Transcript
[music]
[00:00:29]
Bogdan: Hello, everyone, and welcome to a new Novalis Circle Symposium. My name is Bogdan Volcu, I am the director of Novalis Circle and today I have the pleasure to host this online event in collaboration with the annual online ASTRO meeting. For our presentations today, I'm honored to introduce you to two groups, one from Ludwig-Maximilians University in Munich and one from University Hospitals in Brussels that have put together an agenda for talks highlighting clinical and technical considerations for spine and breast treatments. To begin, we'll have Dr. Maximilian Niyazi from LMU Munich discuss disease considerations for spine radiosurgery. Dr. Niyazi is the vice chair of the Department of Radiation Oncology and the deputy coordinator for the Neuro-Oncology Comprehensive Cancer Center.
He will be followed by Dr. Phillip Freislederer who will discuss the first clinical applications of ExacTrac Dynamic. Dr. Freislederer will also discuss general commissioning and implementation considerations for ExacTrac Dynamic. We will follow his talk with Professor Mark de Ridder from University Hospital in Brussels, who will discuss general considerations for breast radiotherapy. Professor de Ridder is the head of the Radiotherapy Department, the vice chairman in the Board of Directors, and also the vice president of the Medical Council. His talk will be followed by that of Professor Thierry Gevaert who will discuss both SGRT and IGRT considerations for breast setups. Dr. Gevaert is a professor and the coordinator of the Medical Physics Group in the Department of Radiation Oncology at UC Brussels.
And as it has become standard by now for online events, please remember to utilize either Google Chrome or Safari, and should you have any internet connection issues, simply refresh the webinar page. Use the chat interface to send us questions, we will answer your questions upon completion of the four lectures. Monitor the polling interface for questions that we might like to ask you, and should you choose to follow us on social media, please use the #NovalisCircle. With this, I'd like to turn it over to Professor Niyazi for the first lecture.
Prof. Niyazi: Thank you so much, Bogdan, for the introduction. Ladies and gentlemen, today I would like to talk about radiosurgical planning and spine disease considerations with ExacTrac Dynamic. So, these are all global disclosures. So, first of all, I will talk about the capabilities of the ExacTrac Dynamic system. And as you probably will note, it has an optical surface scanner with structured light and it has a built-in thermal camera that gives you an additional registration information such as the fourth dimension, if you would like to say, and you can avoid multiple in room cameras that are unnecessary, in this case.
You have the well-known X-ray positioning and the intrafractional monitoring, and the good thing is that you have this additional real-time tracking possibility using the SGRT surface-guided radiotherapy element. That's a nice screenshot of my colleagues actually here. The difference, you have the 2D information, or 3d if you would like to say, and thermal information that's added and gives you another layer of depth about the situation. So, the key benefits of the ExacTrac Dynamic are that you have a super metric accuracy in patients, you can track both very large and very small areas. The respiratory and motion tracking is not too sensitive to room lighting or reflections or the skin tone. There's no imaging or imagery-related blocking of one of the cameras, you have a unique matching algorithm, and you have the accuracy of tracking that's not affected by baseline drifts such as, for example, a change in room temperature.
Our experience so far is that the first patient was treated on June 2nd of this year. So far, we have treated 30 patients with 26 brain cases, 2 spine, and 2 head and neck cases. And the pre-positioning is actually done in this way that you have the automated plan loading from your R&V system, in our case, from Mosaiq. The system itself is fully integrated with the LINAC and the pre-positioning is done using the outer contour from your planning CT and the shifts are automatically sent to the LINAC. The indications from the scratch were multiple brain metastases in our STEREOBRAIN trial and, of course, every kind of hypofractionated SRS in the brain. With fractionated stereotactic radiotherapy with open face masks, for example, meningioma, vestibular schwannoma, or glioma.
And we could easily replace the cone-beam workflow to speed up the treatments in, for example, highly palliative cases, whole-brain irradiation, and other things. And we're now launching our spine SRS program with a prospective registry trial, and of course, you can treat any bone oligometastatic disease anywhere else in the body. Now, I'll tell you something about the conventional radiotherapy with the previous gadgets that we used like, as you all know, 3 gray times 10, 4 grey times 5, or even 8 grey times 1. And the target volume in this situation was traditionally the whole vertebral body including the cord, mostly with one vertebral body above and below, was still mostly the 3D conformal radiotherapy, and it was somehow effective with 60% to 70% of pain response but it was of short duration, and it was, of course, dependent on histology.
So, that's something that still has to be considered, for example, in patients with limited life expectancy but, of course, you will lose local control in the long run. As you can see here in this Shizuoka Cancer Center trial, the local control rate really drops if you have a mass lesion and the bulky tumor patients have been significantly worse one and two-year local control, though it will be relevant for these patients to optimize local control. And one option to optimize local control is SBRT, obviously, because it offers excellent local control and palliation as well. This is a nice analysis from Tseng et al., as you can see here, most of these prospective and retrospective series with a lot of patients altogether has shown control rates around 80% to over 90% mostly even at one year later endpoints and the pain response was mainly impressive.
But what do you have to consider and what's getting more important compared to conventional radiotherapy is that the cord tolerance is important. And in SBRT, it's of course, in the far lower range than you would know and expect from conventional fractionation. And if you add together all these QUANTEC, Katsoulakis-Gibbs, or Sahgal models, then you will end up with something...above 15 gray, it will be a steep increase of myelopathy, and above 12 gray, it will be at least critical. So, you have to consider that if you switch your workflows to spine SBRT. Another risk is the vertebral compression fraction that's obviously dependent on the dose that's delivered to the vertebral body.
That's an example of a fracture that is progressing from initially a lower height deviation to here seven millimeters, that's pre and post-SBRT shown in Cunha, 2013. And this risk is obviously related to several factors and one of the factors is the SINS score that has actually been developed as a classification system for spinal instability. So, this is used by traumatic or neurosurgeons to derive whether the vertebral body addressed to be unstable or not and what goes into that index is the location, whether there's the pain relief with the recumbency and/or pain with movement, bone lesion, radiographic spinal alignment, vetebral body collapse, for example, that is pre-existing or not, and if the posterolateral involvement of the spine elements is given. And according to some of the scores you have a stable situation with 0 to 6 points, or a potentially unstable situation where you should consult your neurosurgeon, and with more than 12 points, it's an unstable situation that's critical.
And as I pointed out, it was initially developed by neurosurgeons but it's also predictive for the vertebral compression fracture and these are potentially problems and here you have to, obviously, contact the neurosurgeon. And concerning the compression rate, it's as well as dependent on the dose. That's a nice work of Arjun Sahgal, where you can see that if you go beyond 20 and beyond 24 gray, then the cumulative incidence of fractures is vastly higher, so you have to consider that. And another factor is the BILSKY epidural disease grade and this goes from 0 like the absence of epidural disease up to 3 where you have a spinal cord compression without visible CSF, and with just the abutment of the thecal sac or the impingement, there are different steps in between.
And critical ones are here with the spinal cord compression, where you have at least to consider whether you need a separation surgery before you have the spine Bilsky treatment and that's substantiated by this nice work from Al-Omair about surgical resection of epidural disease that may improve local control. And one of the reasons, of course, is here, that you have the cord tolerance and if you have like a bit of space to the cord, then you can escalate the dose. And he has shown that patients with the Bilsky score of 2 or 3 benefitted from this kind of separation surgery as they go down with the post-op Bilsky score 2 and 3 in just 10%, dropping from 60%, and here you can see the local control probability for these 48 patients that are presented with a higher score.
And if you can achieve something around 1 or 0 compared to 2, then you have obviously a better local control, so it really makes sense to offer patients this kind of surgery. So, what about the automated target definition? And there are international consortium guidelines, for example, the International Spine Radiosurgery Consensus guideline. And the important aspect about that is that the vertebral bodies are different comparing to the height, of course, and the consensus states that the intention should be to include, of course, the complete extent for the GTV and there's given a consensus for the CTV, where you have to include different parts of the vertebral body according to the extent of the gross lesion within the vertebral body.
So, you can skip, for example, some parts of the posterior elements if, for example, the extension is limited to the vertebral body itself. And here the local anatomy of vertebral bodies again, and according to the location of the tumor, for example, if it's just within the vertebral body, you can skip it to treat, in this case, the pedicle and other regions. So, this is very important, that this consensus guideline leaves you some space to, for example, skip some of the posterior elements, and then it's even easier to achieve the gradient visually to the spinal cord and it actually makes sense compared to the older strategy to have one vertebral body above or below that's really not needed.
These are some of the common fractionation schedules that can be delivered in our SPINES protocol and it's just a collection of different regimens. And our primary aim was to reduce the risk of vertebral compression fracture, but to have a certain range from one to five fractions according to histology and to different anatomic locations and to the size of the tumor. And it goes from like 16, in this case, gray to the elective ISRS volume, up to 24 gray to the SIB boost, so the similar things, integrated boost. And what we're doing here is that we take the gross lesion, we expanded by three millimeters to have the boost volume, and we have the elective volume according to ISRS and add on two to three millimeters, exclude the spinal canal, and have the thecal sac as our POV, it's the cord plus 1 or 1.5 millimeters of the cauda equina.
And these are the organs at risk that are considered. As you can see here, if you look at the different single fraction doses and the cumulative dose then the biological effective dose, for example, if you look at normal tissue with an alpha-beta of 2 gray, that's really vastly different for these different fractionation schedules and you are really several factors above the conventional fractionation paradigm. So, that's very important to know and to consider if you perform these treatments and that you think of, for example, to reduce the vertebral compression fracture with this integrated boost approach.
The inclusion criteria of our SPINES trial are relatively comparable to what's given, for example, by a nice meta-analysis by Sahgal or the ASTRO guidelines. So, we're trying to include less than or equal to three effective vertebral bodies that are untreated or in case of the progression after conventional radiotherapy, but it should at least be a time interval three months or better, five months. So, of course, oligometastatic disease with less or equal to five metastasis, a stable spine with the SINS score of 0 to 6, and the low epidural involvement, the extraspinal tumor situation should be under control, and the life expectancy should be relevantly high. The exclusion is mainly the opposite of that. So, further points that we exclude, radiosensitive histologies where you have good results with conventional fractionation, and, of course, significant spinal cord compression or progressive neurological symptoms where you should consult your neurosurgeon, such as, for example, BILSKY score of more or equal to 2.
These are the normal tissue constraints that should be considered in spine SRS. These are given by, for example, the ASTRO guidelines, but of course, several other guidelines are available as well. And you have here different, for example, Fractionation schedules for the spinal cord, for example, OF 12.2 to 25.3 gray. The max dose is considered, the cord here in the case would or should be, in my eyes, the PRV with an additional margin to represent thecal sac. We have the esophagus here with different doses up to five fractions, and some other relevant at-risk structures right here given by different trials and I think these are very important to be considered to keep the risk as low as reasonably achievable.
And one extra intelligence layer that can ensure the right deformation of the curvature of your MRI that you need to define the cord is given, for example, by the Brainlab software and that's a very nice tool to really match your CT registration. And here you have the planning panel, it can give you your prescription and your constraints, and then this plan is optimized and structures are automatically generated, for example, elective volumes according to the ISRS recommendations. So, I'd like to end up with a clinical example here, it's uterine leiomyosarcoma with metastasis, initially presented in 2015 with an area of the piriforme muscle, just got resected and adjuvant chemotherapy, had a sacral metastasis in 2016, again, resection and adjuvant chemo.
and in 2019 to '20, destructive metastasis of T9 and the humerus, got resected and the palliative radiotherapy with 3 gray times 12. And seven months later, we had a progressive lesion in T9 and applied a spine SBRT with 5 times 4 gray to the elective volume and the boost to 30 gray to this recurrence before resectioning in the sense of the separation surgery. And what did we do right here? We had high margins for the patient surface as we had breathing motion that was detected with very tight margins for X-ray surveillance with 0.7 millimeters and 0.5 degree, and the highest possible frame rate for X-ray imaging with only stereoscopic views in 4 per arc. And the good thing was that the patient was just lying on the table, arms up, no body fixation, no vacuum matters, relatively easy setup, and the good thing is you have the ExacTrac Dynamic keeping track that everything is all right.
And the last thing that I would like to show is this really nice trial from Arjun Sahgal, the SC.24 trial from the Canadian group that was shown in the last ASTRO Congress and you have really an impressive result of 33% complete response compared to 16% with the conventional fractionation randomized Phase II trial. Impressive results, a higher level of quality assurance, and the time point was six months after therapy. So, really after a longer duration, relevant improvement in pain responses, and complete response rate. That's something that's relevantly improved to the RTOG 0631 that use rather lower doses and had three months time interval, and that's probably the reason why these two fractions, so 12 gray times 2 regimen was actually improved compared to the conventional fractionation. So, I think that's a really nice outlook for spine SBRT. I would like to thank you all for your attention, I would like to hand it over to my dear colleague, Phillip Freislederer. Thank you so much.
Bogdan: Thank you for the review of your spine program, Professor Niyzi, and let's go to Phillip Freislederer for a recap of the LMU experience with ExacTrac Dynamic for spine treatments.
Dr. Freislederer: Thank you for the introduction. My name is Philip Freislederer from LMU University Hospital in Munich, and I will be giving some insights from the physical details for spine disease considerations with the ExacTrac Dynamic. So, first of all, let me guide you through the workflow a little bit. At first, the software starts...or the workflow starts with a patient pre-positioning based on the surface only with no thermal information. So, what simply happens is the software calculates the deviation from the planned position from the planning CT and the current live position from the surface camera. As you can see here, this is what the software would look like, the outer contour from the CT, so the treatment position, and the calculated current outer contour from the live position are shown here in this image.
This is what it would look like for real patient, you shift the patient close to the center manually. Afterward, the shift is calculated with the exit track and the shift is then sent to the treatment couch and the patient is automatically shifted to the desired position. What you see here is some residual motion from the patient breathing. After the pre-positioning of the patient, area of interest is selected. In this case for a spine patient, we have a lot of breathing motion, so we want to have an area of interest which is not affected by the breathing motion too much but rather, as you see here, position around the stomach or the abdomen of the patient where you can make sure that you count for any interaction of motion apart from breathing motion.
Afterward, you acquire X-ray images, in this case, you see some red areas where the image is...where we don't want anything, any information to be taken into account from, and the images are automatically registered in 6D. There are some tools for the evaluation of the registration result. What you saw before was the rubber band, right here you have some blending and you can have a really nice view of if the spine is positioned correctly, you can have a look at the ribs, you can have a look at each single vertebra with different methods. What you see here is a spyglass where you can easily, based on the choice that you or your doctor or your therapist would like, can position the patient or control and evaluate the result of the evaluation of the registration result.
Simultaneously, when you capture an X-ray, the thermal surface also captured and stored in the background. So, after you approve this fusion and this shift to send to your six-dimensional table of the treatment couch, also your area of interest is afterward mathematically transformed according to the position of the X-rays. If you look at the camera, here is an image of the camera in the room. There's a stereoscopic 3D data camera or surface camera with a structured light projector and additionally integrated, a thermal camera. Why would you use this thermal camera? What is the reason for it? First of all, when you look at surface data from a rather flat surface like this or like an abdominal treatment or the treatment the patient you saw on the treatment before, it is very tough for any registration algorithm to get a result, a registration result, especially in the longitudinal direction.
The surface not only a surface, it is also warm or cold in this case, and if you add the thermal data, this is what it would look like for the registration algorithm. This is something you don't see but it is something the algorithm sees, so there are a lot of hills and valleys which makes it easy for the algorithm to grab on. And this is the primary use of the thermal camera, to get more registration or a fourth dimension for the image registration, and this makes it unnecessary to have more than one optical surface camera inside your treatment room. Afterward, we will go into monitoring steps. On the right top side, you see the patient breathing very nicely and you can see the same breathing motion on the bottom, and this all happens while the patient is irradiated.
You monitor the surface all-time in this case as there are still a lot of breathing motion left. We have some very high margins for the surface but very, very tight margins for the X-rays. So, once an X-ray is taken, and in this case, you see there in the lateral direction a shift of 1.2 millimeters calculated, afterward, you can review this result, again, with your multiple tools and sense the shift during the treatment to your six-dimensional table. And after you send the shift, the treatment would go on with surface monitoring where you cannot take stereoscopic X-ray images. If you take a...there's also the possibility of taking monoscopic X-ray images, this is just may be seen as another control mechanism of the software, but if you take one, you do not update your reference surface. Only if you acquire stereoscopic images and these are within tolerances, the area of interests of your surface and thermal of the patient's surface are updated and then the region of interest is transformed according to the position of your X-rays.
So, what happens if the thermal surface or a monoscopic X-ray out of tolerance? You can have your settings for different settings. So, you can say if your surface is out of tolerance, the beam is automatically held, the patient would be brought back to a position where you can acquire stereoscopic X-rays. So, with a plus-minus 10 degrees, around 090, 270, or 180 degrees, you can acquire dual X-rays or stereoscopic X-rays, and you reposition the patient according to the values of what the stereoscopic six-dimensional registration result tells you. The fusion always needs to be verified afterward by the user or it can be in particular cases where it needs to be verified by the users with signature, but you have to verify it yourself all the time with all the methods I showed you before in the patient video.
After you send the shift and then approve the fusion, therefore, these shifts are again stored in the thermal surface and the area of interest is again transformed according to these values. All those settings of what you can do, you can edit in a template editor and for each new patient, you can import with a certain template depending on what you want to do. We have certain templates which we mostly use, cranial SRS, cranial and spine, and spine SRS templates. There are certain methods of choosing what to do, if you want to do a...if you want to use your six-dimensional table, if you only want to use your base table, if you want to hold the beam all the time when you're monitoring. So, this is the number one purpose here is holding the beam to detect intrafractional motions, so this is something that should be turned on all the time.
Then you can choose if you want to verify your X-rays after your initial correction and if you want to verify for each new table angle for cranial treatments, for example, and one of the important steps is the decision on your patient tolerances. As I said before, for spine patients, we have some breathing or some breathing motion involved, so we have quite high surface tracking tolerances of five millimeters because shifts like that happen all the time when the patient is breathing. But you don't want to have more shifts happening, for example, the patient standing up or moving in a lateral way or coughing, this will be detected even with a five-millimeter surface tolerance.
For the extra tolerances, we have quite low tolerances, so 0.5 millimeters and 0.7...or 0.7 millimeters and 0.5 degrees for the X-ray positioning tolerances. Additionally, that's another thing you can select in your template or even for each patient individually. If you want to say if you want to trigger X-rays when the surface exceeds the tolerance automatically, if the beam is held if the tolerance is exceeded, and also if you want to...for a 3D conformal plan, for a static beam in this case, if you want to trigger based on a certain amount of treated monitor units, or for VMAT treatment and arc-based treatment, if you want to have only stereoscopic but also monoscopic X-rays.
Now to go a little bit into the QA or the routine QA and the commissioning of the ExacTrac Dynamic. First, let me start off with routine QA, the daily check we perform takes about five minutes in total, most of the times a little bit less. Two different things are checked here. In this case, one is the deviation between the surface camera and your X-ray positioning systems, so your flat panels, in this case. This is a consistency check, so if someone accidentally touched or hit the flat panels or the surface camera, you would receive an error, in this case, daily. Also, in the second step, the deviation from your radiation isocenter is checked and confirmed every day.
Secondly, once a month at the moment, you have to do a calibration of your thermal surface to your three-dimensional surface or your optical surface. This is done with a distinct Brainlab phantom and also takes roughly five minutes. For our stereotactic treatments, if you do a lot of stereotactic treatments or SRS, so cranial or spine, we would recommend a monthly calibration of your radiation isocenter and there is a...you simply place any type of ball bearing inside the radiation isocenter. This could be the ball bearing inside an anthropomorphic phantom or this could also be any type of Winston-Lutz pointer.
When we started with the ExacTrac Dynamic, we switched very quickly from cone Beam CT workflows to ExacTrac-only workflows. And by a very quickly, I mean the first patient we treated was an ExacTrac-only patient, a cranial patient. So, we want to verify the positioning and the verification accuracy of the two systems and just to show you some short examples of this, we use two anthropomorphic phantoms, one from the head and one from the pelvis, we had six random isocenter locations, and validated the setup accuracy from both systems and the verification again. So, the deviation in the initial correction was maybe a little bit higher for the head when it comes to rotation, but for the verification between those systems, so cone beam CT and an ExacTrac image, after you shifted your patient to your planned isocenter was quite low in this case.
And obviously, you don't really know what your ground truth is. Is your cone beam MCT better calibrated to the radiation isocenter or is it your ExacTrac Dynamic? Again, for some short points on QA and commissioning, there are quite a lot of initial commissioning thoughts if you want to install a new surface-guided system and if you have an X-ray system. Additionally, it's still even a little bit more. A couple of things will be the static accuracy, the dynamic positioning accuracy of the surface, in this case, it's possible to test but it's quite hard to store the data. The impact of the region of interest is something very exciting but it's more of a research topic for the future for the surface-guided radiation therapy community.
Some tests on the field of view, the field of view here not only for the camera but also for the X-ray systems, so to test if there was any blocking of the gantry or accessory parts. If you want to get more information about this, you can find a webinar on the Novalis Circle homepage from September 2020 on the initial experiences with ExacTrac Dynamic. Also for the commissioning, especially for spine or dedicated for spine, we have used a...we started to use a new phantom from RTSafe because we could not find any other phantom, in this case, because it's still some type of prototype. In this case, we have less reflectance skin, the reflecting skin from phantoms is a major issue for surface guided systems, so we want something which is as close as possible to the actual patient skin.
You have a patient structure inside, some hip bones and spine, and we filled the phantom with water, this is possible and nothing gets broken inside, and with roughly 40 degrees so we have a warm surface, and we have three ball bearings inside for the radiation isocenter coincidence tests between for the ExacTrac Dynamic. We have looked at the thermal drift of the phantom, the phantom doesn't drift too much, so we have about...after filling the water with 40 degrees...or filling the phantom with 40 degrees hot water, we have roughly a drift of 4 degrees in 60 minutes, so this would still be enough to perform all your commissioning or your monthly or quarterly tests.
You can see this here on the right side, there's the phantom not filled with warm water and we shifted the phantom a little bit and you can see some wobbling in the software, so it's not too easy when the phantom is cold but it's still possible because the thermal camera is still active. And on the left side, you can see the inside of the phantoms, you can detect your ball bearings or the spinal structures very well. If you fill up the phantom with water, we again can monitor the phantom as close as it would be in an actual patient. So, we have...although it seems like a lot of drift, this is 0.1 or 0.2 millimeters in drift calculated from the software. So, this is something we will use for the commissioning in the future and for additional verification and validation tests. Especially if you have three ball bearings inside, we can easily do an isocenter coincidence test between the two systems and get some nice results from it.
To conclude a little bit, there are certain proven, certain potential benefits of the ExacTrac Dynamic at the moment. The first one would be the automatic or the manual X-ray triggering, so on the real system integration, if the deviations between the planned and actual surfaces are detected, you have a direct repositioning when needed. Because of the nice system integration with the LINAC itself, you are very efficient in time. If you want to have a look at it...this is not really positioning from the surface but this is a repositioning from the X-rays, so you quickly evaluate...this is not fast-forwarded, by the way, you quickly evaluate your patient position according to the x-rays and send these shifts to the LINAC.
And once it is sent...one second. Yes, once it is sent to the LINAC, you can continue with your radiation treatment. So, this is quite fast and saves you a lot of time while still...and this is the most important part about it, you really do an intrafractional motion monitoring, not only with the surface, but also with the X-rays. And this is what I've been saying before, again, you have surface motion monitoring throughout the entire fraction. This is an additional patient safety feature, so you all the time have an automatic beam hold when your patient is moving out of tolerance. You have monitoring, you have a pinhole, and this is a safety feature which might come to...which might be part of any type of LINAC in the future, I believe so at least.
Additionally, from the automated workflow, you have to take less cone beam CT, if it is possible if you don't need any type of...if you're not required on any type of soft tissue contrast too much, if you can have a good match on the bony anatomy, so you have less dose to the patient and a faster workflow, much faster workflow. So, you can reduce the amount of cone beam CT when positioning with only anatomy. Additionally, and this is something that's potential or new but this is something which has been proven for every ExacTrac system not only Dynamic, you have X-ray repositioning for each couch angle and you don't do motion monitoring only for couch kicks, you do it with your X-rays.
So, with motion monitoring only, a surface motion monitoring I mean by that, this would mean that you have to do...if you detect some kind of motion at a certain couch angle, what do you do afterward? You have to move back to a couch of zero degree, acquire another cone beam CT, and then reposition the patient and shift the couch back. As intrafractional motion occurs quite a lot for patients, even if you talk about cranial treatments inside the mask, inside every mask, this is something which is not really practical. In addition, you have the tattoo-less or mark-less, tattoo-less workflows, which are introduced more and more in the clinics worldwide, even also with cone beam CT workflows. So, you have a pre-positioning using the patient surface and the possible reduction of a verification cone beam CT if your clinic does that. So, this would be, again, the video from before but this will be a complete tattoo-less workflow if you think about integrating this in your clinic.
Our outlook is in the future, we want to replace the cone beam CT workflows in palliative settings to make it a little faster. Also for the patient benefit, if the soft tissue contrast matters, we want to reduce the frequency of the cone beam CT and implement the ExacTrac Dynamic regularly. In the next step, we want to do deep inspiration breath-hold treatments for left-sided breast cancer and in the future, stuff like liver or lung SBRT would be the fun stuff to look at in the future in this case. So, I would like to thank you for your attention and I will be very happy to get questions from you.
Bogdan: Great review, Dr. Freislederer, and we'll switch now to Professor De Ridder for a review of breast radiotherapy planning considerations.
Prof. De Ridder: Good afternoon, dear colleagues. I, first of all, would like to thank Brainlab for the kind invitation. I will present our work, "Breast Radiotherapy: Towards a Patient- Tailored Approach." What are the benefits of radiotherapy for patients with breast cancer? Well, whole breast radiotherapy reduces the incidence of recurrence after breast-conserving surgery. We see a graph on Fisher and colleagues, with on the y-axis the cumulative incidence of recurrence, you can see that patients undergoing lumpectomy have a 38% of cumulative recurrence incidence after 20 years. If you add all the radiotherapy to this, the local recurrences are reduced to 12%.
Now, radiotherapy also improves disease-free survival and breast cancer-specific survival after breast-conserving surgery. When you look to the panel at the left, you see the any first recurrence on the y-axis and you'll see that after 10 years, patients undergoing breast-conserving surgery showed 35% of recurrences. If you add radiotherapy, this is decreased till 19%. And this is translated into benefit in cancer-specific survival, we see the breast cancer deaths 25.2% after breast-conserving surgery and a reduction to 21.4% of breast cancer deaths when you add radiotherapy. So, we have an improved breast cancer-specific survival of 3.8% and this number is important to remember if we go later on to discuss the late toxicity of breast cancer radiotherapy.
A radiation boost to the tumor beds reduces the recurrence in the ipsilateral breast of the whole breast irradiation. This was nicely shown by Bartelink and colleagues. You can see that they reported 10% of cumulative incidence of local recurrences in the ipsilateral breast of the whole breast radiotherapy and this was reduced to 6% after 8 years with the addition of irritation boost to the tumor bed itself. But this radiation boost will increase the radiation dose to the heart, if the tumor is located in the lower inner quadrant of the left breast. Does internal mammary and medial supraclavicular lymph node irradiation improve the outcome?
To answer that question, we need to refer to the publication by Poortmans and colleagues in the "New England Journal of Medicine" that he poses a randomized study with very broad inclusion criteria, namely centrally and medially located primary tumors irrespectively of axilla involvement, and patients with an axillary located tumor with axillary involvement. And they reported an improvement in distant free survival. So, distant-free survival is the survival without the involvement of metastasis was 75% in the patients without regional irradiation, and it was improved to 78% when adding this regional irradiation to the internal mammary and medial supraclavicular regions, but this was not translated into significant survival benefits.
But we know that internal mammary irradiation will increase the radiation dose to the lung and to the heart and we need to balance the benefits against the toxicity and that is what radiotherapy perception is all about, it's about balancing breast cancer-related benefits versus long-term iatrogenic events. And to do so, we recently published a paper in "The Green Journal," "Estimating Lung Cancer and Cardiovascular Mortality in Female Breast Cancer Patients Receiving Radiotherapy." And in fact, it was Darby and colleagues who reported in 2013, a linear effect between the increase in major coronary events and the mean radiation dose to the hearts.
You can see that women receiving a mean radiation to the heart of 8 gray have an up to 50% increase in the rate of major coronary events. And Taylor and colleagues confirmed this linear relation between the relative risk of heart disease mortality and mean heart dose, and they reported an excess relative risk of 4.1% per gray mean heart dose. Now, these are relative risks. When you want to estimate the absolute risk, we first need to take a look at the score tables. These are tables developed by the cardiologist and they describe the 10-year risk of fatal cardiovascular disease based on gender, on smoking status, on systolic blood pressure, and on the cholesterol levels.
So, what we did we combined the relative risks from the radiation papers with the absolute risk in the score table to estimate the absolute 10-year carryover of mortality in risk in female breast cancer patients based on mean heart dose, as you see here, based on smoking status, based on age, on systolic blood pressure, and cholesterol levels. You can see important differences. If you, for instance, radiate a 50-year-old woman who smokes with a BP systolic performance of 14 millimeters of mercury, and with low cholesterol, you can see that even at mean heart dose of 8 gray only increases the absolute risk of cardiovascular toxicity mortality from 1% to 1.2%. On the other hand, if you radiated a 65-year-old woman with a high blood pressure and high cholesterol, you'll have an increased absolute risk of cardiovascular mortality from 19% to 22.9%.
Now, perhaps this 3.9% does not seem important to you, but it becomes important when you balance it against 3.8% in survival benefits of radiotherapy that I reported at the beginning of this presentation. Now, Grantzau and colleagues report in "The Green Journal" a model of excess risk of lung cancer according to the estimated radiation dose. You can see on the y-axis the excess risk of lung cancer, and on the x-axis, the radiation dose in a certain region of the lungs. And you can see that the region of the lung receiving 10 gray, that these patients have in that region an 85% excess risk of developing lung cancer.
Now in order to translate these towards absolute risks, we use the data from the PLCO-trial, and the PLCO-trial, they describe the probability of smokers who develop lung cancer according to four parameters, namely the age of the patient, the pack-year smoked, and also the fact that the patients committed smoking and since when they quit smoking, and if not, whether they are still smoking and how many years they are smoking. So, we combined this relative risk with this absolute risk on the PLCO-trial and we calculated the absolute 20-year lung cancer risk in female breast cancer patients and you can see huge differences.
For instance, if you irradiate a 55-year-old non-smoking patient, even mean bilateral lung dose of 8 gray increases the risk of developing lung cancer from 0.1% to 0.2%. But on the other hand, if you irradiate a woman from the same age after 20 pack-years that continue smoking, you see an increased risk in developing lung cancer from 15.1% to 39%. So, clearly in these type of patients, you need to pay extra attention in keeping the lung dose as low as possible. So, we develop this novel NTCP-model for estimating lung cancer and cardiovascular mortality and we propose to use it, first of all, for patient-tailored cardiovascular prevention and lung cancer screening strategies after breast radiotherapy.
Secondly, it can help for individualized prescription of radiotherapy, balancing breast cancer-related benefits versus long-term iatrogenic events. And, of course, it can be used to evaluate new radiation techniques that are dedicated to reduce the mean heart dose such as deep inspiration breath-hold. Deep inspiration breath-hold increases the physical space between the heart and the breast. Here you see a CT scan of a patient in shallow breathing where the heart touches the breast. When this patient takes a deep inspiration, you see clearly that the physical distance between the heart and the breast becomes much increased.
And this deep inspiration breath-hold decreases the mean heart dose. We see a patient receiving two-field radiotherapy, so radiation to the whole breast, and you can see that the 2 and the 4 gray isodoses are going through the hearts. When you include an internal mammary lymph node radiation as I discussed in the previous trials, you can see that isodoses are going further into the heart. When you use DIBH, you can spare the heart whether you irradiate the whole breast or whether you irradiate the whole breast and mammary internal lymph node. But of course, to do so you need a very precise positioning and control of the deep inspiration breath-hold. And in order to go to this topic, I would like to give the floor to my colleague, Thierry Gevaert. I thank you for your attention.
Bogdan: Thank you for your talk, Professor De Ridder, and we'll go next to Dr. Thierry Gevaert who will address a rather interesting exploratory outlook into extracranial applications of ExacTrac Dynamic.
Prof. Gevaert: Good afternoon, dear colleagues. First of all, I want to thank Brainlab for inviting me to share our past experience with the ExacTrac Dynamic for breast cases. These are my disclosures. So, when we look back about the ExacTrac system, we all know that there is a success story based on brain and spine. So, in our department, we were using for years, for many, many years already the brain and the spine indications with the stereoscopic X-ray images. And when we were thinking about going to an ExacTrac Dynamic software, we also wanted to see whether or not we could push a little bit the indications and also look into extract cranial cases.
So, when we are looking about the ExacTrac Dynamic, as the previous speaker was already mentioning, there is a service guidance, which is based on light and thermal information. And of course, we maintain the stereoscopic x-rays, which is the IGRT part, and this time, the X-ray can be triggered based on the gantry angle when we are using a dynamic arc or it can be also triggered based on the surface information on the monitor units. So, what was really interesting for us when we were looking into that system is that we have the surface guidance together with the X-ray images that we all know already for years and that we are employing for all our brain and spine indications.
And again, as I was mentioning, we wanted to push a little bit the boundaries and we want to be the first person to also directly see whether or not we can use that system also for extra-cranial indications and more specific, breast cancer patients. As Professor De Ridder the heater was already mentioning, so we want to perform NTCP modeling and, of course, that model will rely on the assumption of correct dose delivery. And when you look about breast cases, you have a good dose gradient towards the heart and the lung, and so it will be, of course, very sensitive to positioning variations and we want also to be sure that you can have a robust evaluation of the dose that was delivered to our patients.
So, we have to find a way that positioning variations and breath-hold variations can become important and we have to find a way that we are sure that those kind of variations will not affect the treatment delivery. And that is why also we wanted to look into the ExacTrac Dynamic as it can monitor reposition based on the surface guidance, but also to anatomical information of the stereoscopic X-ray images. So, when we look about our breast indications, the way we are treating them right now, we believe that there are some points of improvement. And first of all, it's the pre-positioning site, so for the moment, we're using the skin marks in a three-point laser alignment. So, we hope that with the surface guidance, we can start to position the patients in another way.
And then of course, as I was mentioning, the intrafraction motion monitoring, and for the left side of the breast, the breath-hold, the variation on the breath-hold, can we also cope with that new device? So, first of all, we will look a little bit into the surface guidance, how it works, and if it can be reliable for the pre-positioning, and then, of course, we will see whether or not the X-ray images and the X-ray triggering can be an added value for those treatments. So, about surface guidance, well, there are a lot of publications already out, we know that it has a high spatial and temporal resolution and it's an important addition to the patient positioning and the monitoring.
So, we have a broad agreement already when you look into literature about the superiority of that surface guidance over three-point laser alignment, of course, when the surface is able to be used as a surrogate for the tumor positioning. So, that's a little bit of the idea, that's what you find in literature. So, I won't reinvent the wheel and so I wanted directly to see whether or not my ExacTrac Dynamic can lead to a tattoo-free workflow. Considering the emotional burden of skin marks, it's a constant reminder for our patients, so we believe that for our patients, it can be interesting to go to a tattoo-free workflow. And of course, what is really important as well is that we can reduce our patient setup time.
With that in mind, we started to use the workflow for breast indications. In the first 17 patients, we perform a little study after the fifth fraction, just that we are sure that the patient is already used about...just to be sure that the patient is already used to the treatments. And from that point, so from Fraction 6, we pre-positioning our patients with the ExacTrac Dynamic three times and three times, we were pre-positioning it based on the laser alignment, and the residual positioning error was measured with the cone beam CT. When you look about this data, it was not statistically significant due to the small amount of patients, but you see directly that we are in the same order of magnitude between the ExacTrac Dynamic pre-positioning and the laser alignment pre-positioning.
And the RTTs were telling me that they have a feeling of more efficient positioning, moreover for the rotation. So, when you see about the roll and the ease of rotation, they had the feeling that as they can visualize the rotations, that they have a better control over that rotation. But for us, it was important to be sure that what we can do with the ExacTrac Dynamic, so with the surface guidance, it was as good as the well-known laser alignment that we are using already for many years in our departments. And secondly, for efficiency, we went to Reims, we have a close collaboration with them, and as they had one month more experience with the ExacTrac Dynamic, we wanted to use their philosophy and their know-how to see whether or not it was an efficient pre-positioning.
So, we went there at a random day and we were looking into all the kinds of indications they were performing the ExacTrac Dynamic pre-positioning, so it was cranial and extra-cranial indications. And the time that you see on the right is basically the time that the patient is laying down on the couch and the time of the start of the treatment, and you can see that for most of all indications were within the three minutes. So, in that patient population, it's mainly prostate, head and neck, brain cases, lung cases, so it's a diversity of patients, and we see that for most of the patients, it's within that three minutes time slot.
So, with this in mind, we feel really secure and confident that we can use the surface guidance of the ExacTrac for pre-positioning purposes. One drawback is the robustness, so normally, when you are pre-positioning your patients, what is happening is that it's only the structured light that is used for the pre-positioning, so it's not enriched with the thermal information. And what you see normally is those kinds of information, so you see your patient and you see the shifts that you need to perform in order to pre-position your patient. But time to time, certainly what is happening is that the camera is lost and that the structural light is still visible on the patient, but they cannot be seen anymore in three dimensions, so we don't get the pre-positioning shifts anymore.
The good news is that you can just pass it by confirming your positioning and then you start the cone beam CT of the ExacTrac pre-positioning and the X-ray images, and basically, at that time, you can perfectly use back the ExacTrac. So, from time to time, the pre-positioning is getting lost and we feel and it's a feeling that is also a little bit confirmed by literature that, yes, for the Catalyst, that a three-camera system compared to a single camera, it gives a better surface coverage and it's more accurate patient setup. So, keeping that in mind and the fact that we only use the structural light for the pre-positioning, we think that if we can in the next version use also the thermal information for pre-positioning, it will be more robust and also more accurate.
And keeping in mind also that once we pass the pre-positioning step and the thermal camera was switched on, we had a very robust positioning and we never had the fact that the camera was getting lost and was losing the information, as I mentioned, of the position of our patient. So, we think that for pre-positioning purposes, it's for sure a usable system, so let's look now about the intrafraction motion monitoring. So, first of all, when we think about the positioning part and the monitoring part, we need, first of all, to draw a region of interest on our patients in order that the system knows what it needs to track during the treatment course. And so, that's a really important step because you have to know what you have to watch during the treatment.
So, as you can see on those images, so we draw an ROI of the region that we want that the surface scanner and the thermal information is tracking during the treatment. And once you go into the next step, basically it's only tracking the region of interest that was contoured and you have, again, the three shifts and rotations visualized during the old treatment course. So, I was speaking about intrafraction motion, but is it something that is known in literature or not? Well, we had a nice publication of 104 patients and they were only using the optical real-time surface imaging for intrafraction motion monitoring. And you can see that for most of the patients, you are within one millimeter, so you're very accurate, but you can see that sometimes you have intrafraction motion, and time to time, a patient can go out of that one millimeter.
So, basically, surface guidance is interesting because you can have a guidance and a surveyance during the treatment, and if you have one of those patients that goes out of a certain tolerance, you can directly also correct for it. So, intrafraction motion is there and the surface scanner apparently can also watch and see those kinds of variations. So, keeping in mind that we don't have only surface guidance, but we have also the IGRT, so we have also the stereoscopic X-ray images, why not confirming what the surface is seeing with anatomical information? And so, is there any kind of added value by using also an imaging system?
Again, in literature, we have groups that were already performing this kind of study. And so, basically, they were doing a surveyance with the surface guidance and once it's out of tolerance...the threshold was two millimeters, so when the surface guidance was out of that two-millimeter clinical threshold, they were performing a cone beam CT in order to see whether or not it was correct what the surface guidance was mentioning. And basically, you see in the graph that the surface guidance and cone beam CT are more or less always in the same order of magnitude of errors of shifts seen by the systems.
So, keeping that in mind, of course, it's a cone beam CT, so the problem for us, if we want to apply that kind of philosophy in our department, is the fact that every time that the surface goes out of a certain tolerance, we will have to re-perform a cone beam CT. Cone beam CT is time-consuming, so it will take a lot of time, so again, if the time goes up, a patient can maybe again move, and then you will have to recorrect all the time for his positioning. So, that is why we believe that maybe using the X-ray images, it can be an interesting way to go for those kind of surveyance.
So, again, that's what we see during treatments. So, you have the sole variance of the surface and then again, at a certain point in time, if you have an arc, you can use the 0 and the 270 or the 90 degrees to take stereoscopic X-ray images, or you can just ask the system at a certain threshold of surface tolerance to take also X-ray images, or you can base it on monitor units. But basically, we like to do it when we have an arc in the 0 and the 90 degrees or 270, you take X-ray images, and it's directly during the treatment, so you can do beam hold or you can just continue your treatment. But you have directly your surface guidance and your image guidance are just giving you the tolerances or the shift that it's seeing.
So, on the right, it's purely surface, on the left, it's anatomical information with regard to the position of our patient. So, it's combined, it's directly, you can ask whenever you want some images, so that's something that is really interesting because you don't lose time by adding another imaging sequence during your treatment. So, that's a bit of the idea and we've also seen with the first 20 patients that we treated with progress indications that most of the time, what the surface scanner is seeing is also what the X-ray images were seeing with the anatomical information.
So, if we go then to the last step, so what about the deep inspiration breath-hold, because all our left cases are treated in deep inspiration breath-hold? So, first of all, we have a gating window of four millimeters because, for us, it proved to be adequate in the clinical setting. When we take a gating window that is more narrow, the problem was that the inspiration level of our patient was never within that range and we had very long treatment times. So, we think that for us in our case and our philosophy, we take a gating window of four millimeters. And on the right, you can see that you have a breathing pattern, so you see the patient breathing in and breathing out with the thermal camera.
For the moment, unfortunately, we can not rely on the motion of the patient with the breathing pattern, so what we are using for the moment is the SGRT system for monitoring the breathing pattern. So, when we looked then about a complete treatment course, so the idea is basically, first of all, on the right top, we see the breathing pattern of our patients. And so, we'll ask first of all the patient to do a breath-hold, so once a breath-hold is performed, we will come to the region of interest that we want to use during treatment. We take two X-ray images as a reference in the breath-hold phase. And then, first of all, when you see the X-rays, what we will do is we will first block the region of interest. So, we will take the vertebra, the clavicular, and all the moving parts that are not necessary for the fusion, we will block them out with the ROI.
And once that is performed, we will ask to perform a CT registration of the images. And during that time, when you look in the upper right corner, the patient is just breathing in and breathing out. So, once we accept the positioning, we'll just take those images as a reference, so the cone beam will be the complete positioning purpose...the positioning purposes will be performed by the cone beam CT and the first X-ray will just guide us as a reference. So, we'll take now new reference images with our X-ray images, they will be compared to the first X-ray that were taken just prior to being on. So then, we asked again the patient to breathe in and breathe out and then to block when we are in the indicating window and then the treatment is started.
And again, you will see that when we passed by the 0-degree angle, X-ray images will be taken just to confirm the surface guidance information as we see here, so it's perfectly matching. We continue the treatments and like this, we have a guidance with the surface and also an anatomical information in order to be sure that our patient is not moving during the treatment. So, that's a little bit of how we are treating our patients, so now we have a beam hold just because the patient was not breathing again in the gating window. Once again in the gating pattern, we just continue the treatment. So, that's a little bit of an overview of how we are treating our patients.
And so, what we were also doing, so as we have the monitoring, so what we were looking into was for the first five DIBH patients, we were looking into the combination of surface, thermal, and X-ray information. We took an average of two fractions and what you see in the graph on the right is the greatest deviation from the 6D displacements based on every monitor unit that was delivered on our patients. And when you see that we took a threshold of three millimeters, it's not the real threshold that we will apply in our clinical setting, but as we don't have enough information and enough drawback to know what kind of threshold we need to take, we decided to take the three-millimeter. But you see that two out of the five patients went out of that three-millimeter threshold and there was one that's really, really strange.
So, you see that here, all of a sudden, it was out of the four-millimeter tolerance, and then it went back within the three-millimeter. So, if we look into that patient, so again, we ask the patient to do a breath-hold in order to have the information for the region of interest, then we take...sorry, then we take the X-ray images, and from that part, we will start to do the surveyance with the surface guidance and you see that here we are within tolerance. And then, all of a sudden, what was happening is that we are taking X-ray images and suddenly the patient was moving on the couch or something because suddenly we are out of tolerance. You will see now, we have a beam hold because the patient was not within the gating level and now we are out of the tolerance, so of course, we didn't do a beam hold because, for the moment, we are just collecting the data.
But you see the second images, again, the patient is out of tolerance, and then, all of a sudden, he will move himself, and then suddenly it's within the threshold again. So, that's something that we also see. For the moment, we are not correcting for that just because we are treating a lot of breast cases in our institution, up to now we didn't correct for this, we want to know first which kind of threshold we will apply in the clinical setting and then move on with a reliable setting where we will also correct for those positioning issues. And then again, so for those 5 first DIBH patients, what we also trying to do is so on the upper part is RGSC briefing signal and the down part is the average of 5 patients, so it's 5 patients and 2 fractions, so it's 10 information that we will plot against the breathing pattern. And you see that the longer that the patient is on their treatment couch, all of a sudden, you see that you can get to that three-millimeter.
It's also something that you see for other indications, not only for breast, so the longer that patient is on the treatment couch, the more suitable he will be to move, so that's also here something that we could already see that the longer he is on the couch, the more he can move. So, again, that's something also that makes us the stronger feeling that we need to have a kind of intrafraction motion monitoring and we just need now to find the correct threshold in order to correct for those kind of motions. So, to conclude, it's only a proof of principle, of course, but we all know that the surface guidance will correct for patient posture, arm position, and so on. We can go for a tattoo-free workflow, so we can get rid of the skin marks.
And the ExacTrac Dynamic itself will add a combined workflow, surface guidance together with the image guidance, so with the X-ray triggering, which we believe that it can be really interesting because you have the surveyance, you have the surface. But if the surface says that there is something going wrong, well, now we can easily trigger it with X-rays in order to confirm with internal anatomy that what you are seeing with the surface guidance can also be true with the real anatomy of our patients. And we believe that, yeah, it's the way that we can move on to be sure that our patients are well treated during the treatment course, and we hope that soon we can also evaluate the intra-DIBH stability so that we will have a second version where we can also cope with the signal with the thermal information. So, I thank you for your attention.
Bogdan: Thank you for your talk, Thierry. And for those of you interested in seeing a more integrated solution for DIBH setups, please go to the virtual Brainlab booth and request a demo for our ExacTrac Dynamic product. With all four lectures completed, I would like now to start our live question and answer session. Well, thank you all for your presentations and let's see if we have any questions. And Thierry, maybe I will start with you first since we had some questions on your breast workflows. Let's see if we can turn on your camera too. Okay, so I'm not sure if you can hear me, Thierry, but a question for you regarding intrafraction motion, "Before utilizing this first version of ExacTrac Dynamic, what did you use to check for intrafraction motion for your DIBH patients?" You don't hear the question. Okay, is it better now? So, Thierry, the question was, "For your DIBH patients, before ExacTrac Dynamic, what did you use to check intrafraction motion before?"
Prof. Gevaert: Before that, basically, we were doing anything...sorry for that, before that, we were doing anything for intrafraction motion. So, after each tangential field, what we were performing was a kind of EPID image or something else, depending if we saw that on the EPID image the patient was moving, we would take another cone beam or something else. But normally, it was just prior to treatment CBCT images and then the EPID after each tangential field to see whether or not the direct sequel was at the same point that we wanted to perform our treatment. So, we didn't add 3D in intrafraction motion management as our PTV margins were calculated for that as well.
Bogdan: And I think there was a little bit of initial confusion in the audience on how the RPM system is being utilized, so maybe you can briefly summarize again the fact that this is...in the first version, you're basically checking for the IGRT value that ExacTrac Dynamic brings but they're actually triggering based on the RPM. So, maybe to summarize again kind of the workflow that you're doing today.
Prof. Gevaert: Yeah, okay. So, basically, for the moment, we just want to see the added value of the ExacTrac Dynamic for everything about motion m
[00:00:29]
Bogdan: Hello, everyone, and welcome to a new Novalis Circle Symposium. My name is Bogdan Volcu, I am the director of Novalis Circle and today I have the pleasure to host this online event in collaboration with the annual online ASTRO meeting. For our presentations today, I'm honored to introduce you to two groups, one from Ludwig-Maximilians University in Munich and one from University Hospitals in Brussels that have put together an agenda for talks highlighting clinical and technical considerations for spine and breast treatments. To begin, we'll have Dr. Maximilian Niyazi from LMU Munich discuss disease considerations for spine radiosurgery. Dr. Niyazi is the vice chair of the Department of Radiation Oncology and the deputy coordinator for the Neuro-Oncology Comprehensive Cancer Center.
He will be followed by Dr. Phillip Freislederer who will discuss the first clinical applications of ExacTrac Dynamic. Dr. Freislederer will also discuss general commissioning and implementation considerations for ExacTrac Dynamic. We will follow his talk with Professor Mark de Ridder from University Hospital in Brussels, who will discuss general considerations for breast radiotherapy. Professor de Ridder is the head of the Radiotherapy Department, the vice chairman in the Board of Directors, and also the vice president of the Medical Council. His talk will be followed by that of Professor Thierry Gevaert who will discuss both SGRT and IGRT considerations for breast setups. Dr. Gevaert is a professor and the coordinator of the Medical Physics Group in the Department of Radiation Oncology at UC Brussels.
And as it has become standard by now for online events, please remember to utilize either Google Chrome or Safari, and should you have any internet connection issues, simply refresh the webinar page. Use the chat interface to send us questions, we will answer your questions upon completion of the four lectures. Monitor the polling interface for questions that we might like to ask you, and should you choose to follow us on social media, please use the #NovalisCircle. With this, I'd like to turn it over to Professor Niyazi for the first lecture.
Prof. Niyazi: Thank you so much, Bogdan, for the introduction. Ladies and gentlemen, today I would like to talk about radiosurgical planning and spine disease considerations with ExacTrac Dynamic. So, these are all global disclosures. So, first of all, I will talk about the capabilities of the ExacTrac Dynamic system. And as you probably will note, it has an optical surface scanner with structured light and it has a built-in thermal camera that gives you an additional registration information such as the fourth dimension, if you would like to say, and you can avoid multiple in room cameras that are unnecessary, in this case.
You have the well-known X-ray positioning and the intrafractional monitoring, and the good thing is that you have this additional real-time tracking possibility using the SGRT surface-guided radiotherapy element. That's a nice screenshot of my colleagues actually here. The difference, you have the 2D information, or 3d if you would like to say, and thermal information that's added and gives you another layer of depth about the situation. So, the key benefits of the ExacTrac Dynamic are that you have a super metric accuracy in patients, you can track both very large and very small areas. The respiratory and motion tracking is not too sensitive to room lighting or reflections or the skin tone. There's no imaging or imagery-related blocking of one of the cameras, you have a unique matching algorithm, and you have the accuracy of tracking that's not affected by baseline drifts such as, for example, a change in room temperature.
Our experience so far is that the first patient was treated on June 2nd of this year. So far, we have treated 30 patients with 26 brain cases, 2 spine, and 2 head and neck cases. And the pre-positioning is actually done in this way that you have the automated plan loading from your R&V system, in our case, from Mosaiq. The system itself is fully integrated with the LINAC and the pre-positioning is done using the outer contour from your planning CT and the shifts are automatically sent to the LINAC. The indications from the scratch were multiple brain metastases in our STEREOBRAIN trial and, of course, every kind of hypofractionated SRS in the brain. With fractionated stereotactic radiotherapy with open face masks, for example, meningioma, vestibular schwannoma, or glioma.
And we could easily replace the cone-beam workflow to speed up the treatments in, for example, highly palliative cases, whole-brain irradiation, and other things. And we're now launching our spine SRS program with a prospective registry trial, and of course, you can treat any bone oligometastatic disease anywhere else in the body. Now, I'll tell you something about the conventional radiotherapy with the previous gadgets that we used like, as you all know, 3 gray times 10, 4 grey times 5, or even 8 grey times 1. And the target volume in this situation was traditionally the whole vertebral body including the cord, mostly with one vertebral body above and below, was still mostly the 3D conformal radiotherapy, and it was somehow effective with 60% to 70% of pain response but it was of short duration, and it was, of course, dependent on histology.
So, that's something that still has to be considered, for example, in patients with limited life expectancy but, of course, you will lose local control in the long run. As you can see here in this Shizuoka Cancer Center trial, the local control rate really drops if you have a mass lesion and the bulky tumor patients have been significantly worse one and two-year local control, though it will be relevant for these patients to optimize local control. And one option to optimize local control is SBRT, obviously, because it offers excellent local control and palliation as well. This is a nice analysis from Tseng et al., as you can see here, most of these prospective and retrospective series with a lot of patients altogether has shown control rates around 80% to over 90% mostly even at one year later endpoints and the pain response was mainly impressive.
But what do you have to consider and what's getting more important compared to conventional radiotherapy is that the cord tolerance is important. And in SBRT, it's of course, in the far lower range than you would know and expect from conventional fractionation. And if you add together all these QUANTEC, Katsoulakis-Gibbs, or Sahgal models, then you will end up with something...above 15 gray, it will be a steep increase of myelopathy, and above 12 gray, it will be at least critical. So, you have to consider that if you switch your workflows to spine SBRT. Another risk is the vertebral compression fraction that's obviously dependent on the dose that's delivered to the vertebral body.
That's an example of a fracture that is progressing from initially a lower height deviation to here seven millimeters, that's pre and post-SBRT shown in Cunha, 2013. And this risk is obviously related to several factors and one of the factors is the SINS score that has actually been developed as a classification system for spinal instability. So, this is used by traumatic or neurosurgeons to derive whether the vertebral body addressed to be unstable or not and what goes into that index is the location, whether there's the pain relief with the recumbency and/or pain with movement, bone lesion, radiographic spinal alignment, vetebral body collapse, for example, that is pre-existing or not, and if the posterolateral involvement of the spine elements is given. And according to some of the scores you have a stable situation with 0 to 6 points, or a potentially unstable situation where you should consult your neurosurgeon, and with more than 12 points, it's an unstable situation that's critical.
And as I pointed out, it was initially developed by neurosurgeons but it's also predictive for the vertebral compression fracture and these are potentially problems and here you have to, obviously, contact the neurosurgeon. And concerning the compression rate, it's as well as dependent on the dose. That's a nice work of Arjun Sahgal, where you can see that if you go beyond 20 and beyond 24 gray, then the cumulative incidence of fractures is vastly higher, so you have to consider that. And another factor is the BILSKY epidural disease grade and this goes from 0 like the absence of epidural disease up to 3 where you have a spinal cord compression without visible CSF, and with just the abutment of the thecal sac or the impingement, there are different steps in between.
And critical ones are here with the spinal cord compression, where you have at least to consider whether you need a separation surgery before you have the spine Bilsky treatment and that's substantiated by this nice work from Al-Omair about surgical resection of epidural disease that may improve local control. And one of the reasons, of course, is here, that you have the cord tolerance and if you have like a bit of space to the cord, then you can escalate the dose. And he has shown that patients with the Bilsky score of 2 or 3 benefitted from this kind of separation surgery as they go down with the post-op Bilsky score 2 and 3 in just 10%, dropping from 60%, and here you can see the local control probability for these 48 patients that are presented with a higher score.
And if you can achieve something around 1 or 0 compared to 2, then you have obviously a better local control, so it really makes sense to offer patients this kind of surgery. So, what about the automated target definition? And there are international consortium guidelines, for example, the International Spine Radiosurgery Consensus guideline. And the important aspect about that is that the vertebral bodies are different comparing to the height, of course, and the consensus states that the intention should be to include, of course, the complete extent for the GTV and there's given a consensus for the CTV, where you have to include different parts of the vertebral body according to the extent of the gross lesion within the vertebral body.
So, you can skip, for example, some parts of the posterior elements if, for example, the extension is limited to the vertebral body itself. And here the local anatomy of vertebral bodies again, and according to the location of the tumor, for example, if it's just within the vertebral body, you can skip it to treat, in this case, the pedicle and other regions. So, this is very important, that this consensus guideline leaves you some space to, for example, skip some of the posterior elements, and then it's even easier to achieve the gradient visually to the spinal cord and it actually makes sense compared to the older strategy to have one vertebral body above or below that's really not needed.
These are some of the common fractionation schedules that can be delivered in our SPINES protocol and it's just a collection of different regimens. And our primary aim was to reduce the risk of vertebral compression fracture, but to have a certain range from one to five fractions according to histology and to different anatomic locations and to the size of the tumor. And it goes from like 16, in this case, gray to the elective ISRS volume, up to 24 gray to the SIB boost, so the similar things, integrated boost. And what we're doing here is that we take the gross lesion, we expanded by three millimeters to have the boost volume, and we have the elective volume according to ISRS and add on two to three millimeters, exclude the spinal canal, and have the thecal sac as our POV, it's the cord plus 1 or 1.5 millimeters of the cauda equina.
And these are the organs at risk that are considered. As you can see here, if you look at the different single fraction doses and the cumulative dose then the biological effective dose, for example, if you look at normal tissue with an alpha-beta of 2 gray, that's really vastly different for these different fractionation schedules and you are really several factors above the conventional fractionation paradigm. So, that's very important to know and to consider if you perform these treatments and that you think of, for example, to reduce the vertebral compression fracture with this integrated boost approach.
The inclusion criteria of our SPINES trial are relatively comparable to what's given, for example, by a nice meta-analysis by Sahgal or the ASTRO guidelines. So, we're trying to include less than or equal to three effective vertebral bodies that are untreated or in case of the progression after conventional radiotherapy, but it should at least be a time interval three months or better, five months. So, of course, oligometastatic disease with less or equal to five metastasis, a stable spine with the SINS score of 0 to 6, and the low epidural involvement, the extraspinal tumor situation should be under control, and the life expectancy should be relevantly high. The exclusion is mainly the opposite of that. So, further points that we exclude, radiosensitive histologies where you have good results with conventional fractionation, and, of course, significant spinal cord compression or progressive neurological symptoms where you should consult your neurosurgeon, such as, for example, BILSKY score of more or equal to 2.
These are the normal tissue constraints that should be considered in spine SRS. These are given by, for example, the ASTRO guidelines, but of course, several other guidelines are available as well. And you have here different, for example, Fractionation schedules for the spinal cord, for example, OF 12.2 to 25.3 gray. The max dose is considered, the cord here in the case would or should be, in my eyes, the PRV with an additional margin to represent thecal sac. We have the esophagus here with different doses up to five fractions, and some other relevant at-risk structures right here given by different trials and I think these are very important to be considered to keep the risk as low as reasonably achievable.
And one extra intelligence layer that can ensure the right deformation of the curvature of your MRI that you need to define the cord is given, for example, by the Brainlab software and that's a very nice tool to really match your CT registration. And here you have the planning panel, it can give you your prescription and your constraints, and then this plan is optimized and structures are automatically generated, for example, elective volumes according to the ISRS recommendations. So, I'd like to end up with a clinical example here, it's uterine leiomyosarcoma with metastasis, initially presented in 2015 with an area of the piriforme muscle, just got resected and adjuvant chemotherapy, had a sacral metastasis in 2016, again, resection and adjuvant chemo.
and in 2019 to '20, destructive metastasis of T9 and the humerus, got resected and the palliative radiotherapy with 3 gray times 12. And seven months later, we had a progressive lesion in T9 and applied a spine SBRT with 5 times 4 gray to the elective volume and the boost to 30 gray to this recurrence before resectioning in the sense of the separation surgery. And what did we do right here? We had high margins for the patient surface as we had breathing motion that was detected with very tight margins for X-ray surveillance with 0.7 millimeters and 0.5 degree, and the highest possible frame rate for X-ray imaging with only stereoscopic views in 4 per arc. And the good thing was that the patient was just lying on the table, arms up, no body fixation, no vacuum matters, relatively easy setup, and the good thing is you have the ExacTrac Dynamic keeping track that everything is all right.
And the last thing that I would like to show is this really nice trial from Arjun Sahgal, the SC.24 trial from the Canadian group that was shown in the last ASTRO Congress and you have really an impressive result of 33% complete response compared to 16% with the conventional fractionation randomized Phase II trial. Impressive results, a higher level of quality assurance, and the time point was six months after therapy. So, really after a longer duration, relevant improvement in pain responses, and complete response rate. That's something that's relevantly improved to the RTOG 0631 that use rather lower doses and had three months time interval, and that's probably the reason why these two fractions, so 12 gray times 2 regimen was actually improved compared to the conventional fractionation. So, I think that's a really nice outlook for spine SBRT. I would like to thank you all for your attention, I would like to hand it over to my dear colleague, Phillip Freislederer. Thank you so much.
Bogdan: Thank you for the review of your spine program, Professor Niyzi, and let's go to Phillip Freislederer for a recap of the LMU experience with ExacTrac Dynamic for spine treatments.
Dr. Freislederer: Thank you for the introduction. My name is Philip Freislederer from LMU University Hospital in Munich, and I will be giving some insights from the physical details for spine disease considerations with the ExacTrac Dynamic. So, first of all, let me guide you through the workflow a little bit. At first, the software starts...or the workflow starts with a patient pre-positioning based on the surface only with no thermal information. So, what simply happens is the software calculates the deviation from the planned position from the planning CT and the current live position from the surface camera. As you can see here, this is what the software would look like, the outer contour from the CT, so the treatment position, and the calculated current outer contour from the live position are shown here in this image.
This is what it would look like for real patient, you shift the patient close to the center manually. Afterward, the shift is calculated with the exit track and the shift is then sent to the treatment couch and the patient is automatically shifted to the desired position. What you see here is some residual motion from the patient breathing. After the pre-positioning of the patient, area of interest is selected. In this case for a spine patient, we have a lot of breathing motion, so we want to have an area of interest which is not affected by the breathing motion too much but rather, as you see here, position around the stomach or the abdomen of the patient where you can make sure that you count for any interaction of motion apart from breathing motion.
Afterward, you acquire X-ray images, in this case, you see some red areas where the image is...where we don't want anything, any information to be taken into account from, and the images are automatically registered in 6D. There are some tools for the evaluation of the registration result. What you saw before was the rubber band, right here you have some blending and you can have a really nice view of if the spine is positioned correctly, you can have a look at the ribs, you can have a look at each single vertebra with different methods. What you see here is a spyglass where you can easily, based on the choice that you or your doctor or your therapist would like, can position the patient or control and evaluate the result of the evaluation of the registration result.
Simultaneously, when you capture an X-ray, the thermal surface also captured and stored in the background. So, after you approve this fusion and this shift to send to your six-dimensional table of the treatment couch, also your area of interest is afterward mathematically transformed according to the position of the X-rays. If you look at the camera, here is an image of the camera in the room. There's a stereoscopic 3D data camera or surface camera with a structured light projector and additionally integrated, a thermal camera. Why would you use this thermal camera? What is the reason for it? First of all, when you look at surface data from a rather flat surface like this or like an abdominal treatment or the treatment the patient you saw on the treatment before, it is very tough for any registration algorithm to get a result, a registration result, especially in the longitudinal direction.
The surface not only a surface, it is also warm or cold in this case, and if you add the thermal data, this is what it would look like for the registration algorithm. This is something you don't see but it is something the algorithm sees, so there are a lot of hills and valleys which makes it easy for the algorithm to grab on. And this is the primary use of the thermal camera, to get more registration or a fourth dimension for the image registration, and this makes it unnecessary to have more than one optical surface camera inside your treatment room. Afterward, we will go into monitoring steps. On the right top side, you see the patient breathing very nicely and you can see the same breathing motion on the bottom, and this all happens while the patient is irradiated.
You monitor the surface all-time in this case as there are still a lot of breathing motion left. We have some very high margins for the surface but very, very tight margins for the X-rays. So, once an X-ray is taken, and in this case, you see there in the lateral direction a shift of 1.2 millimeters calculated, afterward, you can review this result, again, with your multiple tools and sense the shift during the treatment to your six-dimensional table. And after you send the shift, the treatment would go on with surface monitoring where you cannot take stereoscopic X-ray images. If you take a...there's also the possibility of taking monoscopic X-ray images, this is just may be seen as another control mechanism of the software, but if you take one, you do not update your reference surface. Only if you acquire stereoscopic images and these are within tolerances, the area of interests of your surface and thermal of the patient's surface are updated and then the region of interest is transformed according to the position of your X-rays.
So, what happens if the thermal surface or a monoscopic X-ray out of tolerance? You can have your settings for different settings. So, you can say if your surface is out of tolerance, the beam is automatically held, the patient would be brought back to a position where you can acquire stereoscopic X-rays. So, with a plus-minus 10 degrees, around 090, 270, or 180 degrees, you can acquire dual X-rays or stereoscopic X-rays, and you reposition the patient according to the values of what the stereoscopic six-dimensional registration result tells you. The fusion always needs to be verified afterward by the user or it can be in particular cases where it needs to be verified by the users with signature, but you have to verify it yourself all the time with all the methods I showed you before in the patient video.
After you send the shift and then approve the fusion, therefore, these shifts are again stored in the thermal surface and the area of interest is again transformed according to these values. All those settings of what you can do, you can edit in a template editor and for each new patient, you can import with a certain template depending on what you want to do. We have certain templates which we mostly use, cranial SRS, cranial and spine, and spine SRS templates. There are certain methods of choosing what to do, if you want to do a...if you want to use your six-dimensional table, if you only want to use your base table, if you want to hold the beam all the time when you're monitoring. So, this is the number one purpose here is holding the beam to detect intrafractional motions, so this is something that should be turned on all the time.
Then you can choose if you want to verify your X-rays after your initial correction and if you want to verify for each new table angle for cranial treatments, for example, and one of the important steps is the decision on your patient tolerances. As I said before, for spine patients, we have some breathing or some breathing motion involved, so we have quite high surface tracking tolerances of five millimeters because shifts like that happen all the time when the patient is breathing. But you don't want to have more shifts happening, for example, the patient standing up or moving in a lateral way or coughing, this will be detected even with a five-millimeter surface tolerance.
For the extra tolerances, we have quite low tolerances, so 0.5 millimeters and 0.7...or 0.7 millimeters and 0.5 degrees for the X-ray positioning tolerances. Additionally, that's another thing you can select in your template or even for each patient individually. If you want to say if you want to trigger X-rays when the surface exceeds the tolerance automatically, if the beam is held if the tolerance is exceeded, and also if you want to...for a 3D conformal plan, for a static beam in this case, if you want to trigger based on a certain amount of treated monitor units, or for VMAT treatment and arc-based treatment, if you want to have only stereoscopic but also monoscopic X-rays.
Now to go a little bit into the QA or the routine QA and the commissioning of the ExacTrac Dynamic. First, let me start off with routine QA, the daily check we perform takes about five minutes in total, most of the times a little bit less. Two different things are checked here. In this case, one is the deviation between the surface camera and your X-ray positioning systems, so your flat panels, in this case. This is a consistency check, so if someone accidentally touched or hit the flat panels or the surface camera, you would receive an error, in this case, daily. Also, in the second step, the deviation from your radiation isocenter is checked and confirmed every day.
Secondly, once a month at the moment, you have to do a calibration of your thermal surface to your three-dimensional surface or your optical surface. This is done with a distinct Brainlab phantom and also takes roughly five minutes. For our stereotactic treatments, if you do a lot of stereotactic treatments or SRS, so cranial or spine, we would recommend a monthly calibration of your radiation isocenter and there is a...you simply place any type of ball bearing inside the radiation isocenter. This could be the ball bearing inside an anthropomorphic phantom or this could also be any type of Winston-Lutz pointer.
When we started with the ExacTrac Dynamic, we switched very quickly from cone Beam CT workflows to ExacTrac-only workflows. And by a very quickly, I mean the first patient we treated was an ExacTrac-only patient, a cranial patient. So, we want to verify the positioning and the verification accuracy of the two systems and just to show you some short examples of this, we use two anthropomorphic phantoms, one from the head and one from the pelvis, we had six random isocenter locations, and validated the setup accuracy from both systems and the verification again. So, the deviation in the initial correction was maybe a little bit higher for the head when it comes to rotation, but for the verification between those systems, so cone beam CT and an ExacTrac image, after you shifted your patient to your planned isocenter was quite low in this case.
And obviously, you don't really know what your ground truth is. Is your cone beam MCT better calibrated to the radiation isocenter or is it your ExacTrac Dynamic? Again, for some short points on QA and commissioning, there are quite a lot of initial commissioning thoughts if you want to install a new surface-guided system and if you have an X-ray system. Additionally, it's still even a little bit more. A couple of things will be the static accuracy, the dynamic positioning accuracy of the surface, in this case, it's possible to test but it's quite hard to store the data. The impact of the region of interest is something very exciting but it's more of a research topic for the future for the surface-guided radiation therapy community.
Some tests on the field of view, the field of view here not only for the camera but also for the X-ray systems, so to test if there was any blocking of the gantry or accessory parts. If you want to get more information about this, you can find a webinar on the Novalis Circle homepage from September 2020 on the initial experiences with ExacTrac Dynamic. Also for the commissioning, especially for spine or dedicated for spine, we have used a...we started to use a new phantom from RTSafe because we could not find any other phantom, in this case, because it's still some type of prototype. In this case, we have less reflectance skin, the reflecting skin from phantoms is a major issue for surface guided systems, so we want something which is as close as possible to the actual patient skin.
You have a patient structure inside, some hip bones and spine, and we filled the phantom with water, this is possible and nothing gets broken inside, and with roughly 40 degrees so we have a warm surface, and we have three ball bearings inside for the radiation isocenter coincidence tests between for the ExacTrac Dynamic. We have looked at the thermal drift of the phantom, the phantom doesn't drift too much, so we have about...after filling the water with 40 degrees...or filling the phantom with 40 degrees hot water, we have roughly a drift of 4 degrees in 60 minutes, so this would still be enough to perform all your commissioning or your monthly or quarterly tests.
You can see this here on the right side, there's the phantom not filled with warm water and we shifted the phantom a little bit and you can see some wobbling in the software, so it's not too easy when the phantom is cold but it's still possible because the thermal camera is still active. And on the left side, you can see the inside of the phantoms, you can detect your ball bearings or the spinal structures very well. If you fill up the phantom with water, we again can monitor the phantom as close as it would be in an actual patient. So, we have...although it seems like a lot of drift, this is 0.1 or 0.2 millimeters in drift calculated from the software. So, this is something we will use for the commissioning in the future and for additional verification and validation tests. Especially if you have three ball bearings inside, we can easily do an isocenter coincidence test between the two systems and get some nice results from it.
To conclude a little bit, there are certain proven, certain potential benefits of the ExacTrac Dynamic at the moment. The first one would be the automatic or the manual X-ray triggering, so on the real system integration, if the deviations between the planned and actual surfaces are detected, you have a direct repositioning when needed. Because of the nice system integration with the LINAC itself, you are very efficient in time. If you want to have a look at it...this is not really positioning from the surface but this is a repositioning from the X-rays, so you quickly evaluate...this is not fast-forwarded, by the way, you quickly evaluate your patient position according to the x-rays and send these shifts to the LINAC.
And once it is sent...one second. Yes, once it is sent to the LINAC, you can continue with your radiation treatment. So, this is quite fast and saves you a lot of time while still...and this is the most important part about it, you really do an intrafractional motion monitoring, not only with the surface, but also with the X-rays. And this is what I've been saying before, again, you have surface motion monitoring throughout the entire fraction. This is an additional patient safety feature, so you all the time have an automatic beam hold when your patient is moving out of tolerance. You have monitoring, you have a pinhole, and this is a safety feature which might come to...which might be part of any type of LINAC in the future, I believe so at least.
Additionally, from the automated workflow, you have to take less cone beam CT, if it is possible if you don't need any type of...if you're not required on any type of soft tissue contrast too much, if you can have a good match on the bony anatomy, so you have less dose to the patient and a faster workflow, much faster workflow. So, you can reduce the amount of cone beam CT when positioning with only anatomy. Additionally, and this is something that's potential or new but this is something which has been proven for every ExacTrac system not only Dynamic, you have X-ray repositioning for each couch angle and you don't do motion monitoring only for couch kicks, you do it with your X-rays.
So, with motion monitoring only, a surface motion monitoring I mean by that, this would mean that you have to do...if you detect some kind of motion at a certain couch angle, what do you do afterward? You have to move back to a couch of zero degree, acquire another cone beam CT, and then reposition the patient and shift the couch back. As intrafractional motion occurs quite a lot for patients, even if you talk about cranial treatments inside the mask, inside every mask, this is something which is not really practical. In addition, you have the tattoo-less or mark-less, tattoo-less workflows, which are introduced more and more in the clinics worldwide, even also with cone beam CT workflows. So, you have a pre-positioning using the patient surface and the possible reduction of a verification cone beam CT if your clinic does that. So, this would be, again, the video from before but this will be a complete tattoo-less workflow if you think about integrating this in your clinic.
Our outlook is in the future, we want to replace the cone beam CT workflows in palliative settings to make it a little faster. Also for the patient benefit, if the soft tissue contrast matters, we want to reduce the frequency of the cone beam CT and implement the ExacTrac Dynamic regularly. In the next step, we want to do deep inspiration breath-hold treatments for left-sided breast cancer and in the future, stuff like liver or lung SBRT would be the fun stuff to look at in the future in this case. So, I would like to thank you for your attention and I will be very happy to get questions from you.
Bogdan: Great review, Dr. Freislederer, and we'll switch now to Professor De Ridder for a review of breast radiotherapy planning considerations.
Prof. De Ridder: Good afternoon, dear colleagues. I, first of all, would like to thank Brainlab for the kind invitation. I will present our work, "Breast Radiotherapy: Towards a Patient- Tailored Approach." What are the benefits of radiotherapy for patients with breast cancer? Well, whole breast radiotherapy reduces the incidence of recurrence after breast-conserving surgery. We see a graph on Fisher and colleagues, with on the y-axis the cumulative incidence of recurrence, you can see that patients undergoing lumpectomy have a 38% of cumulative recurrence incidence after 20 years. If you add all the radiotherapy to this, the local recurrences are reduced to 12%.
Now, radiotherapy also improves disease-free survival and breast cancer-specific survival after breast-conserving surgery. When you look to the panel at the left, you see the any first recurrence on the y-axis and you'll see that after 10 years, patients undergoing breast-conserving surgery showed 35% of recurrences. If you add radiotherapy, this is decreased till 19%. And this is translated into benefit in cancer-specific survival, we see the breast cancer deaths 25.2% after breast-conserving surgery and a reduction to 21.4% of breast cancer deaths when you add radiotherapy. So, we have an improved breast cancer-specific survival of 3.8% and this number is important to remember if we go later on to discuss the late toxicity of breast cancer radiotherapy.
A radiation boost to the tumor beds reduces the recurrence in the ipsilateral breast of the whole breast irradiation. This was nicely shown by Bartelink and colleagues. You can see that they reported 10% of cumulative incidence of local recurrences in the ipsilateral breast of the whole breast radiotherapy and this was reduced to 6% after 8 years with the addition of irritation boost to the tumor bed itself. But this radiation boost will increase the radiation dose to the heart, if the tumor is located in the lower inner quadrant of the left breast. Does internal mammary and medial supraclavicular lymph node irradiation improve the outcome?
To answer that question, we need to refer to the publication by Poortmans and colleagues in the "New England Journal of Medicine" that he poses a randomized study with very broad inclusion criteria, namely centrally and medially located primary tumors irrespectively of axilla involvement, and patients with an axillary located tumor with axillary involvement. And they reported an improvement in distant free survival. So, distant-free survival is the survival without the involvement of metastasis was 75% in the patients without regional irradiation, and it was improved to 78% when adding this regional irradiation to the internal mammary and medial supraclavicular regions, but this was not translated into significant survival benefits.
But we know that internal mammary irradiation will increase the radiation dose to the lung and to the heart and we need to balance the benefits against the toxicity and that is what radiotherapy perception is all about, it's about balancing breast cancer-related benefits versus long-term iatrogenic events. And to do so, we recently published a paper in "The Green Journal," "Estimating Lung Cancer and Cardiovascular Mortality in Female Breast Cancer Patients Receiving Radiotherapy." And in fact, it was Darby and colleagues who reported in 2013, a linear effect between the increase in major coronary events and the mean radiation dose to the hearts.
You can see that women receiving a mean radiation to the heart of 8 gray have an up to 50% increase in the rate of major coronary events. And Taylor and colleagues confirmed this linear relation between the relative risk of heart disease mortality and mean heart dose, and they reported an excess relative risk of 4.1% per gray mean heart dose. Now, these are relative risks. When you want to estimate the absolute risk, we first need to take a look at the score tables. These are tables developed by the cardiologist and they describe the 10-year risk of fatal cardiovascular disease based on gender, on smoking status, on systolic blood pressure, and on the cholesterol levels.
So, what we did we combined the relative risks from the radiation papers with the absolute risk in the score table to estimate the absolute 10-year carryover of mortality in risk in female breast cancer patients based on mean heart dose, as you see here, based on smoking status, based on age, on systolic blood pressure, and cholesterol levels. You can see important differences. If you, for instance, radiate a 50-year-old woman who smokes with a BP systolic performance of 14 millimeters of mercury, and with low cholesterol, you can see that even at mean heart dose of 8 gray only increases the absolute risk of cardiovascular toxicity mortality from 1% to 1.2%. On the other hand, if you radiated a 65-year-old woman with a high blood pressure and high cholesterol, you'll have an increased absolute risk of cardiovascular mortality from 19% to 22.9%.
Now, perhaps this 3.9% does not seem important to you, but it becomes important when you balance it against 3.8% in survival benefits of radiotherapy that I reported at the beginning of this presentation. Now, Grantzau and colleagues report in "The Green Journal" a model of excess risk of lung cancer according to the estimated radiation dose. You can see on the y-axis the excess risk of lung cancer, and on the x-axis, the radiation dose in a certain region of the lungs. And you can see that the region of the lung receiving 10 gray, that these patients have in that region an 85% excess risk of developing lung cancer.
Now in order to translate these towards absolute risks, we use the data from the PLCO-trial, and the PLCO-trial, they describe the probability of smokers who develop lung cancer according to four parameters, namely the age of the patient, the pack-year smoked, and also the fact that the patients committed smoking and since when they quit smoking, and if not, whether they are still smoking and how many years they are smoking. So, we combined this relative risk with this absolute risk on the PLCO-trial and we calculated the absolute 20-year lung cancer risk in female breast cancer patients and you can see huge differences.
For instance, if you irradiate a 55-year-old non-smoking patient, even mean bilateral lung dose of 8 gray increases the risk of developing lung cancer from 0.1% to 0.2%. But on the other hand, if you irradiate a woman from the same age after 20 pack-years that continue smoking, you see an increased risk in developing lung cancer from 15.1% to 39%. So, clearly in these type of patients, you need to pay extra attention in keeping the lung dose as low as possible. So, we develop this novel NTCP-model for estimating lung cancer and cardiovascular mortality and we propose to use it, first of all, for patient-tailored cardiovascular prevention and lung cancer screening strategies after breast radiotherapy.
Secondly, it can help for individualized prescription of radiotherapy, balancing breast cancer-related benefits versus long-term iatrogenic events. And, of course, it can be used to evaluate new radiation techniques that are dedicated to reduce the mean heart dose such as deep inspiration breath-hold. Deep inspiration breath-hold increases the physical space between the heart and the breast. Here you see a CT scan of a patient in shallow breathing where the heart touches the breast. When this patient takes a deep inspiration, you see clearly that the physical distance between the heart and the breast becomes much increased.
And this deep inspiration breath-hold decreases the mean heart dose. We see a patient receiving two-field radiotherapy, so radiation to the whole breast, and you can see that the 2 and the 4 gray isodoses are going through the hearts. When you include an internal mammary lymph node radiation as I discussed in the previous trials, you can see that isodoses are going further into the heart. When you use DIBH, you can spare the heart whether you irradiate the whole breast or whether you irradiate the whole breast and mammary internal lymph node. But of course, to do so you need a very precise positioning and control of the deep inspiration breath-hold. And in order to go to this topic, I would like to give the floor to my colleague, Thierry Gevaert. I thank you for your attention.
Bogdan: Thank you for your talk, Professor De Ridder, and we'll go next to Dr. Thierry Gevaert who will address a rather interesting exploratory outlook into extracranial applications of ExacTrac Dynamic.
Prof. Gevaert: Good afternoon, dear colleagues. First of all, I want to thank Brainlab for inviting me to share our past experience with the ExacTrac Dynamic for breast cases. These are my disclosures. So, when we look back about the ExacTrac system, we all know that there is a success story based on brain and spine. So, in our department, we were using for years, for many, many years already the brain and the spine indications with the stereoscopic X-ray images. And when we were thinking about going to an ExacTrac Dynamic software, we also wanted to see whether or not we could push a little bit the indications and also look into extract cranial cases.
So, when we are looking about the ExacTrac Dynamic, as the previous speaker was already mentioning, there is a service guidance, which is based on light and thermal information. And of course, we maintain the stereoscopic x-rays, which is the IGRT part, and this time, the X-ray can be triggered based on the gantry angle when we are using a dynamic arc or it can be also triggered based on the surface information on the monitor units. So, what was really interesting for us when we were looking into that system is that we have the surface guidance together with the X-ray images that we all know already for years and that we are employing for all our brain and spine indications.
And again, as I was mentioning, we wanted to push a little bit the boundaries and we want to be the first person to also directly see whether or not we can use that system also for extra-cranial indications and more specific, breast cancer patients. As Professor De Ridder the heater was already mentioning, so we want to perform NTCP modeling and, of course, that model will rely on the assumption of correct dose delivery. And when you look about breast cases, you have a good dose gradient towards the heart and the lung, and so it will be, of course, very sensitive to positioning variations and we want also to be sure that you can have a robust evaluation of the dose that was delivered to our patients.
So, we have to find a way that positioning variations and breath-hold variations can become important and we have to find a way that we are sure that those kind of variations will not affect the treatment delivery. And that is why also we wanted to look into the ExacTrac Dynamic as it can monitor reposition based on the surface guidance, but also to anatomical information of the stereoscopic X-ray images. So, when we look about our breast indications, the way we are treating them right now, we believe that there are some points of improvement. And first of all, it's the pre-positioning site, so for the moment, we're using the skin marks in a three-point laser alignment. So, we hope that with the surface guidance, we can start to position the patients in another way.
And then of course, as I was mentioning, the intrafraction motion monitoring, and for the left side of the breast, the breath-hold, the variation on the breath-hold, can we also cope with that new device? So, first of all, we will look a little bit into the surface guidance, how it works, and if it can be reliable for the pre-positioning, and then, of course, we will see whether or not the X-ray images and the X-ray triggering can be an added value for those treatments. So, about surface guidance, well, there are a lot of publications already out, we know that it has a high spatial and temporal resolution and it's an important addition to the patient positioning and the monitoring.
So, we have a broad agreement already when you look into literature about the superiority of that surface guidance over three-point laser alignment, of course, when the surface is able to be used as a surrogate for the tumor positioning. So, that's a little bit of the idea, that's what you find in literature. So, I won't reinvent the wheel and so I wanted directly to see whether or not my ExacTrac Dynamic can lead to a tattoo-free workflow. Considering the emotional burden of skin marks, it's a constant reminder for our patients, so we believe that for our patients, it can be interesting to go to a tattoo-free workflow. And of course, what is really important as well is that we can reduce our patient setup time.
With that in mind, we started to use the workflow for breast indications. In the first 17 patients, we perform a little study after the fifth fraction, just that we are sure that the patient is already used about...just to be sure that the patient is already used to the treatments. And from that point, so from Fraction 6, we pre-positioning our patients with the ExacTrac Dynamic three times and three times, we were pre-positioning it based on the laser alignment, and the residual positioning error was measured with the cone beam CT. When you look about this data, it was not statistically significant due to the small amount of patients, but you see directly that we are in the same order of magnitude between the ExacTrac Dynamic pre-positioning and the laser alignment pre-positioning.
And the RTTs were telling me that they have a feeling of more efficient positioning, moreover for the rotation. So, when you see about the roll and the ease of rotation, they had the feeling that as they can visualize the rotations, that they have a better control over that rotation. But for us, it was important to be sure that what we can do with the ExacTrac Dynamic, so with the surface guidance, it was as good as the well-known laser alignment that we are using already for many years in our departments. And secondly, for efficiency, we went to Reims, we have a close collaboration with them, and as they had one month more experience with the ExacTrac Dynamic, we wanted to use their philosophy and their know-how to see whether or not it was an efficient pre-positioning.
So, we went there at a random day and we were looking into all the kinds of indications they were performing the ExacTrac Dynamic pre-positioning, so it was cranial and extra-cranial indications. And the time that you see on the right is basically the time that the patient is laying down on the couch and the time of the start of the treatment, and you can see that for most of all indications were within the three minutes. So, in that patient population, it's mainly prostate, head and neck, brain cases, lung cases, so it's a diversity of patients, and we see that for most of the patients, it's within that three minutes time slot.
So, with this in mind, we feel really secure and confident that we can use the surface guidance of the ExacTrac for pre-positioning purposes. One drawback is the robustness, so normally, when you are pre-positioning your patients, what is happening is that it's only the structured light that is used for the pre-positioning, so it's not enriched with the thermal information. And what you see normally is those kinds of information, so you see your patient and you see the shifts that you need to perform in order to pre-position your patient. But time to time, certainly what is happening is that the camera is lost and that the structural light is still visible on the patient, but they cannot be seen anymore in three dimensions, so we don't get the pre-positioning shifts anymore.
The good news is that you can just pass it by confirming your positioning and then you start the cone beam CT of the ExacTrac pre-positioning and the X-ray images, and basically, at that time, you can perfectly use back the ExacTrac. So, from time to time, the pre-positioning is getting lost and we feel and it's a feeling that is also a little bit confirmed by literature that, yes, for the Catalyst, that a three-camera system compared to a single camera, it gives a better surface coverage and it's more accurate patient setup. So, keeping that in mind and the fact that we only use the structural light for the pre-positioning, we think that if we can in the next version use also the thermal information for pre-positioning, it will be more robust and also more accurate.
And keeping in mind also that once we pass the pre-positioning step and the thermal camera was switched on, we had a very robust positioning and we never had the fact that the camera was getting lost and was losing the information, as I mentioned, of the position of our patient. So, we think that for pre-positioning purposes, it's for sure a usable system, so let's look now about the intrafraction motion monitoring. So, first of all, when we think about the positioning part and the monitoring part, we need, first of all, to draw a region of interest on our patients in order that the system knows what it needs to track during the treatment course. And so, that's a really important step because you have to know what you have to watch during the treatment.
So, as you can see on those images, so we draw an ROI of the region that we want that the surface scanner and the thermal information is tracking during the treatment. And once you go into the next step, basically it's only tracking the region of interest that was contoured and you have, again, the three shifts and rotations visualized during the old treatment course. So, I was speaking about intrafraction motion, but is it something that is known in literature or not? Well, we had a nice publication of 104 patients and they were only using the optical real-time surface imaging for intrafraction motion monitoring. And you can see that for most of the patients, you are within one millimeter, so you're very accurate, but you can see that sometimes you have intrafraction motion, and time to time, a patient can go out of that one millimeter.
So, basically, surface guidance is interesting because you can have a guidance and a surveyance during the treatment, and if you have one of those patients that goes out of a certain tolerance, you can directly also correct for it. So, intrafraction motion is there and the surface scanner apparently can also watch and see those kinds of variations. So, keeping in mind that we don't have only surface guidance, but we have also the IGRT, so we have also the stereoscopic X-ray images, why not confirming what the surface is seeing with anatomical information? And so, is there any kind of added value by using also an imaging system?
Again, in literature, we have groups that were already performing this kind of study. And so, basically, they were doing a surveyance with the surface guidance and once it's out of tolerance...the threshold was two millimeters, so when the surface guidance was out of that two-millimeter clinical threshold, they were performing a cone beam CT in order to see whether or not it was correct what the surface guidance was mentioning. And basically, you see in the graph that the surface guidance and cone beam CT are more or less always in the same order of magnitude of errors of shifts seen by the systems.
So, keeping that in mind, of course, it's a cone beam CT, so the problem for us, if we want to apply that kind of philosophy in our department, is the fact that every time that the surface goes out of a certain tolerance, we will have to re-perform a cone beam CT. Cone beam CT is time-consuming, so it will take a lot of time, so again, if the time goes up, a patient can maybe again move, and then you will have to recorrect all the time for his positioning. So, that is why we believe that maybe using the X-ray images, it can be an interesting way to go for those kind of surveyance.
So, again, that's what we see during treatments. So, you have the sole variance of the surface and then again, at a certain point in time, if you have an arc, you can use the 0 and the 270 or the 90 degrees to take stereoscopic X-ray images, or you can just ask the system at a certain threshold of surface tolerance to take also X-ray images, or you can base it on monitor units. But basically, we like to do it when we have an arc in the 0 and the 90 degrees or 270, you take X-ray images, and it's directly during the treatment, so you can do beam hold or you can just continue your treatment. But you have directly your surface guidance and your image guidance are just giving you the tolerances or the shift that it's seeing.
So, on the right, it's purely surface, on the left, it's anatomical information with regard to the position of our patient. So, it's combined, it's directly, you can ask whenever you want some images, so that's something that is really interesting because you don't lose time by adding another imaging sequence during your treatment. So, that's a bit of the idea and we've also seen with the first 20 patients that we treated with progress indications that most of the time, what the surface scanner is seeing is also what the X-ray images were seeing with the anatomical information.
So, if we go then to the last step, so what about the deep inspiration breath-hold, because all our left cases are treated in deep inspiration breath-hold? So, first of all, we have a gating window of four millimeters because, for us, it proved to be adequate in the clinical setting. When we take a gating window that is more narrow, the problem was that the inspiration level of our patient was never within that range and we had very long treatment times. So, we think that for us in our case and our philosophy, we take a gating window of four millimeters. And on the right, you can see that you have a breathing pattern, so you see the patient breathing in and breathing out with the thermal camera.
For the moment, unfortunately, we can not rely on the motion of the patient with the breathing pattern, so what we are using for the moment is the SGRT system for monitoring the breathing pattern. So, when we looked then about a complete treatment course, so the idea is basically, first of all, on the right top, we see the breathing pattern of our patients. And so, we'll ask first of all the patient to do a breath-hold, so once a breath-hold is performed, we will come to the region of interest that we want to use during treatment. We take two X-ray images as a reference in the breath-hold phase. And then, first of all, when you see the X-rays, what we will do is we will first block the region of interest. So, we will take the vertebra, the clavicular, and all the moving parts that are not necessary for the fusion, we will block them out with the ROI.
And once that is performed, we will ask to perform a CT registration of the images. And during that time, when you look in the upper right corner, the patient is just breathing in and breathing out. So, once we accept the positioning, we'll just take those images as a reference, so the cone beam will be the complete positioning purpose...the positioning purposes will be performed by the cone beam CT and the first X-ray will just guide us as a reference. So, we'll take now new reference images with our X-ray images, they will be compared to the first X-ray that were taken just prior to being on. So then, we asked again the patient to breathe in and breathe out and then to block when we are in the indicating window and then the treatment is started.
And again, you will see that when we passed by the 0-degree angle, X-ray images will be taken just to confirm the surface guidance information as we see here, so it's perfectly matching. We continue the treatments and like this, we have a guidance with the surface and also an anatomical information in order to be sure that our patient is not moving during the treatment. So, that's a little bit of how we are treating our patients, so now we have a beam hold just because the patient was not breathing again in the gating window. Once again in the gating pattern, we just continue the treatment. So, that's a little bit of an overview of how we are treating our patients.
And so, what we were also doing, so as we have the monitoring, so what we were looking into was for the first five DIBH patients, we were looking into the combination of surface, thermal, and X-ray information. We took an average of two fractions and what you see in the graph on the right is the greatest deviation from the 6D displacements based on every monitor unit that was delivered on our patients. And when you see that we took a threshold of three millimeters, it's not the real threshold that we will apply in our clinical setting, but as we don't have enough information and enough drawback to know what kind of threshold we need to take, we decided to take the three-millimeter. But you see that two out of the five patients went out of that three-millimeter threshold and there was one that's really, really strange.
So, you see that here, all of a sudden, it was out of the four-millimeter tolerance, and then it went back within the three-millimeter. So, if we look into that patient, so again, we ask the patient to do a breath-hold in order to have the information for the region of interest, then we take...sorry, then we take the X-ray images, and from that part, we will start to do the surveyance with the surface guidance and you see that here we are within tolerance. And then, all of a sudden, what was happening is that we are taking X-ray images and suddenly the patient was moving on the couch or something because suddenly we are out of tolerance. You will see now, we have a beam hold because the patient was not within the gating level and now we are out of the tolerance, so of course, we didn't do a beam hold because, for the moment, we are just collecting the data.
But you see the second images, again, the patient is out of tolerance, and then, all of a sudden, he will move himself, and then suddenly it's within the threshold again. So, that's something that we also see. For the moment, we are not correcting for that just because we are treating a lot of breast cases in our institution, up to now we didn't correct for this, we want to know first which kind of threshold we will apply in the clinical setting and then move on with a reliable setting where we will also correct for those positioning issues. And then again, so for those 5 first DIBH patients, what we also trying to do is so on the upper part is RGSC briefing signal and the down part is the average of 5 patients, so it's 5 patients and 2 fractions, so it's 10 information that we will plot against the breathing pattern. And you see that the longer that the patient is on their treatment couch, all of a sudden, you see that you can get to that three-millimeter.
It's also something that you see for other indications, not only for breast, so the longer that patient is on the treatment couch, the more suitable he will be to move, so that's also here something that we could already see that the longer he is on the couch, the more he can move. So, again, that's something also that makes us the stronger feeling that we need to have a kind of intrafraction motion monitoring and we just need now to find the correct threshold in order to correct for those kind of motions. So, to conclude, it's only a proof of principle, of course, but we all know that the surface guidance will correct for patient posture, arm position, and so on. We can go for a tattoo-free workflow, so we can get rid of the skin marks.
And the ExacTrac Dynamic itself will add a combined workflow, surface guidance together with the image guidance, so with the X-ray triggering, which we believe that it can be really interesting because you have the surveyance, you have the surface. But if the surface says that there is something going wrong, well, now we can easily trigger it with X-rays in order to confirm with internal anatomy that what you are seeing with the surface guidance can also be true with the real anatomy of our patients. And we believe that, yeah, it's the way that we can move on to be sure that our patients are well treated during the treatment course, and we hope that soon we can also evaluate the intra-DIBH stability so that we will have a second version where we can also cope with the signal with the thermal information. So, I thank you for your attention.
Bogdan: Thank you for your talk, Thierry. And for those of you interested in seeing a more integrated solution for DIBH setups, please go to the virtual Brainlab booth and request a demo for our ExacTrac Dynamic product. With all four lectures completed, I would like now to start our live question and answer session. Well, thank you all for your presentations and let's see if we have any questions. And Thierry, maybe I will start with you first since we had some questions on your breast workflows. Let's see if we can turn on your camera too. Okay, so I'm not sure if you can hear me, Thierry, but a question for you regarding intrafraction motion, "Before utilizing this first version of ExacTrac Dynamic, what did you use to check for intrafraction motion for your DIBH patients?" You don't hear the question. Okay, is it better now? So, Thierry, the question was, "For your DIBH patients, before ExacTrac Dynamic, what did you use to check intrafraction motion before?"
Prof. Gevaert: Before that, basically, we were doing anything...sorry for that, before that, we were doing anything for intrafraction motion. So, after each tangential field, what we were performing was a kind of EPID image or something else, depending if we saw that on the EPID image the patient was moving, we would take another cone beam or something else. But normally, it was just prior to treatment CBCT images and then the EPID after each tangential field to see whether or not the direct sequel was at the same point that we wanted to perform our treatment. So, we didn't add 3D in intrafraction motion management as our PTV margins were calculated for that as well.
Bogdan: And I think there was a little bit of initial confusion in the audience on how the RPM system is being utilized, so maybe you can briefly summarize again the fact that this is...in the first version, you're basically checking for the IGRT value that ExacTrac Dynamic brings but they're actually triggering based on the RPM. So, maybe to summarize again kind of the workflow that you're doing today.
Prof. Gevaert: Yeah, okay. So, basically, for the moment, we just want to see the added value of the ExacTrac Dynamic for everything about motion m