 1.5 and a prototype of version 2.0, due for release. So, I don't need to mention this paper again, but since 2014, I think we've all been more comfortable with offering radiosurgery for five or more brain metastases. And actually, we had a Novalis STx after this time point, so we went straight in. And this is a patient of ours, with 9 brain metastases, a 50-year-old lady wild type, lung cancer. Total cubic of 7.9, who retreated with 1.20 gray to each metastasis.
The software basis for Elements 1.5 we've heard is the single isocenter dynamic conformal non-modulated arc technique. Some metastases were treated in one pass of the arc, the others on the way back. And so far, we've treated 47 patients with 106 brain metastases with this software. I must say that 24 patients had a single metastasis, and 23 had multiple metastases ranging between 2 and 9, a total of 49 brain metastases. We have clinical follow-up data so far on 23 patients, 14 clinical, and 9 with an MRI, at least 3 months after treatment.
And we've been able to show 89% local control rates, so 8 out of 9 patients, low numbers, but the sort of control rates that we would expect with radiosurgery for single metastases, so we're confident that we are actually meeting our target. The demographics, I think, are very standard in terms of age, gender, diagnosis. And as has been alluded to previously, the most frequent outcome is local control of the treated metastases, but then with new brain metastases at three months. And a lot of patients then come back for a further round of radiosurgery.
One patient had questionable radiation necrosis or progression, and it was a superficial temporal lesion that was then resected. And due to leptomeningeal disease on the follow-up MRI, did receive whole-brain radiotherapy. But we also had a complete remission in a patient with gastric cancer. You'll see that we don't always use just one isocenter, and I'll go into the reasons for this subsequently, but some patients have two or three isocenters. We've also combined version 1.5 with iPlan in a couple of cases. The reasons for these, in 8 out of 14 patients, we clustered the metastases into 2 or 3 isocenters. Because at present, we're unable to combine hypofractionation with single prescription doses in the software.
And hypofractionation in our center, as indicated in the post-operative setting, and for particularly large or cystic lesions or those in very eloquent regions, or if we were unable to meet our brain constraints, or V10 of 10 cubic, or to avoid organs at risk, such as the chiasm, or unable to consider previously treated metastases with version 1.5. And then there's always the geometric constraint of the distance between metastases regardless of the software.
Just to discuss a little further, this distance of the metastasis from the isocenter, I think a very useful rule of thumb from the paper from [inaudible 00:04:05] is that with every centimeter from the isocenter, there's a 0.1 millimeter per centimeter rotational uncertainty. So, with our ExacTrac system and the 6 degrees of freedom table, a distance of up to 4 centimeters in the isocenter can be confidently corrected with our 0.5-millimeter and 0.5-degree tolerances. But over 4 centimeters, our choice so far has been to then use multiple isocenters. A different approach is also to increase the margin for smaller lesions more peripherally, and we will be exploring that as a next step. Oh, and just to mention in version 2.0 now, when you look at the DVH, it does specifically give you the distance of the metastasis from the isocenter which is extremely helpful. It's otherwise very difficult to extrapolate in 3D.
I'm only going to show one case, but it's a very typical case. This was a patient with non-mutated lung cancer, who had declined all systemic disease treatment, who had no chemotherapy for the primary and then presented with neurological symptoms, raised intracranial pressure symptoms for three brain metastases, wanted an effective treatment that wouldn't have any side effects. And this was quite early on in our history with the Novalis STx, and we actually only gave 1.15 gray in this case.
Two months later, she had a repeat MRI which showed some partial remission of these lesions. As is the way three months later, three further lesions, again, treated with a single isocenter and another round of 1 times 15 gray because we turn this to be clinically effective in her case so far. And then she succumbed to extracranial disease six months later, at which 0.6 lesions were controlled to be no new lesions. And I say 1.1 times 20 gray is our standard, this was exceptional.
Moving on to the second part. The new features in version 2.0 of the software include the option to optimize coverage of the individual metastases. There's a slider to increase sparing of the normal tissues. There's the option to avoid organs at risk. There's a workaround that you can use previously treated metastases as organs at risk, and you can also add additional arcs to individual metastases. And these are the features that are then iterative planning or dose iteration. The current 1.5 is extremely user-friendly, as we've already heard. One click, you get a plan, but then you need to accept it or reject it. Whereas now you can actually modify any features that aren't to your liking. And these are aspects that we've evaluated so far.
We've also done a trial plan with up to 20 metastases, which is an increase from the 15 that can allegedly be treated with version 1.5. There's also now an option to prescribe to the isodose. I must say in our center, our interpretation of iso 91 is more that we apply 100% of the dose to at least 99% of the volume with a Dmax of 125%. So roughly equivalent to 80% prescription isodose. And in this software, there's also an option for the Monte Carlo dose calculation, but we have not explored that yet. And in our department, this would be future work.
So, to summarize our findings, we have planned in 1.5, with a single isocenter at 1 times 20, and replanned in version 2.0, also with the single isocenter with 1 times 20. Twenty patients with 2 to 9 brain metastases, totaling 85 brain metastases. And as the maximum dose relation or prescription isodose decreases from 83 to 75, you see, as expected, a parallel increase in mean PTV dose. The conformity was relatively unchanged, but the gradient index also improved. And this is with a slider for normal brains varying between low standard and high.
I think very interesting was this finding. I mean, we know that normal brain dose, so V10, increases with PTV size. We've also seen charts along this line this morning. But then as you use your slider to iterate the planning to spare normal brain between low, medium, and high settings, there is then less V10 dose. So, you can spare more normal brain through this dose iteration process. And probably if we did this further, we'd be able to reduce it yet again.
This was just a trial of a patient with 19 brain metastases. A melanoma patient was not treated, had a total volume of 20 cubic. I'm not sure that I would treat, but really just to confirm that a single click or with minimal input other than a little iteration does produce acceptable plans if we think that we have the dose metrics for normal brain for 19 metastases. And I think a really big improvement with version 2.0 compared to version 1.5 is the option to spare organs at risk. This was a reason why we rejected a few plans with version 1.5. You'll see here in the top left that when the arc covers the target but doesn't pass through an organ at risk, the MLC is conformed to the target. However, when the arc then passes over the chiasm, the MLCs will then conform to shield the chiasm.
And you might be able to see bottom left this was the one-click fully automated plan with version 1.5 giving 13 grays the chiasm, for us that was too high. And then in version 2.0, you have the slider, and if you set it too high, it's possible to reduce the dose to the chiasm by compressing these isodoses in fairly and reducing the dose to 6 gray. And, again, that could probably be reduced further should you want to in a different situation by further iteration.
So, in summary, the Multimets version 1.5 is routine in our clinic. Even for one metastasis, we really like the fusion algorithm. The auto contouring is very reliable, and the speed of plan generation and then the whole process, the plan checking, and the reporting is also very popular with the Medical Physics Department. With version 2.0, we found that we were able to produce single isocenter plans that we were not able to achieve with version 1.5. And that's due now to the ability to manipulate the conformity/gradient index trade-off for individual metastases, and also for the organs at risk. And we were able to show that you can reduce the dose to normal brain further by increasing normal brain sparing by improving the gradient index most probably, and this is likely to enable the treatment of a greater volume of brain metastases in the future.
Thank you for your attention. Thank you to our department back in Aarau, Switzerland. And a special thank you to Niki Lomax sitting in the audience who's done all the work for the planning and the replanning, without whom we wouldn't have a talk today. Thank you.