Factors of Brain Aging in People With HIV
In this video, Kalen J. Petersen, PhD, discusses his team's recent study examining the impact of multiple factors on brain aging among individuals living with HIV when compared with individuals without HIV, including traditional clinical factors, comorbidities, and more.
- Petersen KJ, Lu T, Wisch J, et al. Effects of clinical, comorbid, and social determinants of health on brain ageing in people with and without HIV: a retrospective case-control study. Lancet HIV. 2023;10(4):e244-253. doi:10.1016/S2352--3018(22)00373-3.
Kalen J. Petersen, PhD, is a postdoctoral fellow in the Department of Neurology at Washington University in St. Louis, MO.
Kalen J Petersen, PhD: Hi, my name is Dr Kalen Peterson, and I'm a postdoctoral fellow at Washington University in St. Louis. I happen to work in the Department of Neurology, and I'm in the lab of Beau Ances, who is a neurologist. We study among other things, HIV and our lab studies, Alzheimer disease, and other neurodegenerative conditions. So there's sort of an infectious diseases side to the lab, which I'm on, and also a side that's more interested in aging and neurodegeneration. But there's a lot of overlap between those two areas so I think that science has become very collaborative these days in a way that I think is very beneficial. Everything we do we do in teams, with groups both within our institution and outside. So I think that to say you belong to one department or even one lab is actually sort of underselling the collaborative nature of science now.
Consultant360: Could you give us an overview of your study and discuss what prompted this research?
Dr Petersen: This study is about the different ways that people can age while living with HIV. So my mentor has been working on imaging the brains of people living with HIV at this institution for going on 14 years now. And that means that he's seen a lot of people in clinic and seen a lot of people in the research setting who have various degrees of either cognitive impairment or cognitive resilience. Because one of the things about HIV is that it affects the brain. It infects the brain. And it causes to a greater or or lesser extent, it can cause cognitive problems. It can cause problems with the speed or accuracy of thinking. And that is not to say that everyone who has HIV is going to have cognitive impairment, and the contrary, most people do not.
But in the bad old days before very effective combination antiretroviral therapy, people living with HIV unfortunately did not in general live to old ages. And so we have what you might call a good problem in that we have people living healthy long lifespans. But now we have an aging population who is new. We don't know what it looks like to grow old with HIV for large numbers of people. And we don't know what the variability within that population is in terms of the virus's effect on the brain, but also the effect of other things that sometimes accompany having HIV. Some of the life factors or challenges that can go along with HIV positivity.
So the background is that we work in an magnetic resonance imaging (MRI) lab. In neuroimaging, the goal is to answer questions about disease through what we can see on a scanner, either an MRI scanner or a PET scanner. And this lab has been acquiring pictures of people's brains who have HIV and who don't have HIV, going back more than a decade, and so we actually have a very large data set of individuals who are also well-characterized outside the scanner. We've been able to give them neuropsychological and cognitive testing and been able to do bloods to look at, for instance, inflammatory biomarkers in the blood. And we have other characterizations of people's lives and how those may affect health. For instance, measures of whether people are using substances, legal or illegal. So the question that we fundamentally wanted to get at with this study is what do those things that we can measure with MRI tell us about how people are aging with HIV?
So we already know some background. We know that people with HIV on average and in general do have some neuroimaging differences from people who do not. For instance you might see slightly thinner cerebral cortex or you might see a change in the microstructure of white matter that you can measure with diffusion imaging. But the problem is a lot of these differences are really only apparent at the group level, when you look at hundreds of people in one group vs hundreds of people in another. And they're also kind of diffuse. They're not very specific to a brain region. It's not like Parkinson disease where you can look at the substantial nigra and say, "there, there's degeneration right there." So the question is, what technique could we apply that would help us to understand this kind of diffuse and general change to the brain that may be happening in some people living with HIV.
And so that's where we turn to the approach of machine learning. So machine learning is another way of saying artificial intelligence (AI) or a complicated computer science model to help us make sense of a large a big data input in a way that is easily interpretable as an output. And I wanna be clear that I'm not a computer scientist. I'm not a coder by training. And so the fact that I'm able to utilize some of these publicly available machine learning models just goes to show that they're increasingly becoming accessible to people in various fields. So the concept of our study is that we wanted to do something called brain-predicted age. So the way brain-predicted age works is you teach a machine learning model to recognize an individual's chronological age, that is how many years they've been alive based on the MRI picture of their brain. And you do that by showing it a lot of data. So the learning model that we have been using, called Deep Brain Net, was trained on over 11,000 individual MRI scans. And this is a normative model so it learns what aging typically looks like in the brain and there are characteristic patterns that it can identify. Things like the ventricles get bigger, the fluid filled spaces inside the brain, the cerebral cortex, the gray matter of the cortex shrinks. And so the model can get very good at guessing a person's age from the physical characteristics of their brain that you can measure with MRI.
So then what you do is you show the model people it's never seen before, once it's trained, and you see how it performs. And the idea is if the model, which is pretty good at guessing most people's ages, is systematically overestimating when it's shown people with disease of interest, in our case HIV, but you could also say Parkinson disease or something like that. It will guess high in general. Again, not for every individual person because there's so much variability in how we age. But we can say, okay, if you look at older people living with HIV people who may have had a legacy effect of having lived for a while in the era before people had access to good antiretroviral drugs, especially. They may be virally suppressed today. You may not be able to detect HIV in their plasma, but nonetheless there's probably a legacy of early damage. And there are probably also other things going on either lifestyle or comorbidity that may also affect how their brain looks right now. And what we actually had seen in a past study that we published in Clinical Infectious Diseases that use this approach. And what we saw is that, yes, older people living with HIV as compared to younger people with HIV or people without HIV show an older appearing brain on average.
But the other thing that we noticed about this that was very striking is that this is a group difference, but if you look at individuals, there is this high degree of variability. Some people show very resilient patterns of aging. Some people have a brain age, a predicted age from their image, that looks as young or younger than they are, as opposed to older and pathological. So the question that led to this study is accounts for this variability? Why do some people show advanced brain aging, accelerated brain aging and other individuals don't? And are the factors that matter for people living with HIV the same as those of the general population? Or are there specific areas of concern that may be affecting brain health? So that's what led us to the development of this study.
C360: Could you discuss the results of this study and how these results can be applied to clinical practice?
Dr Petersen: One of the strongest outcomes of the study actually is that there are three main categories of predictors that help to explain the variants in brain aging for people with and without HIV. So to begin with, I'll talk about HIV related factors directly. Things that will typically be measured in clinic from a blood draw. So one of the things that we found is that individual viral load, which is simply the number of viral RNA genomes that can be picked up in the bloodstream, which is a measure of how suppressed or unsuppressed the HIV infection is at the moment, was correlated with brain aging.
So people who had a detectable viral load, which these days is maybe 50 copies per milliliter of blood or above, a very low number, do in general have a higher brain age, a brain-predicted age as compared to their chronological age. So that's indicating to us that control of HIV infection, which is very possible these days, is extremely important in maintaining cognitive function and brain health. And right alongside that, we also saw a relationship with CD-4 or helper T-cells. So the number of helper T-cells in the blood is a very good indicator of how intact your immune system is. HIV infects these cells, these T-lymphocytes. And if you have a low number of t white blood cells that is an indication that HIV is bringing down the potency of your immune system. And we saw that, again, the higher the number of good CD-4 T-cells the less your risk was of having an increased brain age. So it seems like some HIV related factors, some traditional clinical factors, do matter here.
Secondly there are comorbidities, things that aren't HIV but often go along with it. And one of the comorbidities that was most critical actually turned out to be cardiovascular health. And there is an increasing focus on heart health in the infectious diseases and HIV field today because people are recognizing that individuals with HIV are at greater risk for coronary disease. And in fact, what we saw in this study, we looked at something called the Framingham score which is a calculator that was developed in this this long term study in Framingham, Massachusetts.
The Framingham score tells you your risk of developing hard coronary artery disease within 10 years. So a Framingham score of 33 is a 33% chance of developing disease. What we saw is that the higher your Framingham score, the higher your likelihood of heart disease, the older appearing your brain was on average. And that was quite a strong effect among people with HIV, highly significant with a big effect size. So it seems like something that clinicians could really focus on is once we've established viral suppression which is the first and most important thing, to ask about risk factors for cardiovascular health. Things like exercises, smoking, the kinds of lifestyle factors that actually we can intervene in. We can't cure HIV in the very vast majority of patients, but we can actually address some of the factors that can exacerbate the effects of HIV on the brain and that includes cardiovascular health.
It also includes one of the other comorbidities we found to be very important, which is hepatitis C co-infection, which is quite common among people living with HIV and is also treatable. So we found that people who have HIV and are co-infected with hepatitis C as opposed to HIV alone have an older appearing brain on MRI by about 4 years. And that was another highly significant effect. In terms of comorbidities that's not necessarily exclusive. There are probably other factors that matter very much too, but the ones we identified in this study as important were heart health and co-infection. So that is something that could be very relevant for the treatment of people with HIV in the clinic.
Then the final category that seemed to be particularly of interest was this broad area that we call social determinants of health. Social determinants of health are factors in a person's lived environment that affect their personal health. For instance, socioeconomic status or exposure to violence and so on. And this is kind of a hard thing to pin down, right? Because there are a lot of social determinants of health and a lot of them are hard to quantify. But what I'm really excited about in this study that that is kind of new and a new approach for us is trying to understand social factors in terms of geography. So what we have been able to do, and I've worked with Dr Julie Wisch on this, is to take participants' addresses within the city of St. Louis, which is where this study was conducted, and turn them into latitude and longitude and plug those into databases from the US Census Bureau that enable us to find out things about what life is like in their neighborhood.
What we looked at in particular in this study was what's called the Area Deprivation Index. And the Area Deprivation Index is a measure that's a composite of various factors from education, to housing, to employment, and it's pretty fine grained. It goes down to the census tract level. And Julie has actually shown in another paper in Alzheimer disease that there is a relationship between the apparent brain age that we get from MRI and the Area Deprivation Index in one's neighborhood.
I think that we can include potentially in the show notes a link to an interactive map where you can zoom in on a city or state and look at the distribution of Area Deprivation Index. I'll just describe it briefly in St. Louis. St. Louis is a historically segregated city. And what you'll see is that there are very discreet regions of high and low deprivation in the city that correlate to areas of historical redlining. So north of a certain street in St. Louis, Delmar approximately, it tends to be predominantly African American as a result of historical housing policies. And these, unfortunately, are the areas today that have the highest Area Deprivation Index score. Also areas in the south of the city. And right in between, there's a central corridor, which is a little more diverse and much more affluent. And then you can also look out in the counties to the West and see very affluent and predominantly White areas.
It appears that this actually influences brain health when we put it together with other comorbidities and clinical factors in a combined model, Area Deprivation index came out as a significant predictor of the appearance of brain health. Now, this isn't deterministic. It doesn't mean that where you live is going with one-to-one correlation match how healthy your brain is. Of course not. Everyone has different levels of individual resilience but it is a thing that appears to matter. And so this is a hard thing for clinicians to respond to because it's not the sort of thing that can be addressed very easily at the individual level. Things like quality of the urban environment or degrees of social support. Those are things that come through community and through policy rather than through clinical practice.
But I do think it's important for clinicians to be asking questions about things like social stressors, things like stigma, things that may affect a person living with HIV in a way that is especially problematic, over and above how those factors might affect a person without HIV. So I think that clinicians at a minimum should be aware of the effects that social determinants of health and socioeconomic status can have on brain health and cognitive function, how things like quality of education or exposure to stressors in early life or in adulthood can affect brain health.
C360: What are the next steps for research in this area?
Dr Petersen: I think there are a lot of pathways we could take. This is a study that had a lot of moving pieces and was a bit of a jigsaw puzzle. But one of the areas that I'm most excited about is to ask more specific questions about different regions of the brain. In this study we looked at the brain as a whole, we looked at it holistically through this deep neural network. And what I would like to do to expand and extend this study is to, instead of just looking at the structure of the brain, which probably reflects a lot of legacies of the early phase of infection, I'd like to be able to look at functional brain activity. So we've been talking about structural MRI, and I think that in the future, this is, this is a study that would benefit from the application of functional MRI, which people are generally aware of as something that can look at areas of the brain lighting up in response to different stimuli or different conditions.
But one of the things that we like to do in MRI world is to have study participants just lay in a scanner and do nothing in particular except allow their mind to do what it does naturally. And then to look at the spontaneous activity, the unprovoked activity of the brain using resting state functional MRI. And what you can do from that is to draw inferences about what regions of the brain tend to be active together spontaneously. And you can parcellate the brain up into different networks that way. For instance, the default mode network which tends to be most active when a person is not focused on a particular task. If the brain aging that we observe is really the explanatory variable for cognitive impairment that happens in some people with HIV, that's because it's being mediated through some network, some process that's happening in the brain. And it's definitely a region specific in a way that we don't understand yet. So I'd like to know what brain functional networks can tell us about cognitive aging and impairment. I think that the application of FMRI would be a great direction to take this research.
C360: Is there anything else you would like to add?
Dr Petersen: I think that one of the things that I'm also interested in is the question of what's happening in the earliest stages of HIV infection. And that's something that with our current data, we're not able to get at. We're looking at people in general years, even many decades, after their initial infection. And after they started on antiretroviral therapy. And I think it would be very interesting to know what happens in the acute phase of infection. And that's something that I would think would be possible with collaborations, with groups that study this question in particular, including in international populations. So then I would give a final, I guess, takeaway for what we've talked about today which is that aging is complex to begin with. HIV-related aging is doubly complex.
There are a lot of factors that work here besides the effect of the virus itself. There are social determinants of health. There are comorbid disease states, and there are the clinical effects of HIV. It is difficult to disentangle those. It's difficult to attribute the changes that we see in thinking or in the structure of the brain itself to one factor or another. But I think it's increasingly important to think in those terms, to think in terms of embracing the complexity of aging and trying to understand it with detailed characterization of individual life histories and individual circumstances to look beyond just people with and without HIV and to begin to understand how an individual's life experience affects the health of their cognitive aging.
Thank you very much. We’re very excited to talk about these results and I very much appreciate being invited on.