Epigraph Vol. 24 Issue 4, Fall 2022
Alzheimer’s and epilepsy: Dr. Andrew Cole and Dr. Alice Lam
Dr. Patricia Saletti interviews Drs. Cole and Lam about potential connections between these two conditions. They discuss recent studies that have revealed intriguing associations, including bidirectional increased risk, anti-seizure medications for people with Alzheimer’s, and machine learning techniques to identify subclinical seizures.
Listen below or download the episode.
Dr. Saletti: Hi, I’m Patricia Saletti, neuroscientist and member of the Young Epilepsy Section of ILAE. Today we're going to talk with Dr. Andrew Cole and Dr. Alice Lam about the relationship between Alzheimer’s disease and epilepsy. Hello Dr. Cole, hello Dr. Lam, it’s a pleasure to have you both here on our podcast. I would like to ask you both to introduce yourselves.
Dr. Cole: Sure. Hi, thank you for inviting us. It's fun to be here. I’m Andrew Cole, I run the epilepsy service at Mass General and I'm professor of Neurology at Harvard Medical School and I've worked mainly and epilepsy and clinical neurophysiology for about uh, 30-some-odd years now.
Dr. Lam: Hi, I’m Alice Lam. I’m assistant professor of neurology at Harvard Medical School and I'm a neurologist, I specialize in epilepsy as well as memory disorders, and I have a research program that looks at the intersection between epilepsy and neurodegenerative issues.
Dr. Saletti: Alzheimer’s disease is a type of dementia that occurs later in life; it’s common in the elderly, while epilepsy is characterize by seizures and affects young children to older adults, so we can find the disease in a very large range of age. What is the relationship between both diseases?
Dr. Cole: I think Alzheimer’s disease and epilepsy are completely different diseases at some level. One is a progressive, cognitive disturbance, a dementing illness, that’s characterized by a very specific neuropathology. And the other, epilepsy, is a collection of many different illnesses, some genetic, some acquired, that cause recurrent seizures. So epilepsy is a bit of a basket of many different things. Whereas Alzheimer’s at some level is a relatively stereotyped degenerative process.
Now it is true, and Alice will talk more about this in a moment, that Alzheimer’s has clear genetic components and there are some probably clear genetic forms of Alzheimer’s disease. There may or may not be other forms; I’m not an expert at that. I think the thing of interest is how these two conditions involve some similar areas of the brain on many occasions. What got us interested in this question of seizures and Alzheimer’s? Alzheimer himself studied the brain pathologically and the hippocampus in particular, and had observations of specific morphological changes, and ultimately others found biochemical changes in those structures.
Epilepsy, while it can come from different parts of the brain frequently comes from the hippocampus as well, and epilepsy scientists have studied the hippocampus also for many years, but not so much with respect to its biochemistry and anatomy but with respect to its physiology. And the insight that has led us to where we are today was that this same structure that’s affected in different ways by very different diseases may play a role in linking the two conditions. That’s sort of how I got interested in this problem, and some observations that we made that we can talk about in a few minutes.
So, can they occur at the same time? Yes of course. Older patients with Alzheimer’s disease can develop seizures related to the Alzheimer’s disease or unrelated. For example, after a head injury or stroke or development of a brain tumor. We don’t see a lot of Alzheimer’s disease in young people, very young people, so in that sense there’s some distinction. Alice, you have perspectives on this from a different perspective – what do you think about this question about Alzheimer’s disease and epilepsy?
Dr. Lam: I think you did a really nice job addressing that they are two pretty different diseases, or probably conditions is the best way to put it. I will say that I think Alzheimer’s disease has more heterogeneity than you alluded to. It’s characterized by specific brain pathologies – amyloid and tau pathologies, and things that people are starting to look at more now in vivo using approaches like PET imaging. And using these approaches, I think people are starting to realize there’s a lot more heterogeneity than we originally appreciated.
That said, thinking back to this idea that in older adults, you can have epilepsy and Alzheimer’s disease occurring at the same time, that’s something that I’ve kind of been interested in, at least from a clinical standpoint, having clinics in epilepsy as well as memory disorders. There is a substantial number of people who present with new-onset epilepsy in their older years—late 50s, 60s, and afterwards, where they present with new-onset epilepsy for unknown reasons. You can’t see a stroke, you don’t see a tumor, they don’t have a history of brain injury, but then shortly after they develop seizures they develop cognitive decline, cognitive impairment. And when you start to study these patients more closely, do a diagnostic workup, a fair number of these people will have Alzheimer’s pathology. So the idea is that their seizures may have been an early manifestation of Alzheimer’s disease.
There are some people who think that maybe there is an epileptic prodromal variant of Alzheimer’s disease that presents initially with seizures and is followed by cognitive impairment. That’s something that’s still very early in our, we’re very early in our understanding of this process.
Dr. Saletti: Does one of the diseases increase the chances of the other one – the occurrence of the other one in these patients?
Dr. Lam: That’s a great question. There’ve been some epidemiologic studies in the past few years that have highlighted that there’s a bidirectional association between these two conditions. So if you have epilepsy, regardless of age of onset, you have a twofold to threefold increased risk of developing dementia later in life. Whereas if you have dementia, you also have a two to threefold increased risk of epilepsy. Why that bidirectional relationship exists, whether there are shared risk factors, we’re still trying to sort through that.
Dr. Cole: That’s interesting, I wasn’t aware of that, Alice. That’s fascinating.
When we first started thinking about epilepsy in the context of Alzheimer’s disease, we sort of speculated about different arrangements, or different relationships. You know, without good epidemiologic data. One relationship that’s been highlighted by older thinking or adopted in older thinking as that epilepsy is just a byproduct of Alzheimer’s disease. Alzheimer’s disease is a degenerative disease and whenever anything goes wrong with the brain, sometimes people have seizures because that’s one of the things the brain does when it's unhappy – it makes seizures. Becomes electrically unstable. It’s really a, not a symptom, but a secondary consequence of the disease and has nothing to do with the biology of the disease itself.
Another possibility is that there’s something about the Alzheimer pathology that actually produces hyperexcitability in a specified way – we can’t specify the way, but that there is a specified path, and this accounts for the occurrence of epilepsy in people with Alzheimer’s. But as far as we know, epilepsy is not universal among Alzheimer’s patients and if it was a consequence of a particular pathology, you might expect it would be extremely common, and we’re not sure how common it is.
Regardless of the relationship between Alzheimer’s disease and epilepsy, there are other questions. For example, does the epilepsy contribute to the symptoms of Alzheimer’s disease? Is some of the cognitive dysfunction actually due, not to the plaque or the tangle itself but to the electrical instability? Of course the corollary would be, if so, perhaps we could modulate that hyperexcitability pharmacologically or some other way. We could at least improve some of the symptoms of Alzheimer’s disease even if we don’t change the natural history, protein deposition, plaque formation, cell loss and other things.
So there’s the issue of whether epilepsy contributes to the progression of Alzheimer’s disease, that’s sort of the third issue. Is it possible that the epilepsy somehow interacts with some of the mechanisms underlying Alzheimer’s pathology, and I’m sort of talking about the canonical plaques and tangles and amyloid and tau. And I realize that many Alzheimer’s scientists feel that the collection of things we call Alzheimer’s disease probably has more diversity than that formulation allows for. Is it possible that the actual electrical events somehow speed that process, somehow make that process worse, and that even if we couldn’t reverse the process, we could slow it down or change the course of decline of patients who are fated to have Alzheimer’s dementia? So there are these different time scales and different relationship perspectives that we all have to jostle.
And the hardest thing to do would be to show that somehow stopping seizures would slow Alzheimer’s disease. You’d need a huge population to be able to see a significant change in the slope that would make you think wow, this is really important. It’s easier to know whether stopping seizures provides some symptomatic benefit, that the patient is not as dysfunctional, at least for a while, as they would otherwise be. And then the causation story is yet another one.
There is biochemical data, and Alice can speak to this more than I can, from David Holtzman and others, suggesting that certain kinds of electrical activity can promote the release of soluble amyloid and perhaps modulate tau deposition and tau phosphorylation. That’s not well demonstrated as far as I know, but there are some model systems in which there appears to be a relationship between activity, activity-dependent pathologies essentially, and that obviously would be a great target, a therapeutic target if that were robust.
Dr. Lam: You know, most of the data we have in terms of how epileptiform activity and hyperexcitability might affect the not just clinical, but also biological, progression of Alzheimer’s disease comes from animal models. It’s the animal models of Alzheimer’s disease that really kind of started this field, right, in terms of mouse models where this idea of silent seizures and silent epileptiform activity was first discovered, and it was in mouse models where people started using antiseizure medications to see if they could stop these seizures and stop the spikes, and found that you could actually improve cognition if you did that in a mouse model of Alzheimer’s disease.
And so in these models as well, a lot of work in cell cultures and brain slices, very careful biochemical and electrophysiologic work, has shown that soluble amyloid and soluble tau can cause neurons to become hyperactive, really. And that in turn there’s a feed forward pathway here, where neural hyperactivity was found to drive the release of soluble amyloid and tau from neurons and actually drive downstream deposition of these proteins into plaques and tangles.
Whether those pathways exist in humans – not really clear yet. Alzheimer’s disease in mice is quite different from Alzheimer’s disease in humans, of course. But you know it provides sort of that foundation for thinking about these kinds of processes in human Alzheimer’s disease and thinking about whether we can manipulate these processes and affect clinical outcomes.
Dr. Saletti: I would like to talk a little bit about silent seizures in Alzheimer’s patients. I had access to very interesting work from both of you that you published in 2017 in Nature Medicine, in which you record the activity with intracranial electrodes in two patients with cognitive decline who also had normal scalp EEG, and then you found some silent seizures happening. What are your perspectives for the future considering the silent seizures that these patients can have?
Dr. Cole: So the fundamental question was brought to us in conversation with Jeffrey Noebels and Lennart Mucke, who had identified hippocampal seizures in a particular amyloid transgenic mouse. They asked the question, “Do you think this happens in people, and how would you look for it?” And we had this little thing going on in our epilepsy surgery program where we used these particular electrodes called foramen ovale electrodes, which are actually not intracerebral but they are intracranial. They’re inside the skull but outside the brain. They end up right along the hippocampus. I said, “Well, we have the perfect tool to answer this question, we just need to find some patients with Alzheimer’s disease to allow us to put these electrodes in and see if we can actually see seizures.”
And lo and behold, over some years we looked and we came across a couple of individuals who had a very curious, funny thing about their dementia phenotype, at least as best as I could recognize, which is that they had a lot of fluctuation -- they had good days and bad days, and a regular amount of fluctuation, which the Alzheimer’s doctors are familiar with is maybe like this, and these patients had just slightly higher amplitudes of fluctuations, almost as if odd things were happening sometimes to them. And because of this we asked the question could they have seizures going on that give them these bad days.
And eventually, someone agreed to allow us to do this procedure and lo and behold the first evening that she went to sleep with these electrodes, she had a bunch of clear, crisp, electrographic seizures coming from the hippocampus with no sign on the scalp to speak of that would have given us a clue that this was going on. And little clinical change – she was asleep, apparently. The only thing that would happen is that at the end of each of these, she would arouse and wake up a little bit, just like anyone would wake up in the middle of the night – it wouldn’t attract any particular attention, but that was sort of a phenomenon at the end.
Then we found a second patient with a similar pattern, and we did the same thing, and we found the same result. So suddenly in two out of two patients we had seen this phenomenon of silent nocturnal seizures that you could record with these hippocampal electrodes or foramen ovale electrode. So that’s what started this whole thing – it was based on animal observation and traying to translate those to people.
Did we just get lucky – were these the only two people in Boston that had this and we just happened to study them? But what’s the actual frequency of this – is it rare, common, or universal? We don’t know. I suspect it’s common but not universal. How can we detect this in another way? We can’t put these electrodes into everybody.
I initially thought people would want to have this test to find out. But it turns out patients are much smarter than we are: “You’re not going to put those electrodes in my head! Absolutely not! Especially because you can’t demonstrate that giving me anti-seizure medications would make me think any better.” And of course with an n of 2, we couldn’t demonstrate that. So then began the quest of trying to find a way to recognize this phenotype without actually having to do the instrumentation. And Alice took that challenge on a couple of years ago and has done some amazing stuff that she’ll tell you about.
Dr. Lam: Thanks, Andy. Yeah, this is a challenge – where do we go from this discovery and how do we translate that into the wider population of patients with Alzheimer’s disease and understand the significance, whether it’s treatable, modifiable, things like that. So what we’ve been doing in my lab the past few years is trying to use datasets that come from people with temporal lobe epilepsy in order to develop machine learning algorithms that allow us to look at their scalp EEGs and infer from their scalp EEG whether there is this silent hippocampal activity going on deeper in the brain, that we can’t actually see with our naked eye with these EEGs, but using signal processing and these machine learning techniques, we can identify.
We’re lucky at [Massachusetts General Hospital] to have this large data set of a combined scalp EEG recordings along with foramen ovale electrode recordings – the same kind of electrodes we used in those patients with Alzheimer’s disease. We had a bunch of patients with temporal lobe epilepsy that had previously been studied at our institution -- they’d come into the epilepsy monitoring unit, be monitored for days to weeks at a time and we have a lot of this data.
So the idea is that we can train these algorithms to recognize the scalp signatures of these silent spikes. We were able to do this recently – we published a paper in JAMA Neurology earlier this year, describing a deep learning algorithm that can identify and detect these hippocampal spikes from just scalp EEG alone, with pretty good accuracy. This was just in temporal lobe epilepsy patients; it’s pretty promising, especially in these patients, and now the hard work is in trying to apply this in Alzheimer’s or older adults.
So that’s what we envision, that’s one approach that we could take to try to understand this phenomenon better, in a more widely accessible way. I think scalp EEG is certainly feasible in people with Alzheimer’s disease, and it would be a noninvasive away to be able to look for this activity and try to correlate that activity with again clinical outcomes, cognitive function, things like that.
Dr. Cole: Aside from the machine learning component of this system you’ve developed, do you think it’d be interesting to talk briefly about some of the more common EEG variants that have always thought to be normal that seem to be overrepresented in this population?
Dr. Lam: Sure, yeah. Andy, I think you’re referring to our Neurology paper in 2020. We’d studied people with Alzheimer’s disease using 24-hour scalp EEG, we captured overnight sleep in these patients and spent a lot of time actually visually reviewing these studies very carefully. One of the things we found that was interesting in these patients was that, you know, there’s this so-called benign variant on scalp EEG called small sharp spikes, or benign epileptiform transients of sleep, BETS – they’re these small, tiny little spikes, basically, that you could see during early stages of sleep. It’s thought to be benign, not to have an association with epilepsy.
I got interested in these because when we were reviewing the foramen ovale electrode recordings from our Alzheimer’s disease patients from the Nature Medicine paper, we were looking at both the scalp EEG as well as the foramen ovale electrodes at the same time and one of the things we noticed was that some of these spikes you saw from the foramen ovale electrode actually had a small scalp correlate – it looked exactly like one of these small sharp spikes, or BETS – something we would look at as an EEGer and just kind of blow over as a normal variant. But I got interested in whether that might be a scalp signature of these hippocampal spikes.
We did find a fair number of patients with Alzheimer’s disease who had much higher numbers of these small sharp spikes than we would expect to see, or that we saw really in older adult controls who were cognitively normal. Again, the clinical significance of this isn’t clear, but there does seem to be an association where patients who had more of these small sharp spikes and who had them unilaterally – either just the left side or the right side – tended to be patients who had Alzheimer’s disease and also had epileptiform activity or epilepsy.
What is the potential for using EEG biomarkers, like these epileptiform abnormalities, in Alzheimer’s disease?
Dr. Cole: I don’t think I could answer that question. If we recognize ultimately a tight association between the occurrence of these benign epileptiform transients of sleep and the development of Alzheimer’s disease, that would be very interesting, but I think there are a lot of reasons why people in the Alzheimer age group or pre-Alzheimer age group can have seizures or electrical abnormality on their EEG. Alzheimer’s may be one of those but it’s not the only one, and whether there’ll be a distinctive marker or not I don’t know.
There is an interesting experiment that’s doable – Alice can help me fill in the blanks here, but oneof our colleagues here is very interested in the genetically determined Alzheimer’s presenilin mutation of Alzheimer’s disease that’s seen in a very large extended family in Colombia.
The strength of having this gene, this gene which is in this family, is almost 100% predictive of developing Alzheimer’s disease at a very specific age, I may have it slightly wrong but it’s typically between 40 and 45. So this familial Alzheimer's in this presenilin mutation family is so predictable – there are only a handful of patients who should have developed the disease who haven’t, out of thousands, that you know exactly when it’s going to start.
So if you’re going to try to figure out, is there a good biomarker for who’s likely to have this gene, and you didn’t have the genetic information, you could do EEGs on 20-year-olds and 30-year-olds and 35-year-olds, years before they’re unfortunately going to develop this phenotype of the disease, and understand whether there are clear EEG predictors, or biomarkers, if you will. But it would also create an opportunity to do some pretty clever, focused trials to understand whether modulating the activity, the electrical abnormality, would actually change the latency to presentation of the first cognitive symptoms.
Of course, that’s….that points out the other side of the problem that we talked about at the beginning of this thing, and that’s that probably not all Alzheimer’s disease is the same. So here's a very specific genetic mutation in a specific family with a specific phenotype, we don’t know, or at least I don’t know as an epileptologist, how representative that is of the general Alzheimer’s population. I’m sure we could find a predictive biomarker in this family – of course we have a great one now, we can just do a genetic test and have pretty much 100% certainty in this particular family. We could find the physiological biomarker there that is of no relevance of anyone else with any other variant of Alzheimer’s disease. So there are a lot of controls and a lot of work that has to be done. But it seems like a golden opportunity to address the question you’ve asked, about EEG biomarkers, because you know so much in advance of when it’s going to happen.
Dr. Lam: Yeah, that’s a great example, Andy. I was going to say that when we talk about biomarkers, it’s worth saying a biomarker for what, right. Andy, you’re talking about a biomarker as a diagnostic biomarker for Alzheimer’s disease. When I think about these epileptiform abnormalities in Alzheimer’s disease, I agree with Andy that I don’t think it’s a universal phenomenon. I think we’re taking about a subgroup of patients that might have this type of hyperexcitable profile. I tend to think about the biomarkers in terms of prognostic biomarkers. So do these epileptiform abnormalities portend a more rapid clinical course, like a more rapid cognitive decline? And there’s some data that that might be the case.
Keith Vossel published a paper looking at these types of scalp EEG and MEG subclinical spikes in people with Alzheimer’s disease and showed that patients who have these spikes actually decline at a faster rate than those who don’t have the spikes. So that would be one interesting use, to be able to prognosticate who’s going to decline faster or not. I think that would be useful information for patients and families, but also for something like stratifying for clinical trials, where there’s so much disease heterogeneity.
And then also I’m thinking about targeted therapies for Alzheimer’s disease – we’ve talked about how it’s a heterogeneous disease. Whether treatment of hyperexcitability, whether hyperexcitability is a modifiable risk factor for cognitive decline in Alzheimer’s disease. Should we treat everybody with Alzheimer’s disease with seizure medication? That’s often a question we get. I don’t know, but I would guess that if not everybody has this hyperexcitability, you’re going to be treating a lot of people who aren’t going to benefit. So how do we select which people might be most likely to respond to that kind of treatment? I think EEG as a biomarker would be very good for that.
And actually Keith Vossel – there was a study he published, I think late last year, the LEV-AD trial, and this was a trial looking at people with Alzheimer’s disease being treated with levetiracetam, very low doses, like 120 mg bid.
One of the interesting things about this trial was before they treated anybody, they did 24-hour EEGs and one hour MEG on all the participants, and stratified participants based on whether or not they had this epileptiform activity. When they looked at the results overall, of all participants, it was a negative study – they didn’t see any effect of levetiracetam on the outcome of interest, which I think was an executive function measure.
But in their subgroup analyses, when they stratified by which patients had epileptiform activity and which didn’t, they actually did see an effect – only in the patients who had the epileptiform activity. So that’s you know, that’s sort of a promising indication that this kind of biomarker could be useful to identify a group of people with Alzheimer’s disease that could benefit from this kind of therapeutic approach.
There’s a lot of caveats of course. It was a very small trial, and it was trial done in a very early-onset Alzheimer’s disease population, so how that will translate into a more typical Alzheimer’s population isn’t clear yet, but I think the idea of using this tool to stratify participants is a really important one. There are a number of clinical trials ongoing right now that I think will be reporting in the next year or two looking at levetiracetam in the early stages of Alzheimer’s disease in people who don’t necessarily have seizures. It’ll be interesting to see, because I don’t think a lot of these studies use that kind of stratification – some of the studies just take all comers. So interpreting the results will be kind of interesting depending on whether they stratified or not.
Dr. Cole: That’s interesting. One question that with the LEV-AD thing is whether in the subgroup analysis there’s enough statistical power to really make a meaningful statement in the subgroups. It does point to an interesting set of questions. You know I gave a talk recently about EEG biomarkers, at a scientific meeting, and what they might be used for. And there’s been a long history in the psychiatric world of trying to use quantitative analysis of EEG to make diagnostic and predictive statements about psychiatric disease, which was largely thought by many people not to be terribly helpful. But with some of the machine learning things that Alice was describing earlier and different ways to look at large sets of EEG and pulling out statistically robust findings, that world may be changing.
There’s a very nice paper by Bob Kaliff, who’s now the head of the FDA, he was a cardiologist at Duke, on biomarkers specifically, from the FDA’s point of view. I think the most interesting thing about the paper is that he defines, two, three, four, five six, at least seven different types of biomarkers – from diagnostic markers to monitoring markers to prognostic markers to safety markers to susceptibility and risk markers.
So you can think about how EEG or other tools might fit in to a set of questions around an Alzheimer population or an epilepsy population and pick and choose, so one thing that’s kind of interesting if you think about the Colombian family that I talked about a moment ago – imagine if people went and did the proper EEGs and found that 20 years before they’re going to become symptomatic there are characteristic EEG changes in the affected versus the unaffected. And said, “Let’s give a medicine to see if we can prevent this disease from happening, or happen later.” Maybe an anti-seizure medication, for example. I’m not saying that should be the case. But you might be able to monitor that EEG over time as part of your dose finding approach. What dose would be the right dose to give? Well if this is the biomarker I’m seeing, maybe I should give a dose that’s large enough to suppress that biomarker, so can I reduce spike frequency by at least 50% with 200 mg of compound X and that will be our hurdle point.
So you can imagine a lot of ways of using these observations as biomarkers and we just discussed a small subset of them; there are many others.
I think there’s a future for EEG biomarkers, not in the old way but maybe in a new way.
Dr. Saletti: I want to thank both of you and congratulate both of you for the amazing work you’ve been doing in the field. Thank you for participating in our episode today.
Dr. Cole: thanks for having us – it was kind of fun to chat about this.
Dr. Lam: Thanks, Patricia.
Research mentioned in this episode
Noninvasive Detection of Hippocampal Epileptiform Activity on Scalp Electroencephalogram – Abou Jaoude M, et al. (JAMA Neurology 2022)
Towards a coherent view of network hyperexcitability in Alzheimer’s disease – Lam et al. (Brain 2022)
Neuronal synchrony abnormalities associated with subclinical epileptiform activity in early-onset Alzheimer’s disease – Ranasinghe KG, et al. (Brain 2022)
Effect of Levetiracetam on Cognition in Patients with Alzheimer Disease With and Without Epileptiform Activity: A Randomized Clinical Trial Vossel K et al. (JAMA Neurology 2021)
Association of epileptiform abnormalities and seizures in Alzheimer disease – Lam et al. (Neurology 2020)
New Approaches to Studying Silent Mesial Temporal Lobe Seizures in Alzheimer's Disease – Lam et al. (review, Front Neurol 2019)
Biomarker definitions and their applications – Califf RM (Exp Bio Med, 2018)
Silent hippocampal seizures and spikes identified by foramen ovale electrodes in Alzheimer's disease – Lam AD et al. (Nature Medicine 2017)
Neuronal activity regulates extracellular tau in vivo - Yamada K et al. (J Exp Med 2014)
Neuronal activity regulates the regional vulnerability to amyloid-β deposition – Bero AW et al. (Nat Neurosci 2011)
Subscribe to the ILAE Newsletter
To subscribe, please click on the button below.
Please send me information about ILAE activities and other
information of interest to the epilepsy community