Slides
Transcript
Hello again. I’m going to be talking about dynamic CT myelography and why I love this. I don’t have any disclosures. What I would say is that no matter what type of myelography you choose, whether it’s DSM or dynamic CT myelography, it is so much more than just locating the leak. It really matters because this is about providing targeted treatment. And it’s not just here we found the leak, you can stop just there. You can derive so much information from your myelography that can help guide your treatment, even the treatment the patient might have, as well as potentially risk stratifying that patient. It can give you a real rich minefield of information.
This is what I’m going to cover today. I’m going to talk about some of the basic techniques for those just starting out in CT myelography, the advantages and disadvantages of the technique, and then some pearls and pitfalls that I’ve learned along the way.
Just to understand some basic techniques. We heard a lot yesterday about the different types of CTM, about the type 1, type 2, type 3 leaks, and then other types of leaks we talked about. And if we brought these basic principles, we already know what we should find when we’re approaching dynamic CT myelography. We can broadly group these leaks into those that have epidural fluid and those that don’t when we think about the types of dynamic CT myelography.
And we look at the positioning of the patient guided by what we’re expecting to find. These are the broad types of CTM: prone dynamic for those ventral tears, decubitus when we’re thinking about a more lateral leak, and then decubitus dynamic when we’re looking for CSF-venous fistulas. Obviously, there are other types of leak, and we can modify our CTM technique, but these are the broad groupings.
And what the basic technique around this entails is that when we’re talking about dural tear type leaks, we want some kind of quick or repeated scanning that is going to capture that leak in real time.
At the CSF-venous fistulas, we consider that it’s much more important to have a dense layer of contrast. But this is also a dynamic technique. We keep mentioning dynamic, dynamic. What does dynamic actually mean? Well, I take this to mean detecting the change in the contrast that you’re putting in over time. But I think it’s much more important that dynamic refers to your operator, that you are adjusting what you’re doing in real time and making sure that the myelography that you perform on the table is adapted to what you’re seeing.
I don’t think that you can just think that you’re going to put contrast in and find a leak. You’ve really got to be looking at what you’re finding. And we’re going to show you some examples of that. Just like DSM, I think CTM is an operator-dependent technique. And really applying some of the pearls and the pitfalls and the things that I’ve learned along the way hopefully will help those just starting out. I do think it’s a more forgiving modality than DSM. Basically, you don’t have to be as good. So when you’re starting, I think this is a really easier modality to start with.
These are the two most common positions that we use for CT myelography: prone when we’re looking mainly for those ventral tears, and then lateral decubitus for those lateral type leaks and CSF-venous fistulas. Now, I work in the NHS predominantly, so this is the setup I have. If you work in some fancy institutions in the US, Andy Callen invented this table which enables you to position the patient nicely and dynamically.
These are the types of myelography approaches that we’ve known and learned about in the last few years. When we talk about this ultra-fast dynamic CT myelography, we’re talking about injecting and scanning the patient while we’re injecting. This involves multiple passes of the spine while we’re scanning. But clearly, that’s a very high radiation dose.
I don’t really use that technique anymore because it’s a high radiation dose. And Mark Mamlouk came up with this incremental dynamic CTM, which is where we inject small boluses, we scan, and then we quickly repeat only if we need to, if we haven’t seen the leak. We really love this method.
Here are these sort of classic examples that we know. Usually, these ventral tears occur most often higher up in the thoracic spine, T12, T2–T3, T3–T4. These are classic appearances that we see of this split in the contrast column and the egress, which usually occurs in a median or paramedian location, but they can occur more laterally. Usually in the upper thoracic spine that we see these.
I’m going to give you these really nice examples in talks, but sometimes they’re just not always like this. Sometimes they’re much more subtle, and this really comes down to being ruthless in interrogating your CT myelogram to find these points of leaks, because they can be very subtle. What we’ve noticed is that it’s often not correlated to the size of the hole. Sometimes you can have a large hole at surgery and it can still not be that obvious on the CT myelogram.
Then to those type 2 lateral leaks, we’re going to position the patient on the side that we think is most likely, and we’re going to again capture it in a dynamic way. Now for CSF-venous fistulas, you need to examine both sides of the spine. There are different ways of doing this. You can inject the contrast, you can turn the patient, you can do it in two different sittings.
We like to do this in one sitting because we find that we’re pretty successful with this. We do a staged contrast bolus. We turn the patient with the needle in situ, and we get really nice layering of the spine. The great thing about this is because we have really limited capacity, we have learned to examine the whole spine in one day. We do this simply by keeping the contrast in the spine. We want to make sure the patient’s head is up, and we are able to tilt the patient in various directions so that we can get a whole-day coverage of the skull base to the sacrum in one sitting. We really like this because it’s really convenient for patients and our department.
Diogo Edelmuth and Peter Kranz published on the safety and technical performance of contrast density and using higher volumes of contrast. This was really important because there had been very historical data suggesting that we couldn’t go up in dose, but we have found that it is incredibly important. We have a poster outside about how we do our exact technique, but we have been able to go up in those contrast volumes per side, and we find that we’re getting great pickup with this.
But I would say that it’s important to elevate your head. It keeps the contrast in the spine and it also keeps your patient happy. When the contrast goes to the head, they’re going to be pretty unhappy with you. So we give intraprocedural monitoring and also just make sure that they’re having adequate analgesia, but keeping that head up makes it more pleasant for them. We love this technique for fistulas. We find a lot of fistulas with this and we’re able to characterize them fully, and that is what I think is really important.
Let me talk about some of the advantages of CTM. I do use DSM sometimes, but this is often what I’m doing when I’m looking at DSM. I’m not as talented as Marcel. I can often really struggle to find these leaks, especially when it’s higher up in the thoracic spine where a lot of these ventral leaks occur. With CTM, you don’t have that problem. You can be very standardized about how you’re seeing this and you know exactly where you’re seeing it.
But Niklas Lützen did recently publish on a modified technique of using the arm upper and a pump inflator to see those upper thoracic leaks. So if you’re using predominantly DSM, I would strongly suggest you read that paper. The other thing about DSM is that I’ve often had problems with identifying the levels because of parallax. You can use a needle to help you or external markers, but sometimes that can take time.
With CTM, you can have absolute confidence in levels. So before every patient, any patient that’s going to surgery, I prepare one of these PowerPoints for my surgeon, and we go through it together because he’s going to say to me, what’s the segmentation, show me exactly, so that we can be really, really clear about the levels.
Not just that, CTM provides you a lot of cross-sectional detail, so it can be really helpful for when the leak is slightly off midline or when you’re getting a slight degree of spinal cord herniation. In this case, this was a very obvious spinal cord hernia, but we sometimes find that with those upper thoracic leaks there is a little bit of cord herniation, and that information can be very useful to tell the surgeon.
In this case, for example, if you watch this video, this was a really nice example. On the CTM you can see that there’s a nerve root that is going out to a party in the epidural space. You can see that nerve root just jumping out, and that’s again useful information to be able to tell your surgeon.
Also, we use cross-sectional anatomy for treatment planning. I told you yesterday how we do a same day treatment after the myelography, so we find the CT technique really, really useful.
Now what I would say is really important is we reconstruct the image so that it’s on a very small field of view. Basically, I don’t want to see all of this because I might find stuff that I don’t really want to find. But also we want to reconstruct on the raw data, and that is important that you do that at the time of acquisition or within the few hours afterwards because that raw data goes from your scanner after a week or so, or sometimes shorter.
And you can see here that you would probably miss the leak if you hadn’t reconstructed that into a really small field of view. Whereas here it’s obvious. And sometimes you have to repeat an examination of the same region.
So in this exam we’d given a bolus of 3 mL. We’d been given great layering. I actually really suspected that this patient was going to be leaking. I was like I can’t see the leak where I think I see this flow void sign. And then we gave a repeat bolus, and you can see that there was a dense column of contrast, but it just took time to be able to, and maybe perhaps a little bit more weight of the additional contrast, to do this.
So that’s what I would say is that there’s no way that the CTM can compare to DSM in terms of temporal resolution, but if you give a quick repeat bolus you can still capture some of that information.
Now sometimes with these type 2 leaks they sometimes leak very fast. This was all within the space of 15 seconds, and you can see that there’s a very rapid leak, a torrential leak from this type 2 leak. But actually in our experience, we find that these type 2 leaks are often slower fast leaks. So this one, there’s barely evident epidural contrast in this region and that’s at 1 minute, and then after 2 minutes it becomes more obvious. So we don’t do this fast technique with lateral leaks anymore. We just give a bolus and we scan over that focused region. We wait, and then if we need to, we’ll scan again.
Now, Niklas Lützen published this amazing paper on the sensitivity of CT myelography for CSF-venous fistulas versus DSM and found that it basically picked up three times as many fistulas. I think that’s because CT has an inherently higher contrast resolution for tiny opacified veins, and also veins that are sometimes not in, that are directly in front of the vertebral body or behind, can sometimes be missed if you’re not acquiring a biplanar acquisition. We’ve certainly found this as well.
In our series, over a 24-month period, we found that we can find the fistula if it’s there in pretty much 91% of patients where they’re appropriately selected, and that’s in one sitting. I think that that is important for the patient, right? If you’re bringing a patient in, you want to make sure that you can try and get as much information out of that myelogram as possible without having to bring them back. We think it’s more convenient for the patient, and it’s certainly more efficient for my department.
The great thing about CSF-venous fistulas with CTM is that you can scan as much as you need of the spine. You’re not limited by the size of your detector. And once you’ve done that run, you can do it again. You can give another bolus if you need to and increase that density. We have found that we have bilateral fistulas in 11% of our patients. As I talked to you about yesterday, we are finding multiple fistulas in more and more of these patients as well. That’s the great thing about the same day technique. You will find more fistulas, and that’s important when you think about how you’re going to treat the patient and how you’re going to risk stratify them for the future.
This paper, as I told you, is hot in press from Ajay Madhavan and our international collaborators throughout several of these groups that are here today, where we’re finding more and more of these bilateral fistulas. In this case, I sort of had to stop looking because we just kept finding more and more and more fistulas. These cases are really challenging to treat, as I mentioned yesterday, but I think they’re probably underdiagnosed. Just finding the leak is not enough. I think it’s really important that you get as much information that can guide your treatment.
So, a few years ago, I might have seen this and thought that the fistula was at this level or maybe even at this level, but here you can see that the fistula is clearly at the level below. You’ve really got to identify the fistula site. There’s no point treating the drainage. And also mapping those drainage patterns can help you in terms of how you do your future drainage, whether it’s approach for how you’re going to treat with fibrin glue or embolization or your surgical approach.
When we go through things as a team, we always talk about the fistula drainage pattern so that we can talk about how we’re going to treat this patient. I mentioned these extra spinal findings. Often we don’t see this, and there was a paper from the Duke group which didn’t really help with this, which helped as a field. But what I’m saying is that if you don’t see it, it’s not that helpful. But when you do see it, and you’ve maybe not seen the fistula—so this was an external myelogram we were asked to review—you can see that there’s increased collection in the kidneys. This was called negative, but actually when we went through it with a fine-tooth comb, you can see that there is a fistula here. It’s one of these things that if you haven’t found the fistula and you see it, go back and really interrogate your scan again.
Well, CTM is not perfect. And I think the biggest issue that we’re facing is that it’s a high radiation dose. And there’s no way of getting out of that. It is a high radiation dose. And if you’re going to do multiple runs of the spine, it’s an even higher radiation dose. And in the UK at least, they really don’t like radiation, so they really want us to keep our radiation doses low. This is important, and we’re going to talk about how we can reduce that radiation dose in a minute.
The other thing with CT myelography is that you really have to think about how quickly you’re doing this because sometimes you can be too late when you give the contrast. Here is an example where we’ve given the contrast, the contrast has flooded the collection, and we’ve really obscured the point of leak. But I think doing that small incremental dose can really prevent that from happening.
One of the ways that we decrease the radiation dose is we play around with our parameter settings, and we found by doing this we can over half the radiation dose per acquisition but also limit the coverage. We really interrogate our MRI studies to really help us try and predict where we might see the leak, and we don’t tend to do so many passes if it’s not necessary.
I mentioned yesterday about this bottom branch sign that Niklas Lützen published on this year. This is a really, really good sign. When you see this, we find it really sensitive and specific, and we see it now on so many of these cases. For example, in this case they had two lateral leaks, but in that you can really just reduce what you’re going to scan to that area. We really only scan that area because we are so convinced we’re going to find the leak. Obviously, sometimes we may be wrong, but we can extend our range out. You’re not limited by this on CT myelography. You can repeat bolus if you’ve kept your bolus low.
Okay. I’m just going to finally finish up with a couple of difficult cases. I think this is the strongest point that I want to give to anybody that’s starting out in CT myelography, which is you’re only as good as the team you work with. And I think it’s super important that you engage your radiographers, your techs on this journey with you, because they are as important to helping you find that leak as you are. And engage your patient. And as Ian Carroll said to me, he said, “It’s like you’re doing a choreographed dance with your patient. It’s really important that an engaged patient will also help you in terms of whether you’re going to be successful in finding your leak.”
What I mean by this is we are super anal about how we position our patient. Sometimes the techs at the beginning were like, “Oh, come on. Let’s get on.” But honestly, this makes the difference sometimes between capturing the leak and not. So, spend your time positioning. You’ve got to have enough cranial flow of contrast if you’re looking for those ventral leaks. Make sure your patient is not rotated. This I don’t like to see when we do that first capture of when we put the needle in. We straighten out the patient. We want to get a nice layering dose and avoid this.
We often see CTMs that are like this at the beginning or external reviews. You really want to make sure you’ve got them in that exactly lateral position, because you can see here that if you look at where this contrast column is, and if you have a fistula here, you’re totally going to miss it with this.
And make sure that you’ve got enough coverage. This is a patient we gave—actually, we only gave 1 ml bolus here. And you can see we haven’t got a really dense layer the full way across the theca. And you can see on the run two, I gave another bolus. We’ve got a really nice coating there, and you can see that at that point it’s a more lateral leak. So make sure you’re not missing that.
Sometimes you may see appearances like this. We often see this in the upper thoracic spine when we can’t get a really nice dense coating at the front, and we don’t know why that is. Maybe it’s secondary to the membranes. Maybe we’re just not getting a great contrast column up there. Sometimes we just need to give another bolus repeat. One of the things that we’ve found really helpful is a Valsalva maneuver, because sometimes the cord can be adherent to the front of the defect. Doing that Valsalva lifts the cord from the front of the defect and helps that contrast come out.
Some leaks are just really difficult to find. This was a patient who had siderosis, a very shallow ventral collection. We did three CTMs in this patient and we found this really difficult to find. On the second CTM, on the last image, we noticed that there was this tiny little puddle here, and that helped us localize it to one vertebral body. But that’s not enough for me. I want to see the leak. I could then bring them back, be really focused, and then I don’t know if this is projecting this well with the lights, but I can tell you that there is definitely a streak here and you can definitely see that egress there. We only saw this on one of the runs because it was an intermittent leak.
Andy Callen has published on the timing characteristics. This is why we do two runs, because we do feel that we get additional information. Sometimes you don’t have a great first run. But what Peter Kranz showed and Diogo showed with this great paper was that actually they think that the contrast density is much more important, not time. If you look between these two runs, you have better density on the second run. It’s important as an operator you’re evaluating what’s going on and thinking, what do I need to do?
Here you can see the fistula is more obvious on the first run, but the density is better on the first run. Sometimes timing does matter as well, though, because look, the density is better here and you can see it better on the first run. So this is why we do find this useful.
But really I’m going to stress this point. As an operator, you can’t just give the contrast and then hope for the best later. You need to be evaluating that scan as it’s happening. We gave this, this was one of our earlier ones, and I thought this density here is terrible. We’re going to give a little bit more bolus and we’re going to lift this patient up a little bit more because what I saw here was this, and I didn’t know whether that was going to be a fistula. And look what happens when we’ve got better density. This was a CSF-venous malformation.
I’m just going to finish up by saying I don’t use photon. We’re actually getting one, it’s very exciting. But this was kindly given to me by Ajay Madhavan who’s an absolute expert in photon CT. Clearly this is going to be a lot of the future in finding a lot more of these subtle leaks.
And finally, it’s not just about making your myelogram great. It’s also about how you convey these results to the patients. We’ve had these two papers in the last six months which help us clarify how we can report these findings to patients.
So I think CTM is great for localization, but remember it can give you a lot more information than that which can help your treatment. But really, your meticulous technique matters. Engage your team and your patient.
Okay, and thank you to my amazing team.
Controversies in SIH: Dr. Marcel Maya on Digital Subtraction Myelography (DSM)
That was an amazing lecture. Thank you. Wow, that was superb. When Wouter showed me the program last winter, I felt like an NBA coach, Laker coach, and I marked my calendar for this game, Lakers against Celtics, and today is the day. So, I’m excited about this. I hope you are too.
So the basics, talk about background, how we do it, show some examples, and talk about really the challenges and potential solutions with DSM. We’ve been doing this for a while and we’ve done many patients, and we obviously like to do DSM because we find it to be useful, effective, and it works in our system.
Just a brief history before that. This is a review that is actually still valid. I think there’s one update to this JAMA review where DSM was superior in a meta-analysis, and that’s as Lal mentioned, Niklas’ paper about comparing it head-to-head.
Brief history: 2002 is when the first DSM was published for a pseudomeningocele at the skull base and obviously showed the leak, an obvious leak there. And then Hoxworth really was the groundbreaking paper, and that was in 2009, showed it in a ventral leak, a spontaneous leak, and we quickly adopted that. We started doing DSMs and then we discovered the CSF-venous fistulas in 2013. But a major break came with Rich Farb up in Toronto, who advocated a lateral decubitus scanning with DSM, which markedly increased the rate of positives in our hands, at least from 15% to 75%. That was an awesome step in the right direction.
You can see here with a positive brain MRI we had results of 75%. Negative brain MRI, it’s 10% is still the quote, and it’s lower. Even though their brain MRI is positive, with this subset of FTD patients it is still lower.
A few examples of ventral leak and DSMUs. Here you have a patient who has an extensive leak from C1–C2 to lumbar spine, and obviously it’s important to select the right area to examine. Typically we go for the money, which is in the thoracic spine, and this shows an obvious leak, no problem. But patient position, anatomy is really critical to address, and here you have mild kyphosis with positioning of the patient, and then you have a leak that is seen over a few segments on the lateral view.
Even seeing the leak in one or two segments is not good enough. You have to localize to the exact level. It can’t be, oh it’s at T3 or T4. Well, what is it, T3 or T4? He wants to know and we need to be able to figure it out.
Here is an example where AP really helped out. You can see on the right side that middle weathervane-like midline defect that’s kind of poking at us and showing us where the leak is, just to the right of midline on the AP view.
This is a patient actually showing the same thing. She was operated at Cedars and then unfortunately she had a persistent leak after surgery. She came back and we repeated the DSM, and again it was very difficult on the lateral. You can see it over two or three segments on the lateral view, but when you look at the AP you can clearly see which level it is at. So that helped us, and this time the leak was appropriately addressed by Wouter.
Obviously, many leaks are high up at T1–T2. That’s a common area of ventral leak due to microspurs and disc herniations, and this is a difficult area in X-ray because of shoulders. In this patient, it was quite broad-shouldered, and we were looking for this source. You can see how kyphotic, lordotic the lumbar spine is. This created a problem for us where we couldn’t see the T1–T2 area no matter what we did. We tried from the cervical spine, from the thoracic spine. We ended up relying on the CT with the microspur. Luckily it was the only one and it was the right level. So it’s not perfect. It can fail, but we have a CT myelogram to fail back on. Mind you, this is not a dynamic CT, it’s just a regular CT done 40 minutes or half an hour after the DSM.
Let’s see, so examples of lateral leak. Obvious classic example, again circumferential collection mostly on the sides and on the back. And here you see the myelogram showing a nice finding that Lal showed earlier. And the leak is obviously opacified on the AP view. And then secondarily you see the collection up and down alongside the dural sac.
I think we talked about this — these lateral leaks that have some variations in them depending on where they are. But the more interesting and the more recent realization is the presence of these small collections which at first were kind of suspicious whether they’re real or not, but turned out, as was shown yesterday as well, that they’re real. They’re either small tears or small veins, CSF-venous fistulas that are right below the nerve root sleeve, and mostly they occur in the lower thoracic spine.
Now, I’m not claiming that they cannot be seen with the dynamic CT, and they can be seen, and they have been seen even in our cases, but DSM certainly masks out everything and brings it to fore.
Here are a few examples of fistula visualization with DSM. Timing is important, as Lal mentioned, and we know from literature that these occur early on, most of them within 30 seconds. Here you have a fast one within a couple of seconds of arrival of contrast.
It is important to check the AP and the lateral. Sometimes you don’t see it clearly on the AP, you see it on the lateral. This really is important for us, as I think Lal alluded to in the early literature in the paraspinal vein with CTM, because of the presence of the vein that was seen in the CT. There have been some cases where the localization was not accurate—one level up, or one level down, or across on the other side. But this temporal resolution does allow for precise localization right and left end levels.
Rare leaks in the sacrum — examples mostly alluded to as a leakage just from the cyst. However, in this patient, you can see it’s not the cyst. It’s actually a real fistula. This was a difficult case, as you can imagine. In the end, through a combination of fibrin glue, surgery, etc., we were able to give her some relief and reduce the syrinx in this patient.
Here’s a subtle fistula up in the cervical spine, and you can see a small vein coming up. Here’s a more subtle one that I think you can just be forgiven for not seeing. It’s a tiny vein right below the nerve root, but it’s important to compare to CT or MR myelogram that you have, so that these obviously do not show that vein because they’re temporally not related. But the extra density there represents a small subtle leak.
This is a recent patient that I had trouble with. This is a 75-year-old woman with an onset of symptoms. She googled the symptoms, she diagnosed herself with a CSF leak, and she went to the ED. They discovered CT bilateral subdural hematomas, and obviously she had a CSF leak.
This is her spine. She had childhood scoliosis correction. She’s completely solid, diffused up and down except for L4-5 where she has severe stenosis. At L5-1 is an opening, but L4-5 is tight. We tried with a puncture at C7-T1, did a myelogram, and we were able to determine the C7-T1 leak, as you can see here, and put some Onyx on it, and she did well.
This is an example of a discrepant level on CT myelogram, more recent, from a different center. She had this discovered while undergoing MRI for cervical spine MS follow-up. She was not symptomatic. However, because of the risk of siderosis, as she had beginnings of it, she was referred for surgery outside. They determined T8-9 was the culprit. But when we did it, we saw that T6-7 level, two levels up.
Post-embolization is also an opportunity for DSM to shine. CT-based techniques are more prone to artifact, and with DSM there is removal of that whole Onyx or clip so that you can see the leak if there’s a recurrent leak or residual leak. These are examples of such cases that were operated on subsequently.
So DSM has some challenges with shoulder artifact. You can use AP to eliminate that if you don’t see it on the lateral. Kyphosis, lordosis positioning is important. Multiple days, yes we do need multiple days. Sometimes we use the two-station technique, which we can eliminate some multiple-day studies. Opportunities and advantages are the precise and accurate level determination, the temporal resolution, the artifact as I mentioned. Radiation is significantly lower, and here I disagree with Lal. I think the CTM, the way Lal does it, requires a lot more skill and dedication from the radiologists. I feel like DSM is really pretty standard and probably less operator dependent.
Let’s see, I think both techniques have been proven effective. It depends on where you are, the availability, equipment, use of time slots. I mean, you have a CT room that you can easily access, you do CTs. Your experience in DSM, you do DSM. In California, Tesla is very popular, and I personally love my Tesla, but a lot of people don’t like Tesla and certainly this was a deal breaker when Trump decided to buy this Tesla. But all hope is not lost. They decided to sell it, so yeah, we’re okay.
All right, bottom line: sensitive, specific for leak site determination, accurate, safe, well tolerated, and radiation dose is better than the dynamic CT. So that’s all. Thank you.