Good afternoon. Thank you to Brainlab for sending invitation for me to come and share some of our experiences with the group at ISRS. The biggest challenge so far I've seen from presenters is how to use the control over here. So hopefully we'll figure it out. Yeah. Okay. All right. So before, what is this one? Right? There you go. Okay. So this is the disclosures to share it with you. And if we look at the ACR data from 2018, we recognize that 1 person in 100,000, or about 10,000 people a year in the U.S. will develop spinal tumors. And we also know that 15% to 20% of the central nervous system tumors occur in the spine. So we do have a large population that we are called to help that have been diagnosed with spine disease. And we do know there is a role for radiation therapy. We just heard the great talk to that effect.
There's two constraints really that we have that we need to consider. One is the plant quality that we offer when we consider radiation therapy treatment, and the delivery efficiency. We know the patients are in pain, and they don't wanna be laying on the table for much longer than they have to be. So looking at our quest of who does it better, faster, provides the plant quality using metrics such as conformity index, gradient index, minimization of dose for the organs are risk, and also for the efficiencies looking at the time and monitor units that we need to deliver the treatment, of course, under image guidance, so we can make sure that we're treating that are area, that is really the challenge that we have to find what's the best way to do it as we embark in that race.
So in terms of an outline and what I'm hoping to share with you in the next 15 minutes is specifically talk to you about the spinal Elements and a quick overview of some of its unique features, look at a treatment planning study that we performed in our institution, speak about commissioning, testing, and clinical implementation, and then close with some final remarks. The Elements that UT Health in San Antonio, which is part of the MD Anderson Network System, was installed in 2018 for clinical use, and we had a research version, pre-clinical version since 2015. We're also called the Mays Cancer Center, so don't let that confuse you. It's all part of the same institution. We concluded with the acceptance of the system, and then we did the commissioning, and we did that in two stages. We first implemented them Multiple Brain Mets, and then the Spine SRS module.
This is how the screen of the Spine Elements looks like. As you go in the columns that we see here, has a viewing option, you can view the data, you can do the co-registration, which includes image fusion and the curvature correction for the spine, the contouring features, the dose planning, and the review, which includes also the physics QA. The Spine Elements is built on a platform that specifically looks at the unique needs of the spinal anatomy, and it's, by design, aimed to give you an optimized plan and efficient delivery specifically for treating spinal radiation therapy.
Some of the unique features I'd like to share with you, first is the spinal curvature correction that is offered as part of the Elements software. We do know that MR images are critical for us to delineating the spinal targets particularly in the retreatment setting. And we also recognize that CT is a primary imaging modality that we use to calculate the dose. We have to do that to take advantage of the change in the density. So there is a need to co-register, fuse together the CT and the MR images. And in doing so, recognizing the fact that the CT and the MR typically obtain on a different day in a slightly different anatomical position of the patient on the treatment table, we have to have a smart core registration.
And that's exactly what the spinal curvature correction does in Elements. It respects the fact that the bone is not deformed, whereas the tissue around the vertebral body is deformed, and it gives you, as you see here, the image on the right, a fusion that has this deformable registration only where it's applicable. This is also very useful when you're treating multiple vertebral bodies, we have mets that you treat into different parts of of the spine.
Another unique feature is the spine contouring that comes with Elements. The user draws the GTV, the growth tumor volume, and the clinical target volume is automatically generated following the consensus guidelines contouring. So we do get the vertebral bodies automatically segmented for us, and specifically for the GTV to CTV, we do follow as it's built into the software the consistent guidelines for that. So the generation of the CTV is very fast and it's very consistent for that reason.
Then we have the spine VMAT optimization, also unique to the implementation of Elements. It recognizes the fact that the vertebral bodies are very...have a large concavity, so it allows what is called the arc splitting. So it will segment the vertebral body for different elements. And it has a smart optimization to decide if we have to have arc splitting enabled so that we can get a highly conform distribution with high avoidance of the spinal cord as well. And you can see how that's demonstrated on the bottom left with the segmentation of the spine in four elements, four areas, and the distributions that we typically get from the calculation of the bottom right screenshot.
Two additional features that aren't necessarily unique to Elements, but also worth noting, is that it has user-defined protocol-based planning, so you can script a lot of what you wanna do including prescription, beam arrangement, etc. as part of a protocol, and you can play that for a very quick start and finish of your planning session. And it also has a Monte Carlo based dose calculation, which is very important, especially when we deal with tumors in the spine and have to consider the soft tissue and bone heterogeneity, but also coming through the lung as well.
I'll share with you now a treatment planning study that that we did. It was to evaluate basically the Elements against the Pinnacle and Monaco to other planning systems that we in our clinic. And for that, we had a cohort of 10 patients with spinal lesions. The GDV was defined both by the treatment team, and the CDVs were automatically generated using the Elements auto-contouring feature. Each case was planned to deliver 20 gray. We used two arcs with collimator angles of 30 and 90 respectively, and the gantry and start/stop rotations almost a full 360-degree arc. The exact geometry that we produce from the Elements plan, we copied it to the Pinnacle and the Monaco planning systems so we can have an exact same patient, exact same beam geometry to compare against.
The dose constraints were such that the PTV D5 was less than 25 gray, and the spinal cord, to almost the point, was less than 14 gray. Our optimization strategy was to cover the PTV at the 95%, that was our constraint. And the spinal cord sparing, and dose falloff were optimized for maximum conformity while enforcing the dose constraints that we had in our protocol. The plan quality was assessed by looking at the conformity index, V20 divided by the PTV, the dose gradient index, the V10 divided by the PTV, the dose, the D5, 5% of the volume, and the physical dose gradient, which we assessed as being the distance between the 20 gray and the 10 gray isodose line in the same slice for all the 3 planning environments.
This is an example of how 3 of our 10 patients' plans look like. You can see in the three rows, we have the three patients identified as patient one, two, three. In the columns, we have the Elements, the Pinnacle, and the Monaco calculations respectively for the slides shown. Looking at it, you don't see big differences between the three planning platforms. And I know the isodose lines are a little thin, especially for those that are sitting in the back, but they look very fairly comparable. When one looks, however, at the dose-volume histograms, we have here again for the same three patients, the orange line is the one from Elements, solid line, dashed orange is Monaco, and the the light blue line is from Pinnacle. What we do see here is the dose-volume histogram for the cord is the best for all the 3 patients that were showing here, but that was true for all 10 that we did in the study. And for the CTV, the dose-volume histogram, it's very comparable for all three, most notably the one from Elements was a little bit more sharp, and since it had a hard constraint of meeting the planning requirement, it was always met.
This is a table that really summarizes our findings, and what you see here is that for the monitor units, the PTV coverage, and D5, between the three different platforms, the results are relatively similar. Most notably the coverage and the D5 are very much almost identical. For the monitor units, there was some variation, but not statistically significant, about 7700 for the Elements, 6800 for the Pinnacle, and 9100 for Monaco. What is different, however, is the bottom half of the table, namely the spinal cord maximum dose, was the least for Elements, the gradient index was the best for Elements, the conformity index also the best for Elements, and the distance from the 20 to 10 gray isodose line was also the least for the Elements plan.
So that led us to believe that based on these metrics that were statistically significant, the Elements had a significantly smaller gradient index and spinal cord maximum dose. Arc splitting, which actually was engaged automatically in 50% of our patients, namely 5 of the 10 plans, resulted in improved spinal cord sparing, and improved dose gradient index and the physical dose gradient was smallest in Elements. So going back to the race that we showed in the first slide of who's doing better at least based on how we validated the comparison between the three planning platforms over here, we felt pretty comfortable that Elements would've been the better one to use in our case.
For the commissioning, we actually followed the MPPG 5.a guidelines from AAPM, and followed to the script the test that we're to perform there. Here, you see three fields, a small field, a large field, and off access field. There's some overlap I see here in the ledger, but you can see what I mean. And we can see that we have both qualitatively and quantitatively a very good agreement between what was measured and what was calculated as we commissioned the Elements, specifically in this case, software. The 2D gamma analysis showed results of better than 98% for all the points in all cases, and the criteria were 2% and 2 millimeter.
This is the IROC Accreditation Lung Phantom that also includes the spine inserts. We've seen that before here in the meeting. And here you see the profiles between the film and the institutional values for the film that was scanned and processed by IROC also showing a good agreement. In fact, we end up repeating that after we tweaked a little bit, the output factors in Elements, and we got an even better agreement. The second time, it was 98% passing rate of the gamma with the IROC criteria. For the patient-specific QA, we also used the Octavius 1000 SRS unit, and we did actually treat all those 10 patients that are noted in our planning study. We did deliver the QA, and this is a representative example here, what type of QA that we obtained. All the passing rates were above 95% for 2% in 2-millimeter analysis. And here's the two orthogonal axis of the analysis using the high resolution Octavius 1000 SRS. I'm sorry, in this instance, the gamma was 3% in 1 millimeter.
And finally, we also did IMRT QA pre-treatment verification with two different phantoms. On the left, you see the Delta4, and on the right you see the RTsafe pseudo in vivo measurement that was obtained with this anthropomorphic phantom that was printed from the patient's DICOM CT that I set. In this systems, contrary to what some of the other presentations that shown the use of the phantom, we did not have a gel, the 3D gel in the phantom, although we've done that before in a different discussion that we present in the past, but here we used two ion chambers, A16 ionization chambers, one in the tumor for that particular patient, and one in the spine. So we can assess the dose to the tumor and the spine as two absolute dose measurement points.
This is the results here for the Delta4 on the left, and right on the top half of the slide, you see the measurement against the planned dose, and qualitatively and quantitatively the agreement was very good. The red line really kind of follows the green dots quite nicely. And when we look at the analysis of the gamma, a 3%, 1-millimeter analysis gives us a 98.3 passing rate. If this was true for all the patients that we did, they all had passing rates better than 97.5% in their analysis. On the bottom, you see the RTsafe pseudo in vivo measurement in the two distinct points that we have the A16 ionization chambers, and the difference was less than 2% in the CTV, and almost zero at the spinal cord. So very good agreement in all different ways that we engaged in QA-ing, both as an end to end process, but also speaking looking at other dose distribution profiles and the doses of the Elements as was implemented.
So in conclusion, the Brainlab Spinal Elements is, we found to be uniquely suited environment for planning SRS. It has some features that specifically address spine anatomy, and that includes the spine curvature correction, the automatic target delineation following the consensus guidelines, optimization for complex spine target shapes to remove, or to account rather better for the concavity of the vertebral bodies, it leverages the use of standardized protocols to expedite the treatment and treatment planning, and it also has the Monte Carlo calculation engine, which is critical to be able to deliver what you plan, especially in the context of strong bone homogeneity in the lung, through which the IROCs are passing to treat the spinal cord. The plans from Spine Elements exhibited improved cord sparing and target conformity in the planning study that we did, and for the commissioning, which included end-to-end testing, we had a very good agreement between what we calculated and what we measured with high passing rates for the MPPG test suites, the IROC spine phantom, the RTsafe phantom, and the patient pre-treatment QA with a Delta4 phantom.
I want to especially thank my colleague and former resident, Daniel, that assisted with this project, and two of our doctoral medical physics students that are now graduated as of last week, George and Rodrigo, the Brainlab team, most notably Bogdan, Rebecca and Corey in engineering. They have always been instrumental in the past five or six years we've been working together as a clinical site with Brainlab, and RTsafe for providing the spine phantom. Thank you.
