Dr. Ian Carroll at the 2023 Cedars-Sinai Intracranial Hypotension Conference
Dr. Ian Carroll, associate professor of anesthesiology, perioperative and pain medicine, and chief of the Stanford Headache Program, presented this talk on long-term outcomes of EBP in patients who fail to meet ICHD-3 diagnostic criteria at the 2023 Cedars-Sinai Intracranial Hypotension Conference on July 9, 2023. The conference was hosted by Cedars-Sinai with generous support from the Spinal CSF Leak Foundation in Kohala Coast, Hawaii.
Transcript
[00:00:12] All right, I’m going to be talking about the approach we’ve been doing at Stanford since about 2016 when we decided that the sensitivity of imaging really wasn’t enough for us to make decisions about whether patients should be offered patches or not. I got help from a lot of people, including Stanford neuroradiology and the headache clinic. And the truth is when I started doing this, Wouter, very helpful and supporting me.
[00:00:37] So I’m going to show you data from patients treated between August of 2016 and November, 2018. As Dr Krantz was saying, we need more prospectively gathered cohorts. Here’s one measured with validated outcome metrics prospectively.
[00:00:54] We started with 243. We were looking for a cohort of patients that had completed the metrics before getting patched, had at least one patch, and completed the metrics at least one time after being patched. We went back and identified 243 patients that we’d been looking at and considering offering patches. And there were, of those 243, 85 that actually had completed the measures first before ever getting a patch, got at least one patch, went on to complete the measures at least once afterwards. The data about who made up the cohort is shown here.
[00:01:32] You can see that they’re two-thirds, roughly, female. Average age was 42. The average duration of symptoms is 8 years. These people had their problems for a very long time. You can see that 98 percent had head pain or pressure. They all had, it was very high symptoms of all the core symptoms of leak: neck pain, tinnitus, cognitive impairments, high rate of fatigue, which is poorly described in CSF leaks.
[00:02:01] 84 percent reported that they had orthostatic features. We dug into the features of of their orthostatic headache, measuring time before their head pain started when they were upright, head pain severity after being upright for an hour, head pain severity after being flat for an hour or being flat overnight, and this is what some of their validated metrics show.
[00:02:25] The first is this PROMIS global health measure. This is a quality of life measure. The PROMIS metrics have been put forward by the NIH to to be metrics that can be used across studies and across different conditions. The core finding with these PROMIS metrics is the score reported as what’s called the T score, where the T score of 50 represents the U.S. population mean. So the mean global physical health of the U.S. population would be 50, and a standard deviation is equal to 10 points. So, when you’re 10 points higher or lower, you are a full standard deviation away from the mean. The cohort’s mean physical global health was 32, suggesting that their global health, their global physical health, was at the 4th percentile, putting these patients in line with people with AIDS and end-stage cancer. These people are devastated by their condition.
[00:03:26] Like Dr. Friedman’s data, the headache impact test scores were in the severe range. We were measuring fatigue, and their score of 66 puts them actually in the 95th percentile of the US population in terms of fatigue. They were more fatigued than all but 5 percent in the population with scores that are significantly worse than patients with MS and in line with average scores reported for people with chronic fatigue syndrome or my myalgic encephalitis.
[00:04:01] Of the 85 patients, we were able to ascertain ICHD-3 criteria for 83 of them; problems with two. And what we found is, in fact, of the 85 patients we evaluated, 16 met ICHD-3 criteria, conformed to that; and 80 percent did not. And so the question is, did that 80 percent just not have a leak, or did they have a leak, but they were imaging-negative? Remembering that these people are eight years into their illness, and it’s possible that the sensitivity of some of our imaging is a function of time.
[00:04:39] So here is their ICHD-3 status again. 16 were positive. They underwent an average of 3.6 epidural patches. This is the cohort as a whole. And these are the outcomes in their metrics, pre versus post. The first thing I’m pointing out is these are durable outcomes. These are changes from pre-patch to the last assessment, which is a mean of 521 days after their first patch and 377 days after their last patch, remembering they had an average of 3.6. What you see is that their global physical health statistically significantly improved to a modest degree, but highly statistically significant, as did their mental health, as did their HIT-6 and their neck pain and their fatigue and their nausea.
[00:05:32] But what really didn’t get much better are their neurocognitive symptoms or their vestibular cochlear symptoms. And so the question becomes, is this modest statistically significant impact clinically significant? The PROMIS global health measure has minimally clinically significant difference published of 1.7 points. And so we went and looked at what percentage of people would meet that previously published minimum change to be a responder. And what you see is that, among the patients who conformed to ICHD-3 criteria, the percentage that had clinically meaningful improvement is quite high. 81 percent versus the three out of those 16 that did not have a clinically meaningful improvement.
[00:06:28] But this raises the question: what about the 81 percent of patients who did not conform to ICHD-3 criteria? In many trials and retrospective cohorts, they’re just not included. But we felt that these were patients who should have a trial of epidural patching. And so here’s what we found with those patients.
[00:06:49] So when you, in fact, look at those patients, 36 out of those 67 wound up with a clinically meaningful improvement. And while that is 54 percent ,that is less than the 81 percent improvement in the people who met criteria. So, clinically, when you look at clinically meaningful response in terms of global physical health, it’s not quite statistically significant. But if you want to maximize your success rates, you should really only be patching those who are ICHD-3 positive. That will give you the higher rate of 81%. But if you want to help the highest number of people that can have a clinically meaningful benefit, you should not do that, because what you see is 73 percent of the clinically meaningful responders would have never received the treatment that helped them if the patching had been contingent on positive imaging or a low opening pressure.
[00:07:56] Three quarters of the people who responded came from the group that did not meet the diagnostic criteria. So what can help us identify patients most likely to have a clinically meaningful response among patients not conforming to the ICHD-3 criteria? We went back and looked at features specifically associated with CSF leak. And these are the orthostatic features that we looked at. And you can see, first of all, that a patient’s simple report of “I have postural head pain” wasn’t associated with the likelihood of being a responder, nor was how long they could be upright before their head started to hurt, nor was how bad their head pain was after being upright for an hour.
[00:08:39] However, the absolute decrease in pain, how low their pain got after an hour flat, was highly statistically significantly associated with the likelihood of being a clinically meaningful responder, with an odds ratio of 0.74 per point. That’s a 25 percent change in odds of being a responder for every one point lower your head pain was after an hour of being flat.
[00:09:09] And similarly, the conceptually related decrease in head pain score when not being flat for an hour, but being flat overnight was also predictive. So this actually starts to provide real outcome predictive validity to specific features of orthostatic headache. It’s not how bad your head pain is. It’s not how quickly it gets bad when upright. It’s not the worst severity of it. It’s how fully it resolves when you are flat. That predicts the likelihood among the ICHD-3 negative subgroup.
[00:09:44] Let me show you what that data actually looks like. So what you see here is a graph that has improvement in global physical health on the y axis against the pain intensity 0 to 10 on the x axis. And so what you see here is, first of all, the line of no change. The graph actually Is constructed in a way that doesn’t emphasize the benefits. The line of no change is the green line. All the red dots are the people who met clinically significant improvement thresholds. The black dots are the ones that did not.
[00:10:24] And what you see here is lots more red dots above the line of no change compared to the black dots beneath it. If you look at a full standard deviation of change in global physical health for the U.S. population, lots of red dots of people who improved more than a full standard deviation of the U.S. population. In contrast, nobody got worse more than a full standard deviation. And if we look at the people whose pain went all the way down to a zero or a one after staying flat for an hour, almost all of them were clinically significant improvers. Very highly statistically significant and clinically actionable in our practice.
[00:11:13] This suggests that there is a specific physical finding or clinical sign that we can ask patients about to help predict, among the ICHD-3 negative patients, the people who had negative imaging and did not have a low opening pressure, something that specifically predicts who’s more likely to have a clinically meaningful response durably, despite being sick chronically. The other thing I think that’s useful about this data is it highlights an aspect of the patients with chronic symptoms. I’m not sure how many we would have said were better if we weren’t using a validated metric to follow this.
[00:11:57] It would be easy to be discouraged If you were a treater or a proceduralist patching, and you weren’t using a validated metric like this showing you that they were having a clinically meaningful response, because the home run responders, the people over that, that 10 point line is much lower than the number of people who actually had a clinically meaningful benefit. The fact that we can’t restore all of these people to normal health should not stop us from trying to get them the clinically meaningful benefits that we can clearly measure when we use validated metrics.
[00:12:35] So, in summary, number one, under the null hypothesis that I’m just engaging in the placebo effect or regression to the mean, I’m doing procedures on people and they’re feeling better because they’re having procedures done on them rather than a specific treatment effect. If that were true, it would be very hard to explain how a specific, two specific orthostatic features before being patched were predictive of long-term outcome. That is hard to explain under regression to the mean or placebo effect.
[00:13:12] Among patients with chronic (eight years), disabling, fourth percentile symptoms suggestive of CSF leak in whom imaging does not confirm a leak, 54 percent of the patients showed a clinically meaningful improvement at long-term follow up after an average of 3.6 epidural patches. Two factors specifically associated with CSF leak. More complete resolution of head pain upon reclining for an hour or upon being flat overnight significantly predicted the likelihood of being a clinically meaningful responder.
[00:13:50] Most of the patients who experienced a meaningful clinical improvement came from the ICHD-3 negative group by a ratio of 3 to 1, which gets at this issue that we were talking about. Are we interested in our batting average? Or are we interested in the runs that we create? So a higher rate of response is seen among the ICHD-3 positive patients. So you have to think about your goals. Is it to have a high hit rate? Or to help the most people who come into your sphere of influence? Thank you very much.