Measuring gait through brain metabolism in Parkinson's disease: How it can help understanding and treating gait disorders

[00:00:43] Dr. Michele Matarazzo:
As we were saying before, gait deterioration is one of the most disabling features of Parkinson's disease. And often the response to dopaminergic medication is quite poor. One of the problems we had when studying gait disorder is how do we measure it objectively? And you tackle that problem in a very innovative way.

In recent years, most of the research in the field focuses on gait analysis, biosensor accelerometers and so on, but you choose to look at the brain. Why did you think this was a better way to do this?
 

[00:01:12] Dr. Hilmar Sigurdsson:
So, I don't think necessarily it was a better approach, but it was just mostly a way to look at the same problem, but from a different angle. So we know that both older adults and people with Parkinson's have gait problems and they fall, and this leads to quite reduced independence in many cases. For example, approximately a third of community dwelling older adults, age 65 in years and over due to postural instability. And this increases with age.

So I think it's very important to not only look at movement activity, but we also need to know why it's happening, essentially. And my lab focus is heavily on the use of biosensors and accelerometry. And we know from previous studies using these methods that you mentioned that gait is an important marker for future progression of Parkinson's. But like I said, these method do not necessarily tell us exactly what's going on in the brain. And if we develop any good interventions that improve gait and postural stability, we wouldn't necessarily know why they're working. So what this essentially means to me is that there are few, if there are any, robust imaging markers of gait dysfunction in both aging and in people with Parkinson's disease.

So this was really an area that was completely unexplored. We see a lot of studies that focus on, on freezing of gaits and or only those that measure very basic gait characteristics like gait speed. So here in Newcastle, we really want to answer these questions that have quite evaded us for quite a while.

Putting someone inside the scanner is quite expensive and it is difficult. You need to lie perfectly still. So it's not very suitable for everybody, specifically those in the very advanced stages of the disease who have quite severe tremors. It's really difficult to image them. But like I say, in my opinion, we need this initial information to support other methods that are possibly cheaper, such as the bio sensors and the accelerometers.
 

[00:02:57] Dr. Michele Matarazzo:
Great. And actually, you did PET scans and the methodology to look at covariance pattern in brain metabolism is quite complex. This has been done before now. Can you explain to our listeners, in a few words, how this works and briefly discuss previous evidence of this approach? Not in, in gait, of course, but in general.
 

[00:03:17] Dr. Hilmar Sigurdsson:
Yeah, of course. So this method that we use in the paper is developed somewhere in the late eighties and early nineties by Muller and colleagues. And this essentially identifies distributed regions of the brain that shows some sort of a relationship with each other.

So it's called the Scaled Subprofile Model and it's based on Principal Components Analysis. So this Principal Components Analysis, or PCA, is quite a robust approach in statistics to show how a collection of variables that you're measuring and how they group together. So this is quite commonly used in questionnaires to see what kind of questions ask about the same specific field, like how your feelings towards something.

So first, because of all of our brains are quite different when we image them, they're not in the same space and they don't overlay off each other. So we need to put them all in the same space. So we do some technique, which is called normalization, to a standard space. So there is a one to one correspondence of the same voxels in the brain between all the participants.

We can then use this method to take a group of images. So in our case, we wanted to look at that there's an increase of blood flow in one region, which is related to an increased blood flow in another region. Similarly, there's a decreased blood flow or glucose metabolism. So like corresponds with each other. If there's a decrease in one region, there should be a decrease in another region, how these two interact. So this methods is quite similar to the one that we use on resting-state fMRI, when we use independent components analysis to get out these non-resting state networks, like the default mode network So from this principle components analysis, you get different components of brain regions that work together or show some sort of a relationship with each other.

And the first component explains most of the variants in the data. The second component explains anything that is left unexplained by the first one, and so forth. So we restricted our analysis to the first 10 components, which account for the 75% of the variants in the brain of everybody. For each component, we compute something called participant specific expression score, which shows how much a participant expresses this specific component from the group.

And then we want to look at how these components, either in combination or singly, how do they relate to a specific behavioral variable, like gait speed, or step length, or gait variability. Then to see which components or a group of components relate to these variables, we use just simple regression. And this gives us the combination of the variables that explain most of the variants in the participant's data.

And then, perhaps most importantly, we use these expression scores to see if they differ between two groups, like people with Parkinson's disease and control groups. So, this is in essence what the method does to get these networks or patterns out of it. And this has been applied quite extensively on a set of brain images from people with Parkinson's disease to develop so-called Parkinson's disease specific motor pattern, by David Eidelberg and his group. And this is an extremely robust pattern. And it's considered quite a strong imaging marker. And it's been replicated all over the world. So this motor pattern is made up of the first component and it correlates really well with the UPDRS-3 motor scores.

But they've also developed cognitive-specific patterns in people with Parkinson's. David developed tremor networks and networks specific to atypical Parkinsonism, like multiple system atrophy and progressive supranuclear palsy. But it's not only restricted to Parkinson's. There are Huntington's disease specific patterns, there's Tourette syndrome specific patterns. And there's also a network specific to normal aging, which was reported by a Muller.
 

[00:06:55] Dr. Michele Matarazzo:
That's quite interesting. So it's, it's very nice that you can look at how the brain uses glucose, and looking at that, you can actually have an idea what is happening and what parts of the brain are functioning well or not, and why they're functioning together or not. And then you can get to a final maybe diagnosis, or maybe you can use it as a biomarker to follow the disease progression. So that's very interesting.

Now you decided to use this very same approach to look at gait. That's very interesting. Now, can you summarize the main results of the study?
 

[00:07:30] Dr. Hilmar Sigurdsson:
Yeah, of course. So the study uses data from a large-scale study that was run between 2009 and 2011. It's called the ICICLE study, or the incidence of cognitive impairment in cohorts with longitudinal evaluation. So this was run here in Newcastle or in a collaboration with Cambridge as well. And both gait and imaging data were collected as well as cognitive data. So my study focused primarily on those who were recently diagnosed. And in this instance, we had 55 participants with Parkinson's disease who completed both PET imaging and gait assessment, which was collected a few months from each other.

And for our tracer, like PET uses this neuro tracer or radio tracer, we used fluorodeoxyglucose, which is a glucose analog. And glucose is the main metabolic substrate of the brain. And when this substance enters our cells, it gets false related, but it doesn't get metabolized. So it becomes sort of trapped inside the brain. But eventually it breaks down and leaves our body.

So the first thing we did in our study, we looked at the differences in different gait characteristics between the two groups. And we found that in this sample of people with Parkinson's disease, we found that they had slower gait speed, they took shorter steps and they had increased variability gait characteristics. And this is quite consistent with findings from other studies with much larger samples. And we found that in this study that there were two unique spatial covariance brain networks that underly gait dysfunction in Parkinson's.

The first network correlated with gait speed and step length. And it showed that as these two variables increase when actually the speed gets faster and the steps get longer, there's an increased glucose metabolism in multiple brain areas. For example, the frontal cortex and the cingulate and insula, and parts of the thalamus. And these areas are quite specific for motor control, cognitive control and sensory processing.

So the other network correlated with our gait variability characteristics. And variability is calculated just as the variants between the left and right steps. And we call this the temporal variability gait network. And it showed that as the variability increases, basically, so as their gait gets worse, there's mainly a decrease in glucose metabolism in multiple regions, such as the caudate, the nucle us accumbens, the red nucleus and the hippocampus.

And when we looked at these networks and these regions of the brain, we realized that all of these regions receive quite dense cholinergic inputs from the basal forebrain and the PPN. And perhaps most importantly, was that we found that these networks were expressed stronger in people with Parkinson's disease. So that was the most important bit.

And then we validated the fact that these patterns were related to gait by also correlating the expression of these networks with other variables that we measured, like disease duration, medication, UPDRS-3 motor scores. But we did not find any relationships between our networks and these variables.

So this really ensured us that these patterns were specifically related to gait and not necessarily other motor impairments.
 

[00:10:30] Dr. Michele Matarazzo:
That's quite interesting. So apparently those patterns are not related with the disease itself, but they are related with the gait problems related to Parkinson's disease. Now I had a question while I was reading the paper, which is well, normally people with Parkinson's have gait problems and they affect many parts of the gait. But you did find two different patterns. So why didn't you find one unique pattern that explains why the brain is not functioning well, and then it's provoking gait disorders in PD? Do you think there are different aspects of the gait that are managed by different parts of the brain? How do you explain this?
 

[00:11:06] Dr. Hilmar Sigurdsson:
Yeah. Yeah. Excellent question. That's exactly it. And this... what we found was very relevant to the great work that has already been done in our lab. So you ask, why didn't we just combine it into one unique pattern? Or the same thing you can say about gait, why not just measure gait speed? So gait speed is a very, very good initial indicator of potential problems with your gait, but it doesn't really tell the whole story.

So, this is the reason that previous people in my lab created this conceptual model of gait in both healthy, older adults and in Parkinson's. And this encompasses a wide range of gate characteristics that we've grouped into five independent domains, like pace and rhythm and variability.

So for me, there was no guarantee that a single brain network can control all of these different discrete gait features. And so, for example, if we manage to develop some novel technique to improve gait speed, it doesn't necessarily mean that there's a correspondence that they also improve gait variability.

And this also mostly comes down to the method that we used. We use regression to find which gait features correlate with each gait network. So we weren't really surprised to see that there were like two or even more networks that came out. Some of them weren't significant, so we didn't really report them. But this really confirmed our initial hypotheses, that there are multiple brain networks that control our gait.
 

[00:12:23] Dr. Michele Matarazzo:
Perfect. Now another thing that I was thinking when I was reading the paper is that the FDG PET were done in the on state. So people had taken the dopaminergic medication before the PET. Now, do you think that the results would've been different scanning the patient in the off state? And do you think it is worth studying the responsiveness of the patterns to the pulmonic medication?
 

[00:12:44] Dr. Hilmar Sigurdsson:
Yeah, it's very possible that it could have been different. And this was picked up by one of our reviewers. So this is the normal clinical approach, to ask them to withhold the medication for few hours before they were scanned.

But for our purposes, it's actually provided quite a unique opportunity to image the brain networks and these pathways pathways that function abnormally when you are optimally medicated. So in essence, we've accounted for the restored function of the dopaminergic system with the medication. So here we've shown metabolic networks that are possibly abnormal and related to these gait problems in Parkinson's. And we could absolutely try to study the responsiveness of these patterns to dopaminergic medication, but there's no guarantee that their medication is working optimally to improve gaits. We see that some gait features continue to decline, even though all of these patients are optimally medicated.
 

[00:13:37] Dr. Michele Matarazzo:
Yeah. Just for you to know, I was not one of the reviewers, so it was not me.
 

[00:13:41] Dr. Hilmar Sigurdsson: 
[laughs] Okay.
 

[00:13:42] Dr. Michele Matarazzo:
Can you help me understand how these results will help us understand better the gait disorders in PD, or do you think that we could use these patterns in clinical practice or maybe as outcomes in clinical trials?
 

[00:13:54] Dr. Hilmar Sigurdsson:
Yeah, absolutely. And this is our plan moving forward. So I think the most important aspect of this method is that once you've developed your patterns to a specific condition, it can be applied to a single brain image. So you can see, for example, someone who has a strong suspicion that they had Parkinson's disease, you can scan them and you can take this network and you can sort of prospectively apply it to their brain, and see how strongly they express this network.

So, for example, with the PDRP, developed by David Eidelberg, they 've shown that perhaps was someone with a rapid eye movement sleep behavioral disorder, which is thought to be a condition that precedes the onset of Parkinson's disease, they see that they start to express these motor related patterns quite early on.

So it can definitely be something that you can use and then try to intervene as early as possible. We definitely want to look at if we can use these gait-related metabolic networks as so sort of a conceptual framework for future studies targeting gait characteristics in people with Parkinson's that are different from healthy controls.

And we would like to look at if the expression of these networks can be reduced or reversed by novel interventions, and see if this coincides with improvement in gait features, definitely.
 

[00:15:03] Dr. Michele Matarazzo:
Fantastic. We've discussed about the paper, now let's take a look at the future. What do you plan to study next to keep unraveling the gait dysfunction of Parkinson's disease?
 

[00:15:13] Dr. Hilmar Sigurdsson:
Yeah. So the primary aim of my research is to characterize changes in the way we walk, not only in Parkinson's disease, but also just in healthy aging. And how these two differ specifically when we scan the brain and look at what regions of the brain control walking. And we want to identify these regions, connections, and different networks that are involved in walking so we can develop new treatments to prevent further decline and falls.

So we recently completed the study here where we wanted to develop a novel protocol that is capable of separating the neural activity specifically related to walking from the neural activity that is specifically related to posture control. And we used our hybrid PET MR scanner here in Newcastle and we asked healthy older adults to complete two tasks. So first we asked them to simply stand for 15 minutes, and then we asked them to walk for 15 minutes. And before each task, we injected them with our glucose tracer.

And then we put inside the scanner and we wanted to look at which regions of the brain are mostly active during walking, accounting for anything that is required for posture control. So we are really excited about these results. We're sort of looking at them right now and we're writing our manuscript and hopefully it will come out sooner rather than later.

But in addition to this, we're also working on novel interventions. So we are looking at something called the vagus nerve stimulation. So, this is sort of a non-invasive stimulation that targets the cholinergic vagus nerve. And first and foremost, we want to look at if this is feasible in people Parkinson's disease and if this treatment is, is, is tolerated. So we are looking at also gaits, both in the laboratory and, and free-living gait features. We are assessing cognition and autonomic function and motor dexterity. So we're looking to recruit for the participants. Half of them will receive the active stimulation and half will receive sham stimulation.

So this is led by Dr. Allison Y in our lab, and we are sort of halfway with recruitment. We're very excited to what this leads to. Hopefully we can complete the study and next year. So this is something that we are looking at. We're looking at all angles right now.
 

[00:17:18] Dr. Michele Matarazzo:
Oh, yeah, definitely plenty of work to do. I mean, you're looking at the neural basis of this and possible intervention to change this. So really excited to see future results of your studies.

Thank you very much for your time. It has been a pleasure to have you on the podcast.
 

[00:17:31] Dr. Hilmar Sigurdsson:
Thank you.
 

[00:17:33] Dr. Michele Matarazzo:
We have had Dr. Hilmar Sigurdsson, and we have discussed the article "Gait-Related Metabolic Covariance Networks at Rest in Parkinson's Disease" from the Movement Disorders journal. Don't forget to download and read the article from the website of the journal, and thank you all for listening.