Slides
Transcript
Good morning, everyone. I’m going to be talking about CT myelography for spinal CSF leak. I don’t have any relevant disclosures.
So, you’ve just heard from Dr. Lützen about DSM, and I’m going to talk to you about dynamic CT myelography. This is the main method I use to locate spinal CSF leaks, and I find it to be very successful.
Why do I use it? Well, partly it’s the imaging test that I have the most access to, so that’s the one I’ve been able to build up the best experience with and evolve and refine the technique. But there are other reasons too, and we’re going to cover that in this talk.
I just want to start out by showing a CTM that I did in 2019 versus 2024. This is on the same CT scanner. I don’t have access to a photon CT scanner, and I don’t have access to ultra-high resolution cone beam. So, what’s changed in this time is probably my technique and my understanding of the nuances of how best to detect CSF fistulas, and that’s important because CTM, just like DSM, is an operator-dependent technique. Training, experience, and an understanding of the different types of leaks matter. These can be difficult procedures to conduct successfully, and importantly, patients and operators may have had multiple bad past experiences, so it’s important to really bear that in mind when we have our patients in front of us.
In this talk, I’m going to explain what a CT myelogram is, what the different types of CTM that I do are, what I tell patients about the procedure, and what happens afterwards. There’s going to be some overlap with what Dr. Lützen already said, and hopefully some reiteration is useful.
In its simplest terms, CT myelography involves doing a lumbar puncture, injecting contrast medium to opacify the cerebrospinal fluid, and then taking images with a CT scanner. All this is to try and identify the point at which contrast is leaking out of its normal CSF space through a spinal CSF leak of whatever cause. But we often hear this term “dynamic CT myelography” – but what does dynamic actually mean? I interpret this to mean visualizing contrast moving or changing over time and how we interpret that to make the diagnosis. But it can also mean that as an operator, I’m reviewing the images in real time and adapting my technique to the procedure that I’m performing in order to get the most benefit out of that procedure. That can mean moving the patient or changing the procedure during it.
So, we’ve just heard about the main types of leaks, and the exact way in which I perform a CTM depends really on what kind of leak I think the patient has. Most, but not all, patients will have one of these three kinds of leaks as we’ve just heard, each of which I approach in a slightly different way. That’s mainly depending on how they’re going to be positioned as well as how we give the contrast. I’m going to talk about the nuances of this a bit later. But there are other types of leaks. We just heard about these sacral dural tears. I wasn’t picking up many of these, but after Dr. Lützen’s paper on sacral dural tears, we started to see a lot more of these. But there are also other types of leaks, including dural ectasia leaks, arachnoid blebs, and we just heard about this new leak that just came out, CSF to lymphatic fistula as well.
So, what needs to happen before we do a CT myelogram? So before a patient has a CT myelogram, I, as an operator, should have a good idea of what kind of leak I’m dealing with. By that, we mean whether the patient has a dural tear-type leak, associated with epidural fluid, and we see on the spinal MRI – or whether the patient has a CSF-venous fistula type leak. There are obviously exceptions to this, but as a guiding principle, it works most of the time.
So I’ll have been reviewing the MRI, and I’ll go through this with the patient to show them whether they have epidural fluid and where that epidural fluid is located, is incredibly useful. If it’s located predominantly at the front, then I’m thinking more that the patient has a ventral dural tear, whereas if it’s at the side or dorsally, I’m thinking that they may have more of a lateral leak. Sometimes it can just be pretty non-localizing on the MRI. Dr. Lützen mentioned this paper we just published we see this flow void sign in various forms over half the time, and we use this to change our myelographic strategy and reduce the range, and therefore reduce radiation exposure.
When I meet the patient before a procedure, I’m going to explain what we’re doing the CT myelography for – in order to hopefully find a site of spinal CSF leakage in order to plan treatment. I’m going to be telling them what does the procedure entail, how long it takes, what it will feel like, and special instructions that I might want them to do during the procedure, as well as what is going to happen afterwards. What I usually tell them is that it will be anywhere from around 30 minutes to an hour, sometimes a little bit longer depending on the type of myelogram I’m doing. We do this under local anesthetic and we often give light sedation. We find lorazepam to be really useful because it’s also protective against seizures.
I’m going to tell them what position they’re going to be lying in, which can be dependent on what kind of leak they have. I’ll often explain any special maneuvers I have, and I often warn them that you can get an injection headache, and that that can feel quite bad, but we try and take measures to avoid this. And then any time when I think you’re going to need to stay still, I try to reiterate the importance of how we’re going to tell people that. And then whether they’ll experience any back ache afterwards. And then usually what patients most want to know, is “When am I going to hear from you as to what I find?” Most of the time, I’ll say, “Look, if I find a leak on the table, I’ll tell you straight there and then. But I’m also going to come and see you afterwards and talk about what that means or any treatment.”
In terms of consent, Dr. Lützen mentioned a lot of these, but I tend to discuss those related to the lumbar puncture. We warn that there can be headache, bleeding, infection, and nerve injury. Most CSF leak patients worry about a new leak developing at the lumbar puncture site. And we talk about the use of non-cutting needles, small-gauge needles to try and minimize that risk. There are these rarer complications that can occur, an allergic reaction – which is incredibly rare to contrast – and seizures, which we haven’t had happen yet, but we always warn about that.
So, how do I actually do these procedures? I just want to caveat that by saying these are the techniques that work for me. There’s no one best way to do a CTM, and there are widespread variations on it. But essentially, when we think about the three main kinds of leak, we’re thinking about dural tear-type leaks, these are types 1 and 2 leaks. These leaks generally leak relatively fast; sometimes these lateral leaks can be a bit slower, but they require imaging that is more time-dependent, so quick, repeated scanning to capture the moment of the leak. Whereas for CSF-venous fistula-type leaks, we’re wanting different aspects of our myelography. And mainly I find that timing does matter, but also the density of the contrast really, really matters. These alter how we position the patient.
So, when we think the patient has a ventral dural tear, they’ll be prone – lying on their tummy on a wedge. Decubitus basically means that they’re lying on their side, and for CSF-venous fistulas, we have to do both sides. So, there are common features to these types of CT myelograms, but there are also specifics in each. So, the common feature is we do the lumbar puncture under CT guidance, we usually puncture around L3-4, but if there’s a collection that extends further than that, we make sure that we go below the level of the collection.
We check the needle is in place by giving a really tiny test injection of contrast. We do that because, sometimes, the needle can look like it’s central, but in patients with spinal CSF leak, they can often be under low tension, and so it’s important to make sure your needle is completely actually within the sac.
I position the patient so that the likely site of the CSF leak is downward, because contrast is denser than CSF and sinks dependently. We can’t tilt the CT table, so we have to tilt the patient to get contrast to run down the dependent portion of the spine under gravity. These are the two most common positions, and there are a variety of positioning devices out there, and these are some of the ones that we use. And what I tell patients is, this feels uncomfortable, especially lying like this for sometimes up to an hour. And I think telling patients that is often what my patients say was the worst thing about the procedure was lying in this position for a protracted period of time.
So, this is often the position we’re using when we’re looking for ventral dural tears, and this is often the position we’re using when we’re looking for fistulas. Note that we raise the head to try and avoid a headache, to try and create a trough in the spine. Sometimes, you only need a very gentle slope, so you can just sometimes use pillows. Sometimes wedges can be too harsh and create too much of a downward slope. Patients often say to me that “You spent so much time positioning.” But actually, what I’ve learned over the last 5 years of doing this, is that the more time you spend at the beginning – it’s less time wasted later on.
So, when there is ventral fluid, we’re thinking about doing a prone myelogram. When there’s more asymmetric fluid, and we’re thinking about a lateral leak, so one of these leaks that arises from the nerve root sleeve, usually at the “armpit” of the nerve root sleeve or the “shoulder”, then we will position the patient on the side. As I said, time spent positioning is often time well spent.
This patient had a very difficult spine to actually get the contrast over, and we thought we’d got a really good slope here, but as it turned out, we hadn’t, and the contrast just stayed there. So, I often spend a lot of time on this in the beginning, and I just warn patients about that.
So Dynamic CT myelography uses multiple scans to detect changes in that contrast density over time. We want to capture that earliest leakage into the epidural space, and this can be done in one of several ways.
Firstly, you can acquire multiple back-and-forth scans while injecting contrast during a long breath hold. Now, I find that this is actually quite difficult for a patient to sustain a breath hold for that long, whilst it is fast. If you don’t do enough passes, you can even miss a leak in the upper thoracic spine. But I find that this is a better technique where we inject a smaller volume of contrast, see if we see the leak, cause we’re imaging quickly, and then inject another bolus if we need to.
So, this is the typical appearances that we’re looking of a ventral dural tear. We’re seeing contrast dropping out of the sac into this epidural collection, and it produces this characteristically forked appearance on the sagittal recon of that.
This is an example where we’ve done it in a prone incremental way. But you can see that this was scanned quite rapidly, and you’d see that with this, you’d miss the point at which this leak is happening. You only really see it clearly on these initial runs. Sometimes, the chronic epidural collections can be really difficult, and we often see this appearance where the contrast doesn’t layer very well across the midline at the front. I’m not sure why that is. It could be due to the membranes; it could be the cord falling against the defect. But we’ve found that if we ask the patient to strain, like they’re going to pass stool, we do a Valsalva maneuver, that can often lift the cord away, and it can show you sometimes these very subtle jets that can be used to identify the leak.
For lateral nerve root sleeve tears, you can see here this patient had asymmetric fluid on their MRI spine, so we were already thinking we know where this leak might be coming from. We positioned the patient on their right side, and you can see here the exact point of the leak, where there’s irregular opacification of that right T10 nerve root sleeve and this epidural leak of contrast. Here’s one that’s more torrential, you actually see the jet of this leaking contrast into this characteristic puddle around that nerve root sleeve that then develops.
But when there’s no epidural fluid, this is when we see a CSF-venous fistula as the most likely cause, and this is when we’re going to be doing the lateral decubitus CT myelogram. There are lots of different ways to do this, but the key is you do have to examine both sides, because you can have – you don’t know which side it’s going to be on from the MRI, and you can have bilateral fistulas. We do it in a way where we always start with the left side, actually. We find that our fistulas are more common on the right, so we start with the left. We’ve been pretty consistent about doing that now. And that’s because there’s always a bed of contrast density on the right.
We do the lumbar puncture, we give contrast, we turn the patient with the needle in situ. Our team is really good at doing this now. We check our needle is in position afterwards, and then we give another bolus. We find that that’s really useful because you get another fresh layer of contrast density. It’s not the only way to do it. You can take the needle out, and you can give your bolus again by re-injecting.
There are sometimes additional techniques to try to increase the chance of finding a CSF-venous fistula, so we now always standardly inject saline at the start of the procedure to increase the CSF pressure. There is some limited evidence that this can improve detection rates from this paper from Dr. Lützen. Similarly, we have asked the patients to breathe in through a drinking straw or small syringe, which can sometimes help visualize CSV venous fistulas.
So, this is our typical appearance of a CSF-venous fistula. You’ll have all seen examples of this, where we’ve got the diverticulum, some foraminal veins around this, and a classic paraspinal draining vein. But there are variable appearances, and I think this is really important to map because it’s important for treatment planning, but also this can explain why I might not tell you the answer straight away. I want to go and look at the CTM and be able to show you some pictures of your myelography and say, “look, this is the treatment we’re now going to see.”
So, after the procedure, you’re going to have a two-to-four-hour bed rest with your head elevated. You can have some bathroom privileges, and then we give you analgesics and antiemetics as required. Now, how do I communicate what the results are? So, if there’s epidural fluid, we know there’s a dural defect, and if I found the leak, I’ll say, “The good news is I found your leak, and this is how we can treat it.”
But sometimes, we get diffuse opacification of the epidural collection, and that prevents the localization of the leak, so I’ll say, “I’ve shown your leak is active, but we need to do more work to pin it down.” Or, sometimes, we just don’t see the leak, so then I have to say to patients, “Look, we have to repeat your myelogram, and we might have to change our technique slightly or maybe try a different modality.” Or maybe no leak has occurred. And sometimes for that, that may be because there are membranes preventing contrast getting into the defect. Sometimes, we pre-pressurize, we try these strain maneuvers, or sometimes we got it wrong, we did the wrong position, and we have to change and repeat that.
Now, with CSF-venous fistulas, if we have unequivocal evidence like in this case, I can say, “Great, I found your CSF-venous fistula, and this is how we can treat it.” Or maybe the findings are a bit more indeterminate, and these subtle densities that might represent abnormal veins or just be artifacts. So I’ll say, “Well, there’s some suspicious areas, but it’s not definite.” Or maybe we see nothing at all, and because there’s no objective spine MRI evidence yet of a CVF, unlike with epidural fluid, I can’t be sure – that’s because there is no CVF or maybe we just haven’t demonstrated it. So, this is what I’ll say: “I haven’t been able to show you have a CVF as the cause of your symptoms.”
So finally, I love CT myelography. I think it’s one of the best techniques, and there are several reasons why. We can scan pretty much all of the spine, and often I will move up or down the contrast bolus in real time to look at the whole spine. It’s less prone to artifact, and I think it’s really good for leaks in the upper thoracic spine. And the cross-sectional anatomy of the CTM enables you to treatment plan. And importantly, for me, I think that these really small CSF-venous fistulas – really, I think you can only see them with CT myelography. And you can assess the quality of your myelogram dynamically. So, when I see this, that’s not good, because I’m not covering the whole nerve root sleeve, and I’ll move the patient in real time to get a better layering. So, I’m going to try and avoid that rotation, and I find that several phases can be helpful.
These are a couple of our patients. We usually see it on the early run, but here’s one where it’s on the later run – maybe that’s because there’s increased density. And here’s one where there was a CSF venous malformation. We saw this, and the more that we let it increase, we can see that it becomes this big malformation. So, I think both timing and density matter, and we do take steps to reduce our radiation dose. We do things like we play around with our parameters; we limit the coverage based on the MRI findings, and we do fewer passes of the spine.
So, in conclusion, dynamic CTM is great for capturing leaks, but your operator experience is important, and technique matters. Be patient with patient positioning and try to reduce your radiation dose where possible, but it may take more than one attempt. So, thanks to my team and thanks to my many international friends who have enabled me to improve my myelography over the last few years.