Transcript
Yael: So, good afternoon. I would like to thank the organizers of this meeting for this really excellent meeting. My name is Yael Mardor. I'm from Israel, and I represent a group of researchers that developed the TRAMs methodology for differentiating tumor from treatment effects. We have licensed this methodology to Brainlab and... Which one is it?
Man: That was... Yeah.
Yael: Okay. Just... Where will I go? Can you help me with advancing the slides? Go back? Okay. I'll go back. It's working opposite maybe because it's Israel going from right to left instead of left to right.
So we licensed this technology to Brainlab. It's called CCA, Contrast Clearance Analysis. And the way we differentiate tumor from treatment effects, we scan the patient with standard 3D-T1 five minutes after contrast injection. We do the standard brain tumor protocol, whatever you have in your institute. We take this patient out, scan another one or two patients, and then we put the TRAM patients back in the machine for another 3D-T1. And then we simply subtract the two images in order to get our treatment response assessment maps or TRAMs. That's the color slide on the right.
And you can see in the TRAMS that there are two populations, blue and red, and I'll explain. So, if you look at the blood vessels in the brain immediately after contrast injection, the signal is maximal. And then it decays with time. And therefore, if we subtract one hour minus five minutes, we'll get a negative signal showing blue in the maps. For this reason, all the blood vessels in the maps looks blue. And if you look at that cyst in that treated GBM, you can see that gadolinium accumulates after one hour. So, if we subtract one hour minus five minutes, we'll get a positive signal showing red in the maps.
So, if you look at the GBM lesion, you will see that there actually is a mixed blue and red. And what we found out by comparing the presurgical TRAMs to histology is that blue regions in the TRAMs represent morphologically active tumor while red regions represent everything else, including treatment effects. So, we think the rationale stems from the morphology of the blood vessels in these regions.
So if you look at the blue regions, you see tumor vasculature. So you see vessels of different shapes and sizes. But what's typical of these vessels is that they are viable so they are able to clear the gadolinium one hour after contrast injection. On the other hand, if you look at the bottom, at the red regions, you'll see blood vessels at different stages of vessel necrosis. And therefore, gadolinium keeps accumulating.
So the whole idea of the TRAMs is that we wait at least one hour after contrast injection. And by this time, tumor will clear the gadolinium and treatment effects will accumulate gadolinium. So we don't really need to cut off. We have a negative signal for tumor. We have a positive signal for treatment effects, and this is what gives us really the high specificity and sensitivity of this methodology.
So when we compare the presurgical TRAMs to histology in the first 54 patients, we reached a specificity of 93% and 100% sensitivity to morphologically active tumor. And in the bottom part of this slide, you can see a comparison and a subgroup of patients resected en bloc, where we could assess the tumor burden and histology. So we compare the tumor burden and histology to the blue burden and the presurgical TRAMs, and you can see there this nice correlation.
So we'd like to give you a few examples of how we use it in the daily clinical scenario. In Israel, we have been using it since 2012, about 500 patients per year. Now, it's spread all to about 60 hospitals around the world.
So, if you look from left to right, this is an example of a GBM patient. We recruited her three weeks after these two protocols. She had the small lesions. She was only biopsied before, and we monitored her every two months. And you can see that four months after these two protocol, the enhancing lesion started to grow, peaking in volume six months after these two protocol, and then shrinking back. This is classical pseudoprogression.
But when you look at the TRAMs, four months after initiation of these two protocol, what you see is that the increase in the enhancement is mainly due to the red component, the treatment of that component, not the tumor component. And therefore, assuming pseudoprogression, the patient was indeed continued with the adjuvant temozolomide.
When you look at six months, you see that it's still mainly the red component that is growing. There's a little bit growth of the tumor component as well. But since the red component is growing faster, we still assume this is pseudoprogression. And as I said, it tracked nicely later, GBM is GBMs, progressed a couple of months later.
This is the opposite example of progression. A patient recruited one week after these two protocol, with this huge lesion growing into the previous surgery site. And you can see in the second row, the TRAMs showed about 40% blue, 60% red, so we were not sure which one of these components was rapidly growing and causing neurological deterioration. So we scanned the patient again two and a half weeks later. T1 FLAIR diffusion and perfusion, there was no change on the standard MRI. But when you look at the TRAMs in the bottom row, you can see that the tumor component, the blue component increased by nearly a factor of two. And therefore, the patient was resected the following day and the neurosurgeon took for as the piece en bloc, showing the typical morphology of this tumor. So you see a central red region surrounded by a thick blue rim, which is a nice correlation with this histological slide showing central necrosis surrounded by a thick rim of morphologically active tumor.
And as we always do at Sheba Medical Center four weeks post-surgery, looking at the top row, we scan the patient. And you can see that the enhancing regions surrounding the surgery site in the TRAMs, in the second row is mostly red, suggesting post-surgical changes over residual tumor. The patient was not started on adjuvant temozolomide because he was hospitalized due to meningitis. So we scanned him again three and a half months after his second surgery. And again, you can see that the enhancing region surrounding the surgery site is mostly red, so the patient was started on adjuvant temozolomide. And two months later, that's the bottom row, he comes feeling well for a standard followup. And this time, you can see a thick blue rim surrounding the surgery site that was not there before. The patient was determined to progress and switched to Avastin.
We do the same thing with brain metastasis. So on the left is a breast cancer brain metastasis. It was treated by SRS. Two months later, it nearly disappeared. And then at five months post SRS, it recurs with a question of tumor versus radiation necrosis. So when we scanned the patient at this point, the left part of the slide, you can see that there were a couple of blue pixels within this red region in the TRAMs. But we have to remember that blue in the TRAMs may also be blood vessels. So this can be angiogenesis of healing around the necrotic region. So we waited another couple of months. And by this time, the lesion has grown. You can see that on the right-hand side of the slide. And in the TRAMs, we see these blue chunks, suggesting there is tumor there. And therefore, the patient was resected. And as usual, we asked the neurosurgeon, "Please take her as a biopsy from the red region," showing in the middle, histology of radiation necrosis. And on the bottom, a biopsy from the blue region in the TRAMs, showing morphologically active tumor.
And we scanned again the patient three weeks after surgery. You can see on the left side that there is post-surgical changes in red but there's also a thin blue rim. But as we said, a thin blue rim can still be angiogenesis of wound healing, of the surgery, so we waited another couple of months. And when we scanned the patient at this time, two and a half months post-surgery, you can see a blue mass suggesting tumor recurrence in the surgery site. So the patient was re-radiated, and you can see three weeks post-surgery, a very nice response, blue turning into red. We monitored this patient for another two years and she was clean.
This was a patient that arrived to us on a Wednesday night with significant headaches. He had three brain metastasis of yolk sac carcinoma. He told us that he was already treated by whole brain RT and SRS, and that it was not helping so he was scheduled the following week to undergo another whole brain RT session. But when we scanned him, with the TRAMs, you can see that all three lesions are very red, suggesting radiation necrosis. And therefore, the largest lesion was taken out en block. And you can see again the nice correlation between this red lesion with a thin blue rim surrounding it, and the histology showing radiation necrosis with a thin rim of morphologically active tumor.
What's nice about the TRAMs is that we calculated from standard T1-weighted MRI so it's not very sensitive to susceptibility artifacts. And for this region, we can look for active tumor within hemorrhages. So these are examples of three patients with brain metastasis of melanoma that were bleeding. In the two left cases, you can see that there is no blue, there is no tumor on the TRAMs, so we just waited a few months for it to be observed. But in the right case, you can see within the bleeding, you can see a round blue sphere with a hook-shaped extension. This was taken out en bloc. And you can see a round tumor with a hook-shaped extension in histology.
We talk a little bit about residual tumor post-surgery. This was a patient with a violent non-small cell lung cancer metastasis. The very left column, she was scanned. They decided to resect her. For some reasons, they did not resect her for four weeks. And since the TRAMs are only valid for two weeks, they scanned her again just one day prior to surgery. And you can see a thick blue rim surrounding a red region. And that was a nice correlation to histology, a central necrosis surrounded by a thick rim of morphologically active tumor.
Four weeks post-surgery, that's the third column. You can see a very thin blue rim surrounding the surgery site. It's really hard to say if this is angiogenesis of wound healing or a tumor. There may be a little bit of mass in the midline that suggests a tumor. Anyhow, she was irradiated to the surgery site. And still, three months later, full-blown tumor. So she was resected again. So, if we look at the second column on the right, we see the post resection TRAMs. So you can see some residual tumor in the midline, which the neurosurgeon knew that he's gonna leave there. But you also can see on the right-hand side or the left-hand side of the surgery site, some residual tumor on the edge of the surgery bed. She was re-radiated and, again, the tumor recurred. She died a few weeks after the last MRI.
What's nice about the TRAMs also is that you can control the sensitivity to tumor. I told you that we scanned the TRAMs by scanning the first MRI after five minutes after contrast injection, and the second one somewhere between an hour and two. But if we change the timing of the first MRI, we are able to decrease or increase the sensitivity to tumor. So if you are looking for a very small new metastasis, or if you're looking for a small residual tumor post-surgery, you may scan for the TRAMs also at a later time. Let's say, not in 5 minutes, but 15 minutes post contrast injection. And then, you are artificially increasing the sensitivity to tumors so you will get a larger blue volume in the TRAMs. You can see that in these two examples where we moved the first time point and you can see more and more blue, i.e., more and more tumor. Of course, each tumor has its own characteristics in this sense. But we found that on average, the peak signal is at 25 minutes post contrast injection, and this is the region we need an hour to wait for the clearance of the tumor when doing the TRAMs.
The TRAMs are, as I said, calculated from 3D-T1. It's naturally registered to your 3D-T1 so we usually recommend to use your navigation sequence, this way if you want to resect the patient later or treat him, you already have the TRAMs registered to the T1. You can upload them to your neuro-navigation sequence or your treatment planning and you can target biopsies at blue regions or you can boost radiation to residual tumor post-treatment, etc.
And the last thing I wanted to note is a paper from 2017 from a German group, which has actually done a very similar study but only on brain metastasis. They used the delay of 55 minutes, not an hour as we do, but they got the same result. They showed that tumor clears gadolinium after 55 minutes and radiation necrosis accumulates gadolinium after this time.
So I'd like to summarize by saying that the TRAMs or the CCA product enables efficient separation between tumor and treatment effect. We use our vast 3D-T1 MRIs so we don't have a problem with susceptibility artifacts. It's model-independent. We don't have to make any assumptions. We also don't have to put a cutoff. We don't have to normalize to contralateral ROIs as you may have to do with other methods. It's easy to interpret. Blue is tumor, red is treatment effects, and it may be applied for studying response pattern to novel and advanced type of treatments. Hopefully, we would like to collaborate with anyone who's interested on immunotherapy, convection-enhanced right to the very lit, etc.
Thank you very much.
Man: I think we have time for some questions.
Man: That was... Yeah.
Yael: Okay. Just... Where will I go? Can you help me with advancing the slides? Go back? Okay. I'll go back. It's working opposite maybe because it's Israel going from right to left instead of left to right.
So we licensed this technology to Brainlab. It's called CCA, Contrast Clearance Analysis. And the way we differentiate tumor from treatment effects, we scan the patient with standard 3D-T1 five minutes after contrast injection. We do the standard brain tumor protocol, whatever you have in your institute. We take this patient out, scan another one or two patients, and then we put the TRAM patients back in the machine for another 3D-T1. And then we simply subtract the two images in order to get our treatment response assessment maps or TRAMs. That's the color slide on the right.
And you can see in the TRAMS that there are two populations, blue and red, and I'll explain. So, if you look at the blood vessels in the brain immediately after contrast injection, the signal is maximal. And then it decays with time. And therefore, if we subtract one hour minus five minutes, we'll get a negative signal showing blue in the maps. For this reason, all the blood vessels in the maps looks blue. And if you look at that cyst in that treated GBM, you can see that gadolinium accumulates after one hour. So, if we subtract one hour minus five minutes, we'll get a positive signal showing red in the maps.
So, if you look at the GBM lesion, you will see that there actually is a mixed blue and red. And what we found out by comparing the presurgical TRAMs to histology is that blue regions in the TRAMs represent morphologically active tumor while red regions represent everything else, including treatment effects. So, we think the rationale stems from the morphology of the blood vessels in these regions.
So if you look at the blue regions, you see tumor vasculature. So you see vessels of different shapes and sizes. But what's typical of these vessels is that they are viable so they are able to clear the gadolinium one hour after contrast injection. On the other hand, if you look at the bottom, at the red regions, you'll see blood vessels at different stages of vessel necrosis. And therefore, gadolinium keeps accumulating.
So the whole idea of the TRAMs is that we wait at least one hour after contrast injection. And by this time, tumor will clear the gadolinium and treatment effects will accumulate gadolinium. So we don't really need to cut off. We have a negative signal for tumor. We have a positive signal for treatment effects, and this is what gives us really the high specificity and sensitivity of this methodology.
So when we compare the presurgical TRAMs to histology in the first 54 patients, we reached a specificity of 93% and 100% sensitivity to morphologically active tumor. And in the bottom part of this slide, you can see a comparison and a subgroup of patients resected en bloc, where we could assess the tumor burden and histology. So we compare the tumor burden and histology to the blue burden and the presurgical TRAMs, and you can see there this nice correlation.
So we'd like to give you a few examples of how we use it in the daily clinical scenario. In Israel, we have been using it since 2012, about 500 patients per year. Now, it's spread all to about 60 hospitals around the world.
So, if you look from left to right, this is an example of a GBM patient. We recruited her three weeks after these two protocols. She had the small lesions. She was only biopsied before, and we monitored her every two months. And you can see that four months after these two protocol, the enhancing lesion started to grow, peaking in volume six months after these two protocol, and then shrinking back. This is classical pseudoprogression.
But when you look at the TRAMs, four months after initiation of these two protocol, what you see is that the increase in the enhancement is mainly due to the red component, the treatment of that component, not the tumor component. And therefore, assuming pseudoprogression, the patient was indeed continued with the adjuvant temozolomide.
When you look at six months, you see that it's still mainly the red component that is growing. There's a little bit growth of the tumor component as well. But since the red component is growing faster, we still assume this is pseudoprogression. And as I said, it tracked nicely later, GBM is GBMs, progressed a couple of months later.
This is the opposite example of progression. A patient recruited one week after these two protocol, with this huge lesion growing into the previous surgery site. And you can see in the second row, the TRAMs showed about 40% blue, 60% red, so we were not sure which one of these components was rapidly growing and causing neurological deterioration. So we scanned the patient again two and a half weeks later. T1 FLAIR diffusion and perfusion, there was no change on the standard MRI. But when you look at the TRAMs in the bottom row, you can see that the tumor component, the blue component increased by nearly a factor of two. And therefore, the patient was resected the following day and the neurosurgeon took for as the piece en bloc, showing the typical morphology of this tumor. So you see a central red region surrounded by a thick blue rim, which is a nice correlation with this histological slide showing central necrosis surrounded by a thick rim of morphologically active tumor.
And as we always do at Sheba Medical Center four weeks post-surgery, looking at the top row, we scan the patient. And you can see that the enhancing regions surrounding the surgery site in the TRAMs, in the second row is mostly red, suggesting post-surgical changes over residual tumor. The patient was not started on adjuvant temozolomide because he was hospitalized due to meningitis. So we scanned him again three and a half months after his second surgery. And again, you can see that the enhancing region surrounding the surgery site is mostly red, so the patient was started on adjuvant temozolomide. And two months later, that's the bottom row, he comes feeling well for a standard followup. And this time, you can see a thick blue rim surrounding the surgery site that was not there before. The patient was determined to progress and switched to Avastin.
We do the same thing with brain metastasis. So on the left is a breast cancer brain metastasis. It was treated by SRS. Two months later, it nearly disappeared. And then at five months post SRS, it recurs with a question of tumor versus radiation necrosis. So when we scanned the patient at this point, the left part of the slide, you can see that there were a couple of blue pixels within this red region in the TRAMs. But we have to remember that blue in the TRAMs may also be blood vessels. So this can be angiogenesis of healing around the necrotic region. So we waited another couple of months. And by this time, the lesion has grown. You can see that on the right-hand side of the slide. And in the TRAMs, we see these blue chunks, suggesting there is tumor there. And therefore, the patient was resected. And as usual, we asked the neurosurgeon, "Please take her as a biopsy from the red region," showing in the middle, histology of radiation necrosis. And on the bottom, a biopsy from the blue region in the TRAMs, showing morphologically active tumor.
And we scanned again the patient three weeks after surgery. You can see on the left side that there is post-surgical changes in red but there's also a thin blue rim. But as we said, a thin blue rim can still be angiogenesis of wound healing, of the surgery, so we waited another couple of months. And when we scanned the patient at this time, two and a half months post-surgery, you can see a blue mass suggesting tumor recurrence in the surgery site. So the patient was re-radiated, and you can see three weeks post-surgery, a very nice response, blue turning into red. We monitored this patient for another two years and she was clean.
This was a patient that arrived to us on a Wednesday night with significant headaches. He had three brain metastasis of yolk sac carcinoma. He told us that he was already treated by whole brain RT and SRS, and that it was not helping so he was scheduled the following week to undergo another whole brain RT session. But when we scanned him, with the TRAMs, you can see that all three lesions are very red, suggesting radiation necrosis. And therefore, the largest lesion was taken out en block. And you can see again the nice correlation between this red lesion with a thin blue rim surrounding it, and the histology showing radiation necrosis with a thin rim of morphologically active tumor.
What's nice about the TRAMs is that we calculated from standard T1-weighted MRI so it's not very sensitive to susceptibility artifacts. And for this region, we can look for active tumor within hemorrhages. So these are examples of three patients with brain metastasis of melanoma that were bleeding. In the two left cases, you can see that there is no blue, there is no tumor on the TRAMs, so we just waited a few months for it to be observed. But in the right case, you can see within the bleeding, you can see a round blue sphere with a hook-shaped extension. This was taken out en bloc. And you can see a round tumor with a hook-shaped extension in histology.
We talk a little bit about residual tumor post-surgery. This was a patient with a violent non-small cell lung cancer metastasis. The very left column, she was scanned. They decided to resect her. For some reasons, they did not resect her for four weeks. And since the TRAMs are only valid for two weeks, they scanned her again just one day prior to surgery. And you can see a thick blue rim surrounding a red region. And that was a nice correlation to histology, a central necrosis surrounded by a thick rim of morphologically active tumor.
Four weeks post-surgery, that's the third column. You can see a very thin blue rim surrounding the surgery site. It's really hard to say if this is angiogenesis of wound healing or a tumor. There may be a little bit of mass in the midline that suggests a tumor. Anyhow, she was irradiated to the surgery site. And still, three months later, full-blown tumor. So she was resected again. So, if we look at the second column on the right, we see the post resection TRAMs. So you can see some residual tumor in the midline, which the neurosurgeon knew that he's gonna leave there. But you also can see on the right-hand side or the left-hand side of the surgery site, some residual tumor on the edge of the surgery bed. She was re-radiated and, again, the tumor recurred. She died a few weeks after the last MRI.
What's nice about the TRAMs also is that you can control the sensitivity to tumor. I told you that we scanned the TRAMs by scanning the first MRI after five minutes after contrast injection, and the second one somewhere between an hour and two. But if we change the timing of the first MRI, we are able to decrease or increase the sensitivity to tumor. So if you are looking for a very small new metastasis, or if you're looking for a small residual tumor post-surgery, you may scan for the TRAMs also at a later time. Let's say, not in 5 minutes, but 15 minutes post contrast injection. And then, you are artificially increasing the sensitivity to tumors so you will get a larger blue volume in the TRAMs. You can see that in these two examples where we moved the first time point and you can see more and more blue, i.e., more and more tumor. Of course, each tumor has its own characteristics in this sense. But we found that on average, the peak signal is at 25 minutes post contrast injection, and this is the region we need an hour to wait for the clearance of the tumor when doing the TRAMs.
The TRAMs are, as I said, calculated from 3D-T1. It's naturally registered to your 3D-T1 so we usually recommend to use your navigation sequence, this way if you want to resect the patient later or treat him, you already have the TRAMs registered to the T1. You can upload them to your neuro-navigation sequence or your treatment planning and you can target biopsies at blue regions or you can boost radiation to residual tumor post-treatment, etc.
And the last thing I wanted to note is a paper from 2017 from a German group, which has actually done a very similar study but only on brain metastasis. They used the delay of 55 minutes, not an hour as we do, but they got the same result. They showed that tumor clears gadolinium after 55 minutes and radiation necrosis accumulates gadolinium after this time.
So I'd like to summarize by saying that the TRAMs or the CCA product enables efficient separation between tumor and treatment effect. We use our vast 3D-T1 MRIs so we don't have a problem with susceptibility artifacts. It's model-independent. We don't have to make any assumptions. We also don't have to put a cutoff. We don't have to normalize to contralateral ROIs as you may have to do with other methods. It's easy to interpret. Blue is tumor, red is treatment effects, and it may be applied for studying response pattern to novel and advanced type of treatments. Hopefully, we would like to collaborate with anyone who's interested on immunotherapy, convection-enhanced right to the very lit, etc.
Thank you very much.
Man: I think we have time for some questions.