Retinal vessel traits and age-related eye disease in the Canadian Longitudinal Study on Aging

Welcome everyone. I'm Sophie Horrobin, data access officer for the Canadian Longitudinal Study on Aging. Thank you for joining us for this CLSA webinar titled retinal vessel traits and age-related eye disease in the Canadian Longitudinal Study on Aging. Before we begin, I want to acknowledge that the CLSA National Coordinating Center and McMaster University are located on the traditional territories of the Mississaugas and Haudenosaunee nations and within the lands protected by the Dish With One Spoon Wampum Agreement. We pay respect to the Algonquin people who are the traditional guardians of this land. We acknowledge their long-standing relationship with the territory with this territory which which remains unceded. We pay respect to all indigenous people in this region from all nations across Canada who all who call Ottawa home. We acknowledge the traditional knowledge leaders, both young and old, and we acknowledge and honor their courageous leaders, past, present, and future. As attendees of this webinar, I encourage you to continue your learning following the webinar and to acknowledge the original inhabitants of the lands where we currently have the privilege to research, live, and work, wherever that may be. Before we begin, we'll review a couple of housekeeping points. Everyone but the presenters will be muted throughout the webinar. If you need to change or test your audio during the webinar, you can click audio settings on the bottom of the Zoom window. At the end of today's presentation, there'll be a question and answer session. If you have questions for the presenter during the webinar, please post them in the Q&A box located in the bottom toolbar. So, this is different from the chat box. Uh there's specifically a Q&A box and we'll address these questions at the end of the webinar. Your questions will be visible to all attendees. If you have any technical trouble concerning the webinar, please use the chat box, so not the Q&A box, uh to communicate with our webinar team. A feedback survey will be launched at the end of the webinar and we invite you to complete it after exiting the Zoom session. The brief survey provides with important feedback we can use to plan future CLSA webinars. Okay, today's webinar is titled Retinal Vessel Traits and Age-Related Eye Disease in the Canadian Longitudinal Study on Aging. The webinar will be presented by Alexis O'Neill, who holds a Master's in Epidemiology from the School of Epidemiology and Public Health from the University of Ottawa. Alexis O'Neill completed her degree under the supervision of Dr. Ellen Freeman with thesis research focused on retinal vessel characteristics and their relationships to various health and lifestyle factors, as well as ocular and systemic diseases. With a clinical background in ophthalmology, she's interested in advancing techniques for early disease detection and improving patient outcomes to help prevent vision loss. Alexis will begin her PhD in Vision Science this fall at the University of Waterloo. All right, over to you, Alexis. >> Perfect. Thank you so much for that introduction. I'm just going to share my screen here and we'll get started. All right. Okay, so I think everyone should be able to see my screen here now. >> Looks good. >> Perfect. Thank you so much, Sophie. Um so yes, like Sophie said, um my name is Alexis. I just finished up my master's in epidemiology and um I'm very excited today to be here to present uh some of our research from my thesis um based on retinal vessel traits in the age-related eye disease in the Canadian Longitudinal Study on Aging. So to get started, I wanted to give a little bit of background on the retina and the retinal vessels. So here we have an image of the retina, um which is the light-sensitive tissue right at the back of the eye. And the retina is able to convert light around us into signals that the brain can interpret, and that is how we see. And the retina is one of the most met- metabolically active tissues in the body, so it relies very heavily on a constant supply of oxygen and um nutrients from from the blood. And that blood supply is provided in part by the retinal blood vessels. So the retinal blood vessels, they enter at the back of the eye, right where the optic nerve is, and um uh right with the central retinal artery. And then that branches off into smaller and smaller little blood vessels called the arterioles, and then from there that enters into the capillaries, where that nutrient and oxygen exchange happens with the retinal cells. And then the blood travels back through the very, very small veins called venules, and exits back through the back of the eye through the central retinal vein. So when we're talking about the blood vessels at the back of the eye, they are classified as arterioles and venules because they're very, very small. They have a very small diameter. There's no autonomic nerve supply, so they're controlled by the local mechanisms in the back of the eye um alone rather than any other external factors. And And different characteristics that we examined when we looked at the vessels were diameter, so the width of the blood vessels, and also the tortuosity, um, which is a geometrical measurement of the blood vessels, like curvature or the, um, the twistiness of the blood vessels. Um, these two images are directly from the Canadian Longitudinal Study on Aging. So, you can see, um, the image here, which is from the 99th percentile of arteriolar diameter, and these blood vessels are much wider than the blood vessels on the other side here, where you can see very, very narrow blood vessels from the first percentile of arteriolar diameter. And these different variations of the retinal vessel diameter can be impacted by various things, such as aging, genetics, high blood pressure, heart disease, inflammation, and also endothelial dysfunction. Very similarly, we have two images here from, uh, again, the CLSA, showing different levels of tortuosity. So, here we have very tortuous, very curvy, twisty, uh, vessels here in the 99th percentile of arteriolar tortuosity. And then, over here, we have much straighter blood vessels, where they're almost straight lines. Um, and that's from the first percentile of arteriolar tortuosity. So, again, this tortuosity, um, variation can occur due to differences in hypoxia, aging, high blood pressure, endothelial dysfunction, and heart disease, as well. Moving on to our age-related eye diseases, we wanted to look at glaucoma and macular degeneration, as these are two of the leading causes of visual impairment and blindness in the world. Uh, starting with glaucoma, this is a progressive degeneration of the optic nerve, and this is characterized by thinning of different layers of the retina, including the retinal nerve fiber layer and the retinal ganglion cell layer, and also cupping of the optic disc. And this can be measured by the cup to disc ratio, where a larger cup to disc ratio is indicative of glaucomatous changes. And glaucoma can, as it progresses, it can cause these very characteristic changes in the visual field. So, here we have visual field results from someone with glaucoma, and you can see this area of darkness represents an area of blindness where this person cannot see. And as glaucoma progresses, this area can extend all the way into the periphery of the of the visual field or to the point where the person is no longer to see in any area of their visual field. So, complete blindness if it's able to progress. And high intraocular pressure is one of the biggest risk factors for developing glaucoma, and it's one of the only modifiable factors right now that can be targeted by treatment. Moving forward to age-related macular degeneration or AMD, AMD is a progressive degeneration of different layers of the of the retina right in the central area of our vision. And so, as macular degeneration progresses into more severe forms like choroidal neovascularization and geographic atrophy, you have these changes in the vision. So, for example, someone with normal vision would be able to see the photo very very clearly. Someone with mild macular degeneration you start to get these like blurry areas of blurriness or distortion kind of right in the central area of the vision. And then someone with very severe macular degeneration that has progressed further may not be able to see anything in the central area of their vision. So, yeah, that's macular degeneration. And how are these age-related eye diseases and retinal vessel characteristics related? So, there has been evidence to suggest that both of them have been associated with disruptions of the retinal vasculature. Glaucoma has generally been associated with arterial narrowing and decreased tortuosity or curvature of the blood vessels. And then, um, associations with mac- macular degeneration and retinal vessel diameter are either generally absent or can be very inconsistent in the literature. Uh, so, when it came to the knowledge gap that we wanted to address with our study, uh, was that a lot of the population-based studies that have looked at the relationships between eye disease and retinal vessel characteristics, most of these have been cross-sectional. And so, that really limits our understanding of the temporality between the retinal vessel changes and the development of eye disease. So, it's not super clear what is coming first, um, when we are looking at cross-sectional data because both your uh, disease and your retinal vessels are measured at the same time period. Um, so, it's hard to to figure out what's coming first. So, when we look at glaucoma, for example, there's two kind of main pathways, um, that might be happening when we see these cross-sectional associations. So, the retinal ganglion cell loss, um, so, that's cell loss that is characteristic of glaucoma might be happening because of, uh, various, uh, changes in the retinal blood vessels, uh, that is causing the disease. Or, in the opposite direction, less blood flow is required to satisfy the metabolic needs of fewer cells. So, the glaucoma might be causing these retinal vascular abnormalities. So, therefore, more longitudinal studies are really needed to help disentangle these pathways and determine which changes are actually happening first. So, with that in mind, we had two main objectives, one for glaucoma and one for macular degeneration. So, our first objective was that we wanted to look both cross-sectionally and longitudinally to investigate the relationship between retinal vessel characteristics and various glaucoma outcomes. So, the intraocular pressure that we talked about, the cup-to-disc ratio that we talked about, and also glaucoma itself, its prevalence and the 3-year incidence. And in the second objective, we wanted to do the same thing with macular degeneration. So, looking at the relationship between retinal vessel characteristics and the prevalence of macular degeneration at baseline and its 3-year incidence. Based on these um Based on these objectives, we came up with a conceptual framework where our main projectors that we wanted to look at were the retinal vessel characteristics. So, the arteriolar and the venular diameter and tortuosity. And our main outcomes were glaucoma, intraocular pressure, cup-to-disc ratio, and macular degeneration. And then we wanted to address um a couple of different confounders including some socio-demographic factors, some health factors, and lifestyle factors as well. And then we additionally adjusted for a couple of other things which we'll get into a little bit later. In terms of the data that we used, we used the baseline and the 3-year follow-up data from the Canadian Longitudinal Study on Aging Comprehensive Cohort. Um so, we used demographic, health, and lifestyle data that were collected via either a physical assessment at a data collection site or via interview questionnaire. Uh the diagnosis of glaucoma or macular degeneration were collected by self-report from uh, participants and intraocular pressure was also measured during these data collection site visits. For the retinal images, these images were taken uh, using a fundus camera and we sent those images to our collaborators in the UK, so Sarah Barman's team, who used their um, deep learning software called Quartz, which stands for quantitative analysis of retinal vessel topology and size, but I'll just call it Quartz, uh, to extract data including the arteriolar and venular diameter in pixels and also the vessel tortuosity and also the cup-to-disc ratio from each from each image. And then from there, um, from each image we took the mean uh, the average diameter and tortuosity of each arterial and venule and that was summarized and weighted by the vessel segment length for each image. I did want to talk a little bit more about some of the key features about Quartz because I think it's an awesome software and really really interesting to kind of see how exactly it it works behind the scenes. So, Quartz is a fully automated deep learning algorithm that is able to process each retinal vessel or each retinal image in about two uh, two minutes. It's able to remove images of inadequate quality, so it gives like basically a grade to each image and then it disregards um, certain images if they don't meet that grade. Um, so but it is also able to minimize the number of wasted images. So, for example, if there's a partial image where there's a blink in the way or there's an area of distortion in the image, um it's able to analyze the retinal vessels that it's able to actually see, um rather than just disregarding the whole image, which I think is very a very cool little feature. Um it is able to differentiate between arterioles and venules with a high degree of accuracy, so that is what this picture down below shows. You can see the arterioles in red and then the venules in blue. Um and it measures the vessel characteristics over the entire image. Whereas traditional methods, they might only be able to uh look at these factors within a certain diameter around the optic disc. Um we had a more of a global view where we were able to see vessels through the entirety of the image. And the software was trained and validated on multiple data sets including the UK Biobank, the CLSA, um and a couple others as well. And I also wanted to talk about how exactly these factors get measured using QuARTS. So, for the vessel diameter, it uses a special technique called a zero crossings of the second derivative, which is a fancy way to say that um it's able to detect the edges of the retinal vessel just based on changes in the brightness in the image. So, it translate that translate that into a graph and then it takes the first derivative and then the second derivative and then wherever these like little zero um points are, that correlates with the edges of the of the retinal vessel. So, from there it's able to draw a center line through the retinal vessel, which you can see in image C here. And then from there it draws very, very small little perpendicular um little segments um across the the center line. And then from one side of that uh uh perpendicular measurement to the other, that is where the width comes from. Um so, yeah, that is how the diameter of the blood vessel is measured using quarts. Similarly, uh tortuosity is uh measured using a subdivided chord length method, um which again is a fancy way to just say that the actual length of the blood vessel, so all this twistiness here in the actual length, is compared to a straight line that is drawn across a similar path. And then, um that means that a higher tortuosity is uh a higher tortuosity value just means that the blood vessel is more curved or twisted. Um and so, yeah, that is kind of how quarts works. So, from there, we went into our data analysis. So, we assessed the normality of our continuous outcomes, or vessel characteristics, um our intraocular pressure, cup-to-disc ratio, um to make sure that we were um uh just making sure that the assumptions of our models were were all uh correct. We completed a complete case analysis using right eye data only, and we made sure that our models were adjusted for the complex survey design of the CLSA. So, we made sure to account for the in the strata um of the CLSA. We used logistic regression to determine the association between each retinal vessel characteristic and glaucoma, macular degeneration, and also their development over 3 years. And we used linear regression for a continuous outcome, so our intraocular pressure, cup-to-disc ratio, and their 3-year change. And we made sure to adjust each model for any potential confounders, so those sociodemographic, lifestyle, and health factors that we talked about, and also um additionally adjusted for um a couple of different types of medications. So, we adjusted for um systemic steroid use, um systemic antihypertensive therapy, so high blood pressure medication, and also ocular antihypertensive therapy as well. So, eye drops to reduce um ocular intraocular pressure. A couple of special considerations that we wanted to highlight as well were that each model included both the arteriolar and the venular trait because there is some evidence to suggest that they confound each other. So, if someone has narrower arterioles, for example, they're more likely to have narrower venules. So, we wanted to make sure that we were including both of those vessels in our models to see the relationship um between each one individually with the uh with the eye diseases. We also made sure that our models that were um considering the 3-year change outcomes, so the intraocular pressure and the cup-to-disc ratio, were adjusted for the baseline level of those factors to account for the floor and ceiling effects. So, this just means, for example, that someone with really high intraocular pressure at baseline, they have less potential for that intraocular pressure to increase over 3 years versus someone who had much lower intraocular pressure at baseline. They have more potential for that intraocular pressure to increase over 3 years. So, we wanted to account for those factors um within our models. Uh moving on to some of our results. So, um we saw that 87% of CLSA participants had acceptable images in at least one eye. So, we were able to analyze data from 87% of the CLSA. Those with unacceptable image scores were a little bit older, and they were also more likely to be diabetic. And the values for the vessel traits between the right and the left eyes were very highly correlated. So, we felt pretty confident moving forward in our analysis with the just the right eye data. These are some of the key characteristics of those with glaucoma and macular degeneration. So, we had out of those people who had adequate image quality, 4.5% reported a diagnosis of glaucoma. They were older, they were more likely to be female, had higher blood pressure, and were more likely to drink alcohol every day, and have diabetes. They also had higher intraocular pressure and worse cup-to-disc ratio. Which is what we would expect of people who reported a diagnosis of glaucoma. Generally, these people also had wider arterioles and wider venules as well. For macular degeneration, those who had acceptable image quality scores and also a diagnosis of macular degeneration, that was about 3.8%. And they had very similar characteristics to those who had glaucoma, and they generally had wider venular diameter and reduced venular tortuosity as well. These are some of our cross-sectional results between our retinal vessel traits and glaucoma outcomes. So, all of our statistically significant associations are shown in bold in the table here. And so, we saw that those with wider arterioles had a lower odds of having glaucoma at baseline versus venular diameter. We saw that those who had wider venules at baseline had a higher odds of having glaucoma at baseline. When it came to cup to disc ratio, we saw that arteriolar diameter, venular diameter, and venular tortuosity were all inversely associated with cup to disc ratio. And with intraocular pressure, we saw something a little similar where venular diameter, arteriolar tortuosity, and venular tortuosity were all inversely associated with intraocular pressure. So, to give you an example of what this actually means and what these numbers can be interpreted as, we can say, for example, with our venular diameter, we can say that for every 10 pixel increase in venular diameter, intraocular pressure decreased by about 1.93 mm of mercury, which is the measurement for IOP. So, yeah, that is how we would interpret that. These are the results from our supplemental analysis, and I'm just going to point out just the one kind of main thing that we difference that we saw. In model 1A, we adjusted for all of our typical socio-demographic, health, and lifestyle factors. And then in our second model here, we adjusted for those extra types of medication. And when we did that, we saw this cross-sectional relationship between arteriolar diameter and glaucoma disappeared. So, this no longer existed after our additional adjustments. So, that indicates that there is some confounding relationship between those extra medications and the arteriolar diameter in glaucoma. This table is for our longitudinal associations between retinal vessel traits and glaucoma outcomes. So, a couple things to highlight here as well were that we did not see any longitudinal association, statistically significant longitudinal association between arteriolar diameter and the development of glaucoma over 3 years. This was not statistically significant. However, we did see that as venular tortuosity at baseline increased, there was a lower odds of developing glaucoma over 3 years. So, we thought that that was interesting and I wanted to point that one out. Very quickly, I just kind of wanted to go over those differences between our cross-sectional and our longitudinal results. So, first, we saw in our cross-sectional analysis at just baseline that narrower retinal vessel diameter was associated with having glaucoma. And that was statistically significant. So, as previously mentioned, when we're just looking at cross-sectional data, we can't determine the directionality of this relationship. We can't determine what is actually coming first um since both of these things were measured at the same time. So, to help address this, we really wanted to whoops. We wanted to look at our longitudinal associations as well. So, we looked at the uh baseline retinal vessel diameter in relation to developing glaucoma over 3 years. This was not statistically significant. And then we wanted to look at the reverse longitudinal association. So, we looked to see if having glaucoma at baseline was associated with a change in the retinal vessel diameter. And we saw that this relationship was statistically significant. So, together, all of this may suggest that at least part of this cross-sectional association that we're seeing both in our results and results in the literature, may be explained in part by reverse causality, where glaucoma, or maybe its treatment, contributes to the retinal vessel narrowing, rather than the retinal vessel narrowing preceding the development of glaucoma. And lastly, these are the results from our macular degeneration analysis. So, we combined both the cross-sectional and the longitudinal results here. Again, statistically significant results are shown in bold. We saw that venular diameter at baseline was associated with having a higher odds of having macular degeneration both at baseline and developing macular degeneration over 3 years. And then, for venular tortuosity, we saw that having higher venular tortuosity at baseline was associated with lower odds of macular degeneration at baseline. As we went through some of our results, you may have noticed that the vessel diameter was all measured in pixels, and that comes directly from Quartz hour, the software that was used. So, pixels are really useful for image analysis, but they're not super clinically relevant because the exact pixel size can vary depending on the camera system that you use, the image resolution, and different scaling factors as well with different images. So, we wanted to translate these findings into units that were a little bit more clinically meaningful. Usually in the literature, this is done by um for the axial length of the eye, so the length uh uh from the cornea back to the retina. Um we did not have this data, so we just wanted to give just a general estimate. So, to estimate this conversion, we used uh the average vertical optic disc diameter, which was about 418 pixels. And then in a similar population, the average vertical optic disc diameter was around 1,820 microns. So, when we use this ratio, we can estimate that one pixel equals approximately 4.35 microns. When we link this back to our results, we can say that for every 10 pixel or approximately 43.5 micron increase in venular diameter, the odds of having macular degeneration at baseline were about 2.77 times higher. Uh moving forward into our discussion, I just wanted to highlight a couple of our main findings and how they relate to some of the existing literature. So, we'll start with glaucoma, our first objective. So, our main finding here was that near where arterioles were only cross-sectionally associated with glaucoma. And additional adjustment for um intraocular pressure-lowering medication, so eye drops, eliminated this association. And uh near where arterioles were not associated with the 3-year development of glaucoma, but when we looked at the reverse longitudinal association, this was statistically significant. And so, this suggests that uh reverse causality may be one of the reasons for why we see this cross-sectional association both in our own results and in the literature a lot as well. When it comes to previous longitudinal studies, there um are few. There's two, and only one of them has found this um Only one of them has found a statistically significant association between retinal vessel diameter um and the development of glaucoma. Um again, for glaucoma, we also saw that more tortuous or curvy venules were associated with a lower odds of developing glaucoma. And this is pretty consistent with previous findings. Um previous cross-sectional findings show that straighter vessels were associated with glaucoma and also a thinner neuroretinal rim. And so, this may suggest that straighter venules may precede the development of glaucoma. And there isn't a lot of research into tortuosity and eye disease in general, so we were um excited to see some consistency in our results here. Moving on to macular degeneration, uh the main finding that we saw here was that venular diameter was positively associated with both baseline macular degeneration and the its 3-year development. And this is not very consistent with previous longitudinal studies. Uh none of the studies on the screen there uh found a statistically significant association. Uh so, that is something different uh between the literature and our own study. Um however, results are consistent with a couple of studies that has showed that wider venular diameter was associated with early macular degeneration. Um the 10-year development of retinal pigment epithelial abnormality, so these are very early signs of macular degeneration. And also worse response to treatment in patients with neovascular macular degeneration. So, some consistency as well as some inconsistencies as well. And lastly, um our finding for macular degeneration, uh we saw that increased venular tortuosity was associated with lower odds of macular degeneration at baseline. And we're not aware of any prior studies that looked at this association. Um like I said, tortuosity is kind of one of those newer things that um are being studied. So, uh we did not see um any prior previous studies that have looked at this association. So, potentially a novel association there. So, I talked a lot about some discrepancies between our results and some of those in the literature, and there are a couple different reasons for why that might be happening. Um so, there's some differences in which confounding factors are accounted for. So, there are a lot of different things that can be associated with retinal vessel characteristics. Just like there's a lot of different things could that can be um associated with eye disease. Uh and there can be a lot of differences on what data is actually collected by different population-based studies. So, um based on that, there is a lot of differences between what we may have adjusted for versus what other groups have adjusted for. And that might be one of the reasons why we have these discrepancies between our results and others. Uh there's also differences in the types of uh fundus image analysis softwares that are used. So, we use Quartz, whereas other people use other um other image analysis software. Generally, there's already very poor agreement between uh software because there's a lot of differences in how the vessels are identified, how they're segmented, and how the characteristics are actually measured. And lastly, um different confounding effects from differences, um unmeasured confounding factors, ethnic factors like diet or genetic factors that that can also lead to some of the discrepancies in the literature as well. Overall, our findings have several important clinical implications moving forward and also some opportunities for future research. Those mainly being screening, prediction, and treatment. So, in terms of screening, fundus images are very quick to take. They're relatively inexpensive. They're non-invasive. If retinal vessel measurements can help to distinguish patients with macular degeneration with glaucoma, um there's potential that these patients can be flagged earlier. And if we can catch these cases of disease sooner, it might be possible to treat them sooner, manage their diseases sooner, and hopefully to prevent vision loss from these from occurring. Very similarly, there is potential for retinal vessel characteristics to be used in prediction. So, if we can predict people who will develop glaucoma or macular degeneration, these people can be like monitored a little bit more closely and again diagnosed and treated sooner. And then if we can potentially use these retinal vessel characteristics to improve prediction modeling, that would be an important next step. To really see if we can use retinal vessel characteristics for prediction as well. And finally, with the treatment. So, if retinal vascular changes can play a role in disease development or progression, they might also represent a therapeutic target. So, rather than you know, treating uh glaucoma as a whole, potentially targeting the retinal blood vessel characteristics, um that might be a potential area of uh future research. Um and then also, if this is true, if the retinal vessels can be used um as a marker, potentially be used as a marker of treatment efficacy as well, to really help clinicians monitor whether therapies are having uh meaningful uh effects over time. So, altogether, hopefully these areas can be um explored further in research, and our results can be used to uh as like a little springboard into some of these potential areas. So, our research has several strengths. Um we leveraged data from the CLSA. This is a large population-based sample. We had access to longitudinal data with a very high retention rate, which is fantastic for a longitudinal data set. Um and through the CLSA, we had access to comprehensive physical assessment and questionnaire data as well. We also use Quartz um as our uh image software analysis software, which provides data over the entirety of the retina. And um this differs from traditional methods that may only be able to focus on a specific area around the optic nerve, or may only rely on certain blood vessels. Um we were also able to use two different retinal vessel characteristics. We looked at diameter and tortuosity, rather than just diameter, to get a better understanding of how these retinal vessel characteristics are really related to eye disease. And also, prior research has focused on arteriolar diameter only. So, like we I had mentioned at the beginning of this talk, uh um we wanted to make sure that we were assessing both arterial and venular characteristics independent independently. So, we made sure to adjust for both vessels to see their associations independent from each other. So, while there was a ton of really great things about our our research, there are also a couple limitations. So, some participants were excluded due to inadequate image quality. And those individuals were older and they were more likely to be sicker. So, that introduces some selection bias into our results. We also did not have information on axial length or refractive error. So, like we had kind of discussed, we were not able to directly convert our pixel measurements into more a more clinically relevant measurement like microns. We also relied very heavily on self-reported diagnoses of ocular disease and generally ocular disease is underreported when um when we're relying on these self-reported diagnoses. Um reports of glaucoma and macular degeneration were also not assessed at the eye level. So, what we mean by that is that they were assessed at the person level, not the eye level. So, we glaucoma and macular degeneration, they can occur unilaterally. So, in one eye or the other or they can occur asymmetrically between the two eyes. So, when we move forward and we use right eye data only, we're not certain that we were actually capturing all these cases of glaucoma um within our data and in our analysis. So, um potential for introduction of misclassification bias in our results there. And we also did not have data on the type or the severity of glaucoma and macular degeneration. So, we were not able to run any subgroup analysis based on these factors. So, our results may not be generalizable to every type of glaucoma or every type of macular degeneration. Despite all of this, we are very happy with our research. We're very proud of it. Um we were the first to investigate retinal vessel characteristics in relation to age-related eye disease in a Canadian population. We were able to leverage underutilized retinal images from the CLSA alongside its longitudinal data. And um we hope that this research can be used to advance the understanding of the vascular mechanisms underlying glaucoma and macular degeneration. And hopefully moving forward, these results can be used to translate um into uh clinical applications. As we wrap up here, I just wanted um acknowledge a couple different people. Uh my supervisor, Dr. Freeman, all of our co-authors, the whole team at the CLSA has been fantastic. Um and also the primary investigators of the CLSA. Um this research wouldn't have been possible without any of these people. We were also funded by the CIHR. So, thank you so much, CIHR. Um and if anyone is interested in reading more, we do have a published open-access paper um in clinical and experimental ophthalmology, as well. So, that is accessible if you would like to take a look and read a little bit more. These are a couple references from throughout the presentation. And that's about it. Thank you so much for everyone for coming and listening. Thank you so much for to the CLSA for inviting us to share our work. Um and I'd be happy to take any questions now. >> Great. Thank you so much, Alexis. I know I've reviewed papers from your group before and it all makes so much more sense that I've heard you actually speak about it. >> Perfect. >> yeah, really appreciate that and we're going to open it up to questions now. Um so just a reminder to everybody muting will remain on, but you can enter your questions in the Q&A box at the bottom of the Zoom window. So we'll just wait for some of those to come in. Um so I was curious though. So our is that the the vessel tortuosity? I guess that's something then that changes over time? Is that >> Um >> what you're suggesting? >> Vessel tortuosity there can be like minor changes over time. In our first paper that we had published, we saw that there were small changes depending on um certain like lifestyle and health uh factors. Um but the main thing that we were looking at in this paper were changes in the intraocular pressure and the cup-to-disc ratio. Um cup-to-disc ratio is kind of that measure of if it's if the if there's a greater cup-to-disc ratio, it's more indicative that there might be glaucoma-like changes. And then the intraocular pressure, that's a very modifiable factor when it comes to glaucoma as well. >> Okay. >> So we weren't looking too much at how the retinal vessels were changing necessarily in terms of tortuosity. Diameter is something that's a little bit easier to was easier to assess, but tortuosity the they're very minuscule changes and we saw that in our first paper over 3 years. Yeah. >> Okay, great. Thank you. Um so we've had a question come in. Um thanks for a great presentation. When a subject has a repeated retinal images, does the software allow for analysis of the paired images within an individual or just treat some as independent? >> I think they treat them as independent. Um yeah, I don't think as of right now they are able to um like correlate the images by the participant. Um I think that step can come from the actual data analysis where we can look at those change values over time of those retinal vessel traits. I think that's where that kind of time aspect comes into it, but as of right now the software itself does not correlate the images. They just treat them as separate images. >> Right. Okay, thanks. Um next question, are vessel diameter and tortuosity related? >> That's a good question. I don't know exactly. I know that there are some things where if you have a wider blood vessel, it doesn't have as much ability to have more like a higher degree of curvature. So if you have a wider blood vessel, it doesn't curve quite as much in general, but I don't think there's a ton we we didn't look at the the relationship between the the two different characteristics, so I don't know the exact relationship between them, but um they are uh associated with very similar factors. Um so when we talked about the determinants of the the diameter and the tortuosity, they are affected very similarly by the different health and lifestyle factors. Um so it's likely that there might be an association between the two different factors. We just didn't look at that specifically. >> Thank you. Um you showed that venular diameter seemed to be more strongly associated with eye diseases rather than arteriolar changes, especially after adjustments. Do you have a mechanistic hypothesis for this? What factors could change venular diameter? >> Yeah, venular diameter um this is a really great question. Um venular diameter is associated or maybe I should say wider venular diameter is highly associated with um different kind of metabolic disease and inflammation and smoking and those kind of kinds of factors which can also be associated with eye disease as well. So what these exact pathways look like, we're not too sure, but it's likely that these kind of either metabolic or inflammatory pathways um might be involved um somewhere along along those lines. So yes, venular diameter um has been consistently shown to be wider in people with these kind of inflammatory metabolic uh conditions, so um certainly that might play a role somewhere along that pathway. >> Okay, and then it's so generally more the venular than arteriolar. >> Yeah, I I wish that I had a good answer for the arteriolar pathway, um but with the the arteriolar narrowing, it can be associated with like high blood pressure, blood pressure medication, kind of these more like heart disease related things. So, these changes in the like in terms of narrowing in the arterioles, um especially when it came to some of that cross-sectional data that we were looking at, there's it might be possible that those like cardiac factors might be related somewhere along those path that pathway. >> Okay, thank you. Is quartz freely available? >> That's a good question. I'm not certain if quartz is freely available. I think if you are interested, we can give you can always give Dr. Freeman a shout. Her email is on the screen there. And inquire whether or not it could be available and she can always connect you with the UK team potentially as well. But yeah, I'm not I'm not certain about that. >> All right. Um congratulations. Is there a physiological reason or explanation for a person uh that a person would have more or less microvessel tortuosity? >> The in terms of tortuosity, there's a couple of things that are possible. Um as tortuosity increases, the amount of blood flow kind of slows down. So, it's thought that an increase in tortuosity might be in response to like a reduced amount of oxygen or or blood supply. So, by increasing the tortuosity, it gives the blood a little bit more time to release the oxygen to those tissues. So, it slows the blood down a little bit. However, with an increase in tortuosity, you also have more risk of issues like stroke. So, it can really go in either way, whether or not this tortuosity is a good thing or a bad thing. Um but yeah. >> And in terms of where that comes from, like is that something you're born with or like genetically predisposed to one or the other? >> Um genetics, hypoxic effects, uh like heart disease, um all those kind of determinants that we talked about. Um yeah. >> Okay. Um AMD is not always a vascular disease. Can you comment on the limitations of self-reported AMD diagnosis and if non-neovascular versus neovascular was asked? >> Um we didn't have that those subgroups. Um so, we just had a general diagnosis of macular degeneration. We know that macular degeneration is typically underreported um if it is based on a like self-report of a diagnosis. So, we know that. Um in terms of macular degeneration, the mechanisms behind macular degeneration um are not super well understood. One of the theories includes this kind of vascular mechanism and pathway and that has a lot of different um kind of ideas behind it. Um so, you're correct in saying that there are those two different types, but uh that is certainly one of the limitations of our work where we were not able to really look at those two different subgroups. Um and is something that moving forward can be uh a potential like future future area of research to look at the neovascular macular macular degeneration versus the geographic atrophy wet versus dry. >> Right, thanks. So next question, in your images you showed that tortuosity seems to change towards the optic nerve or fundus. Is it a spread feature or always near the nerve? Did you assess it in different fields of view? >> We did not, but I do know that our our collaborators in the UK are looking at the different areas um in the retina. So the very very very small vessels that are further away from the optic nerve, they're looking to see if there are differences in those associations between the very very small vessels versus the larger blood vessels that are closer to the optic nerve. Um in terms of differences in tortuosity between closer to the optic nerve and further away, I'm not too sure. But um again, a really great question and a really great future area of research that I believe is being worked on. Um our collaborators in the UK are on it. They are going to see if there's any differences between those smaller blood vessels further away, for sure. >> Thanks. I think the next question is related, but I'll read it out so um you can determine. Um so very nice presentation. Has there been any effort to look at vessel diameter diameter and tortuosity in a region-specific manner rather than average throughout the retina? For a disease like AMD, one could imagine that vascular changes in the parafoveal region would be more informative than those occurring more peripherally in the retina. >> Uh yes. So um very similar to what I had just mentioned, there is efforts to look at the retinal vessels that are further away from the from the optic nerve. There's also a lot of OCT data, so optical coherence tomography data where they look at the different layers of the retina, and also that like foveal avascular zone to see if there's any differences between like those kinds of factors with the blood vessels and eye diseases. So, there are efforts moving towards that as well. We just simply focused on the like fundus images, but yes, OCT data and all of those like extra factors are being looked at zone-wise. I think it's a it's a up-and-coming, maybe not up-and-coming, but coming. >> Great. Thanks. So, we're just about to wrap up. As the last question, what's next for you in this area of research that you've been pursuing? >> I my my supervisor is currently we're we're publishing another paper based on the retinal vessel characteristics in stroke. So, that is hopefully coming out soon. That's in review. So, that is one of the next things that is coming. She is also continuing to work with the retinal vessel data because it's such an awesome data set to be able to work with. So, there's definitely some things coming out in that sense. Me myself personally moving forward, I'll be doing my PhD at the University of Waterloo, and that research will focus still on retinal imaging, but a little bit different. It'll be on adaptive optics, so a different form of retinal imaging where we can see very very small areas or cells in the retina. Um Um, and so, yeah, that's where I'll be moving on personally, but the retinal vessel data is not going anywhere. It's still going to be in use and my supervisor is is a very happy and proud to be able to use that. >> Great. Well, thanks so much, Alexis. Thank you for participating in our webinar series. Um, and I'd like to remind everyone else or everyone in general that the next deadline for data access applications is July 8th, 2026. So, please visit the data access section on our website to review available data as well as additional details about that process. Um, and please um, just a friendly reminder to keep us updated on any changes that affect your data access agreements. So, whether you've moved or your team has changed, um, please reach us at access@clsa-elcv.ca with any questions or updates. And again, a reminder to complete the anonymous survey when you exit the Zoom session. Uh, so, thanks again for attending and participating in our webinar. Our next webinar, Health Disparities Among Lesbian, Gay, and Bisexual People in the Canadian Longitudinal Study on Aging, Methodological Considerations for CLSA Data Users, will be presented on Thursday, June 18th at 12:00 noon Eastern by Nicole G. Hammond, a postdoctoral research fellow at the community call in that in the College of Community and Global Health from the University of Manitoba. If you or a colleague is interested in presenting a CLSA webinar, please reach out to our webinar team. Um, the and the recording for this webinar and the slides will be available on our website in the coming days. Thanks again for joining us. Have a great day. >> Mhm.