Transcript
Thank you, Dr. Callen. So, this is, you know, another conundrum that we’re all dealing with. And, you know, we’ve put some energy into this over the last few years. You know, generally, when we talk about patients who have a leak and have a normal brain MRI, we generally say, you know, 20% of patients with leaks have a normal brain MRI. As you can see here in the report by Dr. Matharu, where he collected existing literature at the time, that’s about 20%. But these are all patients who have an abnormal spine MRI, right? So, these are people who have extradural spinal fluid but a normal brain MRI.
So, what I’m going to be talking about is patients who have a really good story for SIH – if I may use Dr. Carroll’s lingo. But they have a normal brain MRI and they have a normal regular CT myelogram or spine MRI. So, what do you do in that case, right? Should you go look for a CSF-venous fistula, because of course, with those leaks, there is no extradural CSF. You could go ahead and do a non-directed or blind epidural blood patch, or should you first do an abdominal binder trial? Should you send the patient back to the referring headache doctor, or are there some other things that you should at least think about?
So, this is from a study that we completed sometime just before COVID, I believe, and we looked at 60 individuals who had orthostatic headaches and were referred to us to rule out a spinal fluid leak. They had no extradural CSF, and we did digital subtraction myelograms on both sides, except if we found a fistula on the initial scan, then sometimes we only did it on that one side. And these two groups were pretty similar, right? So, the only thing that was statistically significantly different was that it was much more common in those people who had a meningeal diverticulum. You might say, well, that’s not surprising because that’s where most CSF-venous fistulas arise from, and that makes total sense. As a matter of fact, we found fistulas in 20% of people who had meningeal diverticula but in none of the patients who had no meningeal diverticula.
Now, you have to also remember that these are a highly selected group of patients. The vast majority had had a positive response to epidural blood patching before we even considered doing these invasive tests.
And as you know from Dr. Kranz’s study, that meningeal diverticula are really common, right? So, 40% of the general population have meningeal diverticula, so it’s really a variant of normal. And certainly, in the olden days, generally neuroradiologists would not even mention that in their report unless they were of a certain size.
So, then we looked at something else that we thought might be able to help us predict, you know, which patients will have a CSF-venous fistula. And you know, also when I talk about the yield of finding a leak, right, I mean specifically the yield of finding a fistula. It doesn’t necessarily mean that the patient has or does not have a spinal fluid leak.
So, what we looked at is something that Dr. Beck has studied extensively in the past with ultrasound, and that’s the amount of spinal fluid around the optic nerves, right? And we did a small addition to our MRI protocol – only takes four minutes of time. It’s a heavily T2-weighted coronal MRI with fat suppression. And as you can see on those two lower panels – the ones with the arrows – you can see that there’s a nice ring of spinal fluid around the optic nerves.
And what we found in that study is that those patients who had a fistula had a much smaller ring around the optic nerve, and their optic nerve sheath diameter therefor was significantly less. And it was, you know, there was quite a difference.
And then this was a predicted probability model of how likely it is that we would find a fistula. Now, it’s a little bit deceiving because we are all familiar with 95% confidence intervals. These are 80% confidence intervals, because if we did 95% confidence intervals, it would pretty much take up the entire graph. But as you can see here, the chance of finding a fistula as the optic nerve sheath diameter goes over, let’s say, 4 and a half, becomes really, really tiny. So, we’ve been using that since we did that study quite extensively. And this is just to show that after the fistula is repaired, you get restoration of the perioptic amount of spinal fluid.
One of our former residents, Dr. Tay, who you might have met at our meeting in Hawaii. She did a systematic review. We were particularly interested in studies that have looked at this, right? So, people who have a normal brain MRI, no extra-dural CSF – what is the yield of finding a fistula? And she reviewed over 500 publications, and then actually, she ended up with only eight publications that looked at that. And only two centers have published a study of what most of us just call a normal brain MRI, right. So, normal brain MRI, no extradural CSF. There are quite a few other studies that just look at the Bern score, and they categorize it as a low yield, which is a Bern score of 0-2. But I always find that difficult to interpret because if you have florid meningeal enhancement, you have a Bern score of 2.
So, what that showed is that the yield of finding CSF-venous fistula in a patient with a normal brain MRI varied in the different studies. And on average, it was about 6.8%. For patients with a low Bern score, the chance of finding a fistula was quite a bit higher – right – it’s somewhere between 25% and one out of three, but that also included patients who clearly have an abnormal brain MRI.
Specifically, when we compared our reports using digital subtraction myelography to the reports from Duke using dynamic CT myelography, both done mostly in the lateral decubitus position, there was quite a bit of a difference, right. So, we found it in 10%. Dr. Kranz found it in 0% in this limited time that these studies were performed. So, you might say, well, obviously DSM must be superior to dynamic CT myelography, but that’s definitely not the case. Because when we look at patients who we know must have a fistula – right – so people who have no extradural CSF but a positive brain MRI, the yield is almost identical. It’s 74% versus 73%. So that’s not the reason why. So, there must be a different reason. One of them I alluded to earlier – we generally only do these tests if people have had a good result after epidural blood patching.
And of course, you know, I like to bring up this batting average of referring headache doctors. Certainly, don’t ever want to insult anybody, but you know, it’s kind of a fun thing to do. I have this hanging in my office. And last time I talked about this, the highest batting average was .429. But then, of course, this is a Dodger player who won the World Series a few days ago. There’s a doctor in Los Angeles, her name is Dorothy Dada – I don’t know if you know her. She has sent me 5 patients in the last 18 months with a normal brain MRI, no extradural CSF, and 4 out of those 5 patients had clear CSF-venous fistula.
But in general, when we have this situation that we talked about – if there’s normal CSF around the optic nerve and there are no meningeal diverticula – we still do some digital subtraction myelograms. But we know that the yield is going to be really low. It’s going to be just a few percentage points. If you have no CSF around the optic nerve, and you have obvious meningeal diverticula, the yield is quite high – it’s higher than 50%.
Now, of course, there could be some other reasons – right – why people have normal imaging. One of them is pooling of spinal fluid. In the case of dural ectasia, you can have these tiny little lateral leaks that you don’t see on MRI or conventional CT myelography, or you can have this narrowing of the azygos vein.
For pooling of CSF, I’m particularly talking about patients who have connective tissue disorders like Marfan syndrome. So, one of the major criteria for Marfan syndrome is dural ectasia. You see some examples here. Mostly that’s in the sacrum that you can see in the first three panels. But on the right side, you also see somebody who’s got quite a bit of dural ectasia in the thoracic spine, and then one of those cysts actually ruptured and caused a hydrothorax.
You also see it quite often in patients with neurofibromatosis type one. Obviously, you really don’t want to operate on these individuals unless you really need to. Ankylosing spondylitis also can have dural ectasia, as you can see here. And then often, there’s herniation of the spinal cord or just the cauda equina into the dural defect.
Then we talked a little bit about these lateral leaks that are not associated with the nerve root sleeve that I call pedicular leaks. This is an example of that. You see on that DSM there’s a very tiny amount of extradural CSF, and in this particular patient, you could see it on her MRI, but you could not see it on the CT myelogram. And then, at surgery, you see that there’s a little tear and there’s a contained arachnoid membrane that’s pouching out, and you really don’t see any CSF outside of that ballooning of the arachnoid.
This is another patient. Also, you know, they all look very similar – also very tiny amount of extradural CSF. And in this young woman, you could not see it on MRI, you could not see it on conventional CT myelography.
Another reason for spinal fluid loss is something that we’ve treated just for the last year and a half or so, and that’s patients who have azygos vein stenosis. We found this out very fortuitously. We were looking just at the venous system, mostly to see if there was internal jugular vein stenosis. Or we thought, oh, maybe they’re going to have CSF-venous fistula embolization. We do a technique called Feraheme® MR venography, which gives a beautiful roadmap of the whole venous system. And in three patients who had frontal temporal dementia and really bad brain sagging, we found a high-grade stenosis of the azygos vein. These are different MR venograms of these three patients, and then Marcel Maya, the neuroradiologist I work with, he stented these three patients. And they all were, you know, pretty debilitated – their SIHDAS or MIDAS score was really high. It was kind of a miraculous recovery, right, because these patients had been suffering for many years. Within a few days, they basically were back to normal, and their score went from, you know, major disability to almost no disability. The third patient that we stented was a little bit better after the stenting but not significantly so.
And the thinking behind it is that when there’s a high-grade azygos vein stenosis, the azygos vein between the area of stenosis and the atrium has abnormally low pressure. So, we think there’s a kind of like a sump effect of the just normal absorption pathways of CSF, or maybe these patients have really tiny CSF-venous fistulas that we were not able to identify.
Coincidentally, I also had a picture of Donald Rumsfeld – we will not go over that. But I think it’s important to know. So, Donald Rumsfeld talked about the knowns, the known unknowns, and the unknown unknowns, but there also are the unknown knowns, right? So, these are things that we would understand, but we haven’t quite discovered that yet. That’s hopefully what the next frontier of CSF leaks will be about. And we’ll talk about that more at our next conference, which will be in Amsterdam the last weekend of June. Thank you.
Click here to see all Bridging the Gap 2024 videos