Andrew Varga, MD, PhD, explains how sleep, cognition and other risks for Alzheimer's Disease.
um, we are very lucky today to have Dr Andrew Varga for his grand rounds on sleep, cognition and risk for Alzheimer's disease. Dr. Fargo completed his PhD in neuroscience at Baylor before going on to attend medical school at New York Medical College. He completed a residency in neurology at B. I Deaconess in Boston and subsequently went on to complete a sleep medicine fellowship at. As a sleep physician and neuroscientist, Dr Varga has had a long standing interest in the mechanisms of learning and memory and the role of sleep in memory consolidation. He has active R 01 in our 21 funding for clinical and translational research, studying the role of sleep energy, general diseases such as Alzheimer's disease and Parkinson's. Dr. Varga currently serves as an assistant professor in the division of pulmonary critical care and sleep medicine here at Mount Sinai. Please join me in giving a warm welcome to Dr Varga. Thank you very much. Thank goodness. Mhm. All right, well, thank you very much for the invitation to be here and for allowing me to share a little bit about what my lab works on in the way of sleep and cognition and Alzheimer's disease. So sleep is thought to be important for memory consolidation. But the validity of this statement depends in part on which stages or features of sleep are being implicated in what types of memory are being tested. So sleep is broadly divided into non REM and REM sleep stages, with non REM divided further into stages 12 and three. The reflect increasing levels of depth and, uh, sleep and wake vigilant states are defined predominantly by the frequency of brain waves that you see in that state. Such the wakefulness has a predominant 8 to 13 hertz, uh, frequency that's called the algorithm. As you go from wakefulness into non REM stage one sleep, you can see that this starts to slow down a bit. Um, the frequency gets more mixed. You can still see some periods of alpha, but then you start to see some slower rhythms in the 4 to 7 hertz range, which are which are known as state of waves. And as you get into deeper sleep, non REM stage to, uh, the rhythm is generally more uniformly slow, but you can get some bursts of high frequency activity in the 12 to 16 hertz range that are called sleep spindles, and you can also get these large amplitude deflections that are low frequency that are called K complexes. And then, as you get into non REM Stage three, you can see that the amplitude of the waves is more uniformly large, and the frequency is the slowest that you'll see during any of the sleep stages in the 0.5 to 3 hertz range. And so these are called delta waves or slow waves and are characteristic of slow wave sleep or non REM Stage three sleep and then REM sleep is very different, no chemically and was often called paradoxical sleep for many years because the rhythm can look a lot like wakefulness. There can be a lot of alpha in there, but there can also be a very predominant 4 to 700 state arrange, and the morphology of these waves is thought to be so much sharper and have the sawtooth appearance, which characterized which is characteristic of REM sleep and in the same way that sleep is not a unitary phenomenon. Memory can also be divided into a variety of subtypes, so you have things like declarative memories for facts, episodic memories for events from across time, emotional memories, implicit memories, things like perceptual memories or procedural motor memories. And in this talk, I'm going to focus quite a bit on spatial memories and in particular on spatial navigational memory, which I define as the ability to recall routes to sailing targets and novel environments. And there's a few reasons for this particular interest in spatial navigational memory with relationship to sleep, and the first has to do with this idea of replay. So it turns out that hippocampal play cells replay patterns of activity during sleep that were recorded during periods of active exploration during prior wakefulness, and this has worked from mostly rodent models. And it turns out that this replay occurs most reliably and slow wave sleep, but you can also see it in REM sleep. And so this is a mechanism by which sleep disruption is likely to have an acute effect when sleep is disrupted. This replay isn't happening as efficiently, and it's, uh, thought that that that disrupts the consolidation or processing of the spatial information that occurred earlier during wakefulness. And I should just point out that the 2014 Nobel Prize for physiology or medicine was actually given to individuals who identified that the camp of place cells or quote for their discoveries of cells, the Constitutive Positioning System in the brain. The other reason we're interested in spatial navigational memory in particular, is because it turns out that spatial memory is one of the earliest forms of memory that's affected in Alzheimer's disease. And so this is actually from an ad for a medicine meant to treat Alzheimer's disease that says, your patients with Alzheimer's dementia can feel like they're trapped in a maze. So as you guys are aware, Alzheimer's disease is a neurodegenerative disorder, Uh, that can be appreciated even on the gross level. So this is a gross picture of an Alzheimer's brain that has widening of the salsa and thinning of the gira suggestive of atrophy and degeneration. And on a neuro pathological level. We know that Alzheimer's disease is characterized by neurotic plaques, which are made up of beta amyloid protein, and these are extra cellular and also characterized by narrow February tangles, which were made up of the tau protein. These are largely intracellular current thinking about Alzheimer's disease is that there can actually be a very long pre clinical phase, last year's to possibly even decades. And that's represented in this figure here, where you see cognition shown by the purple trace going from some normal level at the bottom of the Y axis to some abnormal level at the top of the Y axis and well before that in time, you see that there can be these other changes, such as accumulation of beta amyloid shown in red, and this can be observed, observed both in the spinal fluid and by brain imaging with PET and also changes in the uh in Tao shown in the green tracing So um, metabolism of beta amyloid and tau represents a second potential mechanism by which sleep and spatial memory interact, and it's for the following reasons. One is that sleep is thought to result in decreased production of beta amyloid and tau, which is possibly related to decrease normal activity in certain cortical regions during sleep. Additionally, sleep is thought to result in increased clearance of neuronal metabolites, including things like beta amyloid, and so this is a mechanism in which sleep disruption is more likely to have a chronic effect. Um, you're not gonna have your memory disrupted due to the small amounts of accumulation of beta amyloid in town from one night. But when this happens over and over and over again across many nights in many years you can get accumulation of beta amyloid and tau, increase the formation of plaques and tangles and impact spatial memory in that way. So what are some of the tools that we use to study spatial memory and risk for Alzheimer's disease in our lab? Well, first, we developed a sleep dependent spatial navigation tasks that I'll tell you more about momentarily. We also do a lot of spinal fluid collection, where we look predominantly levels of things like beta amyloid 42 40 in the spinal fluid. We also look at levels of both total and fast food related tower in the spinal fluid, and then finally, we do a fair amount of brain imaging, and this includes some standard structural imaging that allows us to assess cortical volumes. We also do some pet imaging primarily to look at amyloid plaque load in the brain. And so today I'm going to tell you guys to stories about sleep, spatial memory and risk for Alzheimer's disease. The first, really, is about what happens in the absence of significant obstructive sleep apnea. And here you're going to hear a story about the role of slow wave sleep and its particular influence on memory and beta amyloid, and then a separate story on the role of sleep spindles in its relationship with Tao. And then I'm going to switch gears a little bit and talk more specifically about the influence of obstructive sleep apnea on both memory and risk for Alzheimer's by virtue of modulation of some of these biomarkers. So, um, we know that there's a number of changes that happened to sleep with aging. Um, we know that slow wave sleep declines with age. We know that mark is a sleep fragmentation increased with age. Cortical atrophy increases with age, and slow waves are thought to generate in the frontal cortex. And some prior work indicated that slow wave sleep might be particularly important for processing of some of these special navigation of memories. Um, and this was work. This, um, this showed that the degree of cerebral blood flow specifically to the right hippocampus correlated with the degree of improvement across the night on a special navigation task and a small number of people. So this is only six subjects. But this at least suggested that slowly sleep might be important for spatial navigational memory processing and set up the possibility that declines of slowly sleep with age might then impact the ability of older people to appropriately processed facial navigational memories. And so, in this first work, we aim to do a couple things which refers to ascertain the consequences of aging on sleep, architecture and sleep fragmentation, to examine the relationship between prefrontal, cortical volume and slowly of activity, and to determine whether the consolidation of spatial navigation of memory that normally occurs across sleep is diminished as a function of slowly sleep disruption with age. So in order to do this, we had subjects common Explorer, a three D computer generated spatial maize and then complete three time trials before and after sleep. With these trial capped at 10 minutes, so shown here on the left hand side is the first person view of the maze. This is something subjects might actually see while they're exploring the maze, and shown on the right is a bird's eye view of the maze, and I should point out that subjects never actually see this but there, perhaps able to imagine it in their minds, I by virtue of their exploration. So in this work we had a group of older subjects who were cognitively normal by neuro psychiatric testing and had no sleep complaints. The number 13 total with eight women and five men and had an average age of 68 years. We also had a group of younger subjects who also had no sleep complaints and were cognitively normal. They numbered 18 total with 10 women and eight men, and they had an average age of 20 years. And this work subjects were excluded if they had more than mild sleep apnea president, defined by an H I 4% greater than 15 per hour, or if subjects failed to find the target two or more times during their pre sleep trials. So the experimental timeline looks something like this. We had subjects come in the evening, explore the maze for the first time and then complete three time trials on on the maze, and then subjects were connected to policy ethnography and had a standard diagnostic. Alexander Graham overnight, about one hour after awakening in the morning, they would complete a 20 minute cycle mortar vigilance test for attention. And then, at the completion of the second water vigilance test, they complete their three final test trials on the maze and all subjects completed a three test three Tesla structural brain MRI prior to the experimental night, and prefrontal cortical volumes were calculated using free surfer. So showing here are measures of sleep sleep architecture between younger and older subjects. And we saw, of course, the expected decrease in non REM slow wave sleep in older subjects as compared to, uh, to younger subjects. And this was compensated for in part by, uh, smaller. But nonetheless, statistically significant increases in non REM Stage two and REM sleep in older subjects as compared to younger subjects. Now, in addition to a slow wave, sleep amount were also interested in the power in these slow waves, which is called the Slow Wave Activity. And we looked at this in a topographic sort of fashion and noticed that, um, the slow of activity was particularly decreased across the front of leads F three and F four and older subjects as compared to younger subjects. And finally, we're interested not only in amount and power of the slow ways, but also some measure of sleep continuity in these different stages. And this is represented in these tracings where the probability on the Y axis is shown of sleep, of any particular stage lasting the duration shown on the X axis. And what we see is that for REM sleep and non REM Stage one and non REM Stage two sleep that these curves are actually entirely overlapping for younger and older subjects. But for non REM stage three or slow wave sleep, we see that there's a significant leftward shift of this curve for older subjects as compared to younger subjects suggesting the slow wave. Sleep is the one stage of sleep that is particularly more fragmented in older individuals. And when we actually looked at the mean about length of slow wave sleep, we saw that it was significantly different between older and younger subjects, with younger subjects having a mean about length of slow receive of nearly six minutes, where it was much shorter at just two minutes for older subjects. So shown here is the pre sleep completion time performance on the maze between older subjects shown in gray and younger subjects shown in white. And perhaps not surprisingly, we see that older subjects are taking longer, on average, to complete their priestly trials as compared to younger subjects. And this was statistically significant. But when we actually looked at how performance change its compared to their baseline, we actually saw that older subjects in younger subjects seem to get relatively better, uh, equivalently across the first three trials, just that older subjects got about 27% better between trials one and three before asleep. And younger subjects got about 24% better between trials one and three before sleep. And this is important because it suggests that both younger and older individuals are able to appropriately encode the special material and and act on what they've learned to improve to, um, relatively similar degrees. So now what happens after sleep? So even though older subjects are starting from a much higher baseline and you'd expect there to be a lot more room for improvement, we see that other subjects actually get a very small 4.8% performance improvement that was not statistically significantly different than their evening performance. In contrast, younger subjects get actually 31% better in the morning, and this was statistically significantly improved compared to their uh, evening performance. Additionally, when we look at the percent change in completion time as a function of this power in the frontal slow waves, we see that there is a significant positive correlations suggesting that the higher the relative frontal slope of activity, the more likely that you're going to get better across the night on the special navigation task in terms of your completion time. Importantly, we saw no differences in, uh, the second motor vigilance testing in the morning between younger and older subjects. So there are no differences in mean reaction time and no difference in the number of lapses on the TV T. And this is important because it suggests that any change in performance in the morning between younger and older subjects is unlikely to be due to any sleepiness or an attention that may have resulted from differences and sleep architecture between the two groups. Now this measure of frontal slow of activity also very significantly correlated with medial prefrontal cortical volumes across all subjects. And when we look just in the older subjects, we saw that measures of slow wave sleep either By this frontal, slow wave activity is shown in red or simply looking at just total slow wave. Sleep duration shown in green had a significant inverse correlation with levels of beta amyloid in the spinal fluid of the older subjects. And so what does this mean? Um, well, it's known that levels of beta amyloid in the spinal fluid are low by the time someone has, uh, full blown Alzheimer's dementia as compared to age matched controls, Um, and people who are cognitively normal. Um, but we and others think that there may be this model of how beta amyloid changes over time in the spinal fluid shown in red, where it has this inverse U shaped curve where initially levels of beta ambulance pops would go up before reaching some tipping point and then going down. So at cross section, what we think we're seeing is people on this side of the curve where having low slow of activity is associated with having higher beta amyloid, 42 levels in the spinal fluid and kind of pushing people towards this tipping point, where people with higher slow of activity or perhaps protected by having lower levels of beta amyloid 42. Now, some evidence for this model, uh, comes from work looking at associations between beta amyloid and tau in the spinal fluid. So fortunately for our field has a lot easier and more straightforward. How pretty much only goes up over time and it goes up to a higher degree and people that are, uh, prone to developing Alzheimer's disease. And so this was work where levels of tau and beta amyloid 42 in the spinal fluid were amalgamated from a few different cohorts and what they what they did was they broke, uh, the ages of these individuals down into into portals. And so when they looked at the earliest portals from ages 45 to about 63 from 63 to 70 they saw that there's actually a positive overall association between beta amyloid 42 tests that as Tau goes up, so does beta amyloid 42. But when they looked at the older, uh, portals of age from age is, uh, 70 to 76 from 76 to 90. They saw that overall, there was this negative relationship between beta amyloid 42. And how suggesting that as time continues to go up, that actually beta amyloid starts to go down. And so this was the direct conclusion from this work that says we found that the relationship between Bad Emily, 42 how changes with age between ages, uh, 45 70 years. There's a positive linear association. Where is between 71 90 years? There is a negative linear association between 8 to 42 town, so some of the conclusions from this work includes the fact that aging is associated both with decreased slow wave sleep and increased low. A sleep fragmentation aging is associated with decreased consolidation of spatial navigational memory. Without significant changes in mourning cycle motor vigilance, overnight performance changes positively correlated with frontal slow wave activity, and frontal slow of activity is positively correlated with medial prefrontal cortical volume. So taken together, we think these data contribute to an emerging model in which medial prefrontal cortical atrophy with aging leads to slow wave sleep disruptions that subsequently lead to impaired memory consolidation acutely and can also lead to elevated levels of beta amyloid acutely, which over time they predispose someone to developing plaques and impairing spatial memory in that way. So, so far, I've been telling you a lot about this relationship between beta amyloid and, uh, and slow waves, and we're wondering whether there might be some marker of sleep. Uh, that predicts levels of tau in the spinal fluid, because we actually looked at whether Tao is associated with slow of activity and found that there were not any significant relationships. And so this is work that's spearheaded by Corey Cam, who's a postdoc in my lab. And there's a couple of reasons why we thought that, uh, spindles might be a marker of sleep architecture that associates with talents for the next couple of reasons. So first, we know that spindle density is a rough correlative intelligence and involved in forms of sleep dependent memory processing, brain towel load correlates with cognitive capacity and actually does so better than beta amyloid does. So both spindles and Tao are linked by their connections to to cognition, and we know that spindle density is something that also decreases with age and so, sure enough, When we looked at levels of, uh, CSF uh, tile levels, we saw that there was this inverse correlation with spindle density in non REM Stage two sleep. And we looked at a couple other measures of spindles in addition to this density. And so we saw that, uh, some other features of sleep sleep spindles such as just count, uh, total count shown in panel B spindle duration showing in panel C or the frequency of fast spindles, which are, um, spindles, generally in the 13 to 16 Hertz range. All showed this negative association with levels of Tao in the spinal fluid. And this correlation remained significant when controlling for age and sex. And these correlations were also present but somewhat weaker. Between Tao and spindles and other stages of sleeves. We looked at spindles in stage three non rem sleep or look at spindles and all of the non REM sleep as a whole. Uh, these correlations also exist at all, albeit somewhat weaker than a non REM Stage two. Now, having defined this relationship between spindle density and town, we really wanted to vet this relationship a little bit more, and we wanted to vet it in two ways. We wanted to first of all, really ascertain that it's Tao that's really specific to the association with Spindles. Um, and not that it might be some other Alzheimer's disease biomarker and conversely, were also really interested in making sure that there's something specific about spindles. That's the predicts tower. And it's not just some measure of poor sleep in general. So we did some regression modeling to try to suss this out a little bit. And in the first regression models, we used, uh, spindle density as the outcome variable. And then we plugged a bunch of predictor variables into these models. And in the first model, we just included age sex in April 4 genotype as the predictors for spindle density. We saw that age sex in April before combined to predict 12.8% of the variance in spindle density. And then on top of that, we added one at a time. Uh, the concentrations of either a beta 42 phosphor related to our total Tao and saw that adding a beta 42 contributed an additional 8.8% of the variants to spindle density. The first related towel added an additional 18 1% of the variance of spindle density, and the total towel predicted an additional 25.9% of the variance in spindle density. And then we took pairs of these 80 mile markers together. And when we added Total Tau and Beta 42 together, we actually saw the tower remained a significant predictor of spindle density. But a beta 42 no longer did. And similarly, when we added a bit of 42 44 related to together, we saw a similar pattern, which was a bit of 42 was no longer a significant predictor of spindle density but phosphor related to what was so taken together. The data indicated to us that a couple of things one was that, uh, Tao is really the strongest predictor of spindle density, much more so than beta amyloid 42. And then, in fact, when you when you account for both of these together, that, um, the degree to which beta amyloid 42 alone predicts presidency is probably a result of its intrinsic associations with with Tao and then when you add these things together, a beta 42 falls out Okay, So now, conversely, were interested. Having established the Tao is the predominant 80 bio market that predicts spindles. We wanted to, um, determine whether you know the extent to which spindles were, uh, really the specific specific predictor of of CSF towel levels. And so we again did some regression modelling where, uh, initially we included just age sex in April, before genotype as predictors for now CSF Tao, and saw that those three together predicted 7.7% of the variants in CSF toe. And then similarly, we we went and added one sleep variable at a time and saw that of all the sleep variables that spindle density and stage and to uh, contributed an additional 28.4% of the variants and CSF tie levels. And then we went and added these other sleep variables and found that none of them were significant predictors at all. And they included a bunch of different things, including things like slow of activity, the amount of time of wake after sleep onset sleep efficiency, two different markers of apnea severity, total sleep time measured in the lab with polish stenography and then also a total sleep time at home over over seven days. That was measured with active graffiti. So So none of these other sleep variables were, uh, significant predictors of CSF Tao levels. So now, having established this relationship between Tao and and sleep spindle density and non REM stage two, we wondered whether this might have some some functional implications for older people. And, uh, unfortunately, we didn't have a very high number of these people that completed an actual sleep dependent memory task. But we did have all of them do, um, some pretty standard daytime neuropsychological testing, including these tests that are shown here. So one of the tests we had people do with something called the digit symbol substitution test. And this is just simply a test where numbers digits one through nine are encoded by these different symbols. And then you're given a little time to practice this and put the symbols in. You know, these boxes here and then once you reach the sick black line, you're actually, uh, time. So you get 90 seconds to fill in as many of these as you can, and the more you can do in 90 seconds, the better your performance. So that's That's the digits. Simple substitution test. And then we also had people do task called Trails A and B, which are essentially a versions of Connect the dots. So trails a literally is connect the dots. It's just connect 12 to 2 to 33 to 4, and so on. There's 24 total items, and you're just time dressed to do this as quickly as you can. So lower time is better on this test and trails B is similar, except that now numbers are interdigital did with letters. So you have to connect one to A A to To to to be be 23 and so on. And, uh, and again, there's 24 total items in this and you're just you're tested for speed so similarly lower, Uh, time is better performance. And so relationships between, uh, spindle density and some of these, uh, neurocognitive performance measures are shown here. So the cognitive performance measures are the trails, trails being digit symbol substitution test. On the left hand side of each of these tests are the just sort of robbed by variant on adjusted correlations, and then on the on the right hand, side. We list the correlations that are where the cognitive task is actually adjusted for age, sex, race and years of education. And so we saw some interesting things, Um, which included the fact that both total spindle density across the full frequency range of spindles in the 11 to 8 16 hertz, as well as just fast spindles in the 13 to 16 hertz range, uh, significantly correlated with trails, a performance both unadjusted and adjusted. Additionally, fast enough density significantly correlated with the digit symbol substitution test at the unadjusted level, but not after adjusting for the co variants. And then, somewhat interestingly, we actually saw that there was this significant but inverse relationship between slow spindle density and trails be such that higher levels of slow spindles seem to be bad. It actually took longer to perform trails be the more slow spindles you have. So to summarize this portion of the work, we demonstrated that low spindle density is associated with high CSF tao concentrations and the CSF total. Tao explains greater variants and spindle density than, uh, levels of beta amyloid 42 in the spinal fluid. Do we demonstrated that spindle density and non REM Stage two Sleep explains explains. The greatest variants and CSF total tell levels among multiple sleep variables. And, uh, the directionality of this relationship remains to be explored. So we don't know whether it's how this driving spindles or spindles is driving tower or possibly neither. I could be just an epi phenomenon. We do think there may be some functional relevance, and that's middle density does seem to predict at least some, uh, neurocognitive performance outcomes. So now I'm going to switch gears a little bit and talk about obstructive sleep apnea. So, as you guys likely no obstructive sleep apnea occurs when soft tissues in the upper airway collapse uh, repetitive lee leading to, uh, to main consequences which are shown here in this, uh, Polly sonogram on the screen. So, um, this is what a pretty standard PhD looks like. The e G activity is shown here on a 32nd time. Window breathing is shown in this blue trace, also on the 32nd time window, and then more of the breathing variables are shown in this expanded time window that it's two minutes overall. And so in this individual who is having repetitive avenues we see the breathing shown in this blue trace goes from some pretty normal. You know, Sonya, soy to basically an absence of flow. And at the same time, we see that there are efforts still going on in these, uh, effort belts demonstrating that these events are in fact obstructive. And we see that, of course, along with these, uh, cessation in breathing, that levels of oxygen saturation shown in this red tracing here are are fluctuating. They're going from normal levels and in the nineties, down in this case, the pretty significantly low levels in the seventies and and cycling again. So that's what we call intermittent hypoxia. And the other thing that happens with these ethnic events is shown in the egg, where, as you can see, when the breathing stops and then this person takes a recovery breath, we see that there's a sudden burst of high frequency activity in the E G, along with a burst of activity in the in the muscle tone in the chin, suggestive of an arousal. So this is, uh, the other hallmark of Sleep Avenue, which is sleep fragmentation. And, uh, just to get give you guys a little primer on the nomenclature of respiratory events. Apnea is pretty easy to define. Its just no breathing for 10 seconds or more, and you don't have to have any other qualifying criteria to define that. Hi, papa. On the other hand, uh, is simply reduced breathing, and the American Academy of Sleep Medicine requires that there's a reduction of the amplitude of breathing by at least 30% or more from some baseline level. And this reduction in breathing has to last at least 10 seconds, and it has to be associated with some outcomes. So, um, if that hypothesis is associated with a drop in blood oxygen saturation of 4% or more, we call that a hypothesis of 4%. However, if it terminates in a 3% or more drop in blood, oxygen saturation or terminates in arousal, it's called the Hypothalamus three A and the app in High Poverty Index, or H. I is simply just the number of apnea is, and the number of hypotheses added together across sleep and divided by total sleep time. But because there's two different definitions of high papa, you end up with two different ages that we like to define as an H I 4% which is the metric that really captures more of the intermittent hypoxia and then something that we call an H I all or sometimes in H I three a. That's reflective of either the the saturation zor the arousal. And it turns out that if you have an H I 4% greater than five per hour or an H I all greater than 15 per hour, plus daytime symptoms and typically that's daytime sleepiness. Then you're said to have the obstructive sleep apnea syndrome, or O S, A s, and the estimated prevalence of the obstructive sleep apnea syndrome is about 4% across the U. S population. Now, if you have just an h I 4% greater than five per hour or an h I all greater than 15 per hour, you know, whether you have daytime symptoms or not, you're said to have always a somewhat more generically and in this case, the prevalence of always A is much higher. So it's about 20 to 30% across the population. Um, but it's even higher in people that are older. So we think that the prevalence of this kind of more generic obstructive sleep apnea. Prevalence is about 30 to 50% and older individuals and some of this work comes from this hip and allow study, which was a large epidemiological study coming out of Switzerland that had more than 2000 subjects with an age range of 40 to 85 years. And so, uh, apnea, by severity categories is shown in these bar charts here, separated by gender between men and women, and the light blue bars are indicative of older individuals those that are greater than equal to 60 years old. So we see that in men, there's actually 66% of men have an H I, uh, 4% of at least 15 per hour, half of whom are 33% have an H I, all between 15 and 30 which would qualify as mild obstructive sleep apnea and then 33% having an h I all greater than or equal to 30 per hour, which we categorizes more moderate to severe. And in women the numbers are about 40%. So we get about 23% of older women that have an H I in the 15 to 30 range and then about 15% that have an H II greater than 30. So we've been following a cohort of cognitively normal older individuals. They're cognitively normal by virtue of having clinical dementia rating scores of zero. They're screened for depression and found to be non depressed. Um, in our cohort, they have an average age of 68 years with a range of 55 to 90 years. And these older people have done a bunch of things for us. So they complete, um, a Home Polly Sonogram at Baseline and the Home Policy. Sonogram is really good for determining apnea severity. It's not so great for determining sleep Architecture changes. They also do a lumbar puncture for us. And then they do some brain imaging, which includes both structural armories as well as Emily Pet Imaging. And so here are some of the characteristics of this cohort divided by apnea severity. So this is kind of ongoing work. But at the time we did this analysis, we had 208 of these older subjects, and we saw that 46.6 of them had normal sleep breathing. But of course, that means that greater than 50% have at least mild USA in our in our hands, which is similar to what's seen in hymnals. So in our hands we see that about 36.5% of these people have mild obstructive sleep apnea and 16.8% have more moderate to severe obstructive sleep avenue. Now, despite this high prevalence of obstructive sleep avenue, we actually see that there's not a lot of daytime sleepiness going on. So shown in this row is the Epworth Sleepiness score or E. S s. This is a 24 point questionnaire that asked people about the likelihood of falling asleep in a variety of different situations. And clinically, we think that scores of 10 and higher are indicative of clinically meaningful levels of sleepiness. But these subjects have very low upward sleeping the score. So the entire group has an average upward sleeping a score of five, and it ranges only from 4 to 6 across the different the essay severity categories. Additionally, these older people are sleeping a pretty normal amount of time. So on average, the group is sleeping seven hours, and it ranges from 7 to 7.5 hours across the different OS A severity categories. So we were interested in asking questions about what obstructive sleep apnea in these older people, uh, might be doing, uh, military sleeve for memory. And so, um, we wanted to look at how these people did on the same spatial navigation task that I described earlier. And so this is work that's really spearheaded by mastery. Williams, who is a second year medical student here at Mount Sinai, as well as Anna Mullins, who's a post doctoral fellow in my Love. So in this work we had a group of older subjects who had no obstructive sleep apnea. These are subjects who are cognitively normal by neuropsychiatric testing and had no sleep complaints. So I should point out that these are people that were essentially recruited from the community, not recruited from the sleep clinic. So no sleep complaints. They numbered 30 total with 19 women and 11 men, had an average age of 67 years, and they had a median h I all of five per hour, which puts them well within the normal range. And then we compared them to a group of subjects that were older and had obstructive sleep apnea who also had no sleep complaints and were cognitively normal. The number 12 total with seven women and five men, had an average age of 66 years and they had a median h I all of 21 per hour. So this puts them kind of well in the middle of the mild range of obstructive sleep apnea. And in this work, subjects were excluded if they failed to find the target two or more times during their pre sleep trials. So shown here are some of the sleep architecture measures between individuals with and without sleep apnea from this older group and of course, by design. We saw that those subjects with obstructive sleep avenue had significantly higher degrees measured by either an HIV 4% or an H I all. We also saw that this total arousal index was higher in subjects with obstructive sleep. Apnea is compared to those without in terms of the sleep architecture. We saw that there was a significant increase in the amount of light non REM stage one sleep in subjects with obstructive sleep apnea and a significant decrease in the amount of non run stage two sleep in subjects with obstructive sleep apnea, but, uh, notably no differences and no significant differences in non REM stage three or slow wave sleep or in REM sleep, although on average the numbers were a bit lower, as as you might expect compared to those people that had no obstructive sleep apnea. So what happens with this overnight changing completion time? Well, for the subjects without obstructive sleep apnea, we see that there's about this 5% performance improvement overnight. This is very similar to some of the data I showed you earlier, Uh, when the older subjects were compared to younger subjects. When we looked at the subject with obstructive sleep apnea, there actually were getting worse overnight on average, so other performance went about negative 10%. So they're getting about 10% worse across the night as compared to okay compared to their pre sleep trials. Um, as you can see, there's a lot of scatter in this data. So it turns out that although people would always say are performing worse in terms of the percent change in completion time, this is not significantly different than the performance of people with without osa. We wanted to look at this little more carefully by looking at their trial by trial performance to get a more nuanced picture. Perhaps what's going on with the people that have sleep at in this older group and so shown here is the trial by trial performance, uh, for individuals without obstructive sleep apnea shown in the in the white circles. And this performance conforms to my you know, preconceived expectations of how people probably should be performing on this task. So showing here is completion time, and what you see is that across the first three pre sleep trials, people are getting faster, meaning they're doing better. And then they sleep. And then they come back in the morning, and overall, they continue to get better and better by and large, and their sixth trial is their fastest and best trial overall. So that seems to make sense when you look at people with obstructive Sleep Avenue. If you look at their precinct trials, there's really no significant difference. They also get a little bit better across their first three trials, uh, overnight, and this is not scientifically significantly significantly different than the way people without obstructive sleep apnea perform what happens in the morning is that they perform in a way that's very different than people with normal sleep breathing. So people with obstructive sleep apnea is one of the great dots actually get worse and worse and worse across the morning trials such that their sixth and final trial in the morning is actually the worst trial among all trials. And this was statistically significantly different pattern of performance than people without obstructive sleep apnea. Similarly, we looked at second border vigilance measures, including me in reaction time and number of lapses between individuals without and with obstructive sleep apnea, and saw that in this case there were no differences. Um, again, this is important because it suggests that this pattern of performance in the morning on these Maze trials is unlikely to be due to, uh, significant sleepiness or inattention that may have resulted from the model level of apnea in this group. So now, in addition to looking at acute effects on memory, were also interested in understanding potential longitudinal effects of having obstructive sleep apnea on risk for Alzheimer's disease. And as I mentioned, we had this cohort that number 208 subjects total, and we actually had 109 subjects come back for a second LP. That was performed about 2.5 years after the first. And we also had a smaller subset of 34 subjects that ended up completing a second amyloid pet scan. And again, this is about 2.5 years after the first. And so what we observed was that apnea severity measured either by an H i all or an H I 4% uh, predicted significant decreases in the levels of beta amyloid 42 in the spinal fluid over time. So again, to take you back to this model of how we think beta amyloid 42 is changing in the spinal fluid. We think we're capturing people who are now on this side of the curve and are starting to decrease their levels of beta amyloid 42 over time. And of course, if this is true for right about this, then we should also see increases in cortical amyloid deposits in the brain. And so that's actually exactly what we saw when we looked at levels of amyloid in the brain itself by pet imaging. So, uh, h I all or h I 4% was associated with a significant increase in the levels of, uh, amyloid by pet imaging over time in this group. Now, one of the other ways we're interested in understanding the relationship between, uh, sleep apnea and, uh and cognition over time was to, uh, do some data mining from this Alzheimer's disease Neuroimaging initiative or add Nico. So add me is, uh is an active, ongoing cohort. At the time we did the analysis, there were 2470 subjects, and there followed for 2 to 3 years. And all this data is publicly available. And it turns out that there was a subset of 767 subjects that were available for sleep analysis. And these are subjects who completed sleep questionnaires with really sort of simple and rudimentary questions. Like, Do you have sleep apnea? Yes or no? And if you have sleep apnea, do you see path to treat it? Yes or no. And so, even though the level of sleep data from this group was was, you know, self reported and relatively simplistic, uh, one of the nice things about this group was that they were followed very carefully for their cognition and in those people that ended up converting from normal cognition to mild cognitive impairment. Just thought to be the early harbinger for Alzheimer's disease. The precise age at which that conversion happened was known in this acne cohort. And so when we looked at onset of Moloch, cognitive impairment between individuals with and without self reported sleep disordered breathing, we saw that there was a significant difference between these groups. So individuals with self reported sleep disordered breathing shown in the green tracing had a much earlier age of conversion to mild cognitive impairment that did people without any self reported sleep disordered breathing shown in the green trees and the difference was pretty substantial was about 11 years. So individuals without sleep disordered breathing had an average age of onset of about 83.6 years, whereas those with self reported sleep disordered breathing had an average age of onset, uh, tamale cognitive impairment of 72.6 years. And even though the number of individuals reporting use of CPAP was very low when we looked at the survival curve for conversion to M. C. I and those people using CPAP shown in the Red Trace, we saw that their curve actually looked fairly similar to what seen in individuals with no sleep disordered breathing at all. And so the average age of onset to mild cognitive impairment and individuals who are CPAP users was actually 82.1 years, so fairly similar to those without any sleep disordered breathing at all. Now, additionally, in this Agni cohort we could look at not only, uh, age at which people converted to M. C I. But we could actually look at how some of these 80 biomarkers are changing over time. And this is work that was really spearheaded by Michael Boo Boo, who's a post doctor fellow and, uh, Ricardo Sorrows Lab, who's a colleague of Iris. So when we looked at some of these changes in 80 biomarkers, what Michael found was that, uh, levels of beta amyloid by Penn imaging were increasing to a greater degree in those uh, with self reported obstructive sleep apnea shown in the red tracing that's compared to those without levels of beta amyloid in the spinal fluid were decreasing at faster rates again and and the people that have self reported obstructive sleep apnea. And then we saw similar patterns for Tao Tao. Both total towel levels and phosphor related to levels were increasing at faster rates across about a three year period, and those people were self reported obstructive sleep apnea as compared to those without so to summarize this portion of the work. We've demonstrated that obstructive sleep apnea is very common in elderly subjects, although often without daytime sleepiness, obstructive sleep apnea, alter spatial navigational memory processing and the elderly acutely and greater OS A severity is associated with longitudinal decreases in CSF beta amyloid 42 levels and longitudinal increases in cortical amyloid deposits, both of which are thought to increased risk for the development of mild cognitive impairment and Alzheimer's disease. The presence of obstructive sleep apnea is associated with an earlier age of onset of mild cognitive impairment. And, perhaps most crucially, obstructive sleep apnea is a potentially modifiable risk factor for cognitive decline. So with that, I'd just like to thank many of my collaborators that made this work possible, including David Rapaport and in Dubai Opera here at Mount Sinai, Ricardo Osorio at the Center for Brain Health and a really talented cohort of students and postdocs, including Cory Cam Park and Mullins Michael Boo Boo and mastery Williams, most of whom are here in the audience today and finally like to thank my sponsors. And you guys for your attention, I'll take any question questions, please. Mm. One of the major divisions of memory impairment in old, healthy people is obviously loss of immediate recall or loss of something from the past. And educating people, as you know, have given the name. Oh, my God. I've had a senior moment when they're giving a lecturer and they forget something that happened a moment ago. So my question for you is do you study the difference between the two types of memory impairment that seem obvious to me? The older persons senior moment when they get something, But they have good recall of things that happened 50 80 years ago. Have you studied the difference of these two entities with all your measurements? Yeah. So the short answer is No, we haven't. Um you know, the longer answer is, um, you know, I think you bring up a good point, right? I mean, we know that you know, you can have some, uh, cognitive changes with aging. Some of that can be normal. As you said, I think it's normal to have you know what's called a senior moment. What you really you know, I think I want to pay more attention to is if if those are becoming progressive, right, if they're becoming more frequent. If there If there, uh you know, becoming, uh, or switching categories right. It's not just for sort of forgetting a person's name, but it's, you know, other things, right? And, uh, and I think you hit another important point, which is that, uh, long term. Memories are really tend to be very preserved. So by the time those are starting to be affected, there's usually it's usually indicative that there's a bigger problem going on. It's it's the, you know, prospective memory tends to be affected earlier, right? Stuff that you sort of recently memorized. Or, you know, I want to go to the store to get X y and Z items, and by the time you got there, you can't remember. That's the kind of prospective memory. So, um you know, like said we haven't we haven't delved into this, too much. I mean, I was talking with somebody before the talks started, and I was talking about the fact that, in my opinion, I think more can be done with some of the neuropsychiatric testing that's done. And people, you know, neuro psych testing is is relatively straightforward, actually. And because of that, by the time people develop impairments and frank neuropsychiatric testing there often like pretty significantly impaired. So it's my opinion that, you know, you should potentially push people with harder tasks on neuro psychiatric testing. You should take into account things like level of education, right, because we know the cognitive reserve is an important protective factor, uh, for for cognitive decline and and again because I'm asleep guy. I also think that, uh, it's potentially worth looking at cognitive tasks that are sleep dependent, meaning that, um, it would be great to have people encode something before sleep and then give them a period of sleep even if it's a daytime nap, and then ask them to recall it later. Because I think that's a paradigm that, um, pushes the system a little bit more, and I think you can bring out cognitive deficits that way. okay? Yes. Thank you for an excellent presentation. Um, were all of these data collected independent of whether or not the patients were on medication to induce Lee? Or is there a correction for that in this presentation? Yes. All all these patients were The patients were excluded if they were on any, uh, any sleep medicines, any north psychiatric medicines? It's a little sort of project specific. I mean, some of the tests were looking at sort of pure cognitive things. Um, in some cases, if you're on a single sorry, you can still be included, but if you're on multiple necessaries or any other strong neuropsychiatric medicine, stimulants, sedatives, that medicine, have you tested any of the rejected group? Yeah, but France, Come on. Thank you. We have, please. Mhm. By far the largest. Can you predict progressive changes in the basement? Um, in terms of publishing. Yeah, So you know we're doing that. So we're like I said, this work was with some longitudinal work that we reported the results after 2.5 years. Um, you know, we're following these people still and have some funding to continue following these people for another five years or so. So we think the changes in things like beta amyloid and tau biomarkers, as I showed in the early slide, are likely to be early harbingers for for actual cognitive change. Um, in terms of the number of people in our cohort who actively converted from normal cognition to my colleague impairment, it's been very, very small. But of course, we think that as we follow these people, uh, longitudinal, that we're gonna get, you know, higher numbers of people converting. So, you know, hopefully we'll be able to see that and and demonstrate that things like apnea it's a very to have an impact on actual rates of conversion to different cognitive classes. Right? So you're that genealogical study suggested that the C patch actually helped prevent fine. If you were to institute the sea path later. How much of that could be recovered? Is any of this reversible? Yeah, that's that's a million dollar question. And, uh, you know, there's surprisingly been not a lot done on this, actually, So, um, you can look at it from two different angles, right? One angle is the CPAP do anything to improve cognition, right? That's not been looked at in older people. Specifically, it's been looked at as a whole in, you know, a wide age range of people. This was a large clinical trial that was called apples, and they saw some pretty modest effects. Um, they saw, you know, these people are put on CPAP and followed for, you know, something like nine months or something. And, um, they saw some early improvements, Uh, like a three or six months or the time point, especially in domains like executive function that then seemed to kind of normalize at the at the later time point. Um, that studies criticized a little bit for a couple of things. I mean, one was that, um, people who had sleep apnea who were randomized to see pepper not in that trial were had very mild sleep avenue overall. So people said, Well, you know, maybe it's really the people that have more severe moderate is seriously bad news. They're gonna be more effective. And then the other thing is that these were you know, like I said, relatively youngish and cognitively normal people. So if you have a pretty cognitively normal baseline, you know, where is there going to be room to improve, right? So some people have said you really need to take people who are actually starting to show some cognitive deficits to see that there can be some improvement. Right, So that's on the cognitive side. And then the other thing you'd like to sort of see is, you know, does cpap actually, you know, change the course of accumulation of some of these 80 biomarkers and again that that work, um hasn't really been done. I mean, there was there was a very small study. Uh, if I'm remembering, collect correctly of people that were put on CPAP and and they looked at some of these farm workers and there did seem to be some improvement. But I think a lot more work needs to be done. And and I'm not sure, really What the what? The result is going to be because, you know, um, people don't develop, you know, sleep apnea overnight, right. They tend to have sleep apnea for long periods of time. And if you're someone who's you know, let's say 70 and being diagnosed with sleep apnea for the first time. I mean, there's a chance you might have had sleep apnea from the time you were 30 and so are there these decades long changes that happen to the brain as a consequence of obstructive sleep apnea that are irreversible? I mean that that may be possible. So, like I said, I think the data right now is pointing toward obstructive Sleep Avenue being a significant risk factor. But the extent to which that's reversible is it's like I said, it's a million dollar question and our group and others are making some headway in this. So, um, sort of remains to be seen, But the N I H has has reached out to our group actually about doing a clinical trial with CPAP and potentially other things for the treatment of obstructive sleep apnea. As you may know, one of the the bugaboos, of course, of any clinical trial for CPAP is that everybody hates it, and nobody really uses it very well. Um, so David Rapoport, who's in our group, has championed this idea that, you know, you should do a clinical trial where the, um, the intervention is anything that reduces the H I whatever that maybe it could be cpap. It could be a oral appliance it could be positional therapy. It could be a combination of all these three. But do anything you can to reduce the H I and then do a clinical trial that way. And and hopefully you get sort of better overall adherence and and can maybe have more quality in the study. Thank Dr Work.