• 817
Comments (0)
Login or Join to comment.
Interviewer: Together, they will present their experience with our Multiple Brain Mets SRS Elements, as well as ExacTrac, and share with us some of the clinical outcomes for the patients they have treated so far. Dr. Girvigian is the assistant chief of the Department of Radiation Oncology and the co-director of the radiosurgery program. Dr. Rahimian is the chief of medical physics in the Department of Radiation Oncology. And both of them have been long-term Brainlab users. As always, we are providing CE credits for all of you. Should you require CAMPEP, MDCB, or ASRT credits, upon successful completion of this webinar, please send us an email at, and we will let you know how you can obtain those credits.

And don't forget to sign up for our upcoming webinar with Dr. Nader Pouratian from UCLA Department of Neurosurgery and yours truly on August 19th. And we will be discussing new technologies in the treatment of functional disorders. And lastly, please remember to log in via either Google Chrome or Safari. And should you have any incident difficulties throughout this webinar, please refresh your window. Utilize the chat interface to send us questions. We will answer your questions upon completion of the two lectures. We may also ask you some questions via the polling interface, so feel free to send us feedback. And as always, if you'd like to follow us on social media, please use the hashtag provided. With that said, I'd like to turn it over to Dr. Girvigian.

Dr. Girvigian: Good morning, everyone. As mentioned, I'm Michael Girvigian. I'm here with my colleague Chief Assistant Dr. Javad Rahimian, and we are with the Kaiser Permanente Regional Radiotherapy Department in Los Angeles. We're gonna talk to you about our experience with Brainlab Elements, multiple brain metastases. I have no disclosures to report. I did not accept an honorarium for this event, so I work for free. Don't tell my med group administrator. So who are we? Kaiser Permanente, Southern California is an integrated nonprofit health plan providing medical services within its own Kaiser foundation hospital network through its contract at Southern California Permanente Medical Group, of which physicians are partners.

Kaiser Permanente Southern California currently provides care to 4.5 million members in Southern California. The Regional Radiotherapy Department has three internally operated locations, and we do have an ACGME accredited Radiation Oncology Residency Program based in Los Angeles. At our Los Angeles center, we have seven linear accelerators. There are six currently operational. One was recently installed. It's not currently operational yet. Ontario has four linear accelerators. Orange County, Anaheim has three linear accelerators. These three centers serve between 450 to 500 patient treatments per day, and these locations are supplemented by contracts for outlying areas. The Los Angeles Medical Center oversees operations at all sites with a shared chief of service, assistant chief of service, director, chief physicists, and an assistant medical center administrator. There are currently plans to develop two new radiotherapy centers in Bellflower and Woodland Hills, California.

We routinely evaluate our geographic access gaps as represented by these red circles. And in so doing, we're able to determine when and where we should build and provide new services, and hence the two new centers that I just mentioned. In 2016, we did see 5,334 unique members present to our departments. And here, the Kaiser Los Angeles Medical Center is shown on the left, the beautiful Department of Radiation Oncology at Kaiser Anaheim on the right, and Kaiser Ontario on the bottom.

So we are long-time partners with Brainlab. We installed the Gen1 Novalis linear accelerator in 2002, and this really revolutionized our radiosurgery program at that time. Prior to that, we were using the old radionics X-9 system. If you're familiar with this, this is a bolt-on system. We were using this coupled to a Clinac 2100C. But the new Novalis, with its micro multi-leaf collimator, and the dynamic conformal arc, was really a game-changer for us at that time. And we did a lot with this machine. We treated approximately 10,000 patients on this device. We had a very skinny radiosurgery team at that time. We had two neurosurgeons led by Dr. Joseph Chen. We had myself, my colleague, Dr. Michael Miller, who's now retired. We had whatever radiologists would return our calls. We had Dr. Rahimian, chief physicist, who's here with me today, and a nurse coordinator.

So we go back to the time when frames were in widespread use. I'd like to think that along the way, we challenged sort of the dominant paradigm with regards to linear accelerator-based radiosurgery. Dr. Rahimian showed that the geometric accuracy was there with linear accelerator-guided systems. And we showed that you could treat trigeminal neuralgia with LINAC-based radiosurgery, which was pretty much taboo in the early 2000s. We then switched over to a frameless system with the 6D robotic couch approximately 11 or 12 years ago. And again, the work with Dr. Rahimian did show that the geometric accuracy was there. And again, I think we challenged the system by showing you could treat trigeminal neuralgia with this frameless radiosurgery delivery. And we ultimately showed that our results were as good with trigeminal neuralgia as previously reported.

So, we commissioned the Brainlab Elements Multiple Mets with the new TrueBeam STX with ExacTrac linear accelerator with the HD 120 high-definition collimator in April 2017 at Kaiser Los Angeles. We've treated quite a few patients on this machine. All told, fractionated, single fraction, benign, malignant, about 1,200 patients have been treated in that machine. In 2018, we expanded our radiosurgery program to Anaheim with the installation of the TrueBeam ExacTrac with the Millennium Collimator and Brainlab Elements. And this year we installed two more machines. We installed the TrueBeam ExacTrac with the Millennium Collimator in Ontario and a second TrueBeam STX with HD 120 high definition Collimator at our LAMC location.

Our radiosurgery team has grown fatter. We've now spread to another center at Anaheim. We have very good neurosurgery support. We have regular and excellent neuroradiology support now, and, of course, the ever-necessary, nurse coordinators. So we evaluated several platforms and systems when we realized it was time to change out our Gen1 Novalis and upgrade our program. We looked at Accuray CyberKnife. We looked at Gamma Knife, Perfexion Icon. We're an integrated health plan. As we know, the incidents of brain metastases is increasing. We needed a system that could provide us good dosimetry, but also increase our throughput and efficiency so that we could expand access to this technology to our members. And Brainlab Elements did fit the bill with that. And I'll explain.

This is some work, again, by Dr. Rahimian. These are three plans for nine lesions. A nine isocenter cone plan, a nine isocenter MLC plan, and the Elements plan. And what we see is there's pretty good dosimetry across the board here. And certainly, Elements with the single isocenter are satisfactory in dose distribution. Again, we have the nine isocenter cone plan, sharper penumbra than the MLC, and the Elements. And in regards to efficiency and throughput, here we have eight lesions planned with eight isocenters with dynamic conformal arc. We've got good conformity indices with a mean of 1.285. We have 35,000 MUs, and we have a treatment delivery time of 240 minutes.

Now, we have the same lesions plan with Brainlab Elements with the HD120 Collimator. We've got a very good conformity index mean of 1.23. The Millennium, a little bit worse in terms of their conformity index, but certainly acceptable. But the treatment delivery time drops to a mean of 25 minutes. And so, that's really unparalleled. So we have shorter overall treatment time. We have similar conformity in gradient indices compared to Gamma Knife, but it may be better at homogeneity index, and significantly lower monitor units than traditional localized or center LINAC-based plan. So we know the incidence of brain metastases is going up. Our patients with metastatic disease are living longer. We're screening them more. We're screening them better. Traditional LINAC-based systems with one isocenter per target provide accurate dosimetry but it may be from 10 to 20 minutes or more per isocenter. Frameless systems certainly are out there, coupled with 60 couches provide good acceptable dosimetry.

So a paradigm shift with our program came with the Brainlab Elements and the TrueBeam STX linear accelerator. Using multiple isocenters prior to 2017, we treated on average 290 to 300 lesions per year. The number of metastatic lesions treated since we commissioned Elements Multiple Brain Metastases from 2017 to July of this year is 1,804. So greater than 600 lesions per year. This is in 382 patients, 114 treated on the Millennium Collimator and 278 on the HD120. In terms of whole-brain radiotherapy, we tried to do an analysis to see if our utilization of whole-brain radiotherapy decreased. It was a little bit hard for us to sort of pull this data for this presentation. It looks like maybe there is a 15% drop in whole-brain radiotherapy usage, but there's really a wide confidence interval there. I think we all know over the last 10 years that the utilization of whole-brain radiotherapy has been decreasing overall, and we're certainly privy to that.

The issue is now the time course of whole-brain radiotherapy is different. So for our patients presenting with multiple lesions instead of getting whole brain radiotherapy at the beginning, they really now are sort of getting that at the end of the line. So our recently-graduated chief resident, Dr. Debra John, analyzed the first two years of our brain development patients with metastatic disease from May 2017 to 2019. She's now working in Tennesse, by the way. These accounted for 204 radiosurgery sessions for 1,101 total lesions. In these patients, we have a median follow-up time of 318 days. There's a wide volume range here. And this is because there are some resection cavities included in this, and we do treat the resection cavities along with intact brain metastasis. And we can talk about how we do that later if we need to.

Lesion number ranged from 1 to 42 with a median of 6 lesions. And remember, these patients are often undergoing multiple treatments over the course of their disease and their illness in their lifetime. We had 49 patients treated with concurrent chemotherapy, 32 patients treated with concurrent immunotherapy, 45 patients treated with other targeted therapies. We had a median whole brain dose of 157 centigrade and a median treatment time of 25 minutes. At the time of this analysis, 35 patients were still alive. There were 98 deaths. Eighty-four lesions ultimately developed recurrence in 39 patients. Twenty-six patients underwent whole-brain radiotherapy after Elements radiosurgery in a median of 274 days after treatment. Eighteen lesions had radiation necrosis, Eleven lesions were symptomatic from the radiation necrosis, and again, a wide confidence interval on the asymptomatic lesions were an integrated health plan. And if the lesions were symptomatic, the radiation oncologists are certainly gonna be pulled into managing those, and it's a little bit easier for us to find those in the data.

For the Chi-Squared test, a total tumor volume greater than 5 cubic centimeters was associated with neurologic death. Total tumor volume of greater than 10 cubic centimeters was associated with radiation necrosis, with a trend towards radiation necrosis, but no effect on symptomatic radiation necrosis. Total tumor volume greater than 10 cubic centimeters was associated with increased toxicity, but not the grade of toxicity. On multivariate analysis, total lesion number was not associated with radiation necrosis, whether symptomatic or not. Total lesion number was not associated with toxicity or toxicity grid. And in regards to targeted therapies, radiation necrosis was only associated with total lesion number with a trend towards the targeted therapy, but not with immunotherapy, which I know is counter to many of our experiences out there in the community. And there's a trend towards higher recurrence with lesion number.

So with this, we see that total tumor volume was more indicative of outcome than lesion number. And certainly, this is similar to what's been reported in prospective data from groups like Yamamoto. That lesion number is not as important as total tumor volume. We know the common theme is that solitary lesions do better than anything more than solitary in terms of outcomes and survival. But two to four lesions does not fair any worse than greater than five. And in terms of our Kaplan-Meier curves, we have freedom from recurrence per lesion. Lesion control at one year was very good. Freedom from recurrence per patient. And again, these patients have multiple lesions that are undergoing multiple treatments. So there's more time and probability of them having a failure a little upwards.

And, again, the time course of whole-brain radiotherapy is now coming more at the end than at the beginning. So in terms of progression-free survival, new distant lesions were common. Overall survival, the trend over the years has been improving. And now the median overall survival is trending more towards one year whereas when I started, it was more like six months. One thing that Elements allows us to do is follow the lesions better. When you treat multiple metastases and you're doing your follow-up scans months later, however long it is, there are remnants left behind. It's sometimes hard to know what's new and what's been treated. You can take their follow-up MRI scan, upload it to the patient's Elements treatment plan, fuse it with the planning MRI that was done at the time of treatment, and you can see the contours for the previously treated lesions, and you can tell very quickly what's new and what's not.

And so, you see that here in the before and after. On the right, you've got a new left cerebellar lesion. Other lesions are responding. The other thing you wanna know is how you're doing with the small targets. You wanna have that confidence that your treatment delivery has the accuracy needed to treat the small targets. And we've shown and discussed amongst ourselves and our tumor boards time and time again that the small targets are responding. And here you see the before and after, the small targets responding again and again. What about radiation necrosis? The more we do this, the more we're going to enter the world of radiation necrosis, which nobody wants to enter, but that's the way it is. So radiation necrosis acutely is primarily a vascular-mediated injury, mediated by vascular factors, most commonly indicated as VEGF. Subacute is primarily a glial injury-mediated phenomenon, and chronic is multifactorial, and that's where nobody wants to be.

Again, with the vasogenic edema associated with radiation necrosis, that's a vascular-mediated phenomenon most commonly indicated as VEGF. And that's why bevacizumab works. And the risk factor as we know is a dose-volume relationship, and related to that are the larger tumors in the prior treatments for cumulative dose in the histology location. And certainly, the targeted therapy is most commonly implicated as anti-PD-1. MRI is not a good diagnostic tool for radiation necrosis. I often tell my residents to look for the bell pepper sign. That's where you take a green bell pepper, cut it in half, look at it edge-on. That amorphous shape is the type of amorphous shape you'll see with the contrast-enhancing lesion of radiation necrosis. However, with that, I would have said figure A here is radiation necrosis, but this caption from textbook says it's not. What do they know?

MR perfusion we've been using for a number of years. Decreased cerebral blood volume is associated with treatment effects. Increase cerebral blood volume is associated with active tumor. Our neuroradiologist will tell us you need to, with the software and the post-processing, calculate the relative cerebral blood volume of the lesion in question and relate that to the relative cerebral blood volume value of the contralateral normal brain. And the delta there will clue you in as to whether you're dealing with radiation necrosis or not. Given that, we're looking here at the imaging. And again, our neuroradiologist will tell us that you can't just look at the imaging. You have to calculate the rCBV. But nevertheless, here on the left top, we've got a bright red lesion, which has increased rCBV suggesting tumor recurrence. And on the bottom over to the right, we've got decreased rCBV, which is more blue, which is related to treatment effect.

We are using the Brainlab Elements contrast clearance analysis, and we thank Dr. Prasad for helping us get our feet wet with this technique and helping us with some of the diagnoses early on. This is a subtraction algorithm. This is the MRI-based treatment response assessment maps. And the software is subtracting two T1 weighted MRIs. The first one is done 5 minutes post-gadolinium injection, and the second one is done 60 to 105 minutes post the same gadolinium injection. And in doing so, the software can figure out what areas have efficient clearance of contrast. And efficient clearance of contrast is associated with progressive or recurrent tumor whereas contrast accumulation is associated with treatment-related effect and it's the opposite of MR perfusion. In this situation, efficient clearance of contrast shows up as blue and that's a tumor whereas in MR perfusion, red is a tumor and here, red is treatment effect.

What really got our attention regarding the Brainlab Elements contrast clearance analysis is that it's not affected by dephasing artifacts. So hemorrhagic blood products like what we will often see in melanoma and renal cell carcinoma and lung cancer where the melanin and the melanoma brain metastasis causing dephasing artifact does not affect contrast in clearance analysis but can render MR perfusion useless. There is a correlation of percentage of the blue TRAM areas with percent active tumor on histopathology. In terms of treatment, I'm sure we treat the way everybody else does. If the patients are asymptomatic, we just follow them. If they're symptomatic from radiation necrosis, depending on how bad it is, we use a carefully designed steroid taper. And if that becomes prolonged or becomes chronic, we certainly jump to bevacizumab. We don't necessarily jump to bevacizumab right away.

We don't have LITT, laser interstitial thermal thermography, at our institution. We send patients out for this. In terms of the targeted therapies, we do treat concurrently with targeted therapies, except for BRAF. We don't treat with BRAF inhibitors. But certainly, with EGFR, ALK, ROS1, which is the same as the ALK drugs, HER2, which is lapatinib and [inaudible 00:20:27]. Immunotherapy. Most commonly implicated certainly is the anti-PD-1. We've run the gamut with the ALK and EGFR inhibitors. Dr. Rahimian is gonna present to you our claim to fame, which was a young man who presented with multiple metastatic lesions with an ALK-positive lung cancer as a teenager. And he ran the gamut from crizotinib to alectinib over his treatment time. It's interesting that Iressa was approved in 2015 when really, it was first on the scene, but the initial trial was not well-designed.

Most of our patients now are started on Osimertinib rather than erlotinib. But can we skip radiotherapy in the EGFR positive patients? That's the ongoing question, isn't it? Well, three years ago, Meg Newson, et. al. reported this retrospective multi-institutional analysis. Prior to this, many medical oncologists and medical oncology groups were espousing that you could start the EGFR TKI and withhold the radiotherapy. And actually, many are still espousing this. But this study did show that upfront whole-brain radiotherapy or radiosurgery did provide better outcomes in terms of survival compared to withholding and starting the TKI. But when you look at this paper, what you're really looking at is a paper of erlotinib or Tarceva. Most of these patients were on erlotinib. This study did not include Osimertinib. We do not, as of yet, have that type of data, at least in multi-institutional form. I know the British Columbia Cancer Agency has started the trial on this, but it'll be a while before we get those results.

So the ALK take-home points, certainly these patients can survive for a long time, probably longer than the EGFR patients can, but in both these patients, once these drugs stop working, they usually go downhill pretty fast. These tumors don't have high tumor mutational burden, and they usually don't respond very well to immunotherapy. I just put this up for discussion. These are the wholesale acquisition costs for targeted therapies that I got from my institution. Your institution may do better or worse than this. I don't know. But primarily, for osimertinib or Tagrisso, all of our patients now are coming in having been started on this drug. And some of them with multiple metastases are being started on double dose Tagrisso. And so, that's double the cost as you can see up at the top and double the side effects. So I just pose this for discussion.

We don't have data on this, but wouldn't it be more cost-effective and maybe a better medical value for the patient if we just go ahead and treat them with radiosurgery upfront and they can do their single-dose osimertinib rather than starting on double dose and withholding the radiotherapy? Certainly, immunotherapy works alone. And certainly, the Nivo Ipi combo works well for melanoma and anti-PD-1 drugs certainly work concurrently with radiosurgery. But as always, we need to be careful of what can happen afterwards. And we know we have a higher rate of MRI changes after treating concurrently, or even before or after with radiosurgery with the anti-PD-1 drugs. And this runs the gamut from, sort of, an asymptomatic more sort of progression to a full-blown symptomatic radiation necrosis. And we do not treat concurrently with BRAF inhibitors. And at this time, I'm gonna go ahead and turn this over to my colleague, Dr. Giovan Rahimian, who's gonna talk more about our workflow with Brainlab Elements and some of the work he's done with optimization. Thank you.

Dr. Rahimian: I just wanna thank Brainlab for invitation to participate in this webinar and present our data. This is my disclosure. The outline of my talk is to talk about the rationale for treating multiple metastases at the... partly Dr. Girvigian talked about it. The workflow with Elements Multiple Brain Metastases Radiosurgery Planning System, the volume of 2Gy optimization, and finally, the dose delivery issues and stereoscopic image-guided radiosurgery parameters. This is a slide from National Cancer Institute published in 2018 that showed the overall cancer death rate has declined significantly between 1991 to 2015 up to 26%. And specifically, the death rate has gone down for men in 2006 to 2015 by 1.8% per year. For the same period for women was the rates has declined by 1.4%. Among young children, between zero to 19 years old for the period of 2011 to 2015, the rate has declined by 1.4%.

These are really encouraging news, but, obviously, we are striving to do better. The smoking cessation has definitely improved and has declined in smokers. However, in our patient population because of aging and obesity, which are the other risk factors, the cancer has risen, specifically like uterine cancer. Back into 2011, 2012 it was like 48,000. Now, it's about 62,000. So you still have works to do. But in terms of cancer, this is one of the patients that he was diagnosed at 16 years old in 2011. And he was ALK-positive with adenocarcinoma of lung and we treated 50 lesions in 9 years with 13 courses of radiotherapy, and the last SRS was 2016. And the last follow-up a year or so ago, he was cancer-free.

And this is also referring to the graph that Dr. Giovigian just presented of freedom from recurrence per lesion is about 90% in 1 year. So this gives us encouraging news that you can treat regardless of numbers of tumors successfully, especially since the patient's systemic disease is controlled by chemotherapy or immunotherapies, as well as the target therapy. The blue color that you see is the blue background is whole-brain dose from all these treatments, which was, like, about 9.5 Gy. Moving on to our workflow, we acquired a high-resolution MRI, and then subsequently with a high-resolution CT scan in 1-millimeter slices and we fuse them.

And then we define the GTVs and we add 1 millimeter to each GTV to get the PTV. And then we defined organs at risk and we planted the multiple couch positions and dynamic conformal arcs. Before treatment, each patient plans are mapped because the MapPhan is a dyed array. We measure each case to make sure that these are passing the test. And then the patient is positioned on the 6D couch with ExacTrac. With 6D robotic couch, we treat the patient. The treatment plans were typically made and optimized using 5 to 7 couch positions with up to 14 dynamic conformal arcs. Treatment plans are done with both 6X and 6X flattening filter-free. And we compare those two and whichever is better, we use it for treatment.

IGRS was delivered using a 6D robotic couch with stereoscopic imaging to position the patient. Corrections were made at each couch position using half a millimeter and a half degrees correction tolerances. Dose was prescribed by treating lesion volumes per our institution prescription volume-dose grid, which we have over 5,000 cases treated between 2002 and 2017 via our classic Novalis system. So this actually works pretty good for us. And our prescription typically is 2 in 90% isolates 9. Delivery of those beams I view is shown here in this video. You can see multiple lesions are treated simultaneously and different couch positions are used as you do a dose painting for each lesion. This is a patient with 17 lesions. We used two isocenters, which look better than single iso because especially if you have a couple of lesions next to the orbits and others are more posterior, the optimization is basically superior if you use two isocenter. We use it occasionally to make sure that our conformity index is better and the V12 is also lower.

For the same patient, you can see that the dose cloud resolution of each lesion. You can see the cross-talks between lesions are minimal. The conformity is pretty good, and the whole brain dose, in this case, is about...the mean dose is 5Gy, which is about the mean grade rule of thumb is typically about 25 to 35 centigrade per lesion. So if you wanna just guesstimate, you could see most of the cases I've done is about 30, 35 centigrade. It depends on the volume. Now, the whole brain dose is obviously dependent on how many MUs you're delivering, and also [inaudible 00:31:29] is a single iso and also single lesion. The Jaws are tracking the MLC. The MLC leakage and interleaf leakage is about 1.2% for HD MLC compared with 1.8% for Millennium MLC for 10XFFF. So if you track those MLCs or you use with cone, you get a slightly better whole Gy dose.

But I think Brainlab announced that they are supporting the Jaw tracking for the Element as well in their next version, which is version three. For each patient, we do Q/A. And that's, basically, we have the MapPhan, which has 5-centimeter solid water buildup and 445 diodes. And we looked at that, and we published some of the Q/As that we did, which a pass rate is about 98%. And we typically use 1 millimeter to 2-millimeter dose [inaudible 00:32:39] agreement, 3% to 5%. dose criteria. It depends on the size of the lesions. If they're tiny, we use 1 millimeter and 5% dose criteria. This is a quotation from Minniti's paper, which talks about V12Gy and radionecrosis that Dr. Girvigian also talked about. SRS represents a feasible option for patients with brain metastases associated with survival benefits. However, a significant subset of patients may develop neurological complications.

Radionecrosis represents the most important late toxicity after SRS with the brain volumes irradiated at 10 Gy, and 12 Gy being the most important independent predictors of brain necrosis. Further, they talk about lesions with volume of 12 Gy. More than 8.5 CC carries a risk of more than a0% for radionecrosis and should be considered for hypofractionation stereotactic radiotherapy, especially when that area is near eloquent area. So for that reason, we've been looking to minimizing the V12 and we have, for instance, used the single iso for smaller lesions, and for larger lesion we fraction it. So we are conscientious about that. And we have a couple of solutions especially if you have, you're treating more than 10 lesions and these are smaller as presented by Dr. Girvigian. Again, the total volume is a risk factor for radionecrosis rather than the total number of lesions.

And I mentioned that the mean contribution for each lesion to the whole brain is about 30 centigrade. And to minimize the V12, we have, especially if you have high-definition MLCs, you can divide the lesions into two hemispheres, like right and left, or superior and inferior, depending on the distribution and your graphic distribution of the lesions relative to each other. And this way, you can treat those left or right lesions separately in a 8 by 8 field size, which is a high-resolution MLC of 2.5 millimeter and like that you can have a higher conformity index. And also with Jaw tracking, you can limit the leakage radiation, which is contributing to V12. Examples of it here. This patient has 32 lesions and, in one setting, we treated with MLCs. And on the top panel, we have two isocenters and the lower is a single iso, which we can see the DVH for a single iso for the whole brain is worse than the dual isocenter. And the V12 drops from 61CC to 53CCs. And you can see the definition of lesions and the dose distribution is better in the two isocenters, especially in this area. And the cross-talks obviously is better and conformity is definitely better.

The second way we can minimize the V12 and optimize it, we define each EGTV and the 1 millimeter regardless of the location of the GTV is relative to isocenter by 1 millimeter. We expand each GTV by 5-millimeter ring, and we add all these, we call it PRVs, and then we subtract that from the whole brain. And we tried to minimize those to this, what we call the OAR. In the Element, you assign it what's for the auxiliary and OAR and you call it Help1. And that way, by using various couch positions, so you could have five, six, or seven, and also moving your couch positions as well. So if you have a set protocol, you can adjust that so it goes to more lesions and you have to play with it. There's no other nice way of doing it to minimize the V12. In our experience, more couch position does not necessarily yield a better plan. It all depends on the number, the volume, as well as the location of each lesions relative to each other.

This is a study I did with 10 patients, overall 118 lesions. A mean of 12 plus minus 8 lesions. And then the total volume of these lesions on the average was 6.8CC plus one is four. Their highest was 13CCs, and their lowest was 2.58. We tried various couch positions and number of arcs, and we lowered the V12 and to some extent, significantly for some cases. But the overall volume dropped from 330CCs to 199CC, which is pretty good. It's not statistically significant, but nevertheless, it's minimized. But if you look at our conformity indices, the average is 1.39. And for the optimized one and regular is 1.55, which is significant, a p-value of 0.039.

And this is actually also reflected in the number of MUs that we use. The min MUs for optimized one is 13,566 versus 16,000. So we know as your MUs are lower, your leakage is less, and conformity is improved. So it has to affect your V12. And you could see it here. This is outpatient number eight. Twelve lesions, conformity index dropped to 1.38. V12 is 12CC and non-optimized was 1.82 conformity index and V12 is 21.5. So it's a significant drop in V12 in this fashion. And you can see that the whole brain, that also shows 12CC for the optimized one compared with regular 21.5 for the 12 Gy volume.

Now, changing the subject from planning into delivery, delivery of the Multiple Met with single iso is very important in terms of the precision. As you know, we always added one millimeter in the margin to PTV because we think that as you go move away from the isocenter at even half a degree or 1-degree rotation, you can easily over-treat or under-treat that tumor. So we then want to add, like, 1.5 millimeter or 2 millimeter or 3 millimeter, depends on the position, and how far it was from the isocenter. So we thought that delivery is actually more important.

So to investigate that, we have the MapPhan Phantom with 445 solid-state detectors, and we defined 5 lesions in this varying from 0.17CC to 8.7CCs at the diameter of 0.7 centimeter to 3 centimeter. And the distance too from the isocenter vary from zero to 10 centimeter. So we defined that, and we did a plan of a single iso using VMAT two arcs. And we give 200 centigrade to each PTV, and you can see those coverage here and the distance from the iso. So the diameter of each lesion. And then, we positioned that isocenter at basically pitch, roll, and yaw as zero, zero, zero. We treated that Phantom. So we did 10 treatments and we also did the dose delivered to the Phantom, varying the pitch, roll, and yaws. And the pitch and roll varied by plus one is zero, 1, 1.5, and 2 degrees each, and the yaw, we changed it from zero, 1, 2, 3, and 4 degrees. We wanted to see the effect of it on our response.

So on a perfect position with 2% absolute dose at 2-millimeter distance to agreement, this is the plan on the right side. And on the left side, we see the measurement. And the agreements is almost 97% pass rate. And the line is...the profile of the plan and the dots are the actual measurement. But when we changed yaw, roll, and the pitch by 1 degrees each, that pass rate dropped to 83%. And that's pretty significant. And you can see that certain areas are underdosed here because of rotational error. And so, if you'd left uncorrected, obviously, you're underdosing in this region, a certain area you're overdosing it. This is the worst-case scenario, and the right is the plan and the left is the Phantom moved by 4 degrees by kicking the couch or yaw position. And you can see the difference here. So the information is gathered and tabularized here. So the perfect pass rate for absolute is 96% versus 95% for relative dose distribution. And the worst case is obviously 58%. And for both of them, we had a yaw 4 degrees.

And graphically, you can see the percent that failed those points versus rotational angle in degrees for absolute and relative dose distribution. And you can see as you're having more error, or if you don't correct your roll, pitch, and yaw there, basically your failed position, or, sorry, the failed dose points significantly goes higher. So it basically, as most 3D rotational correction algorithms perform their required rotational adjustments about the isocenter, inadequate rotation correction causes undertreatment or overtreatment of the lesions far from the isocenter. And we plotted the pass rate versus the degrees. And we wanted to have better than 90% pass rate, which we ended up with half a degrees. And so to have a optimal treatment, you need to set your ExacTrac corrections at a lower limit of the limit of half the degrees and half a millimeter, which we have been doing it.

And that's actually the most important part of this I wanted to emphasize. We can ignore this, let's say, if you add more margins. If you don't do it, or if you have a system that's not able to reach to this accuracy, obviously, you have to add more margins for PTVs or, sorry, for GTVs farther away from the isocenter. But we know that ExacTrac is able to achieve these precisions and especially with 6D corrections and also validating it at every couch positions. And I emphasize that, especially the small lesions and all these Multi Mets, has to be corrected after every couch positions because even though we use by above masks, which is pretty rigid but the patients move, as well as couch positions, can vary. It depends on how old the couch is. And so you have to correct it if you wanna have a high-precision treatment.

So I picked one case to show here to emphasize that this patient had 14 small metastatic lesions to be treated as single isocenter. And the treatment time was about 25 minutes. But the reason I picked this because this is where the 14 lesions were very tiny. The GTVs varied from 0.01CC to 0.26CCs. The PTV obviously was GTV plus 1 millimeter. And the PTV volume was 0.05CC to, sorry, 0.05CC to 0.55CC. And we used the TrueBeam 6C couch with ExacTrac and 6XFFF with Millennium MLC. This patient had their lung cancer primary. The MapCheck Q/A pass with 96% and these are the... On the left side, you see the tumor and the right side is a four-and-a-half-month post-treatment MRI. So we have two lesions here and this is pretty much gone.

These are on the left side. This is another lesion. And you can see it's gone and this is another one and another one, which is pretty much gone. And another tiny lesion. This is another lesion and one here, and one here, and these two, which has totally responded. So looking at this, it means that we have been able, using our planning as well as the delivery algorithms that we develop, to hit all these 14 lesions that are all over the brain with single iso almost perfectly. So in conclusion, cancer will soon become the number one cause of death in the U.S. Current noninvasive radio surgical techniques in combination with either chemotherapy, target therapy, and immunotherapies improve patient survival, quality of life, and local controls.

Technology will change how cancer is prevented, detected, and treated in amazing new ways. For patients with multiple brain metastases, utilizing image-guided radiosurgery with single isocentric technique, using Brainlab's forward plan, dynamic arcs, reduce their planning, and on-table treatment time. The radiosurgical treatment approach helps our patients with multiple brain metastases to have access to this new technology that we would have otherwise given whole brain radiation, or we would have sent them to hospice. The steps can be taken by the physicists to optimize multiple metastasis Element cases to minimize the V12 Gy dose similar to the examples presented here. Lower monitor units allow for lower integral whole brain dose than would be expected with a similar volumetric intensity-modulated arc therapy or VMAT plans, which is definitely more MUs.

The patient cases presented illustrate how technology and innovations are providing new hope for cancer patients and their loved ones. I would like to acknowledge our colleagues, physicists, as well as radiation oncologists, and especially Dr. Jang who did a lot of research in evaluating our 1,100 lesion patients, 133 patients. Thank you. And we are ready to provide the best care for our Kaiser Permanente members. Thank you for your attention. I'll be glad to answer any questions.

Interviewer: Thank you, Dr. Girvigian. Thank you, Dr. Rahimian. If you would like to turn on your cameras, we have some questions for you.

Dr. Rahimian: Can you hear me?

Interviewer: I can hear you. Yes.

Dr. Rahimian: Perfect.

Interviewer: All right. Dr. Rahimian, maybe I'll start with you. There are some technical questions in terms of Q/A and also treatment efficiency. So you've showed a few cases where you are using more than one isocenter. Can you explain how you cluster the tumors? So how do you pick the isocenters and, also, how do you do Q/A for those plans? Do you do it per isocenter, or do you do it per lesion or for the entire plan?

Dr. Rahimian: Good morning. I put...

Dr. Girvigian: The Q/A is really a question for Dr. Rahimian. Is he on?

Interviewer: Yes, he's on.

Dr. Rahimian: Yes. So, yeah, it depends on the distribution of the lesions and volumes. And typically, on single iso, especially if you have lesions more anterior or near the orbital areas, those distribution doesn't look as good as it should with single iso and especially if most of them are postural lesions. So we have done a few cases that we had to switch to two isocenter even to, let's say, 10 lesions with single ISO and then one or two anterior lesions with another isocenter, and the dose distribution V12 is definitely more improved. And with that, we typically do two measurements for each isocenter. And that way, the Q/A is pretty quick. Especially with TrueBeam we can automate, and the beams are basically running. You press a button and you can run it and analyze it.

So in terms of workflow, we have, like, two or three Element cases. It happened that we plan them, and they come on a Wednesday or Friday. We plan them. And we do the Q/A at noon, at lunchtime, and there, we treat them around 1:30-ish or 2:00. So it's extremely efficient. In terms of the, again, pass rate with Map Phantom, we have 445 diodes, and the distance to agreement typically for a smaller lesion is 1 millimeter and then up to 5%. And usually over 97%, 98%. It's definitely more than 95% pass rate.

Interviewer: Maybe just a follow-up question on this. Have you noticed any differences between the unflattened beam profiles versus the flattened profiles in terms of treatment and delivery?

Dr. Rahimian: The delivery time is definitely faster. And when you get used to flattening filter-free, you feel it's too slow to do a 6X. The quality of plans is sometimes even superior with flattening filter-free, and as you know, the mean energy is lower. So that's basically our... I could say over 95% of our cases are done with flattening filter-free nowadays.

Interviewer: Great. Dr. Girvigian, some controversies regarding predictors for radionecrosis and usefulness of evaluating V12 on the chat line. What do you do in your practice? And I know you also mentioned that you are obviously visually reviewing the symmetry around each tumor. So what are some of the practical aspects that you evaluate to try to determine if you'll have a higher probability of developing radionecrosis?

Dr. Girvigian: The lesional dosimetry, I think is one of the more important things. And we come from a time when we used frames and cones. So we look for homogeneity per lesion, which is why I think, as Dr. Rahimian was talking about, we prescribe to the 90% versus maybe 80% or lower. We look for meaning. Whole-brain dose V12 is controversial. I know, certainly, when you start treating multiple lesions, you see V12 being blown out of the water and you have to sort of adjust your expectations, which in my mind is kind of analogous to V5 lung dose for MRT in the thorax.

So we look at V12 and we try to maintain it at a reasonable level, but we realize we're often not gonna get the V12 below 10 cubic centimeters and it really hasn't mattered in our patients. As I reported, the rate of symptomatic radiation necrosis in our Elements patients has been very low, which is a good thing. We are an integrated group. So if patients do develop symptomatic radiation necrosis, we're gonna be pulled into it. But we definitely look at mean whole-brain dose. We definitely look at homogeneity of each individual lesion. I think that's really important to us, and this is why we might go from a single isocenter plan to maybe two or more isocenters, depending on what Dr. Rahimian thinks we can do to improve the homogeneity.

Dr. Rahimian: Can I add something here?

Interviewer: Please.

Dr. Rahimian: Okay. So I think the concept of especially hotspots if I'm coming from Gamma Knife, like, I have tried prescribing to 80% over and 50%, especially in Element. Now, you can plan it similar to Gamma Knife even. Your V12 goes down, but the question is a bit multi-med. Like, we treated a patient yesterday, like, 14 lesions and the overall volume was 2CCs. So if you prescribe something like that to 50% or 80%, you'd definitely get radionecrosis in tiny lesions. So you have to really evaluate what kind of a tumor size you have. And obviously, we prescribe 90% because we have such a huge database from our classical Novalis that's actually safe.

And our number of radionecrosis that we got in that 1,100, at least the recorded ones, it was about 18 cases only. So some could be a loss to follow-ups and things like that, but nevertheless, it's really minimal. So it all depends. If you have a large tumor, central necrotic already, so you can actually prescribe to 80% so you can get a hotspot in the center because of hypoxia. That's advantageous. But in these cases that we are talking about, these are very small lesions. And I don't think that V12 becomes very relevant about it.

Interviewer: Okay. Thank you for that. Dr. Girvigian, another question for you in terms of overall efficiency for treating patients through your department from intake to actual treatment. Have you seen any differences as you switched to Elements? And maybe you can also discuss a little bit how long does it take you for some of your planning steps to be completed such as contouring and plan evaluation, etc.

Dr. Girvigian: Okay. So we'll start with the planning. The auto segmentation in Elements is quite good. So we were iPlan users and before that, we were BrainSCAN users. iPlan has auto segmentation too, but we need to be careful and make sure that we go back and correct that. Elements' auto segmentation is superior to iPlan and we review it, but often don't need to correct it. So that saves a lot of time, especially with the optic apparatus. It does quite a good job. I'm very pleased with that. Then contouring each lesion, that takes time, but there are fast ways to do it with SmartBrush and 3D brush with Elements. The planning, it's a lot faster with Elements than planning each individual lesion like we used to do. That could take us quite a bit of time.

If we were planning, say five lesions, and we had one iso per lesion, and we had to make sure that the arcs didn't intersect, we had to look at the 3D picture because we didn't want dose contribution from one lesion to another, which ruins the lesional dosimetry, that took quite a bit of time, but Elements, we don't have to worry about that. In terms of treatment delivery, when you come from a time where treatment delivery was maybe 10 to 20 minutes per lesion, on a good day, faster with cones maybe, and you're thinking about treating 6 to 10 lesions and so you're looking at an hour plus. Now, it's a mean of 25 minutes per total delivery regardless of lesion number. That's a significant improvement, and that's what we were looking for because we have so many patients that need access to this technology we needed that efficiency, and that's what drove us to make this decision.

Interviewer: We have some Gamma Knife users on the webinars as well. I know there were some questions that you partially answered regarding either dosimetry gains or treatment time gains. What is your take on essentially a individualized-centered treatment strategy versus a single isocenter strategy? You addressed that [crosstalk 01:00:58].

Dr. Girvigian: Right. Right. I did partially answer that on the chat. The treatment delivery gains I think I just mentioned because Gamma Knife, I think is, you know, again, about 10 to 20 minutes per lesion, depending on. Your Perfexion Icon is, it's fast, and I understand that it's a lot more automated than it used to be. But still, it's at least 10 minutes per lesion. So the gains are there. In terms of dosimetric gains with Elements to gamma plan, in my mind, I think of it as looking for more of an equivalency. We know Gamma Knife is unparalleled in terms of its conformity. So what we're looking for is equivalency in the face of significant efficiency gains. That's how I look at this. We looked at Gamma Knife perfection at the time. It's a great device. The lesional dosimetry is fantastic and it's conformal. But if I have to treat 4 patients in an afternoon that I have a total of 40 lesions, how am I going to do that on Perfexion whereas I can be done in three hours with the Elements plan and the TrueBeam STX with FFF mode?

Interviewer: There's was a question regarding the implementation of contrast [inaudible 01:02:31] analysis imaging. Obviously, we've covered this in the previous webinar. But in your practice, maybe you can tell us how do you schedule this with radiology and how long do you have to keep the patients in the department on the scanner for the two scans?

Dr. Girvigian: Well, this is something we're still working out and those logistics have been a bit of a challenge in terms of working out with radiology how to order these tests. We're still trying to work that out with them and come up with an order in our electronic medical record that can be more automated than what we're doing now. But basically, these patients need two MRIs. And the first one, as we mentioned, is 5 minutes after gadolinium injection and the second one is an hour to 105 minutes after. So it does tie up the MRI scanner depending. But if we notify the radiology department ahead of time and they know what's coming, they can think about putting in a couple of quick MRI tests for different patients in between so they don't lose that dead space on the MRI scanner. And in terms of, what was the other part of the question, in terms of how we use this, or...?

Interviewer: Well, in terms of, again, how long do you keep the patient on the scanner, but [inaudible 01:03:57].

Dr. Girvigian: Okay. Yeah, so the patient doesn't need to stay in the MRI scanner. The patient needs the first scan and it just needs to be a T1 post-contract. You don't need the entire series that you would do for a standard MRI. And then the patient can leave and needs to come back in an hour. So that's the logistical issue of the MRI is how to schedule those so that you don't have that dead space on the MRI scanner. And if we notify our radiologists ahead of time, they can plan simple sequences in the interim so that they don't lose that time. I don't know if that answers the question.

Interviewer: Right. Here's a question maybe for both of you. What is, for you, the minimum target size for a brain met that you would treat with the MLC versus the cone, given that you have such an experience in the past with using cones as well?

Dr. Rahimian: I think the smallest, obviously, we have treated and a lot of them also 0.01CC, which is about 1 millimeter or 2 millimeter to 2.5 millimeter. With PTV when you add a millimeter, that actually is basically covered by two high def MLCs and is very conformal. And they respond. I mean, the example I showed you, we have a lot of those examples and that totally responds. So I think that's something that I think the new version, it limits you to 0.01CC. It doesn't actually plan it anymore. So while you could definitely hit these targets, I think, again, I emphasize that the delivery accuracy is really, and the precision of that is very important. The Q/A of the linear accelerator, you're checking your isocentricities, and all of that.

And especially this is actually maybe describing the publications that we have done, you're dealing with multiple parameters, like the resolution of MRI has to be high. Your resolution of CT has to be high. Your fusion has to be...usually, fusion is the weakest link and you have to have a very good fusion. Otherwise, if you emphasize every step from imaging all the way to planning and delivery, you definitely can't hit all these targets. So the Q/A is extremely important and that way, you can actually hit all these targets. I mean, but the other issue with ExacTrac that you might not have it with any other system is that you are able to image every couch angle. You can kick the couch and do multiple arcs, but you have to correct for patient motion. You can have problems with couch because the couch travels, especially laterally. Sometimes, patient move in and out. So if you correct all of that, I don't see any reason not to treat this lesion this, especially with high-definition MLCs.

Interviewer: One question regarding electronic QA and diode arrays. How do you account for angular dependence for the diode response and part of your commissioning? Have you checked any of these readings against film Q/A, for example?

Dr. Rahimian: Well, we have both ArcCHECK and Map Phantom. Obviously, the diodes are directional-dependent and they have directional dependencies. We check with ArcCHECK and ArcCHECK obviously is like a cylindrical [inaudible 01:07:54] from Sun Nuclear and we use it as a comparative. I think my colleague, Dr. Zhang from Anaheim, compared it, and it's very similar. So in case if it doesn't pass with MapPhan, for instance, we have checked it. And it passes definitely with that because of the beams or beams are in form of arcs, and that dependency goes away with ArcCHECK. So we are very confident about our accuracy of those distributions. And now, we have close to almost 400 patients treated so, and the results are really good. So we definitely, we are confident about it.

Interviewer: Right. Well, gentlemen, thank you very much for your presentations and for taking the questions. And thank you all for joining the webinar, and we'll see you on our next one. Thank you, and have a great day.

Dr. Rahimian: Thank you for the invitation.

Dr. Girvigian: Thank you very much, everyone.

Dr. Rahimian: Bye-bye.