Transcript
Bogdan: Hello and welcome, everyone, to a new Novalis Circle webinar. My name is Bogdan Volcu, I'm the director of Novalis Circle. The topic for today involves a clinical and technical review of Elements Spine SRS, and we have the privilege to have both Dr. Shi and Dr. Liu from Jefferson Hospital for Neuroscience in Philadelphia present. Dr. Shi is a radiation oncologist and the medical director of Jefferson Hospital for Neuroscience. He is also the co-director of the Brain Tumor Center and the director of the Jefferson Cutaneous Lymphoma Center. Dr. Liu is the director of radiosurgery physics, Division of Stereotatic Radiosurgery, and he is an associate professor in the Department of Radiation Oncology.
As always, we are providing CE credits upon successful completion of the webinar, you may request credits by emailing us at info@novaliscircle.org and we can provide you with details for CAMPEP, MDCB, and ASRT. Also, please don't forget to sign up for our upcoming webinars. The next one will review clinical utilization for our latest version of Elements Multiple Brain Mets SRS with a clinical team from Alliance Cancer Care in Huntsville, Alabama, and that will be on June 25th. In order to properly visualize the webinar, please remember to use Google Chrome or Safari.
And should you have questions for our two presenters, please utilize the chat functionality throughout the lecture, and we will answer all your questions upon completion of the two talks. We would also like to get feedback from you today on some key issues via our polling infrastructure, so we will let you know when those polls are open. And as always, if you'd like to follow us on social media, you can do that with the Novalis Circle hashtag. With that said, I'd like to turn it over to Dr. Shi and he'll cover a clinical space for spine radiosurgery. Dr. Shi?
Dr. Shi: Thank you, Bogdan. First of all, it's my great pleasure to be here along with my colleague, Dr. Haisong Liu, to present evaluation of Elements Spine SRS plan quality for radiosurgery and SBRT for the treatment of spine metastases. We highly appreciate Brainlab to provide this opportunity. And just a quick note regarding the radiosurgery SRS and sbrt. Conventionally, many people refer for extracranial SRS as SBRT. But for CNS point of view, any single fraction treatments will be referred as SRS for both intracranial and actual cranial disease, and for two to five fraction will be referred as SBRT and this will be used for this presentation. This is my introductory disclosures. In the prior webinar by the Brainlab in May, Dr. Alongi presents excellent overview of the background for spine radiosurgery and also his experience using Elements Spine SRS.
And I encourage everyone to listen to that if you missed it, and it's available through the Nova Circle website. Today, I would like to kick off the webinar basically looking at some of the background information for spine radiosurgery and specifically focusing on some clinical considerations and challenges of using SBRT/SRS technique for spine metastases. And afterward, Dr. Liu will present our data on some plan evaluation for the Elements Spine SRS software. Before we talk about the Spine SRS, for us to better understand the role of spine SRS in the clinical setting, it's good to step back and take a look of oncology in the modern days. So, this is the cancer statistics that's published by American Cancer Society every year.
One of the very exciting findings as you can see here is since 1991, there has been a very steady decline in overall mortality for cancer patient, both in males and in females. And over the past 30 years, that accounts for 29% of decrease and that translates to 2.9 million fewer cancer deaths and those improvements is mainly driven by the overall survival improvement in lung, breast, and prostate cancer. And as you can see, a significant portion of the cancer patients will present or will develop into metastatic disease, so that's really set a background for us to discuss the role of spine SRS for spine metastasis. So, first of all, spine metastasis is common. Bone metastasis is one of the most common area affected by the metastatic disease besides lung and liver, and the incidence of bone metastases is estimated to be around 30% of all cancer patients.
According to the cancer statistics 2020, over 1.8 million new cases will be diagnosed. Of these, lung cancer, breast cancer, and prostate cancer will accounts for almost 700,000 patients. If 30% of them have bone metastases, that will be around 200,000 more. And this was drive by a couple of factors. First of all, as you're showed and I already showed you, the overall survival for cancer patient has improved over the past 30 years, so patients live longer with successful systemic therapy and local treatment. And also, we are facing an aging population, and the incidence of cancer increase with patient age. And obviously, we have a better imaging modality that allow us to detect earlier of the bone involvement. But nonetheless, it's clear that incidence of spine metastasis will continue to increase.
Bone metastases is a clinically important problem because it is very morbid for patient. Pathological fracture can result from the spine involvement, patient will present with pain even without compression fracture, epidural disease can potentially compromise neural structure such as spinal cord and cauda equina and lead to paralysis, and historically, radiation with or without surgery is often offered for the cases for alleviation of the pain and to control or prevent neurological complications. Currently, SRS for spine metastases is very well accepted as a treatment of choice for patient with oligometastases, there are many advantages of using radiosurgery technique for spine metastases. And I don't need to preach to the crier but I want to just emphasize that with SRS technique, we are delivering a definitive dose for local control.
As a result, as supported by extensive literature, our local control is in the range of 85% to 95%. And that is particularly meaningful because we have more data indicating that in the patient who has oligometastases setting, aggressive local control can translate into overall survival benefit, and this is some of the data which has been well presented in our community. With the recognition and acceptance of radiosurgery for spine metastases, utilization for radiosurgery for spine metastasis continue to grow. And this is an NCDB study looking at the utilization of spine radiosurgery from 2004 to 2013. And over 1,000 cases were included in this study and as you can see, the primary site is, not surprising, for lung, kidney, breast, and also prostate. At this time, 80% of the cases are practiced in the academic center.
As you can imagine, this is the beginning of the adoption of Spine SRS. Obviously, the practice patterns significantly changed in the past five years and the spine SRS are readily offered in community centers. Also interesting to see, the most common fractionation pattern is single fraction, almost one-third, but three fraction and five fractions were often utilized as well. And if you look at the overall treatment delivered over the past 10 years from 2004 to 2013, only 2% of the over 1,000 cases were treated in 2004, but in 2013, almost 20% of the cases were treated. So, that's more than a tenfold increase and also, you're looking at the overall adoption of the spine SRS in all the patients with spine metastases. There's also a very clear steady increase of the utilization of 1.4% of all patients to close to 6% of all patients.
So, it is of great importance for us to be able to ensure a high-quality treatment as well as a solution to improve workflow efficiency to adapting for this ever-increasing demand. There are many challenges to implement spine radiosurgery. It's very important to establish a multidiscipline clinic because this involving close cooperation between surgery, either neurosurgery, orthopedic surgery, radiation oncology, medical oncology, radiology, and palliative care. And there are some algorithm has been proposed and the most well known is the NOMS algorithm proposed by Memorial Sloan-Kettering group that provides some guidance to select a proper patient for radiosurgery and that's very well recognized and very well adopted.
From the technical aspects, the most important to start radiosurgery practice is to establish a very reliable and secure way of patient immobilization. Fortunately, our technology really allow us to move into a noninvasive immobilization method for our patient and there are many commercially available systems such as I showed here, BodyFIX system and Accufix system. In our practice, we're using a full-size head and shoulder mask, the Brainlab mask along with custom-made fit net support. It is very important to use custom-made fit net support rather than uniform PET support to secure the disk curvature of the C spine. And for the T spine and the L spine treatment, it's also very, very important and it's strongly recommended for full body immobilization, including the lower extremity.
That really helped to ensure the reproducibility of patient positioning. In our practice, we usually do not use the option lock as well as the compression but the full-body blue bag immobilization is highly, highly recommended. Adequate imaging with high resolution is also important to allow accurate target delineation and dose calculation. For CT simulation, we always use thin cuts and in our practice, we use 1.25-millimeter thickness slices but no more than 1.5 millimeters thickness is recommended, particularly for Monte Carlo calculation. And for MRI scan and the thin cut MRI scan, it's also very important for us to adequate delineate a particular paraspinal involvement as well as the delineation of a spinal cord.
It is often challenging to get axial cuts or forcing cuts in the 1.5-millimeter thickness because the amount of imaging and scanning time. One of the recommendations you maybe consider is using high-resolution SPGR or Stryker sequence to do the sagittal imaging which has a limited number of slices and still will capture the entire acetabular column in high resolution. And the basic imaging including T1 pre, T1 post, and T2 are very important minimum necessary. In a post-op setting, the artifact induced by hardware always make the CT and MRI scan imaging challenging and in many cases, it will be necessary to perform CT myelogram for accurate delineation of the spinal cord.
Imaging fusion is very routinely used in radiation oncology planning such as CNS radiosurgery, however, imaging fusion for spine metastasis is challenging, the reason is because MRI scan and CT scan usually have different patient position and spine curvature. So, even though we have readily available the rigid or deformable registration method available from many software packages, these provided a perfect solution because there's rigid and non-rigid tissue throughout the spinal column including vertebral body and the soft tissue in between. Often for patients using the spinal fusion, we need to define the region of interest focus around the area of target, but if patient has multiple targets such as illustrated here with upper T spine and lower T spine involvement, you may need to do fusion independently to achieve adequate and optimal results.
The Brainlab Elements radiosurgery software does provide a very innovative and perfect solution to address these issues for spine imaging fusion and it utilizes the tissue classification for deformable and non-deformable structures. So, for the vertebral body, it defined as a rigid structure and will not be deformed, but for the soft tissue in between, defined as deformable tissue and will deform to fit the curvature between CT scan and MRI scan. As you can see the illustration here, the red pseudo color indicating the deformation between the imaging sets.
Target delineation is also a major challenge for spine radiosurgery, it is very well recognized that it has high variability in the target definition. In order to address that, there have been multiple publications established in the consortium guideline for both novel treatments including the CT as well as sequence as well as the post-op setting. As you can see, you know, one of the workshop experience showing that even with the availability of the consortium guideline, a short workshop coaching will significantly improve the variability in the target delineation between the audience but this also highlights room for improvement for the target delineation despite the publication available.
Elements Spine Radiosurgery is really helpful clinically in our experience in multiple ways. One is it has the automatic normal tissue segmentation and this is very efficient. And that not only it automatically segment for the normal organs, it also segments all the vertebral body and labels it accordingly. Even though this is very easy, it significantly improves the efficiency of clinical flow and also helps with the accuracy and the safety because most often in the T spine, if the patient's lesion with minimum CT changes, it's very, very important for us to align the proper vertebral body correctly. And this automatic contouring feature along with the numbering of the vertebral body will ensure the accurate localization of the correct vertebral body.
The Smartbrush feature allows the software to automatically generate a CTV once the GTV is defined, and the CTV will be generated according to the International Spine Consortium guideline. And more importantly, this is also a very, very flexible tool because it automatically segments the vertebral body into different segments as recommended by International Spine Consortium guideline. So, based on the physician's decision, you can easily include an extra segment, for example, of the vertebral body but you can include the pedicle if you're concerned that it's involved and need to be treated. And so, it's very customizable and very easy to use.
IGRT is absolutely critical for the accurate treatments and delivery for the spine radiosurgery. Both daily IGRT with intrafraction IGRT is strongly recommended in the needed. And the daily IGRT will help to reduce the intrafraction motion and intrafraction IGRT will minimize the intrafraction movement. And all the modern IGRT system either using cone beam CT or kV system has been evaluated and shown to perform very well and this is one of the study performed by our group looking at the Brainlab ExacTrac system or spine radiosurgery IGRT. As you can see highlighted here, before correction, the 3D vector can easily be over one millimeter, but with correction, the residual error can be kept less than one millimeter, close to the 0.6-0.7 millimeter range. And this has been shown both for T spine, C spine, as well as the lumbar. So, with daily IGRT, we are able to achieve a very high setup certainty in the range of less than one millimeter and less than one degree.
The planning for the radiosurgery or spine SRS is actually particularly challenging, more so than any other SBRT/SRS technique. There are multiple reasons. First of all, the vertebral body structure has many concavities, as illustrated here, and as a result, to try to fit this irregular structure, there's a very high modulation required. And often, because of the spinal cord protection, we need to create a low dose zone in the center of the vertebral body and that requires a very steep dose for us to achieve spinal cord tolerance and this is being very demanding to achieve the optimal modulation. And for value plan-based planning, it really highly depends on a very experienced planner to achieve a high-quality plan and it's very well recognized based on the experience of different planners that plan quality is highly, highly variable.
Because of the challenge to achieve adequate dose around the spinal cord and the epidural space area, that really has been shown in the clinic, this is the main area of failure. And as you can see, with multiple publications looking at hundreds of spine SRS treatment patients, the adjacent vertebral body failure is very rare. However, the majority of the failure pattern, 50% or higher, will happen in epidural space, indicating the challenges to get adequate dose to this area while observing the spinal cord tolerance. As a result, special attention really should be paid to maximize the epidural coverage while observing the spinal cord tolerance. Here I would like to hand over to have my colleague, Dr. Liu. He will further discuss our experience evaluating Elements Spine radiosurgery software for spine metastases.
Bogdan: Thank you for your clinical review, Dr. Shi, and let's hand it over to Dr. Liu.
Dr. Liu: So, good morning, everyone. So, I'll continue with Dr. Shi's presentation and as Dr. Shi introduced, that for spine SRS and SBRT, there is really a very challenging plan due to the concave shape of the spine vertebrae and the optimizer is constrained to spare the concavity surrounding not only the body but also the transverse processes and spinous. The spinal cord interface is thus a challenging trade-off region for any generic VMAT algorithm. We get these Brainlab elements software, so we want to evaluate how these software work. And one of the unique features that we noticed is the so-called PTV splitting. So in this example, due to the complexity of the target between the blue and orange partitions area, Elements will automatically deliver two parts.
The blue arc will deliver dose only to the blue parts, which is the body of the target, and the orange arc will only deliver dose to the transverse process of the target. We selected 20 physician-approved and clinical-treated patients at Jefferson to receive this spine SRS and SBRT. And the clinical plan was done using Eclipse version 11, AAA algorithm, and re-plan using the Elements version 1.5 with the Monte Carlo algorithm turned on for optimization and calculation and the results were retrospectively analyzed. Our plans were normalized such that 95% of PTV received 100% of the prescription dose and they are plan based on Varian Truebeam linear accelerator. The study's results was published on "Frontier in Radiation Oncology" in April 2020.
This is the 20 cases statistics, we have 1 C spine and 2 between C and T, and 6 thoracic spine, 9 L spine, and one sacrum. Target volume range from 15 cc to 150 cc and prescription dose range from 12 gray to 24 gray for a single fraction. We have 10 SRS patients in the list and 7 patients with 3 factions either receive 24 gray or 27 gray, and we also have 3 patients received 5 fractions total of 30 gray. And four of the plans using 6 MV Flattening-Filter-Free mode and others with the standard mode, 6 MV. Elements Spine SRS software uses a preset template for arc geometry. So, this is the only arc geometry that we use in this study, which is two full rotation coplanar arcs, one arc from 185 to 175, which is 350-degree coverage and at 45-degree chronometer angle, and the other one at 315-degree chronometer angle, and both at table zero.
So, these are templates, in the study we didn't change at all, but in the actual planning you can put in the number as you wish to provide more flexibility. And also on the lower part of the two pictures that you can see, which show you the automatic splitting of the PTV into two parts or three parts, which end up with either four arcs or six arcs respectively. The software also use the preset templates for prescription dose and dose and dose constraints for easy plan setup and to achieve a consistent plan. We have three templates used during this study for our one, three, and five fractions, so all the PTV coverage is normalized so that 95% of volume is covered by the prescription dose. And for spinal cord, we have the maximum point dose which is defined as 0.035 cc, is set to be less than 14 gray, 22.5, and 28 gray for one, three, and five fractions. And another point is the 0.35 cc volume dose is set to be less than 10, 16, and 22 gray respectively.
And both these prescription dose and constraints can be adjusted in the plan to provide some flexibility. For example, in the right-hand picture, you can see although the templates show you three fraction, 27 gray, where here we have 24 gray prescription. It also displays the actual values of each point gose of the constraints in the preset and use a traffic light fashion to tell the user if it either meet or violating or even marginal to compare to the objective. So, for example, on the left side of the radiosurgery plan, we have 10 gray here and we have 9.9 gray in actual dose. In 14 gray objective, we achieved the 13.6, and the max dose which is a pixel, we turn it off, and so this one, 17 exceed the 16, so it's in red.
So, there are the available user interfaces in Elements software for plan evaluation. On the left top corner is a so-called overview. It provides isodose distribution in three orthogonal views and the DVH of all the structures and it also listed the CI and GI values and the total monitor units of the plan. On the right top corner is the 3D view, it not only displays a DVH, but also a three-dimensional rendering of the planning image with overlaid MLC shape on top. On the left bottom corner is the beam's eye view. It displays the MLC shape of any selected arc in a step size of about every 20-degree. The last one in the so-called gallery view, which displayed isodose distribution for each slice for user-selected views either axial, coronal, or sagittal.
So, let's see some examples of the Elements plan compared to our clinical plan. So, example one is a T11 radiosurgery, 24 gray prescription, and the distance between the PTV and the cord in the AP direction is about five-millimeter. So, Elements plan does not have a PTV splitting, so it only has two arms. And after the Elements plan, we export the plan and dose to Eclipse so that we can pull them together side by side to show you the comparison. So, on the left-hand side is the Eclipse plan, on the right-hand side is the Elements plan, and in the middle is the DVH combination of PTV and the cord. So, at 24 gray prescription, they are coincidence in 95% volume, and also we can see that the Elements plan is much harder inside PTV and then the cauda, the Elements plan reduce the cauda from 9.9 gray to 8.7 gray.
If we drag a line along the AP direction passing through the PTV and cord to a one-dimensional dose profile as shown here, so the red line is the Elements plan which shows a steeper dose fall off. So, these dose fall off along AP direction is 2.8 gray per millimeter for Elements plan or 2.3 gray per millimeter in clinical plan. So, next example is a T2 spine and 24 gray in three fractions, the gap between the PTV and cord in the AP direction is 4-millimeter. And you can see from this picture, the right top corner, the Elements plan splitting ones and using four arcs. In order to see how this PTV splitting works, we've transferred the plan into Eclipse and in Eclipse, we can actually turn off each individual arc so we can see the dose distribution from each group.
So, as you can see from this picture, Group 1, Arc 1 and 2, mostly treating the left side of the PTV, and Group 2, Arc 3 and 4, mostly concentrated on the right side of the PTV. And this is a side-by-side plan comparison, the left side is the Eclipse and the right side is the Elements. In the clinical plan, the cord max dose is 11 gray and Elements reduced it to 9.2. And the dose fall-off in this picture, it shows that the Elements plan is 3.1 gray per millimeter compared to 2.7 gray per millimeter in Eclipse plan which has a steeper dose fall off. And next example is an L3 radiosurgery, 16 gray prescription. As you can see here, this one not only including the body but also including the right-side pedicle and left-side transverse process. So, because of the complexity of the target, Elements split the PTV twice, so it ended up with six arcs, as you can see from this right top picture.
So, these ones showed the Elements plan was mapped to a body function in the patient's specific QA module in Elements, so that in this module, monitor units of each arc can be turned on or turned off so that we can check the dose distribution for each arc. And in this example, we can see that Arc 3 and 4 are mostly treating the body of the spine, Arc 1 and 2 mostly treating the left side transverse process, and Arc 5 and 6 treat the right-side pedicle and these are the PTV splitting illustration. And then this is the side-by-side comparison, again, left side clinical plan, right side Elements plan. And while we drag the AP line, we can see this line goes through the body and then the cord region and then the transverse processes, the same thing as we see here.
So, the Elements plan have a much steeper dose fall off, it's 1.5 gray per millimeter compared to only 0.6 gray per millimeter in clinical plan. And also, in this case, you can see cord max dose reduced from 15.4 gray to 12.6 gray. And this is an example that we use in our publication, this is a T11 SBRT, 27 gray in three fraction, and theses isodose and DVH comparisons are displayed when we transfer both plants into Eclispe. And the gap here is 4.8 gray and the dose fall-off in this region is 1.1 gray per millimeter in clinical plan and 2.6 gray per millimeter in Elements and max cord dose is 17.5 clinical versus 15.4 Elements. We do see that the D5% PTV dose is 31 gray versus 34 gray in Elements which is corresponding to a harder inside, and also the CI reduced from 1.4 to 1.1 and GI reduced from 4.4 to 3.8.
If we look at a zoomed-in view of this dose distribution, we can see that through a spyglass, the outside is Elements plan and the inside is clinical plan. Prescription dose, isodose line in Elements plan covers more of the target in close proximity to the spinal cord than the clinical plan while also create a ring of 50% isodose line around the spinal cord, while the same isodose line in the clinical plan completely engulfs the spinal cord. So, this is a full list of the results show all 20 cases for each plan. We have CI, GI, max cord dose, and min cord dose. And the analysis of CI and GI, we can see that on average, the CI in Elements was reduced from 1.25 to 1.1 and the GI was reduced from 4.2 to 3.5. So, less CI and GI means more conformal plan and less intermediate dose to the surrounding tissue.
And also we noticed that not only the absolute value of CI and G is reduced, but also the standard deviation is also reduced. So, you can see the CI standard deviation, 0.16, reduced to 0.04 in Elements and GI 0.7 reduce to 0.3 in Elements, and this less standard deviation which means we achieved a more consistent plan within the 20 cases. And this is the max cord dose...both max and min cord dose significantly lower in Elements plan, max was reduced from 14.6 to 11 and min reduced from 6 to 4.3. If we take the ratio of the dose of the Elements plan divided by the dose of the clinical plan, then we got 78% and 73% for max and min respectively, which means the Elements plans reduce both max and min cord dose by about a quarter.
The next comparison is the total monitor unit and the modulation factor because the 20 cases have a different prescription dose, so if we only compare monitor units, maybe we have confusion then. So, we divided the total monitor units divided by the total prescription dose and we define it as a modulation factor. Basically, it's normalized with the prescription dose. And for both total MU and modulation factor, Elements plan is higher, so it's more monitor units and more modulation. So, further, we divided all 20 cases to different PTV splitting scenarios, we want to see if PTV splitting costs the higher MU or modulation factor and the result is showing us, yes. And without the PTV splitting, the modulation factor only increased by about 2% from 3.25 to 3.31, with PTV splitting once, the modulation factor increased from 3.56 to 3.94, so it represents 11% increase. With PTV splitting twice, the modulation factor increased about 30%.
So, therefore, we can see that, comparing to our clinical plan, the Elements may achieve better dosimetry, however, it at the expense of more number of arcs, more monitor units, and thus the longer delivery time. And we also stratify the results by the number of factions and the trends consistent with the overall data sets that were observed, that Elements plans have significantly better CI, GI, max cord dose for single fraction plans. Four three and five action plans, although we observe the similar trends, but we believe since we only have a limited number of cases, we still need more cases to show statistical significance. We also compare the pencil beam algorithm versus Monte Carlo. So, Elements utilize a pencil beam algorithm for initial optimization and dose calculation and then allow for optional final multicolor optimization and dose calculation. All plans in this study were analyzed after the final Monte Carlo dose calculation.
And in order to evaluate the dosimetry change with the results of the Monte Carlo algorithm after the pencil beam calculation, we have to calculate the plan using only pencil beam and then we transferred the plan into the so-called RTQA recalculation and recalculated with the Monte Carlo algorithm using exactly the same plan. And after that, we don't renormalize the PTV coverage, we just get it applied as-is. So, this is an example of T5 lesion with a large amount of lung in the beam path and we can see the PTV coverage, it was 95% because that was what we normalized, and it reduced to 91% after Monte Carlo recalculation, and the cord max dose increased from 6 gray to 7 gray, that's about a 15% increase. And then we calculate another one which is a T11 lesion with very little lung in the beam path and the PTV coverage actually increased from 95% to 96% and max cord dose is reduced from 9.3 to 8.7 gray. So, that tells us that for a very homogeneous tissue region, the user may decide to use pencil beam only to save the planning time.
So, the benefits of Monte Carlo optimization, it has more accurate dose calculation, especially for thoracic spine, the interface between air and bone, and paraspinal and regions with spinal implants. And sometimes, even for cases where there is not a large amount of heterogeneity, it still benefits using Monte Carlo but the cost is longer planning time. So, last, we try to show you the deliverability and QA, patient-specific QA of the two plans. So, this is a summary of the results and we have two patients plan delivered. Case number one is a simple case, only have two arcs with similar modulation factor between Elements plan and clinical plan and they all get a 95% and above gamma passing rates and we used a 3% dose difference, one millimeter DTA, and 10% lower dose threshold for the gamma calculation.
And the second one, the results show you a six arc, which is here, as you can see the six arcs around Delta4 Phantom, and six arcs here. And this is the result and the red line shows you the planned calculated dose and the green dots represent the measured dose and in this six-arc plan, we achieved almost 99% gamma passing rate. So, in conclusion, the Elements Spine SRS plans were better than the current clinical plan in achieving higher conformity, decreasing gradient index, and sparing dose to the spinal cord but it will cost us more number of arcs and longer delivery time. The automation in Elements was shown to generate more consistent plans with quality as good as or better than the current plan.
The average calculation time for the 20 cases was about 28 minutes with Monte Carlo optimization and calculation turned on, however, it also with a large variation depends on the target complexity and PTV splitting times. For example, simple case without PTV splitting, we may gather a plan in 10 minutes, if we get a very complicated case, it may go up to 1 hour to the calculation. Although there is other algorithm and planning system that can also utilize templates to reduce the manual effort, from our initial experience with this Elements VMAT implementation for spine and we can see it's specific to creating a steep dose gradient towards the spinal cord and that can help reduce the probability of one of the primary failures of the spine SRS and SBRT. Thank you.
Bogdan: Thank you, Dr. Shi. Thank you, Dr. Liu. Let's see if I can bring both of you back online to answer some questions. Haisong, you may need to re-enable your camera so we can see you. Maybe I'll start with you, Dr. Shi. So, we polled the audience today on a few topics and regarding treatment, it seems that both external beam as well as single fraction, although utilized, have the lowest numbers. In your practice, what do you consider to be the challenges in offering single fraction radiosurgery to more patients? And how would you think that that particular treatment strategy can be expanded? Nope, we don't actually hear you.
Dr. Shi: So, I think, you know, we have briefly discussed patient selection criteria and right now, we still follow the NOMS algorithm as a primary guideline to select the patients. Obviously, as mentioned earlier, to try to establish a practice for radiosurgery, you really need to have a very close that work together multidisciplinary team because, you know, this is really required inputs from multiple disciplines between neurosurgery, orthopedic surgery, medical oncology, radiology, and based on the patient's status and figure out what is the ideal treatment. Currently, our preference is to use radiosurgery SBRT for patients with oligometastases patient. We have a strong preference for the patient who has pathology that shows relatively radioresistance, for example, renal cell carcinoma or GI primary. Those patient's tumor require a higher dose, so low dose palliation radiation treatment may not have adequate clinical durable benefits. And I think for a patient with a more favorable pathology, for example, breast cancer even with metastatic disease, they tend to have longer survival, those are also patient we have a strong preference to use radiosurgery or SBRT techniques.
Bogdan: Okay, great. Next question was more looking at referral patterns and what could contribute to an enlargement of a spine radiosurgery program and both medical oncologists and neurosurgeons were weighted equally in terms of who should be better targeted in order to grow the referrals. What's your opinion on that and then how does it work in your practice...?
Dr. Shi: Yeah, definitely different between academic practice and community practice, I know, but, for me, a multidisciplinary tumor board will be really helpful. And this is what we do on...so we have a group of people who are able to review all cases and try to [inaudible 00:40:51] the disposition of patients. And in a community practice, I think the tumor board setting is very common for most hospitals, so definitely, this is the main area of focus, so really try to, you know, present the information, share the knowledge, and educate other specialty physicians as we see that we have lots of evidence. Right now, it's very well established, local control is excellent with SBRT technique and we know that in a set of patients, SBRT and this type of aggressive local treatment is very, very meaningful not only for disease local control but also in overall survival setting for those patients.
So, share this information, really bring this knowledge to all the referring physicians and, you know, obviously, make sure they're aware of the technology available. As we also mentioned, you know, this technology is very challenging on a technical point, so it really requires team efforts, and the radiation oncology team really need to have a very high quality and safety procedure and also efficiency. That's why, you know, to implement some of the software, it really ate the quality assurance and also the efficiency because of patient volume, as mentioned early, with increasing incidence and ever-growing cancer patient population and the longer survival, we do anticipate that more and more patient will be eligible or SBRT/SRS surgery or radiosurgery type of procedure. And we really need to have the capacity to absorb more case load and so an automated solution will be tremendously useful.
Bogdan: Great. Well, speaking of technical challenges, Haisong, we have a lot of technical questions for you, so I'll try to cluster some of the questions. But let's maybe start with selection of energy for your treatment plans. We had a question pertaining to why do you still use the regular 6x beam when the Flattening-Filter-Free 6x mode has a sharper gradient? Yeah, we can't hear you, Haisong, so let's try that again. Make sure your mic is on.
Dr. Liu: Can you hear me now?
Bogdan: Okay, I can hear you now.
Dr. Liu: Okay. So, in terms of the entity selection, actually, we didn't select any of the energy because we try to compare the Elements play with clinical plan. And so, these energies, either 6MV or 6 Flattening-Filter-Free are all coming from the original energy selection of the clinical plan. So, these clinical plans are based on our older version of Eclipse, version 11, and at that time...the clinical cases are from 2016 to 2017 and at that time, we only start to utilize 6FFF occasionally, not routinely. And I believe we're utilizing the Flattening-Filter-Free energy mode more and more, but this is just the original energy selection.
Bogdan: Okay, next question, or questions, should be regarding more PTV selections, so, obviously, you're showing in your example a lot of continuous spine PTVs. How does Elements handle non-continuous PTVs? One question. And the second question, "Do you use PRVs for your treatment planning?"
Dr. Liu: So, as far as we noticed, the current version of Elements, which is 1.5 can only plan the continuous despite target, either the single or if it's connected. For example, we have example cases from T2 to T2, but if the spines are separated, either T7 and L3, I'm not aware that...we haven't tried that case, I'm not sure if we can plan. Maybe Bogdan you can give more of the update on this question.
Bogdan: So, yeah, you're absolutely right. The current version, which is version 1.5 allows you to create essentially a one isocenter plan, so if you have multiple targets, you would have to create two plans. The next version coming out by the end of the year, the 2.0 version, would allow you to do that in an integrated plan and see the overall dose. So, the ability to treat non-continuous spine levels will come with the next version to make it a little bit more streamlined for everybody. Next question in regard to the selection of arcs and it may be worthwhile to say that the dosimetry comparison was in place to mainly check if the automation generated by Elements was adequate for the treatments. But there's a question pertaining to the fact that some plans done in the Eclipse, so your clinical plans have two arcs, but there were six arcs in Elements because of PTV splitting, what is the delivery time difference between those two plans?
Dr. Liu: Yeah, so, again, the study was not aimed for compare Elements versus Eclipse because the purpose is to say evaluation of Elements can give us adequate clinical approved plan, acceptable plan, and so that is mostly compared to our clinical-accepted plan and that clinical-accepted plan is happened to be planned with Eclipse and at that time, it's the older version. And right now we have a newer version 15.6 with a different optimization algorithm, it's a PO instead of the TRO in the older version, and also the newer version, we have more functionalities like MCO and other features that we can use.
And so if we really want to do apple to apple comparison, then we may want to transfer all the same arc geometry from Elements plan to Eclipse and lose the constraints of the maximum dose so that we allow the Eclipse to kind of give us more maximum dose inside the PTV and see what Eclipse new algorithm can give us and we can definitely do that in the near future.
Bogdan: Okay, there's a question regarding monitor units, and I think you may have answered this so the question may have been asked before you had a chance to cover it, but what difference in monitor units do you see between the Elements plans and the Eclipse plans?
Dr. Liu: So, I showed a table for comparing the monitor units or modulation factor for different scenarios of PTV splitting. And what we found is that without PTV splitting, basically compare two arcs versus two arcs, that with the Elements, the modulation factor only increased by about 2%. But if the PTV splitting once, then the modulation factor increased by 10%-11%. And if the PTV splitting twice, then the modulating factor increased by about 30%, so more monitor units to deliver, and also more number of arcs, and definitely, it will be a longer delivery time. So, that's why we say that. And we can say that the Elements give us a better dosimetry at some cost and the cost is the longer delivery time and more modulation.
Bogdan: We have some questions regarding the dose engines, particularly Monte Carlo. So, how long does a Monte Carlo calculation take and what kind of dose resolution do you use? And would you recommend dose-to-medium or dose-to-water for Monte Carlo calculations? Do you have any experience with how our Monte Carlo compares to Monaco?
Dr. Liu: So, we did the time the difference between run the power with pencil beam only and with Monte Carlo turned on, and I think the average with Monte Carlo turned on is about half an hour, like 28 minutes, but with the pencil beam only, the average is about 10 minutes. But as I said in the conclusion part, the averages does have a large variation. For some simple cases, the calculation even with Monte Carlo turned on, it can be done in 15-20 minutes, but some complicated cases, especially if there are six arcs and two PTV splittings, they will sometimes run about one hour, one and half hour.
Bogdan: And I don't think you've done comparisons to Monaco, I don't know that you have the Elekta software, but for those of you interested, we have presented information on Monte Carlo comparisons with another group in San Antonio, so you can review those results. Regarding QA, Haisong, and gamma calculations, are you looking at local or global gamma values?
Dr. Liu: I believe what we did is the maximum dose at 100% and it's a global gamma instead of local.
Bogdan: And I did miss a question regarding Monte Carlo, what materials are supported by the Monte Carlo dose engine? Would it expect surgical hardware, etc.? I'm not sure if you can answer that question, Haisong. If not, we can provide technical details to the person who asked the question. We have some questions regarding IGRT and maybe both of you can answer this. "Do you utilize online IGRT to monitor intrafraction deviations to guarantee the high precision delivery and overall, for your spine radiosurgery patients, how frequently do you monitor and what do you use?"
Dr. Liu: So, from our previous study, we did notice that the spine, especially the neck region or the T spine region, they tend to...the patients tend to, you know, move during the treatment and that can be one half to two-millimeter. And so, we set up our institutional policy that we will take the ExacTrac verification and make the necessary adjustment before each arc start. And also, if the arc period takes about one minute or two, and when the arc comes to any orthogonal...when the gantry comes to any orthogonal position, either 90, 0, or 270, we can throw in some snap verification just to make sure the patient as stable as possible.
Bogdan: Question regarding modulation complexity and Haisong, I think you've defined your own version of that, the software also gives you a number that we call a modulation complexity score. So, it's kind of a two-part question, what kind of ranges do you see for that modulation complexity score in the plans generated for your 20 case reviews? And how is the QA impacted by an increasing modulation complexity score?
Dr. Liu: Sorry, Bogdan, I didn't hear you.
Bogdan: Okay, so the software gives you a number that provides a modulation complexity score, have you monitored the variation in that number or what kind of range do you see for your 20 evaluated plans? And how does the modulation complexity score impact QA?
Dr. Liu: Yeah, I'm browsing back to see where the questions are. I couldn't hear you, sorry.
Bogdan: Okay, so maybe look in the chat line, the last question from Matt.
Dr. Liu: What dosimetry is used to measure for gamma index? So, the phantom we used is ScandiDos Delta4. The average MCS value?
Bogdan: Yeah.
Dr. Liu: So, I think the MCS value given by Elements itself is kind of confusing to us, so that's why we kind of defined it ourselves, just simply divided the total MU by the total prescription dose and it's about 3.3 for no PTV splitting, 3.5 for splitting once, and over 4 for splitting twice. I'm not so sure about how Elements defined MCS value, so that's why in our study and publication, we didn't use that value at all.
Bogdan: Okay, great. Well, again, Dr. Shi, Dr. Liu, thank you very much for your presentations today and thank you all for joining and we look forward to seeing you at our upcoming webinar. Thank you and have a great day.
Dr. Shi: Thank you.
Dr. Shi: Thank you.
As always, we are providing CE credits upon successful completion of the webinar, you may request credits by emailing us at info@novaliscircle.org and we can provide you with details for CAMPEP, MDCB, and ASRT. Also, please don't forget to sign up for our upcoming webinars. The next one will review clinical utilization for our latest version of Elements Multiple Brain Mets SRS with a clinical team from Alliance Cancer Care in Huntsville, Alabama, and that will be on June 25th. In order to properly visualize the webinar, please remember to use Google Chrome or Safari.
And should you have questions for our two presenters, please utilize the chat functionality throughout the lecture, and we will answer all your questions upon completion of the two talks. We would also like to get feedback from you today on some key issues via our polling infrastructure, so we will let you know when those polls are open. And as always, if you'd like to follow us on social media, you can do that with the Novalis Circle hashtag. With that said, I'd like to turn it over to Dr. Shi and he'll cover a clinical space for spine radiosurgery. Dr. Shi?
Dr. Shi: Thank you, Bogdan. First of all, it's my great pleasure to be here along with my colleague, Dr. Haisong Liu, to present evaluation of Elements Spine SRS plan quality for radiosurgery and SBRT for the treatment of spine metastases. We highly appreciate Brainlab to provide this opportunity. And just a quick note regarding the radiosurgery SRS and sbrt. Conventionally, many people refer for extracranial SRS as SBRT. But for CNS point of view, any single fraction treatments will be referred as SRS for both intracranial and actual cranial disease, and for two to five fraction will be referred as SBRT and this will be used for this presentation. This is my introductory disclosures. In the prior webinar by the Brainlab in May, Dr. Alongi presents excellent overview of the background for spine radiosurgery and also his experience using Elements Spine SRS.
And I encourage everyone to listen to that if you missed it, and it's available through the Nova Circle website. Today, I would like to kick off the webinar basically looking at some of the background information for spine radiosurgery and specifically focusing on some clinical considerations and challenges of using SBRT/SRS technique for spine metastases. And afterward, Dr. Liu will present our data on some plan evaluation for the Elements Spine SRS software. Before we talk about the Spine SRS, for us to better understand the role of spine SRS in the clinical setting, it's good to step back and take a look of oncology in the modern days. So, this is the cancer statistics that's published by American Cancer Society every year.
One of the very exciting findings as you can see here is since 1991, there has been a very steady decline in overall mortality for cancer patient, both in males and in females. And over the past 30 years, that accounts for 29% of decrease and that translates to 2.9 million fewer cancer deaths and those improvements is mainly driven by the overall survival improvement in lung, breast, and prostate cancer. And as you can see, a significant portion of the cancer patients will present or will develop into metastatic disease, so that's really set a background for us to discuss the role of spine SRS for spine metastasis. So, first of all, spine metastasis is common. Bone metastasis is one of the most common area affected by the metastatic disease besides lung and liver, and the incidence of bone metastases is estimated to be around 30% of all cancer patients.
According to the cancer statistics 2020, over 1.8 million new cases will be diagnosed. Of these, lung cancer, breast cancer, and prostate cancer will accounts for almost 700,000 patients. If 30% of them have bone metastases, that will be around 200,000 more. And this was drive by a couple of factors. First of all, as you're showed and I already showed you, the overall survival for cancer patient has improved over the past 30 years, so patients live longer with successful systemic therapy and local treatment. And also, we are facing an aging population, and the incidence of cancer increase with patient age. And obviously, we have a better imaging modality that allow us to detect earlier of the bone involvement. But nonetheless, it's clear that incidence of spine metastasis will continue to increase.
Bone metastases is a clinically important problem because it is very morbid for patient. Pathological fracture can result from the spine involvement, patient will present with pain even without compression fracture, epidural disease can potentially compromise neural structure such as spinal cord and cauda equina and lead to paralysis, and historically, radiation with or without surgery is often offered for the cases for alleviation of the pain and to control or prevent neurological complications. Currently, SRS for spine metastases is very well accepted as a treatment of choice for patient with oligometastases, there are many advantages of using radiosurgery technique for spine metastases. And I don't need to preach to the crier but I want to just emphasize that with SRS technique, we are delivering a definitive dose for local control.
As a result, as supported by extensive literature, our local control is in the range of 85% to 95%. And that is particularly meaningful because we have more data indicating that in the patient who has oligometastases setting, aggressive local control can translate into overall survival benefit, and this is some of the data which has been well presented in our community. With the recognition and acceptance of radiosurgery for spine metastases, utilization for radiosurgery for spine metastasis continue to grow. And this is an NCDB study looking at the utilization of spine radiosurgery from 2004 to 2013. And over 1,000 cases were included in this study and as you can see, the primary site is, not surprising, for lung, kidney, breast, and also prostate. At this time, 80% of the cases are practiced in the academic center.
As you can imagine, this is the beginning of the adoption of Spine SRS. Obviously, the practice patterns significantly changed in the past five years and the spine SRS are readily offered in community centers. Also interesting to see, the most common fractionation pattern is single fraction, almost one-third, but three fraction and five fractions were often utilized as well. And if you look at the overall treatment delivered over the past 10 years from 2004 to 2013, only 2% of the over 1,000 cases were treated in 2004, but in 2013, almost 20% of the cases were treated. So, that's more than a tenfold increase and also, you're looking at the overall adoption of the spine SRS in all the patients with spine metastases. There's also a very clear steady increase of the utilization of 1.4% of all patients to close to 6% of all patients.
So, it is of great importance for us to be able to ensure a high-quality treatment as well as a solution to improve workflow efficiency to adapting for this ever-increasing demand. There are many challenges to implement spine radiosurgery. It's very important to establish a multidiscipline clinic because this involving close cooperation between surgery, either neurosurgery, orthopedic surgery, radiation oncology, medical oncology, radiology, and palliative care. And there are some algorithm has been proposed and the most well known is the NOMS algorithm proposed by Memorial Sloan-Kettering group that provides some guidance to select a proper patient for radiosurgery and that's very well recognized and very well adopted.
From the technical aspects, the most important to start radiosurgery practice is to establish a very reliable and secure way of patient immobilization. Fortunately, our technology really allow us to move into a noninvasive immobilization method for our patient and there are many commercially available systems such as I showed here, BodyFIX system and Accufix system. In our practice, we're using a full-size head and shoulder mask, the Brainlab mask along with custom-made fit net support. It is very important to use custom-made fit net support rather than uniform PET support to secure the disk curvature of the C spine. And for the T spine and the L spine treatment, it's also very, very important and it's strongly recommended for full body immobilization, including the lower extremity.
That really helped to ensure the reproducibility of patient positioning. In our practice, we usually do not use the option lock as well as the compression but the full-body blue bag immobilization is highly, highly recommended. Adequate imaging with high resolution is also important to allow accurate target delineation and dose calculation. For CT simulation, we always use thin cuts and in our practice, we use 1.25-millimeter thickness slices but no more than 1.5 millimeters thickness is recommended, particularly for Monte Carlo calculation. And for MRI scan and the thin cut MRI scan, it's also very important for us to adequate delineate a particular paraspinal involvement as well as the delineation of a spinal cord.
It is often challenging to get axial cuts or forcing cuts in the 1.5-millimeter thickness because the amount of imaging and scanning time. One of the recommendations you maybe consider is using high-resolution SPGR or Stryker sequence to do the sagittal imaging which has a limited number of slices and still will capture the entire acetabular column in high resolution. And the basic imaging including T1 pre, T1 post, and T2 are very important minimum necessary. In a post-op setting, the artifact induced by hardware always make the CT and MRI scan imaging challenging and in many cases, it will be necessary to perform CT myelogram for accurate delineation of the spinal cord.
Imaging fusion is very routinely used in radiation oncology planning such as CNS radiosurgery, however, imaging fusion for spine metastasis is challenging, the reason is because MRI scan and CT scan usually have different patient position and spine curvature. So, even though we have readily available the rigid or deformable registration method available from many software packages, these provided a perfect solution because there's rigid and non-rigid tissue throughout the spinal column including vertebral body and the soft tissue in between. Often for patients using the spinal fusion, we need to define the region of interest focus around the area of target, but if patient has multiple targets such as illustrated here with upper T spine and lower T spine involvement, you may need to do fusion independently to achieve adequate and optimal results.
The Brainlab Elements radiosurgery software does provide a very innovative and perfect solution to address these issues for spine imaging fusion and it utilizes the tissue classification for deformable and non-deformable structures. So, for the vertebral body, it defined as a rigid structure and will not be deformed, but for the soft tissue in between, defined as deformable tissue and will deform to fit the curvature between CT scan and MRI scan. As you can see the illustration here, the red pseudo color indicating the deformation between the imaging sets.
Target delineation is also a major challenge for spine radiosurgery, it is very well recognized that it has high variability in the target definition. In order to address that, there have been multiple publications established in the consortium guideline for both novel treatments including the CT as well as sequence as well as the post-op setting. As you can see, you know, one of the workshop experience showing that even with the availability of the consortium guideline, a short workshop coaching will significantly improve the variability in the target delineation between the audience but this also highlights room for improvement for the target delineation despite the publication available.
Elements Spine Radiosurgery is really helpful clinically in our experience in multiple ways. One is it has the automatic normal tissue segmentation and this is very efficient. And that not only it automatically segment for the normal organs, it also segments all the vertebral body and labels it accordingly. Even though this is very easy, it significantly improves the efficiency of clinical flow and also helps with the accuracy and the safety because most often in the T spine, if the patient's lesion with minimum CT changes, it's very, very important for us to align the proper vertebral body correctly. And this automatic contouring feature along with the numbering of the vertebral body will ensure the accurate localization of the correct vertebral body.
The Smartbrush feature allows the software to automatically generate a CTV once the GTV is defined, and the CTV will be generated according to the International Spine Consortium guideline. And more importantly, this is also a very, very flexible tool because it automatically segments the vertebral body into different segments as recommended by International Spine Consortium guideline. So, based on the physician's decision, you can easily include an extra segment, for example, of the vertebral body but you can include the pedicle if you're concerned that it's involved and need to be treated. And so, it's very customizable and very easy to use.
IGRT is absolutely critical for the accurate treatments and delivery for the spine radiosurgery. Both daily IGRT with intrafraction IGRT is strongly recommended in the needed. And the daily IGRT will help to reduce the intrafraction motion and intrafraction IGRT will minimize the intrafraction movement. And all the modern IGRT system either using cone beam CT or kV system has been evaluated and shown to perform very well and this is one of the study performed by our group looking at the Brainlab ExacTrac system or spine radiosurgery IGRT. As you can see highlighted here, before correction, the 3D vector can easily be over one millimeter, but with correction, the residual error can be kept less than one millimeter, close to the 0.6-0.7 millimeter range. And this has been shown both for T spine, C spine, as well as the lumbar. So, with daily IGRT, we are able to achieve a very high setup certainty in the range of less than one millimeter and less than one degree.
The planning for the radiosurgery or spine SRS is actually particularly challenging, more so than any other SBRT/SRS technique. There are multiple reasons. First of all, the vertebral body structure has many concavities, as illustrated here, and as a result, to try to fit this irregular structure, there's a very high modulation required. And often, because of the spinal cord protection, we need to create a low dose zone in the center of the vertebral body and that requires a very steep dose for us to achieve spinal cord tolerance and this is being very demanding to achieve the optimal modulation. And for value plan-based planning, it really highly depends on a very experienced planner to achieve a high-quality plan and it's very well recognized based on the experience of different planners that plan quality is highly, highly variable.
Because of the challenge to achieve adequate dose around the spinal cord and the epidural space area, that really has been shown in the clinic, this is the main area of failure. And as you can see, with multiple publications looking at hundreds of spine SRS treatment patients, the adjacent vertebral body failure is very rare. However, the majority of the failure pattern, 50% or higher, will happen in epidural space, indicating the challenges to get adequate dose to this area while observing the spinal cord tolerance. As a result, special attention really should be paid to maximize the epidural coverage while observing the spinal cord tolerance. Here I would like to hand over to have my colleague, Dr. Liu. He will further discuss our experience evaluating Elements Spine radiosurgery software for spine metastases.
Bogdan: Thank you for your clinical review, Dr. Shi, and let's hand it over to Dr. Liu.
Dr. Liu: So, good morning, everyone. So, I'll continue with Dr. Shi's presentation and as Dr. Shi introduced, that for spine SRS and SBRT, there is really a very challenging plan due to the concave shape of the spine vertebrae and the optimizer is constrained to spare the concavity surrounding not only the body but also the transverse processes and spinous. The spinal cord interface is thus a challenging trade-off region for any generic VMAT algorithm. We get these Brainlab elements software, so we want to evaluate how these software work. And one of the unique features that we noticed is the so-called PTV splitting. So in this example, due to the complexity of the target between the blue and orange partitions area, Elements will automatically deliver two parts.
The blue arc will deliver dose only to the blue parts, which is the body of the target, and the orange arc will only deliver dose to the transverse process of the target. We selected 20 physician-approved and clinical-treated patients at Jefferson to receive this spine SRS and SBRT. And the clinical plan was done using Eclipse version 11, AAA algorithm, and re-plan using the Elements version 1.5 with the Monte Carlo algorithm turned on for optimization and calculation and the results were retrospectively analyzed. Our plans were normalized such that 95% of PTV received 100% of the prescription dose and they are plan based on Varian Truebeam linear accelerator. The study's results was published on "Frontier in Radiation Oncology" in April 2020.
This is the 20 cases statistics, we have 1 C spine and 2 between C and T, and 6 thoracic spine, 9 L spine, and one sacrum. Target volume range from 15 cc to 150 cc and prescription dose range from 12 gray to 24 gray for a single fraction. We have 10 SRS patients in the list and 7 patients with 3 factions either receive 24 gray or 27 gray, and we also have 3 patients received 5 fractions total of 30 gray. And four of the plans using 6 MV Flattening-Filter-Free mode and others with the standard mode, 6 MV. Elements Spine SRS software uses a preset template for arc geometry. So, this is the only arc geometry that we use in this study, which is two full rotation coplanar arcs, one arc from 185 to 175, which is 350-degree coverage and at 45-degree chronometer angle, and the other one at 315-degree chronometer angle, and both at table zero.
So, these are templates, in the study we didn't change at all, but in the actual planning you can put in the number as you wish to provide more flexibility. And also on the lower part of the two pictures that you can see, which show you the automatic splitting of the PTV into two parts or three parts, which end up with either four arcs or six arcs respectively. The software also use the preset templates for prescription dose and dose and dose constraints for easy plan setup and to achieve a consistent plan. We have three templates used during this study for our one, three, and five fractions, so all the PTV coverage is normalized so that 95% of volume is covered by the prescription dose. And for spinal cord, we have the maximum point dose which is defined as 0.035 cc, is set to be less than 14 gray, 22.5, and 28 gray for one, three, and five fractions. And another point is the 0.35 cc volume dose is set to be less than 10, 16, and 22 gray respectively.
And both these prescription dose and constraints can be adjusted in the plan to provide some flexibility. For example, in the right-hand picture, you can see although the templates show you three fraction, 27 gray, where here we have 24 gray prescription. It also displays the actual values of each point gose of the constraints in the preset and use a traffic light fashion to tell the user if it either meet or violating or even marginal to compare to the objective. So, for example, on the left side of the radiosurgery plan, we have 10 gray here and we have 9.9 gray in actual dose. In 14 gray objective, we achieved the 13.6, and the max dose which is a pixel, we turn it off, and so this one, 17 exceed the 16, so it's in red.
So, there are the available user interfaces in Elements software for plan evaluation. On the left top corner is a so-called overview. It provides isodose distribution in three orthogonal views and the DVH of all the structures and it also listed the CI and GI values and the total monitor units of the plan. On the right top corner is the 3D view, it not only displays a DVH, but also a three-dimensional rendering of the planning image with overlaid MLC shape on top. On the left bottom corner is the beam's eye view. It displays the MLC shape of any selected arc in a step size of about every 20-degree. The last one in the so-called gallery view, which displayed isodose distribution for each slice for user-selected views either axial, coronal, or sagittal.
So, let's see some examples of the Elements plan compared to our clinical plan. So, example one is a T11 radiosurgery, 24 gray prescription, and the distance between the PTV and the cord in the AP direction is about five-millimeter. So, Elements plan does not have a PTV splitting, so it only has two arms. And after the Elements plan, we export the plan and dose to Eclipse so that we can pull them together side by side to show you the comparison. So, on the left-hand side is the Eclipse plan, on the right-hand side is the Elements plan, and in the middle is the DVH combination of PTV and the cord. So, at 24 gray prescription, they are coincidence in 95% volume, and also we can see that the Elements plan is much harder inside PTV and then the cauda, the Elements plan reduce the cauda from 9.9 gray to 8.7 gray.
If we drag a line along the AP direction passing through the PTV and cord to a one-dimensional dose profile as shown here, so the red line is the Elements plan which shows a steeper dose fall off. So, these dose fall off along AP direction is 2.8 gray per millimeter for Elements plan or 2.3 gray per millimeter in clinical plan. So, next example is a T2 spine and 24 gray in three fractions, the gap between the PTV and cord in the AP direction is 4-millimeter. And you can see from this picture, the right top corner, the Elements plan splitting ones and using four arcs. In order to see how this PTV splitting works, we've transferred the plan into Eclipse and in Eclipse, we can actually turn off each individual arc so we can see the dose distribution from each group.
So, as you can see from this picture, Group 1, Arc 1 and 2, mostly treating the left side of the PTV, and Group 2, Arc 3 and 4, mostly concentrated on the right side of the PTV. And this is a side-by-side plan comparison, the left side is the Eclipse and the right side is the Elements. In the clinical plan, the cord max dose is 11 gray and Elements reduced it to 9.2. And the dose fall-off in this picture, it shows that the Elements plan is 3.1 gray per millimeter compared to 2.7 gray per millimeter in Eclipse plan which has a steeper dose fall off. And next example is an L3 radiosurgery, 16 gray prescription. As you can see here, this one not only including the body but also including the right-side pedicle and left-side transverse process. So, because of the complexity of the target, Elements split the PTV twice, so it ended up with six arcs, as you can see from this right top picture.
So, these ones showed the Elements plan was mapped to a body function in the patient's specific QA module in Elements, so that in this module, monitor units of each arc can be turned on or turned off so that we can check the dose distribution for each arc. And in this example, we can see that Arc 3 and 4 are mostly treating the body of the spine, Arc 1 and 2 mostly treating the left side transverse process, and Arc 5 and 6 treat the right-side pedicle and these are the PTV splitting illustration. And then this is the side-by-side comparison, again, left side clinical plan, right side Elements plan. And while we drag the AP line, we can see this line goes through the body and then the cord region and then the transverse processes, the same thing as we see here.
So, the Elements plan have a much steeper dose fall off, it's 1.5 gray per millimeter compared to only 0.6 gray per millimeter in clinical plan. And also, in this case, you can see cord max dose reduced from 15.4 gray to 12.6 gray. And this is an example that we use in our publication, this is a T11 SBRT, 27 gray in three fraction, and theses isodose and DVH comparisons are displayed when we transfer both plants into Eclispe. And the gap here is 4.8 gray and the dose fall-off in this region is 1.1 gray per millimeter in clinical plan and 2.6 gray per millimeter in Elements and max cord dose is 17.5 clinical versus 15.4 Elements. We do see that the D5% PTV dose is 31 gray versus 34 gray in Elements which is corresponding to a harder inside, and also the CI reduced from 1.4 to 1.1 and GI reduced from 4.4 to 3.8.
If we look at a zoomed-in view of this dose distribution, we can see that through a spyglass, the outside is Elements plan and the inside is clinical plan. Prescription dose, isodose line in Elements plan covers more of the target in close proximity to the spinal cord than the clinical plan while also create a ring of 50% isodose line around the spinal cord, while the same isodose line in the clinical plan completely engulfs the spinal cord. So, this is a full list of the results show all 20 cases for each plan. We have CI, GI, max cord dose, and min cord dose. And the analysis of CI and GI, we can see that on average, the CI in Elements was reduced from 1.25 to 1.1 and the GI was reduced from 4.2 to 3.5. So, less CI and GI means more conformal plan and less intermediate dose to the surrounding tissue.
And also we noticed that not only the absolute value of CI and G is reduced, but also the standard deviation is also reduced. So, you can see the CI standard deviation, 0.16, reduced to 0.04 in Elements and GI 0.7 reduce to 0.3 in Elements, and this less standard deviation which means we achieved a more consistent plan within the 20 cases. And this is the max cord dose...both max and min cord dose significantly lower in Elements plan, max was reduced from 14.6 to 11 and min reduced from 6 to 4.3. If we take the ratio of the dose of the Elements plan divided by the dose of the clinical plan, then we got 78% and 73% for max and min respectively, which means the Elements plans reduce both max and min cord dose by about a quarter.
The next comparison is the total monitor unit and the modulation factor because the 20 cases have a different prescription dose, so if we only compare monitor units, maybe we have confusion then. So, we divided the total monitor units divided by the total prescription dose and we define it as a modulation factor. Basically, it's normalized with the prescription dose. And for both total MU and modulation factor, Elements plan is higher, so it's more monitor units and more modulation. So, further, we divided all 20 cases to different PTV splitting scenarios, we want to see if PTV splitting costs the higher MU or modulation factor and the result is showing us, yes. And without the PTV splitting, the modulation factor only increased by about 2% from 3.25 to 3.31, with PTV splitting once, the modulation factor increased from 3.56 to 3.94, so it represents 11% increase. With PTV splitting twice, the modulation factor increased about 30%.
So, therefore, we can see that, comparing to our clinical plan, the Elements may achieve better dosimetry, however, it at the expense of more number of arcs, more monitor units, and thus the longer delivery time. And we also stratify the results by the number of factions and the trends consistent with the overall data sets that were observed, that Elements plans have significantly better CI, GI, max cord dose for single fraction plans. Four three and five action plans, although we observe the similar trends, but we believe since we only have a limited number of cases, we still need more cases to show statistical significance. We also compare the pencil beam algorithm versus Monte Carlo. So, Elements utilize a pencil beam algorithm for initial optimization and dose calculation and then allow for optional final multicolor optimization and dose calculation. All plans in this study were analyzed after the final Monte Carlo dose calculation.
And in order to evaluate the dosimetry change with the results of the Monte Carlo algorithm after the pencil beam calculation, we have to calculate the plan using only pencil beam and then we transferred the plan into the so-called RTQA recalculation and recalculated with the Monte Carlo algorithm using exactly the same plan. And after that, we don't renormalize the PTV coverage, we just get it applied as-is. So, this is an example of T5 lesion with a large amount of lung in the beam path and we can see the PTV coverage, it was 95% because that was what we normalized, and it reduced to 91% after Monte Carlo recalculation, and the cord max dose increased from 6 gray to 7 gray, that's about a 15% increase. And then we calculate another one which is a T11 lesion with very little lung in the beam path and the PTV coverage actually increased from 95% to 96% and max cord dose is reduced from 9.3 to 8.7 gray. So, that tells us that for a very homogeneous tissue region, the user may decide to use pencil beam only to save the planning time.
So, the benefits of Monte Carlo optimization, it has more accurate dose calculation, especially for thoracic spine, the interface between air and bone, and paraspinal and regions with spinal implants. And sometimes, even for cases where there is not a large amount of heterogeneity, it still benefits using Monte Carlo but the cost is longer planning time. So, last, we try to show you the deliverability and QA, patient-specific QA of the two plans. So, this is a summary of the results and we have two patients plan delivered. Case number one is a simple case, only have two arcs with similar modulation factor between Elements plan and clinical plan and they all get a 95% and above gamma passing rates and we used a 3% dose difference, one millimeter DTA, and 10% lower dose threshold for the gamma calculation.
And the second one, the results show you a six arc, which is here, as you can see the six arcs around Delta4 Phantom, and six arcs here. And this is the result and the red line shows you the planned calculated dose and the green dots represent the measured dose and in this six-arc plan, we achieved almost 99% gamma passing rate. So, in conclusion, the Elements Spine SRS plans were better than the current clinical plan in achieving higher conformity, decreasing gradient index, and sparing dose to the spinal cord but it will cost us more number of arcs and longer delivery time. The automation in Elements was shown to generate more consistent plans with quality as good as or better than the current plan.
The average calculation time for the 20 cases was about 28 minutes with Monte Carlo optimization and calculation turned on, however, it also with a large variation depends on the target complexity and PTV splitting times. For example, simple case without PTV splitting, we may gather a plan in 10 minutes, if we get a very complicated case, it may go up to 1 hour to the calculation. Although there is other algorithm and planning system that can also utilize templates to reduce the manual effort, from our initial experience with this Elements VMAT implementation for spine and we can see it's specific to creating a steep dose gradient towards the spinal cord and that can help reduce the probability of one of the primary failures of the spine SRS and SBRT. Thank you.
Bogdan: Thank you, Dr. Shi. Thank you, Dr. Liu. Let's see if I can bring both of you back online to answer some questions. Haisong, you may need to re-enable your camera so we can see you. Maybe I'll start with you, Dr. Shi. So, we polled the audience today on a few topics and regarding treatment, it seems that both external beam as well as single fraction, although utilized, have the lowest numbers. In your practice, what do you consider to be the challenges in offering single fraction radiosurgery to more patients? And how would you think that that particular treatment strategy can be expanded? Nope, we don't actually hear you.
Dr. Shi: So, I think, you know, we have briefly discussed patient selection criteria and right now, we still follow the NOMS algorithm as a primary guideline to select the patients. Obviously, as mentioned earlier, to try to establish a practice for radiosurgery, you really need to have a very close that work together multidisciplinary team because, you know, this is really required inputs from multiple disciplines between neurosurgery, orthopedic surgery, medical oncology, radiology, and based on the patient's status and figure out what is the ideal treatment. Currently, our preference is to use radiosurgery SBRT for patients with oligometastases patient. We have a strong preference for the patient who has pathology that shows relatively radioresistance, for example, renal cell carcinoma or GI primary. Those patient's tumor require a higher dose, so low dose palliation radiation treatment may not have adequate clinical durable benefits. And I think for a patient with a more favorable pathology, for example, breast cancer even with metastatic disease, they tend to have longer survival, those are also patient we have a strong preference to use radiosurgery or SBRT techniques.
Bogdan: Okay, great. Next question was more looking at referral patterns and what could contribute to an enlargement of a spine radiosurgery program and both medical oncologists and neurosurgeons were weighted equally in terms of who should be better targeted in order to grow the referrals. What's your opinion on that and then how does it work in your practice...?
Dr. Shi: Yeah, definitely different between academic practice and community practice, I know, but, for me, a multidisciplinary tumor board will be really helpful. And this is what we do on...so we have a group of people who are able to review all cases and try to [inaudible 00:40:51] the disposition of patients. And in a community practice, I think the tumor board setting is very common for most hospitals, so definitely, this is the main area of focus, so really try to, you know, present the information, share the knowledge, and educate other specialty physicians as we see that we have lots of evidence. Right now, it's very well established, local control is excellent with SBRT technique and we know that in a set of patients, SBRT and this type of aggressive local treatment is very, very meaningful not only for disease local control but also in overall survival setting for those patients.
So, share this information, really bring this knowledge to all the referring physicians and, you know, obviously, make sure they're aware of the technology available. As we also mentioned, you know, this technology is very challenging on a technical point, so it really requires team efforts, and the radiation oncology team really need to have a very high quality and safety procedure and also efficiency. That's why, you know, to implement some of the software, it really ate the quality assurance and also the efficiency because of patient volume, as mentioned early, with increasing incidence and ever-growing cancer patient population and the longer survival, we do anticipate that more and more patient will be eligible or SBRT/SRS surgery or radiosurgery type of procedure. And we really need to have the capacity to absorb more case load and so an automated solution will be tremendously useful.
Bogdan: Great. Well, speaking of technical challenges, Haisong, we have a lot of technical questions for you, so I'll try to cluster some of the questions. But let's maybe start with selection of energy for your treatment plans. We had a question pertaining to why do you still use the regular 6x beam when the Flattening-Filter-Free 6x mode has a sharper gradient? Yeah, we can't hear you, Haisong, so let's try that again. Make sure your mic is on.
Dr. Liu: Can you hear me now?
Bogdan: Okay, I can hear you now.
Dr. Liu: Okay. So, in terms of the entity selection, actually, we didn't select any of the energy because we try to compare the Elements play with clinical plan. And so, these energies, either 6MV or 6 Flattening-Filter-Free are all coming from the original energy selection of the clinical plan. So, these clinical plans are based on our older version of Eclipse, version 11, and at that time...the clinical cases are from 2016 to 2017 and at that time, we only start to utilize 6FFF occasionally, not routinely. And I believe we're utilizing the Flattening-Filter-Free energy mode more and more, but this is just the original energy selection.
Bogdan: Okay, next question, or questions, should be regarding more PTV selections, so, obviously, you're showing in your example a lot of continuous spine PTVs. How does Elements handle non-continuous PTVs? One question. And the second question, "Do you use PRVs for your treatment planning?"
Dr. Liu: So, as far as we noticed, the current version of Elements, which is 1.5 can only plan the continuous despite target, either the single or if it's connected. For example, we have example cases from T2 to T2, but if the spines are separated, either T7 and L3, I'm not aware that...we haven't tried that case, I'm not sure if we can plan. Maybe Bogdan you can give more of the update on this question.
Bogdan: So, yeah, you're absolutely right. The current version, which is version 1.5 allows you to create essentially a one isocenter plan, so if you have multiple targets, you would have to create two plans. The next version coming out by the end of the year, the 2.0 version, would allow you to do that in an integrated plan and see the overall dose. So, the ability to treat non-continuous spine levels will come with the next version to make it a little bit more streamlined for everybody. Next question in regard to the selection of arcs and it may be worthwhile to say that the dosimetry comparison was in place to mainly check if the automation generated by Elements was adequate for the treatments. But there's a question pertaining to the fact that some plans done in the Eclipse, so your clinical plans have two arcs, but there were six arcs in Elements because of PTV splitting, what is the delivery time difference between those two plans?
Dr. Liu: Yeah, so, again, the study was not aimed for compare Elements versus Eclipse because the purpose is to say evaluation of Elements can give us adequate clinical approved plan, acceptable plan, and so that is mostly compared to our clinical-accepted plan and that clinical-accepted plan is happened to be planned with Eclipse and at that time, it's the older version. And right now we have a newer version 15.6 with a different optimization algorithm, it's a PO instead of the TRO in the older version, and also the newer version, we have more functionalities like MCO and other features that we can use.
And so if we really want to do apple to apple comparison, then we may want to transfer all the same arc geometry from Elements plan to Eclipse and lose the constraints of the maximum dose so that we allow the Eclipse to kind of give us more maximum dose inside the PTV and see what Eclipse new algorithm can give us and we can definitely do that in the near future.
Bogdan: Okay, there's a question regarding monitor units, and I think you may have answered this so the question may have been asked before you had a chance to cover it, but what difference in monitor units do you see between the Elements plans and the Eclipse plans?
Dr. Liu: So, I showed a table for comparing the monitor units or modulation factor for different scenarios of PTV splitting. And what we found is that without PTV splitting, basically compare two arcs versus two arcs, that with the Elements, the modulation factor only increased by about 2%. But if the PTV splitting once, then the modulation factor increased by 10%-11%. And if the PTV splitting twice, then the modulating factor increased by about 30%, so more monitor units to deliver, and also more number of arcs, and definitely, it will be a longer delivery time. So, that's why we say that. And we can say that the Elements give us a better dosimetry at some cost and the cost is the longer delivery time and more modulation.
Bogdan: We have some questions regarding the dose engines, particularly Monte Carlo. So, how long does a Monte Carlo calculation take and what kind of dose resolution do you use? And would you recommend dose-to-medium or dose-to-water for Monte Carlo calculations? Do you have any experience with how our Monte Carlo compares to Monaco?
Dr. Liu: So, we did the time the difference between run the power with pencil beam only and with Monte Carlo turned on, and I think the average with Monte Carlo turned on is about half an hour, like 28 minutes, but with the pencil beam only, the average is about 10 minutes. But as I said in the conclusion part, the averages does have a large variation. For some simple cases, the calculation even with Monte Carlo turned on, it can be done in 15-20 minutes, but some complicated cases, especially if there are six arcs and two PTV splittings, they will sometimes run about one hour, one and half hour.
Bogdan: And I don't think you've done comparisons to Monaco, I don't know that you have the Elekta software, but for those of you interested, we have presented information on Monte Carlo comparisons with another group in San Antonio, so you can review those results. Regarding QA, Haisong, and gamma calculations, are you looking at local or global gamma values?
Dr. Liu: I believe what we did is the maximum dose at 100% and it's a global gamma instead of local.
Bogdan: And I did miss a question regarding Monte Carlo, what materials are supported by the Monte Carlo dose engine? Would it expect surgical hardware, etc.? I'm not sure if you can answer that question, Haisong. If not, we can provide technical details to the person who asked the question. We have some questions regarding IGRT and maybe both of you can answer this. "Do you utilize online IGRT to monitor intrafraction deviations to guarantee the high precision delivery and overall, for your spine radiosurgery patients, how frequently do you monitor and what do you use?"
Dr. Liu: So, from our previous study, we did notice that the spine, especially the neck region or the T spine region, they tend to...the patients tend to, you know, move during the treatment and that can be one half to two-millimeter. And so, we set up our institutional policy that we will take the ExacTrac verification and make the necessary adjustment before each arc start. And also, if the arc period takes about one minute or two, and when the arc comes to any orthogonal...when the gantry comes to any orthogonal position, either 90, 0, or 270, we can throw in some snap verification just to make sure the patient as stable as possible.
Bogdan: Question regarding modulation complexity and Haisong, I think you've defined your own version of that, the software also gives you a number that we call a modulation complexity score. So, it's kind of a two-part question, what kind of ranges do you see for that modulation complexity score in the plans generated for your 20 case reviews? And how is the QA impacted by an increasing modulation complexity score?
Dr. Liu: Sorry, Bogdan, I didn't hear you.
Bogdan: Okay, so the software gives you a number that provides a modulation complexity score, have you monitored the variation in that number or what kind of range do you see for your 20 evaluated plans? And how does the modulation complexity score impact QA?
Dr. Liu: Yeah, I'm browsing back to see where the questions are. I couldn't hear you, sorry.
Bogdan: Okay, so maybe look in the chat line, the last question from Matt.
Dr. Liu: What dosimetry is used to measure for gamma index? So, the phantom we used is ScandiDos Delta4. The average MCS value?
Bogdan: Yeah.
Dr. Liu: So, I think the MCS value given by Elements itself is kind of confusing to us, so that's why we kind of defined it ourselves, just simply divided the total MU by the total prescription dose and it's about 3.3 for no PTV splitting, 3.5 for splitting once, and over 4 for splitting twice. I'm not so sure about how Elements defined MCS value, so that's why in our study and publication, we didn't use that value at all.
Bogdan: Okay, great. Well, again, Dr. Shi, Dr. Liu, thank you very much for your presentations today and thank you all for joining and we look forward to seeing you at our upcoming webinar. Thank you and have a great day.
Dr. Shi: Thank you.
Dr. Shi: Thank you.