Longitudinal investigation of spatial memory and retinal parameters in a 5xFAD model of Alzheimer’s disease reveals differences dependent on genotype and sex

15 Significance: The retinal phenotype of Alzheimer’s disease (AD) is poorly understood. The connection between 16 spatial memory and retinal phenotype is poorly investigated. Additionally, the influence of sex on the disease in mouse 17 models is not sufficiently clear and requires further investigation. 18 Aim: To investigate the retina and spatial memory of 5xFAD mouse models of AD by measuring retinal and 19 behavioral parameters. 20 Approach: A custom-built optical coherence tomography (OCT) system is used to image the retina of both eyes of 21 32 transgenic 5xFAD mice and 32 non -transgenic littermates over the course of 6 months (3 -9 months of age) to 22 investigate retinal parameters. The Morris Water Maze (MWM) test was performed to examine correlations between 23 the retinal and spatial memory phenotype of the mouse model. 24

Data were acquired in the form of OCT reflectivity images and OCT angiograms as well as video recordings 25 of the MWM test. Layer thickness and vascular density w ere calculated from the resulting data. Behavioral data was 26 extracted from the videos acquired from the MWM. Total retinal and inner retinal layer thickness increased slightly 27 over the measurement period, while outer retinal layer and retinal nerve fiber layer thickness showed no significant 28 change. The correlation analysis between MWM and layer thickness data revelated a positive correlation between 29 inner nuclear layer thickness and MWM test day parameters. 30

OCT and MWM data revealed sex -based differences in the retinal phenotype of the 5xFAD mouse 31 model, with changes in retinal thickness in different stages of the study and dissimilar correlations between retinal and 32 spatial memory phenotype. 33 34

Optical Coherence Tomography, Retinal Imaging, Alzheimer’s Disease, 5xFAD Mouse Model, Spatial 35 Memory Testing, Morris Water Maze 36 37 *Georg Ladurner, E-mail: [email protected] 38 1 Introduction 39 Alzheimer’s disease (AD) is the most common form of dementia and represents a huge challenge 40 for modern health care systems in an increasingly aging population. 1 AD-related lesions in the 41 brain include the appearance of amyloid beta plaques, neurofibrillary tangles as well as loss of 42 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 2 neurons and synapses on a cellular level. 2 On a behavioral level, clinical signs of AD include 43 memory impairment, irritability, orientational troubles and in later stages also difficulties with 44 basic body functions. 3 Diagnosis of the disease is at this point not fully possible, a definitive 45 diagnosis can only be achieved by postmortem neuropathology.4 Promising diagnostic approaches 46 include magnetic resonance imaging, positron emission tomography and cerebrospinal fluid 47 assays, in all cases in combination with neurological tests. 5 Blood tests have recently emerged as 48 a new alternative for the detection of AD biomarkers such as phosphorylated tau protein, amyloid 49 beta or neurofilaments , a lthough distinction between AD and non - related dementias can be 50 challenging due to similar biomarkers. 6 Treatment options for AD are still limited, due to a lack 51 of methods to stop or reverse the disease progression. 7 New compounds for AD treatment like 52 donanemab and lecanemab recently received FDA approval, although their efficiency and safety 53 have been disputed.8 The National Institute for Health and Care Excellence (NICE) even does 54 reject the use of donanemab due to significant health risk associated with treatment and high costs.9 55 Due to a common embryological origin, the retina and the brain share similar functionalities as 56 well as disease manifestations.10 For many neurodegenerative diseases, as for example Parkinson’s 57 disease11 or amyotrophic lateral sclerosis (ALS),12 retinal pathologies in parallel to lesions in the 58 brain have been reported. In the case of AD, markers such as inflammation, neurodegeneration as 59 well as amyloid beta deposits and hyperphosphorylated tau aggregates have been reported to 60 appear in the retina of patients even at an early stage, 10 although controversial findings have also 61 been published 13. Whether the appearance of AD makers in the retina can be exploited for 62 diagnostics purposes is still disputed.14 63 Given the difficulty of AD treatment7 and the challenges of diagnosis,5,6 it is crucial to increase 64 the understanding of the retina as a potential diagnosis method of the disease.15 Mouse models are 65 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 3 a central component of drug testing and the investigation of disease mechanisms16 and are thus an 66 important subject to increase the understanding of the connection between retinal and cerebral 67 pathologies of AD. The retinal phenotype of different mouse models of AD has been investigated, 68 however reporting inconclusive results.17 69 One candidate technology for retinal diagnostics in AD is optical coherence tomography (OCT) . 70 OCT is a non-invasive imaging technique often used for in vivo retinal imaging, also in the context 71 of neurodegenerative diseases.18 OCT is based on the interference of low-coherent light scattered 72 by the sample with a reference beam to reveal information about the sample. OCT can be used to 73 generate 3D images of tissue, also in real time. 19 Modern OCT technology yields high-resolution 74 images of vasculature20, charts retinal thickness21 and can visualize focal retinal lesions.22 Several 75 mouse models of familial AD have been investigated using retinal OCT imaging. In an APP-PS1 76 mouse model, retinal thinning was observed in one study,23 whereas no changes could be measured 77 by another group. 22 Several studies reported thinning of retinal layers for the 3xTg mouse 78 models.17 In previous investigation s of the 5xFAD mouse model of AD thinning of the retinal 79 nerve fiber layer (RNFL) and thickening of the inner plexiform layer (IPL) was measured with a 80 commercial OCT device (Leica Envisu R2200) in 6–to 17-month-old mice.24 In a study by Kim et 81 al., thinning of the total retina, the inner (IRL) and outer retinal layers (ORL) as well as the RNFL 82 were reported with a spectral domain OCT (SD-OCT) system. Additionally, a decrease in capillary 83 density was reported for this study with female mice.25 A third study using a commercial SD-OCT 84 system reported thinning for the RNFL and thickening for the OPL and ONL for male transgenic 85 5xFAD mice and C57B L/6J controls at 3 months of age .26 Overall, the retinal phenotype of AD 86 mouse models can be considered very controversial and demands careful clarification to enlighten 87 the current landscape of conflicting results. 88 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 4 In addition, the connection of the retinal phenotype with cognitive impairment of the investigated 89 mouse models is largely unexplored and leaves room for investigation. Moreover, the potential 90 differences between female and male mice are often not investigated in the present literature. Here, 91 by using a high-resolution OCT prototype tailored for multi -contrast retinal imaging in mice and 92 combining the data with the assessment of cognitive impairment through behavioral testing , we 93 aspire to gain new insights on the development of retinal parameters and unveil potential 94 connections between the appearing phenotypes. 95 2 Materials and Methods 96 2.1 Animal Model 97 The 5xFAD mouse model (The Jackson Laboratory, strain #008730) is a common transgenic 98 mouse model for the investigation of Alzheimer’s disease and has been used in about 10% of 99 studies.27 The mouse line is based on a C57BL/6J background and overexpresses human amyloid 100 beta amyloid precursor protein related to the Swedish (K670N, M671L), Florida (I716V), and 101 London (V717I) familial Alzheimer's disease (FAD) mutations as well as the human presenilin 1 102 (PS1) protein by harboring two FAD mutations, M146L and L286V. 5xFAD mice have been 103 reported to present amyloid beta plaque formation in the brain as early as two months of age.27 104 2.2 Study Design 105 5xFAD mice (32 transgenic, 32 non-transgenic littermates) at 10 weeks (±1 week) of age were 106 provided by Scantox Neuro GmbH. The animals were housed under controlled light conditions 107 (12 hours dark, 12 hours light) with food and water ad libitum. Overall health status and weights 108 were monitored every week. Animals were longitudinally investigated over the course of 6 109 months. OCT scans of animals were taken on five occasions at 12, 20, 24 and 36 weeks of age. 110 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 5 Spatial memory was tested at 35 weeks of age using the Morris Water Maze (MWM). The exact 111 number of animals and the number of transgenic (tg) and non-transgenic (ntg) animals involved 112 in each test is listed in Table 1. Note that the animal numbers decrease over the course of the 113 study because some animals were used for another investigation, which is not part of the research 114 described in this work, and thus were extracted from the study at 12 weeks and 24 weeks of age. 115 All experiments were performed in accordance with the ARVO Statement for the Use of 116 Animals in Ophthalmic and Vision Research and Directive 2010/63/EU. All experimental 117 procedures and protocols were approved by the ethics committee of the Medical University of 118 Vienna and the Austrian Federal Ministry of Education, Science, and Research (GZ 2024-119 0.044.300). 120 Age [weeks] 12 20 24 35 36 Number of mice examined with OCT 64 48 48 / 27 Transgenic Non-transgenic 32 32 24 24 24 24 / 14 13 Number of mice tested in the MWM / / / 27 / Transgenic Non-transgenic / / / 14 13 Table 1 Number of animals used for each test. 121 2.3 OCT System 122 The polarization-sensitive OCT (PS-OCT) system first presented by Fialová et al., was used to 123 perform the study.28 The system was based on a super-luminescent diode with a central 124 wavelength of 840 nm and a bandwidth of 100 nm, resulting in an axial resolution of ~3.8 µm in 125 tissue (n=1.35). The spectrometer line scan cameras acquired the spectral data with 3072 pixels 126 for the co-and cross polarized channels with an A-scan rate of 80 kHz.28 By aligning the mouse 127 placed on a mount providing three axes of translation and two axes of rotation, the imaged field 128 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 6 of view was centered at the optic nerve head (ONH) and measured approximately 1mm x 1mm. 129 Five repeated B-scans were acquired in 400 positions, resulting in volumes of 2000 B-scans 130 consisting of 512 A-scans. The recorded volumes contained 512x400x3072 spectral voxels. 131 2.4 Anesthesia and Imaging 132 Mice were placed in an anesthesia induction chamber, which was afterwards filled with 4% 133 isoflurane (IsoFlo, Zoetis Österreich GmbH) in oxygen for 4 minutes prior to imaging. 134 Tropicamid drops (0.5%, Agepha Pharma s.r.o.) were applied to dilate the pupils of the animals. 135 For imaging, the mice were transferred to a home-built animal mount and kept under anesthesia 136 using a nose cone applying 2% isoflurane. In some cases, individuals developed resistance to the 137 isoflurane during the study, requiring an increased concentration of 2.5% at later imaging 138 timepoints to ensure proper sedation. To prevent hypothermia, the mouse was blanketed with a 139 heating pad. Oculotect eye drops (Théa Pharma GmbH) were frequently applied to keep the eyes 140 of the animals hydrated during the entire anesthesia sessions. Shortly prior to imaging, excessive 141 eye drop liquid was carefully removed from the mouse eye using a cotton swab to avoid 142 additional lensing effects. Both eyes were imaged with the ONH in the center of the field of 143 view. Several male animals (n(tg) = 3, n(ntg) = 3) deceased during or after anesthesia and could 144 not be investigated for all time points, but no female animals were affected. 145 2.5 Image Processing 146 Image processing was performed to provide images displaying reflectivity, motion and 147 polarization-based contrast, using the pipeline described previously (Augustin et al., 2016). 148 Retinal image data were flattened with respect to the retinal pigment epithelium (RPE) detected 149 via its depolarizing properties in the PS-OCT images.29 Data sets that could not be processed in 150 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 7 this first step due to low signal intensity, strong vignetting or acquisition errors, among others, 151 were excluded from the study. 152 153 2.5.1 Layer Thickness Analysis 154 To measure retinal layer thickness, the algorithm described by Augustin et al. was used 155 (Augustin et al., 2018). The distance between RPE and inner limiting membrane (ILM) was 156 considered as the total retinal thickness.30 Additionally the thicknesses of the outer retinal layers 157 (ORL) and inner retinal layers (IRL) were measured, using the posterior surface of the outer 158 plexiform layer as the boundary. The thickness of the retinal nerve fiber layer/ganglion cell layer 159 complex (RNFL/GCL), the inner (INL), the inner (IPL) and outer plexiform layer (OPL), the 160 photoreceptor complex (PRC) as well as the depolarizing RPE complex were also measured. 161 Layer thickness data were stored as 2D en-face maps. To align measurements between all 162 animals, the center of the ONH in each 3D volume was manually annotated and the central circle 163 with 200 µm in diameter was removed from further analysis. The rest of the volume was divided 164 into four sectors (superior, nasal and inferior, temporal), using the diagonals of the 1x1 mm² 165 square as borders and into radial zones equidistant from the center. Then a zone with an inner 166 radius of 200 µm and an outer radius of 600 µm distance from the ONH is selected. The average 167 thickness of the total retinal thickness and the thickness of the sublayers was calculated in this 168 annular zone. Thickness maps were manually screened during post processing and volumes with 169 insufficient quality to produce a reliable segmentation of the retinal layers were excluded from 170 the study. When two or more acquisitions of the same eye and timepoint were available, the scan 171 with best signal quality or most centered ONH was used. When neither of the scans was 172 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 8 significantly better than their counterparts, the resulting thickness measurements were averaged 173 between both scans. This resulted in the number of thickness maps listed in Table 2. 174 Age [weeks] 12 20 24 36 Number of analyzed volume scans 125 87 86 50 Transgenic Non-transgenic 61 64 41 46 44 42 25 25 Table 2 Number of individual volumetric OCT scans used for retinal layer thickness analysis. 175 2.5.2 Angiography 176 OCT angiography (OCTA) data were computed as described in reference 22. Volumetric OCTA 177 data were divided into three slabs using the layer segmentation described in section 2.5.1: the 178 superior vascular plexus (SVP) in the RNFL/GCL complex, the intermediate capillary plexus 179 (ICP) between the RNFL and the INL, and the deep capillary plexus (DCP) between the INL and 180 the ONL. Using thresholding at 10 dB above the noise floor and a Frangi filter, binarized en-face 181 images of the vascular plexuses were created, and a margin measuring 10 pixels at the border 182 was set to zero to avoid border artifacts. Using the manually selected ONH position previously 183 used for the quantification of retinal layer thickness, a circle around the ONH was cut out from 184 the binarized images. The signal-to-noise ratio (SNR) was calculated for the regions visible in 185 Figure 1(D), and the image was divided into four regions along the diagonals and then again into 186 rings with 100 µm thickness, each of the zones has an individual SNR value. The binarized 187 images were multiplied with the SNR map of the scan shown in Figure 1(D), thus assigning the 188 values to the binarized image as exemplified in Figure 1(E). The regions whose SNR was below 189 a predefined threshold were not used for further evaluation and were thus set to zero (see Figure 190 1(F)). For SVP and ICP, the threshold was set to 20 dB, while for the DCP, the threshold was set 191 to 15 dB since the signal in this region was generally lower. In the case displayed in Figure 1, the 192 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 9 regions in the corners and around the ONH did not display above threshold SNR and were 193 removed. By inverting only the pixels in the regions with above-threshold SNR, a negative 194 image was created. By counting the positive pixels in the regions with above-threshold SNR 195 (Figure 1(G)) and the negative pixels in the same regions (Figure 1(F)), the area vessel density 196 was calculated for each vascular plexus. 197 198 Fig 1: Example for the calculation of vessel density in the DCP. (A) Raw OCTA data after layer segmentation of the 199 3D stack, (B) binarized image with borders set to zero (C) removal of the circular region centered at the ONH from 200 the binarized image, (D) SNR sector map, rings have a thickness of 100 µm (E) application of the SNR on the 201 binarized image, (F) removal of the regions with SNR below threshold and creation of the negative for counting 202 pixels. 203 The OCTA data was sorted similar to the retinal layer analysis. Additionally the data was 204 screened manually, to only include the best available datasets. Table 3 presents the number of 205 datasets that yielded at least one measurement for the 3 different vascular regions and thus were 206 used for the longitudinal OCTA analysis. 207 Timepoint [weeks] 12 20 24 36 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 10 Number of analyzed volume scans 119 84 85 50 Transgenic Non-transgenic 64 55 41 43 43 42 25 25 Table 3 Number of individual volumetric OCTA datasets used for angiography. 208 2.6 Morris Water Maze 209 Spatial memory capabilities of the animals were tested in a custom-built Morris Water Maze 210 (MWM) setup.31 The maze consisted of a water filled pool, one meter in diameter and 211 surrounded by opaque curtains, a translucent platform (8 cm diameter) and four distinct 212 landmarks as illustrated in Figure 2(A). A monochrome camera with 5 megapixels resolution and 213 14 fps (Basler Ace Classic, acA2500-14uc) was mounted 1.5 m above the pool and the 214 landmarks placed in the middle of each of the four sides surrounding the pool, to avoid any 215 additional orientation points for the mice. Movie data were acquired by using the Basler Video 216 Recording software (version 1.3). The northeastern quadrant contained a translucent platform 217 (target zone) about 1 cm below water level, such that it was invisible for the mice. The pool 218 temperature was kept between 21-22°C, and the light intensity at the water surface level was 219 controlled to be around 50 lux. The pool was divided into four quadrants to randomize the 220 starting position (see Figure 2(B)). The starting positions were defined as per Table 4 and were 221 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 11 applied to every mouse to assure an equal distribution of starting positions. 222 223 Fig. 2: Schematic of the Morris Water Maze set-up. (A) Division of the pool into quadrants for the randomization of 224 starting positions. (B) Schematic of the pool set up with landmarks and camera placement. 225 226 Trial 1 2 3 4 5 Task Training Training Training Training Test day Platform position (for all trials) Day 1 NE SE SW SE / NW Day 2 SW NE SE SW / NW Day 3 SE SW NE SW / NW Day 4 NE SE SW NE / NW .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 12 Day 5 / / / / SE / Table 4 Starting positions for each training and the test day. Quadrants: NE = northeast, SE = southeast, SW = 227 southwest, NW = northwest. 228 229 The testing protocol included four consecutive training sessions and one test day. A training day 230 consisted of four trials, where each mouse was placed in the pool (in varying quadrants as 231 outlined in Table 4) for a maximum of 60 seconds and had to find the hidden platform. When a 232 mouse was not able to find the platform within this time, it was placed on the platform and left 233 there for ten seconds to memorize the position. In case the mouse managed to find the platform, 234 the recording was stopped, and the mouse was removed from the pool. The time between daily 235 trials was 10 minutes. On the test day, the hidden platform was removed, and the mice were 236 placed in the pool once for one minute. Each trial was recorded using the top-down mounted 237 camera. 238 For data evaluation an automatic tracking system, namely Noldus EthoVision XT 14, was used. 239 For the training days, latency to find the platform [s], distance traversed [m], as well as the 240 percentage of time spent floating and thigmotaxis (time spent close to the pool walls) was 241 measured. 32 For the test day, the abidance in each of the of the quadrants and the number of 242 crossings of the target zone were evaluated. The resulting data were analysed in GraphPad 243 PrismTM 10. 244 2.7 Statistical Analysis 245 For the longitudinal analysis of the retinal parameters, a mixed intercepts model was applied to 246 investigate age dependent effects and account for multiple measurements per mouse. For 247 comparison between groups at each timepoint, a t-test was applied. P-values smaller than 0.05 248 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 13 were considered significant. 249 To assess whether retinal layer thickness and vessel density data were correlated with spatial 250 memory impairment, the OCT and MWM parameters were analyzed using linear regression. If 251 data for both eyes of a mouse was available, the average of the two values was used as one single 252 datapoint for the analysis. Additionally, the weight of the animals was also included in the 253 analysis. Linear correlation coefficients, R-squared, mean squared error (MSE) intercept and the 254 linear regression coefficient were calculated using a custom Python script. Pearsons’s correlation 255 analysis was used to calculate the correlation coefficients and the corresponding p-values. 256 3 Results 257 3.1 Longitudinal Development of Retinal Layer Thickness 258 The longitudinal analysis of the total retinal thickness reveals an overall thickening of the retina 259 for male animals by 5.7 µm (p=0.0029) from the beginning to the end of the study. Overall 260 retinal thickness increases significantly by 5.8 µm (p=0.0051) for all animals in the study 261 duration. Significant differences between groups were observed between tg male and female 262 animals, as well as for ntg male and female animals at 3 months of age. Tg male animals 263 (225.4±7.1µm) had significantly thicker retinas than tg female animals (221.1±7.8µm) 264 (p=0.014), also for ntg animals, male mice (224.4±7.3µm) displayed a higher total retinal 265 thickness than female animals (218.5±10.7µm) (p=0.0298). Other significant differences were 266 found between male and female ntg animals at 6 months of age, where male animals 267 (223.6±3.6µm) showed significantly higher total retinal thickness than females (220.5±4.5µm) 268 (p=0.0236), and between male and female tg mice at 9 months of age: Male tg mice had 269 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 14 significantly thicker retinas (228.0±6.4µm) than female tg mice (220.0±6.5µm) (p=0.0386) for 270 the latest measurement point. 271 272 Fig. 3: Longitudinal measurement of the total retinal thickness for all mice divided by genotype and sex. Individual 273 points represent singular measurements for one mouse eye. Significant differences between groups are marked with 274 brackets above the boxes and annotated with the corresponding p-values. 275 Following the analysis of total retinal thickness, the retinas were segmented into IRL (RNFL, 276 IPL, INL) and ORL (PRC, RPE and OPL) to evaluate the thickness of these retinal sublayers. 277 Figure 4 shows the time development of the IRL (A) and the ORL (B). The mixed effects model 278 yielded a significant thickness increase of the ORL of 3.8 µm (p=0.0048) for all mice over the 279 course of the investigation. Significant thickening of the IRL in comparison to the measurements 280 at 3 months of age could not be detected (p=0.0503). Significant group differences for the IRL 281 measurements were observed at 3, 5 and 9 months of age. At 3 months of age male, tg mice 282 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 15 (97.3±3.7µm) showed a significantly higher IRL thickness than female tg mice (94.7±5.0µm) 283 (p=0.0215). This was also observed for ntg mice, where male mice (96.7±4.1µm) had thicker 284 IRL than female ntg mice (94.4±3.9µm) (p=0.0228). For 9 months old mouse models, the IRL 285 thickness of tg male mice (98.16±3.19µm) was significantly larger than for ntg male mice 286 (95.4±2.1µm) (p=0.0363), and also in comparison with female tg mice (94.7±4.0µm), male tg 287 mice had significantly (p=0.0331) thicker IRL. For the measurements of the ORL, no significant 288 changes were observed for the last measurement at 9 months of age. 289 290 Fig. 4: Longitudinal measurement of the IRL (A) and ORL (B) for all mice divided by genotype and sex. Individual 291 points represent singular measurements for one mouse eye. Significant differences between groups are marked with 292 brackets above the boxes and annotated with the corresponding p-values. 293 As a next step, the sublayers of the IRL (INL, RNFL and IPL) were individually investigated to 294 assess potential changes over time and between groups. Neither of these retinal sublayers showed 295 significant changes in thickness over the study time of 6 months. For the INL measurements, the 296 analysis yielded significantly different thicknesses for male and female tg animals, where male tg 297 animals displayed a thickness of 26.8 ±1.3µm, significantly thicker than female tg animals with 298 27.2 ±1.4µm (p=0.0109). Significant thickness differences also appeared between ntg male and 299 female, as well as between tg male and female mice at 5 months of age. The investigation of 300 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 16 RNFL thickness showed significantly higher values for male tg animals (33.9±6.2µm) than for 301 female tg mice (29.4±5.6µm) (p=0.0041). For 5 months of age, a thicker RNFL for ntg male 302 animals than for tg male animals was observed as well. For the final measurement with 9-month-303 old animals, no significant differences were observed for the three inner retinal layers. 304 305 Fig. 5: Longitudinal measurement of the INL (A), RNFL (B) and IPL (C) for all mice divided by genotype and sex. 306 Individual points represent singular measurements for each one mouse eye. Significant differences between groups 307 are marked with brackets above the boxes and annotated with the corresponding p -values. 308 The ORL consists of the PRC, the OPL and the RPE. Figure 6 shows the development of 309 thickness over time for these layers. An overall thickening of the PRC for all animals by 310 2.6±1.3µm (p=0.0350) over the study duration was observed for. Significant differences between 311 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 17 groups were found for ntg male and female animals for 3 and 6 months of age. However, similar 312 changes in layer thickness were not observed for the measurements at 5 and 9 months of age. For 313 the OPL, an overall thickening of the layer was measured (1.1±0.3µm) (p=0.0003) for all 314 animals. At the last measured timepoint, female ntg animals showed a significantly thicker OPL 315 (12.7±1.2µm) compared to female tg animals (11.7±0.6µm) (p=0.0128). 316 For the RPE data, the mixed effects models showed an overall lower RPE thickness for tg 317 animals of about 0.5±0.24µm (p=0.036). Changes over time for the RPE were not identified. 318 Similar to the OPL, female ntg animals had a different RPE thickness than tg female animals at 9 319 months of age. With 11.6±1.1µm, the female ntg animals display a significantly thicker RPE 320 than the tg animals with 10.6 ±0.4µm (p=0.0058). Note that the RPE thickness measurements 321 provided here reflect the segmented thickness of the depolarizing layer.28 Further significant 322 changes were measured for PRC thickness between male (115.7±3.5µm) and female 323 (112.5±7.8µm) ntg animals at 3 months of age (p=0.0394), between tg (11.0±0.7µm) and ntg 324 (11.7±1.3µm) male animals at 5 months of age for OPL (p=0.0423) and for tg (10.3±0.6µm) and 325 ntg (11.1±1.0µm) male animals at 5 months of age for the RPE thickness (p=0.0036). 326 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 18 327 Fig. 6: Longitudinal measurement of the PRC (A), OPL (B) and RPE (C) for all mice divided by genotype and sex. 328 Individual points represent singular measurements for one mouse eye. Significant differences between groups are 329 marked with brackets above the boxes and annotated with the corresponding p -values. 330 331 3.2 Angiography 332 Significant differences were observed between male ntg and tg mice at 9 months of age for the 333 SVP and ICP measurements. The SVP density for ntg male mice (19.8±1.6%) was significantly 334 lower (p=0.0063) than for male tg mice (22.1±1.5%). The mixed intercepts model revealed a 335 significant (p=0.025) density decrease of about 2% for male mice over the course of the study 336 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 19 from 3 months until 9 months of age. Significant differences were also observed between ntg 337 male and ntg female mice with lower values for ntg male mice (p=0.0047) at 9 months of age 338 and for female ntg and female tg mice at 5 months of age (p=0.0365, see Figure 7(A)). Male ntg 339 mice also experienced an ICP density decrease over the course of the longitudinal investigation. 340 The mixed effects model yielded a significant decrease of 3.6% in density for all male mice 341 (p=0.039) and 5.4% for male ntg mice (p=0.025) from the baseline measurement to the endpoint. 342 Significant differences between ntg male and tg male mice can be observed at 9 months of age. 343 With 6.8±1.4%, ntg male mice showed significantly lower density than tg male mice with 344 9.9±2.5% (p=0.0023). Male ntg mice also had significantly lower densities than female ntg mice 345 (p=0.0292). Ntg male mice also showed significantly lower ICP density than female ntg mice at 346 5 months of age (p=0.0207, see Figure 7(B)). In contrast to these findings in the SVP and ICP, 347 no significant changes over time or between individual groups were observed for the 348 measurement of the DCP density (Figure 7(C)). 349 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 20 350 Fig. 7: Longitudinal measurements for vascular density in percentage of area (A) SVP density, (B) ICP density, (C) 351 DCP density. Significant differences between groups are marked with brackets above the boxes and annotated with 352 the corresponding p-values. 353 354 3.3 Spatial Memory Testing 355 3.3.1 MWM at 35 weeks of age 356 All remaining animals (n=27) were retested in the same MWM set-up at the age of 35 weeks. As 357 shown in Figure 8, no significant differences between the tg and ntg animals can be observed for 358 the experiment. For female animals, significant differences were observed for the latency to find 359 the platform on the fourth training day (Figure 9 (A)). For the other parameters, no changes can 360 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 21 be observed. 361 362 Fig. 8: MWM results for animals at 35 weeks of age. (A) Latency to find the target platform, (B) distance traversed, 363 (C) floating behavior, and (D) thigmotaxis over the four training days. (E) Abidance in target quadrant and (F) 364 number of target zone crossings on the test day. Whiskers indicate ± the standard deviation (SD). 365 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 22 366 Fig. 9: MWM results for female animals at 35 weeks of age. (A) Latency to find the target platform, (B) distance 367 traversed, (C) floating behavior, and (D) thigmotaxis over the four training days. (E) Abidance in target quadrant 368 and (F) number of target zone crossings on the test day. Whiskers indicate ± the standard deviation (SD). 369 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 23 370 Fig. 10: MWM results for male animals at 35 weeks of age. (A) Latency to find the target platform, (B) distance 371 traversed, (C) floating behavior, and (D) thigmotaxis over the four training days. (E) Abidance in target quadrant 372 and (F) number of target zone crossings on the test day. Whiskers indicate ± the standard deviation (SD). 373 Male animals displayed no significant differences in the comparison with ntg animals in the 374 MWM test. 375 3.4 Correlation between retinal parameters and spatial memory 376 The correlation analysis between retinal parameters and values obtained from the spatial memory 377 testing revealed a significant association between the INL thickness and the MWM parameters of 378 the test day (day 5) for female transgenic mice. The number of target zone crossings (Figure 11 379 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 24 (D-F)) as well as the percentage of time spent in the target sector (Figure 12 (D-F)) correlated 380 strongly with the thickness of the INL (0.87, 0.86), the IRL (0.77, 0.74) and the total retina (0.76, 381 0.69). Pearson’s correlation coefficient was the highest for the INL with 0.87 (p=0.0054) for the 382 number of target zone crossings and 0.86 (p=0.0060) for the time spent in the target sector. 383 These correlations were only observed for female transgenic mice. For male transgenic mice, no 384 correlation between layer thickness measurements and MWM parameters was observed for the 385 test day. Combining the two groups and looking at the correlation for all transgenic animals 386 revealed a statistically significant correlation between the two investigated MWM parameters 387 and the IRL (0.54) and INL (0.62), respectively, as shown in Figures 11 (B, C) and 12 (B, C). 388 389 Fig. 11: Correlation and linear regression between total retinal (left), IRL (middle) and INL thickness (right) with 390 number of target zone crossings for all transgenic mice (A-C), female transgenic mice (D-F) and male transgenic 391 mice (G-I). The respective Pearson correlation coefficient (Corr) and p-value as well as R2 and mean squared error 392 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 25 (MSE) resulting from the linear regression are indicated below each graph. For each plot, the colored line represents 393 the linear regression. 394 395 Fig. 12: Correlation and linear regression between total retinal (left), IRL (middle) and INL thickness (right) with % 396 of time spent in the target sector for all transgenic mice (A-C), female transgenic mice (D-F) and male transgenic 397 mice (G-I). Below each graph, the respective Pearson correlation coefficient and p-value as well as R2 and MSE for 398 the linear regression are indicated. The colored line represents the respective linear regression. 399 Additionally, the ICP density correlated with thigmotactic behavior for female transgenic mice 400 (Corr=0.76, p=0.047) and the PRC thickness correlated negatively with floating behavior (Corr= 401 -0.82, p=0.0126). For male transgenic animals , no correlations between MWM and retinal 402 parameters were found. Furthermore, the weight of the male transgenic animals strongly correlated 403 with the SVP density (Corr=0.96, p=0.008). Tables with all compared parameters for all animal 404 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 26 groups (tg, tg females, tg males, ntg, ntg females and ntg males) including correlation coefficients, 405 p-values, R2 and MSE are provided in the supplementary information (Table 1-12). 406 4 Discussion 407 In this investigation, longitudinal OCT imaging of a 5xFAD mouse model was performed in 408 parallel to spatial memory testing and uncovered pronounced differences in retinal parameters 409 dependent on both genotype and sex . Our analysis of retinal layer thickness revealed subtle but 410 significant thickening of the total retina over the course of the study. Thickening was also observed 411 for the ORL, where in particular the PRC and the OPL thickness increased over time. In the 412 longitudinal investigation, c onsistent thinning was not detectable for any retinal layers from 3 413 months to 9 months of age. However, thickness differences were observed for group comparison 414 at individual measurement timepoints. Our comparison between the groups at the last measurement 415 point at 9 months of age resulted in the detection of several parameters with statistically significant 416 differences. Here the total retina and the IRL of tg male mice was significantly thicker than for tg 417 female mice. The IRL layers for tg male mice were also significantly thicker than for the ntg male 418 mice indicating a swelling of the IRL for male mice. For female mice , OPL and RPE thickness 419 significantly thin in comparison to ntg female mice at 9 months of age. The longitudinal analysis 420 of OCT angiography data did not show changes for neither ntg nor tg female mice. However, for 421 male ntg mice, a significant density loss in the SVP and ICP was observed over time, leading to 422 significant differences when comparing the group to tg male mice. The observed changes in retinal 423 layer thickness and vascular density differed strongly between male and female mice indicating a 424 dependence of the phenotype on sex. Changes in angiographic parameter s depended more on the 425 sex of animals than on their genotype. The differences in retinal layer thickness also appeared very 426 differently dependent on the sex of animals. 427 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 27 The correlation analysis of the MWM data and the retinal parameters revealed an association 428 between the thickness of the total retina, the IRL and the INL with the number of target zone 429 crossings (NTZC) and the time spent in the target sector (TSTS) on the test day. Both parameters 430 for the MWM – although not completely independent of each other – are measures of spatial 431 memory capabilities. The strongest correlation was observed between these parameters and the 432 INL thickness of female tg mice. The correlation between the NTZC and TSTS with the IRL as 433 well as the total retinal thickness for female mice is likely due to their strong correlation with the 434 INL. For male tg mice, no significant positive correlation between total retina, IRL and INL layer 435 thickness was measured. With a correlation coefficient of 0.65 for INL and TSTS, correlation 436 between the two parameters can still be observed . Combining this information with the sex 437 differences in layer thickness, we conclude a strong sex influence on the retinal phenotype of the 438 5xFAD mouse model and its connection to spatial memory. 439 Our results partly differ from data reported in other studies on 5xFAD mice . Lim and coworkers 440 observed RNFL thinning and OPL thickening in the investigation of 32 tg and 38 ntg mice with 441 unspecified sex.24 In contrast, we did not observe RNFL thinning, although our investigation did 442 indicate (non-significant) thickening of the OPL. Given that the sex specifications were missing 443 in that publication, it was not possible to perform a clear comparison of our data with this previous 444 measurement. We do want to stress that the gender composition of the investigated animals is 445 especially relevant given the strong impact of sex on the results measured in this study and in AD 446 mouse models in general.33 The inclusion of more female or male tg or ntg control animals could 447 shift the outcome of the measured results considerably. 448 In another study, Matei et al. performed OCT on 16 male tg 5xFAD mice in comparison with 10 449 C57BL/6J and 6 C57BL/6 mice. In 3-months-old tg mice, the authors do not report changes in the 450 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 28 total retinal thickness measured with OCT .26 Our investigation confirms these findings for male 451 animals. Of note, this study did not use littermates, but C57BL/6J and C57BL/6 as controls, using 452 two different control strains may bias the results.26 453 In their 2021 study, Kim et al. performed an OCT investigation in a female-only cohort of 5 tg and 454 6 ntg animals at 6 months of age.25 The chosen control animals were of the associated background 455 strain (B6SJLF1/J), hence a comparison with our data on female 5xFAD mice should be straight-456 forward. Still, we cannot confirm their findings of retinal thinning in the total retina, the RNFL, 457 the IRL and ORL at 6 months of age. On the other hand, our measurements confirm the absence 458 of changes in vascular density for SVP, ICP and DCP. 25 459 Given that we investigated 44 tg (22 female, n=12) and 42 ntg (19 female, n=12) volume scans for 460 mice at 6 months of age, the lack of overlap with the previously reported results is surprising and 461 can probably be attributed to a number of differences in the study protocols. One potential reason 462 for this is the use of differing anaesthesia agents in the mentioned studies, as ketamine/xylazine 463 was used by Kim et al., Lim et at., and Matei et al., whereas our study used isoflurane. 24,25,26 464 Additionally, genetic drift, breeding, body temperature of animals, anaesthesia time, blood 465 pressure, time of imaging, eye drops as well as diet and housing conditions (e.g., single vs. group 466 housing) could potentially – and in some cases do – differ between the studies and can influence 467 the outcome of experiments. For this reason, we want to stress that there is a lack of standardization 468 in the research field that makes the replication and confirmation of results extremely difficult. 469 Without a description of every detail in the study protocol , results might considerably differ 470 between research groups. This also applie s to the comparison between mouse models. Retinal 471 changes were reported for several other mouse models, some with partially overlapping knock -in 472 genes (3xTg) or modelling similar disease aspects (APP/PS1 mouse for amyloid pathology) as the 473 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 29 5xFAD mouse model. 17 Another factor that should be considered is the regional dependency of 474 retinal thinning, as for example measured in 3xTg mice, w here the longitudinal development of 475 sub-layer thickness differs depending on the distance to the ONH. 34 Given the variety in imaging 476 protocols and aforementioned factors, as well as varying data analysis approaches, a reliable 477 comparison of OCT findings of retinal pathologies in mouse models seems currently only possible 478 if done in the same research group, leaving a significant knowledge gap in the investigation of the 479 retinal phenotype of the AD disease. Since most AD mouse models only model one or few aspects 480 of the disease pattern, comparability between studies and thus models is critical for the research of 481 the retinal pathology of AD and the translatability of the results to human AD diagnostics. 482 Gender-specific medicine is crucial for drug development, diagnostics and treatment. It is therefore 483 also increasingly important to discover sex-based differences in mouse models of disease and take 484 these into account for preclinical studies. The study presented in this work represents not only the 485 most extensive longitudinal OCT investigation of the 5xFAD mouse model to date, but also reveals 486 sex-based differences in comparisons of retinal parameters with the spatial memory phenotype of 487 the mouse models, thus adding more knowledge to further improve targeted research in the field 488 of AD. 489 5 Conclusion 490 The presented study investigated retinal parameters in the 5xFAD mouse model of AD over the 491 course of 6 months providing insights into the progression of retinal pathologies, especially the 492 respective retinal thickness changes for male and female mice, as well as the decay in vascular 493 density for ntg male mice. By correlating retinal parameters with the spatial memory phenotype 494 of the investigated animals, a connection between INL thickness and MWM performance was 495 observed for female tg animals, expanding the knowledge on the interaction between phenotypes 496 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 30 for this particular model of AD. Our investigation proves that a multidisciplinary approach, with 497 retinal imaging and behavior analysis, can provide more comprehensive and complementary 498 information, which may help understanding the concurrent development of different aspects of 499 AD. Additionally, this study shows that a separate analysis of female and male mice, as well as 500 the minute control of experimental conditions are crucial for the generation of unbiased results in 501 the investigation of the retinal phenotype in mouse models of AD. 502 503 Disclosure 504 Magdalena Daurer, Laurenz Jauk, Roland Rabl and Manuela Prokesch are employees of Scantox 505 Neuro GmbH. All other authors declare that there are no financial interests, commercial 506 affiliations, or other potential conflicts of interest that could have influenced the objectivity of this 507 research or the writing of this paper. 508 509 Acknowledgments 510 The authors want to thank And reas Hodul for his help in constructing the MWM set -up and the 511 mouse imaging stage . We are grateful to Sonja Reynoso-De-Leon, Christian Schönauer , Jasmin 512 Rezek and the te am in the animal facility for their irreplaceable assistance with the mice. 513 Additionally, we want to thank Patrick Bilic and Eva Fuchs for their support and guidance on 514 animal welfare related issues , and Robin Ristl (Center for Medical Data Science, Medical 515 University of Vienn a) for assistance with the statistical analysis. Funding for this project was 516 provided by Scantox Neuro GmbH, the FFG grant 900435, the ERC Proof of Concept grant 517 101069344 OPTIMEYEZ and the Austrian Science Fund grant I6092-B. 518 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 31 Code, Data, and Materials Availability 519 All data in support of the findings of this paper are available within the article and as 520 supplementary material. Raw data can be requested from the author at 521 [email protected]. . 522

523 1. Health Organization, W. Global Status Report on the Public Health Response to 524 Dementia. (2021). 525 2. Serrano-Pozo, A., Frosch, M. P., Masliah, E. & Hyman, B. T. Neuropathological 526 alterations in Alzheimer disease. Cold Spring Harb Perspect Med 1, (2011). 527 3. Chatterjee, S. & Mudher, A. Alzheimer’s disease and type 2 diabetes: A critical 528 assessment of the shared pathological traits. Frontiers in Neuroscience vol. 12 Preprint at 529 https://doi.org/10.3389/fnins.2018.00383 (2018). 530 4. King, A., Bodi, I. & Troakes, C. The neuropathological diagnosis of alzheimer’s 531 disease—the challenges of pathological mimics and concomitant pathology. Brain Sci 10, 532 1–23 (2020). 533 5. Leuzy, A., Heurling, K., Ashton, N. J., Schöll, M. & Zimmer, E. R. In Vivo Detection of 534 Alzheimer’s Disease. YALE JOURNAL OF BIOLOGY AND MEDICINE vol. 91 (2018). 535 6. Alawode, D. O. T. et al. Transitioning from cerebrospinal fluid to blood tests to facilitate 536 diagnosis and disease monitoring in Alzheimer’s disease. Journal of Internal Medicine 537 vol. 290 583–601 Preprint at https://doi.org/10.1111/joim.13332 (2021). 538 7. Thangwaritorn, S. et al. A Review of Recent Advances in the Management of Alzheimer’s 539 Disease. Cureus 16, e58416 (2024). 540 8. Espay, A. J., Kepp, K. P. & Herrup, K. Lecanemab and Donanemab as Therapies for 541 Alzheimer’s Disease: An Illustrated Perspective on the Data. eNeuro vol. 11 Preprint at 542 https://doi.org/10.1523/ENEURO.0319-23.2024 (2024). 543 9. Mahase, E. Donanemab in Early Symptomatic Alzheimer Disease: The TRAILBLAZER-544 ALZ 2 Randomized Clinical Trial. JAMA 330, 512–527 (2024). 545 10. Mirzaei, N. et al. Alzheimer’s Retinopathy: Seeing Disease in the Eyes. Frontiers in 546 Neuroscience vol. 14 Preprint at https://doi.org/10.3389/fnins.2020.00921 (2020). 547 11. Zhang, Y., Zhang, X., Yue, Y. & Tian, T. Retinal Degeneration: A Window to Understand 548 the Origin and Progression of Parkinson’s Disease? Frontiers in Neuroscience vol. 15 549 Preprint at https://doi.org/10.3389/fnins.2021.799526 (2022). 550 12. Soldatov, V. O., Kukharsky, M. S., Belykh, A. E., Sobolev, A. M. & Deykin, A. V. 551 Retinal damage in amyotrophic lateral sclerosis: Underlying mechanisms. Eye and Brain 552 vol. 13 131–146 Preprint at https://doi.org/10.2147/EB.S299423 (2021). 553 13. den Haan, J. et al. Amyloid-beta and phosphorylated tau in post-mortem Alzheimer’s 554 disease retinas. Acta Neuropathol Commun 6, 147 (2018). 555 14. Ashraf, G. et al. Retinal imaging biomarkers of Alzheimer’s disease: A systematic review 556 and meta-analysis of studies using brain amyloid beta status for case definition. 557 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 32 Alzheimer’s and Dementia: Diagnosis, Assessment and Disease Monitoring vol. 15 558 Preprint at https://doi.org/10.1002/dad2.12421 (2023). 559 15. Chiquita, S. et al. The Retina as a Window or Mirror of the Brain Changes Detected in 560 Alzheimer’s Disease: Critical Aspects to Unravel. Molecular Neurobiology vol. 56 5416–561 5435 Preprint at https://doi.org/10.1007/s12035-018-1461-6 (2019). 562 16. Zhong, M. Z., Peng, T., Duarte, M. L., Wang, M. & Cai, D. Updates on mouse models of 563 Alzheimer’s disease. Molecular Neurodegeneration vol. 19 Preprint at 564 https://doi.org/10.1186/s13024-024-00712-0 (2024). 565 17. Sánchez-Puebla, L. et al. OCT Imaging in Murine Models of Alzheimer’s Disease in a 566 Systematic Review: Findings, Methodology and Future Perspectives. Biomedicines vol. 12 567 Preprint at https://doi.org/10.3390/biomedicines12030528 (2024). 568 18. Minakaran, N., de Carvalho, E. R., Petzold, A. & Wong, S. H. Optical coherence 569 tomography (OCT) in neuro-ophthalmology. Eye (Basingstoke) vol. 35 17–32 Preprint at 570 https://doi.org/10.1038/s41433-020-01288-x (2021). 571 19. Aumann Silke, Fischer Jörg, Müller Frank & Donner Sabine. High Resolution Imaging in 572 Microscopy and Ophthalmology New Frontiers in Biomedical Optics Forewords By. 573 (2019). 574 20. de Carlo, T. E., Romano, A., Waheed, N. K. & Duker, J. S. A review of optical coherence 575 tomography angiography (OCTA). International Journal of Retina and Vitreous vol. 1 576 Preprint at https://doi.org/10.1186/s40942-015-0005-8 (2015). 577 21. den Haan, J., Verbraak, F. D., Visser, P. J. & Bouwman, F. H. Retinal thickness in 578 Alzheimer’s disease: A systematic review and meta-analysis. Alzheimer’s and Dementia: 579 Diagnosis, Assessment and Disease Monitoring 6, 162–170 (2017). 580 22. Harper, D. J. et al. Retinal analysis of a mouse model of Alzheimer’s disease with 581 multicontrast optical coherence tomography. Neurophotonics 7, 1 (2020). 582 23. Georgevsky, D., Retsas, S., Raoufi, N., Shimoni, O. & Golzan, S. M. A longitudinal 583 assessment of retinal function and structure in the APP/PS1 transgenic mouse model of 584 Alzheimer’s disease. Transl Neurodegener 8, (2019). 585 24. Lim, J. K. H. et al. Retinal Functional and Structural Changes in the 5xFAD Mouse Model 586 of Alzheimer’s Disease. Front Neurosci 14, (2020). 587 25. Kim, T.-H., Son, T., Klatt, D. & Yao, X. Concurrent OCT and OCT angiography of retinal 588 neurovascular degeneration in the 5XFAD Alzheimer’s disease mice. Neurophotonics 8, 589 (2021). 590 26. Matei, N. et al. Retinal Vascular Physiology Biomarkers in a 5XFAD Mouse Model of 591 Alzheimer’s Disease. Cells 11, (2022). 592 27. Forner, S. et al. Systematic phenotyping and characterization of the 5xFAD mouse model 593 of Alzheimer’s disease. Sci Data 8, (2021). 594 28. Fialová, S. et al. Polarization properties of single layers in the posterior eyes of mice and 595 rats investigated using high resolution polarization sensitive optical coherence 596 tomography. Biomed Opt Express 7, 1479 (2016). 597 29. Augustin, M. et al. Multi-functional OCT enables longitudinal study of retinal changes in 598 a VLDLR knockout mouse model. PLoS One 11, (2016). 599 30. Augustin, M., Harper, D. J., Merkle, C. W., Hitzenberger, C. K. & Baumann, B. 600 Segmentation of Retinal Layers in OCT Images of the Mouse Eye Utilizing Polarization 601 Contrast. in Lecture Notes in Computer Science (including subseries Lecture Notes in 602 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint 33 Artificial Intelligence and Lecture Notes in Bioinformatics) vol. 11039 LNCS 310–318 603 (Springer Verlag, 2018). 604 31. Morris, R. G. M. Spatial Localization Does Not Require the Presence of Local Cues. 605 LEARNING AND MOTIVATION vol. 12 (1981). 606 32. Löffler, T. et al. Impact of ApoB-100 expression on cognition and brain pathology in 607 wild-type and hAPPsl mice. Neurobiol Aging 34, 2379–2388 (2013). 608 33. Sil, A. et al. Sex Differences in Behavior and Molecular Pathology in the 5XFAD Model. 609 Journal of Alzheimer’s Disease 85, 751–774 (2022). 610 34. Batista, A. et al. Normative mice retinal thickness: 16-month longitudinal characterization 611 of wild-type mice and changes in a model of Alzheimer’s disease. Front Aging Neurosci 612 15, (2023). 613 614 615 616 Georg Ladurner has been a PhD student at the Medical University of Vienna since March 2023. 617 He received his BS and MS degrees in physics from the LMU Munich in 2020 and 2021, 618 respectively. His current research interests include optical coherence tomography, Alzheimer’s 619 disease, retinal diseases and retinal imaging techniques . He is a member of SPIE. 620 621 Biographies and photographs for the other authors are not available. 622 623 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655771doi: bioRxiv preprint