Integrative Severity Scale for Diabetic Macular Atrophy and Ischemia Using Structural OCT and OCT Angiography

Integrative Severity Scale for Diabetic Macular Atrophy and Ischemia Using Structural OCT and OCT Angiography | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Integrative Severity Scale for Diabetic Macular Atrophy and Ischemia Using Structural OCT and OCT Angiography Miyo Yoshida, Tomoaki Murakami, Kenji Ishihara, Yuki Mori, Akitaka Tsujikawa This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7000479/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 11 Dec, 2025 Read the published version in Scientific Reports → Version 1 posted 10 You are reading this latest preprint version Abstract We investigated the characteristics of neurovascular degeneration in diabetic retinopathy (DR) using optical coherence tomography (OCT) and OCT angiography (OCTA). En-face 3 × 3 mm OCTA images were obtained from 327 eyes of DR patients without macular edema. Nonperfusion squares (NPSs) were defined as 15×15-pixel regions lacking vascular signals. Neurovascular parameters were extracted from five subfields of the Early Treatment Diabetic Retinopathy Study grid. High-dimensional data were embedded into a two-dimensional space using Uniform Manifold Approximation and Projection, and clustering revealed three distinct groups: Mild , Intermediate , and Severe . Eyes with central subfield thickness (CST) < 246 µm were classified as having diabetic macular atrophy . The Mild group exhibited lower NPS counts, while the Intermediate group showed increased deep-layer ischemia. The Severe group had the highest NPS counts and the lowest CST, with a significant negative correlation between CST and superficial NPS counts ( ρ = − 0.252, P = 0.039). Eyes with diabetic macular atrophy in the Severe group demonstrated higher NPS counts, worse visual acuity, and more frequent ellipsoid zone disruption ( P < 0.001). These findings suggest a pathological relationship between macular ischemia and retinal atrophy, offering new insights into DR progression. Health sciences/Diseases Health sciences/Endocrinology Health sciences/Medical research diabetic macular atrophy diabetic macular ischemia diabetic retinopathy neurodegeneration uniform manifold approximation and projection Figures Figure 1 Figure 2 INTRODUCTION Diabetic retinopathy (DR) is a leading cause of vision loss among working-age populations. Degenerative processes and inflammatory responses reciprocally promote DR pathogenesis, leading to vision impairment. 1 In particular, vascular endothelial growth factor (VEGF) exacerbates angiogenesis and vascular permeability, both of which are hallmarks of disease activity, and anti-VEGF therapies stabilize active vascular lesions. 2 In contrast, degenerative processes, such as neurodegeneration and capillary nonperfusion, still exacerbate vision loss and remain inadequately understood. 3 – 5 The concept of diabetic retinal disease (DRD) encompasses the full spectrum of pathophysiological changes. 6 , 7 While DR is diagnosed based on vascular lesions, several clinical and biological studies suggest that neuronal dysfunction may precede the development of these vascular abnormalities, supporting the impairment of the neurovascular unit in DRD. 6 , 8 – 14 Hyperglycemia appears to exacerbate pathological processes in both vascular and neuronal cells, potentially mediated by glial cells and the extracellular matrix. Despite these findings, the clinical characteristics of neurovascular degeneration in DR remain poorly characterized. Advances in imaging technologies, such as optical coherence tomography (OCT) and OCT angiography (OCTA), have enabled detailed assessments of both retinal structure and the three-dimensional retinal vasculature. 15 – 17 OCT provides objective measurements of retinal thickness, facilitating the diagnosis of center-involving diabetic macular edema (DME). 18 In addition, OCT can be used to assess retinal degeneration in age-related macular degeneration and retinitis pigmentosa. OCT images offer a layer-by-layer view of retinal structural lesions. 11 , 14 , 19 , 20 In particular, photoreceptor integrity is represented by the external limiting membrane and ellipsoid zone (EZ). 16 , 21 OCTA allows us to evaluate capillary nonperfusion and diabetic macular ischemia (DMI) objectively. 5 , 22 – 28 These multimodal imaging techniques have enhanced our understanding of the complex pathogenesis of DME, DMI, and neurodegeneration. 6 , 13 , 29 – 33 Uniform manifold approximation and projection (UMAP) is a method for dimensionality reduction and visualization and has been used to analyze high-dimensional data in the biomedical fields. 34 , 35 Since the curse of dimensionality does not allow us to recognize high-dimensional data fully, dimensionality reduction improves our understanding of such datasets. In particular, UMAP has been used for single-cell transcriptome analysis to explore the organization of cell clusters, and has been recently applied to characterize the patterns of clinical features with high-dimensional data in retinopathy of prematurity, macular telangiectasia, and DR. 28 , 36 – 38 In this study, we investigate the simultaneous degeneration of the neurovascular unit in DR on structural OCT and OCTA images, based on statistical approaches and dimensionality reduction using UMAP. 34 METHODS Participants In this prospective study, we enrolled patients with DR without macular edema, who were examined at the Department of Ophthalmology, Kyoto University Hospital. The study adhered to the principles of the Declaration of Helsinki and was approved by the Ethics Committee of the Kyoto University Graduate School and Faculty of Medicine. Informed written consent was obtained from all participants prior to their inclusion in the study. We included eyes with DR that had swept-source (SS-)OCTA and spectral-domain (SD-)OCT images of sufficient quality. The inclusion criteria were as follows: the presence of DR, the absence of center-involving DME, acquisition of 3 × 3 mm SS-OCTA images centered on the fovea, acquisition of three-dimensional SD-OCT images centered on the fovea, acquisition of horizontal and vertical OCT sectional images through the fovea, and written informed consent. Exclusion criteria included axial length less than 22 mm or greater than 26 mm, severe media opacities, the presence of other ocular diseases that could affect visual acuity, including glaucoma, epiretinal membrane, and vitreomacular traction, or previous ocular treatments within 6 months prior to image acquisition, such as intraocular surgery, anti-VEGF injections, ocular steroids, or photocoagulation. Additional exclusion criteria included poor image quality (signal strength index of 7 or less), significant segmentation errors in the superficial slab of OCTA images, and inadequate quantification of retinal thickness in OCT images. If both eyes met the inclusion criteria, the right eye was selected for the study. Imaging Best-corrected decimal visual acuity (VA) was measured and then converted to the logarithm of the minimum angle of resolution (logMAR) for statistical analysis. A comprehensive ophthalmic examination was performed, and the severity of DR was classified according to the International Clinical Diabetic Retinopathy Severity Scales. 9 Axial length was measured using partial coherence interferometry (IOL Master, Carl Zeiss Meditec, Inc., Dublin, CA). SD-OCT images were acquired using the raster scan and cross-hair modes of the Spectralis OCT (Heidelberg Engineering, Heidelberg, Germany). The retinal thickness in five subfields of the Early Treatment Diabetic Retinopathy Study (ETDRS) grid (center [within 1 mm diameter], nasal, superior, temporal, and inferior parafoveal subfields [1–3 mm diameter]) was quantified based on the two-dimensional OCT map. Eyes with a central subfield thickness (CST) greater than 320 µm for males or 305 µm for females were diagnosed with center-involving DME. 18 According to previous studies, the CST in eyes with no or minimal DR is 270 ± 24 µm, so we selected the mean minus one (246 µm) or two (222 µm) standard deviation as the threshold. 18 In this study, we observed that eyes with a CST < 246 µm were less frequent and had a high percentage of reduced VA (Supplementary Fig. S1 ). Therefore, we chose 246 µm as the threshold to define diabetic macular atrophy . For qualitative analyses, the status of the EZ line was assessed on cross-sectional OCT images, as previously described. 21 SS-OCTA images were obtained within a 3 × 3 mm region centered on the fovea using the Plex Elite 9000 system (Carl Zeiss Meditec, Inc., Dublin, CA). To standardize the orientation, images of the left eye were horizontally inverted to ensure the nasal side was consistently aligned. The nominal 3 × 3 mm area was scanned with 300 × 300 A-scans, which were then digitally converted to a 1024 × 1024 pixel array for further analysis. Nonperfusion squares (NPSs) were defined as follows: 1) creation of en face images of the superficial and deep retinal layers, 2) vessel detection using edge detection algorithms, and 3) identification of squares (15 × 15 pixels) without visible retinal vessel edges. 28 The superficial and deep layers were defined from the inner limiting membrane to the boundary between the inner plexiform layer and inner nuclear layer, and from the inner border of the inner nuclear layer to 70 µm above the retinal pigment epithelium, respectively. Vessel edges were automatically detected using the Canny Edge Detector plugin in ImageJ software (NIH, http://imagej.nih.gov/ij/) . 39 The circular area with a 2.5 mm diameter centered on the fovea was divided into 15 × 15 pixel squares, and those without visible vascular edges were defined as NPSs. We counted NPSs within five subfields of the ETDRS grid, including the central 1-mm area and the four parafoveal sectors (1–2.5 mm). Uniform manifold approximation and projection We investigated the impairment of the neurovascular unit in DRD based on capillary nonperfusion and retinal thicknesses using complex and high-dimensional datasets. We applied UMAP for dimensionality reduction and visualization and subsequently demonstrated the arrangement of eyes according to the similarity of nonperfusion areas and retinal thinning in a two-dimensional space. Briefly, we employed four steps for dimensionality reduction and subsequent clustering: 1. quantification and normalization of 15 variables related to capillary nonperfusion and retinal thicknesses, 2. principal component analysis (PCA) for variance-based dimensionality reduction, 3. UMAP for mapping each eye onto a two-dimensional space based on similarity, 4. subsequent k-means clustering. We selected 15 parameters for each eye: the NPS counts in the superficial and deep layers for each of the five ETDRS subfields (a total of 10 parameters), and the mean retinal thickness in the five subfields (a total of 5 parameters). These parameters represent both vascular and structural aspects of the retina, allowing for an integrative analysis of neurovascular degeneration. We first applied PCA, retaining over 90% of the total variance, which resulted in seven principal components. We then employed UMAP to project eyes with these components into a two-dimensional space. 34 UMAP is designed to preserve the local and global structure of high-dimensional data, making it suitable for identifying patterns and relationships in complex datasets. 34 Each eye was thus represented as a point in the two-dimensional UMAP space, with proximity between points indicating similarity in neurovascular parameters. After the optimal number of clusters was determined utilizing the elbow method, we performed k-means clustering on the two-dimensional UMAP representation to categorize eyes into groups with similar neurovascular characteristics. The clusters were referred to as Mild , Intermediate , and Severe based on the NPS counts and retinal thickness values. Statistics All values were expressed as the median (interquartile range). Statistical significance was set at P < 0.05. Normality was assessed using the Shapiro-Wilk test. For continuous variables, the Mann-Whitney U-test or Kruskal-Wallis test with Bonferroni correction was used, while categorical variables were analyzed using the Fisher’s exact test or Chi-square test. Spearman's rank correlation coefficient was used to assess statistical associations. All statistical analyses were conducted using SPSS software (version 24; IBM). RESULTS Severity scale of neurovascular pathology based on UMAP In this cross-sectional study, we analyzed 327 eyes from 327 DR patients without macular edema. The patients’ characteristics are shown in Supplementary Table S1 . All DR eyes were mapped onto the two-dimensional UMAP space according to the similarity of neurovascular parameters (Fig. 1 ). The arrangement of DR eyes based on DMI and retinal thinning in the macula suggested a continuous severity scale of neurovascular pathology (Fig. 2 A-C). Cases in the lower left corner had smaller amounts of NPSs in both layers and normal CST. Deep NPSs increased in eyes located in the upper areas of the UMAP space. Heatmaps demonstrated that eyes in the right regions of the UMAP space had higher NPS counts in both layers and simultaneously smaller CST values. Additionally, these eyes also exhibited poorer logMAR (Fig. 2 D). Interestingly, CST decreased in a subgroup of eyes located in the center of the UMAP space, despite no increase in superficial NPS counts (Fig. 2 C). DR severity was also illustrated in Fig. 2 E. NPS counts and mean retinal thicknesses for each subfield were shown in Supplementary Fig. S2 -S4. Subsequent clustering using the k-means method divided the 327 cases into three distinct clusters (Fig. 2 F; Table 1 ), representing discrete grades referred to as Mild , Intermediate , and Severe groups based on NPS counts. The number of superficial and deep NPSs differed between the three groups (Table 1 ). The central subfield of the superficial layer showed similar trends. In the parafoveal sectors of the superficial layer, there was a slight difference in NPS amounts between the Mild and Intermediate groups, whereas the Severe group exhibited much higher NPS counts (Supplementary Table S2 ). CST decreased in a stepwise manner from the Mild to the Severe group (Table 1 ). Eyes in the Severe group had thinner retinal thickness in the parafoveal sectors compared to those in the Mild and Intermediate groups. In contrast, there were no significant differences or only minimal differences between eyes in the Mild and Intermediate groups (Supplementary Table S2 ). Table 1 Comparisons of clinical parameters among three groups based on uniform manifold approximation and projection and subsequent clustering. Variables Mild (n = 131) Intermediate (n = 129) Severe (n = 67) P -value Age (years) 64 (51–71) 66 (55–74) 70 (59–77.5) * 0.009 Sex (male/female) 82/49 78/51 47/20 0.401 Hemoglobin A1c (%) 7.8 (6.8–9.0) 7.6 (6.9–8.5) 7.1 (6.6–8.2) 0.122 Duration of diabetes (years) 15 (8.5–23) 17 (10–24) 14.5 (10–22.8) 0.730 Systemic hypertension (present/absent) 73/58 88/41 46/21 0.067 Dyslipidemia (present/absent) 61/70 58/71 30/37 0.957 LogMAR -0.079 (-0.128–0.000) 0.000 (-0.079–0.097) 0.301 (0.097–0.699) *† < 0.001 Phakia/pseudophakia 74/57 68/61 12/55 < 0.001 International DR severity grade (eyes) Mild NPDR 11 2 0 0.004 Moderate NPDR 72 65 35 Severe NPDR 13 7 2 PDR 35 55 30 Diabetic macular atrophy (present/absent) 8/123 24/105 37/30 < 0.001 Central subfield thickness (µm) 283 (268–297) 272 (254–289) * 240 (190–265) *† < 0.001 NPS counts in the superficial layer 207 (167–278) 368 (284–423) * 871 (665–1007) *,† < 0.001 NPS counts in the deep layer 630 (507–713) 1000 (878–1127) * 1384 (1230–1644) *† < 0.001 Prior PRP (present/absent) 50/81 68/61 63/4 < 0.001 Prior STTA (present/absent) 11/120 7/122 11/56 0.036 Prior anti-VEGF injection (present/absent) 15/116 14/115 9/58 0.864 Prior vitrectomy (present/absent) 14/117 16/113 20/47 < 0.001 Data are shown as numbers or median (interquartile range). Abbreviations: DR = diabetic retinopathy; NPDR = nonproliferative diabetic retinopathy; PDR = proliferative diabetic retinopathy; PRP = panretinal photocoagulation; STTA = subTenon’s injection of triamcinolone acetonide. * P < 0.05 vs. Mild; † P < 0.01 vs. Moderate. Clinical characteristics in each cluster In the Severe group, CST was negatively correlated with NPS counts in the superficial layer, but not in the deep layer (Table 2 ). LogMAR showed a positive correlation with NPS counts in the superficial layer and a negative correlation with CST, while no significant association was observed between logMAR and NPS counts in the deep layer (Table 2 ). Table 2 Associations between neurovascular parameters within each group based on uniform manifold approximation and projection and subsequent clustering. Mild Intermediate Severe Association with central subfield thickness NPS counts in the superficial layer ρ =-0.111 P = 0.206 ρ = 0.068 P = 0.441 ρ =-0.252 P = 0.039 NPS counts in the deep layer ρ =-0.054 P = 0.537 ρ = 0.004 P = 0.962 ρ =-0.235 P = 0.056 Association with logMAR NPS counts in the superficial layer ρ = 0.285 P = 0.001 ρ = 0.286 P = 0.001 ρ = 0.435 P < 0.001 NPS counts in the deep layer ρ = 0.136 P = 0.121 ρ = 0.219 P = 0.013 ρ = 0.159 P = 0.200 Central subfield thickness ρ =-0.005 P = 0.953 ρ = 0.094 P = 0.290 ρ =-0.351 P = 0.004 Data are shown as numbers or median (interquartile range). Abbreviations: logMAR = logarithm of the minimum angle of resolution; NPS = nonperfusion square. In the Mild and Intermediate groups, logMAR was positively correlated with NPS counts in the superficial layer, but not with CST (Table 2 ). In the intermediate group, there was a modest association between NPS counts in the deep layer and logMAR. Two subclasses of diabetic macular atrophy Eyes with a thinner CST were mapped to two distinct areas in the UMAP space: the center and the right region of the UMAP space (Fig. 2 C). The Severe group (55.2%) had a higher percentage of eyes with diabetic macular atrophy (CST < 246 µm) than the Mild (6.1%) and Intermediate (18.6%) groups (Table 1 ). Based on these results, we proposed the existence of two subgroups of diabetic macular atrophy and compared several parameters between these two groups (Table 3 ). Eyes in the center of the UMAP space belonged to the Mild and Intermediate groups and had fewer NPS counts and a continuous EZ line more frequently (Table 3 ). In contrast, eyes mapped to the right areas were included in the Severe group, exhibiting higher NPS counts, worse logMAR, and more frequent disruption of the EZ line. Alternative thresholding of CST (< 222 µm) showed similar results (Supplementary Table S3). Table 3 Comparisons of clinical parameters between two subgroups of eyes with diabetic macular atrophy . Mild + Intermediate (n = 32) Severe (n = 37) P -value NPS counts in the superficial layer 383 (326–456) 931 (696–1051) < 0.001 NPS counts in the deep layer 931 (721–1120) 1480 (1232–1767) < 0.001 Central subfield thickness (µm) 229 (210–237) 192 (163–216) < 0.001 EZ line (continuous/not continuous) 26/6 10/27 < 0.001 logMAR 0.023 (-0.079–0.111) 0.398 (0.222–0.699) < 0.001 Data are shown as numbers or median (interquartile range). Abbreviations: CST = central subfield thickness; EZ = ellipsoid zone; logMAR = logarithm of the minimum angle of resolution; NPS = nonperfusion square. DISCUSSION In this study, we investigated capillary nonperfusion and retinal thicknesses in the macula of DR eyes without macular edema. Statistical analyses revealed modest associations between NPS amounts and CST, while dimensionality reduction using UMAP highlighted various pathological patterns in the neurovascular unit. Some eyes exhibited greater amounts of deep NPSs without a decrease in CST, while others showed CST reduction with few NPSs. Additionally, certain cases presented with high NPS counts in both layers, retinal thinning, and concomitant VA reduction. We hypothesize that integrative analyses of neurovascular structures provide a more comprehensive understanding of retinal pathogenesis affecting visual function in DRD, compared to assessments based solely on a single parameter, such as capillary nonperfusion or neurodegeneration. Our study utilizing UMAP highlights the clinical features of DRD, in which both vascular and neuronal pathological mechanisms contribute to vision loss. 6 Previously, it was believed that vascular hyperpermeability and capillary nonperfusion lead to neuronal dysfunction and subsequent visual impairment. 22 , 40 Investigations using structural OCT have demonstrated novel imaging biomarkers of retinal neurodegeneration, such as disorganization of retinal inner layers and EZ disruption, in eyes with DME. 20 , 21 Recent statistical analyses have also reported associations between capillary nonperfusion and neuronal biomarkers in eyes with DMI. 4 , 29 – 32 These findings raise a key question: which comes first, vascular lesions or neurodegeneration? Or what are the common mechanisms for neurovascular degeneration? Our results suggest novel clinical entities, e.g., diabetic macular atrophy, DMI alone, and both atrophy and ischemia, in DRD. Furthermore, the distribution of cases in the two-dimensional UMAP space may represent a continuous and integrated severity scale of neurovascular degeneration. Interestingly, CST was negatively correlated with NPS counts in the superficial layer in eyes of the Severe group. Photoreceptors, which are the main component of the central subfield, are nourished by the deep retinal capillaries and the choroid. We applied the default setting for the superficial layer to create en face OCTA images, which include all retinal capillary layers at the fovea. The capillaries in the superficial and deep layers are fused in the perifoveal capillary network, which is visualized in superficial en face OCTA images. 41 This suggests simultaneous loss of both vascular and neuronal components at the fovea, and is consistent with previous findings that photoreceptor damage is associated with deep capillary nonperfusion. 30 In this study, we defined eyes with CST less than 246 µm as diabetic macular atrophy , which were divided into two subgroups. One subgroup was mapped to the center of the UMAP space, while the other was located in the right regions. These two groups exhibited differences in several neurovascular parameters and logMAR. Eyes mapped to the right regions of the UMAP showed significant capillary nonperfusion in both the superficial and deep layers as well as photoreceptor disruption. This suggests that detrimental effects may mutually promote the impairment of the neurovascular unit or that common mechanisms could lead to simultaneous degeneration. In contrast, we could not identify definite retinal layer loss in eyes mapped to the center of the UMAP. These eyes demonstrated proportional retinal thinning and an enlarged foveal pit, which led us to hypothesize that the loss or volume reduction of Müller cells contributes to retinal thinning. Several studies have documented glial fibrillary acidic protein expression in Müller cells of diabetic retinas, 42 but the pathogenesis in these eyes remains to be clarified. In addition, we selected the mean minus one standard deviation, because of the frequency and the percentages of VA reduction in each group of retinal thicknesses in our single center (Supplementary Fig. S1 ). Future multi-center studies should determine the better thresholding to define diabetic macular atrophy . Regarding the question of defining thresholds for each severity scale for NPS counts in the superficial layer or the deep layer and retinal thicknesses, our study takes a different approach. Traditionally, diseases have often been diagnosed using a single parameter, e.g., CST. However, neurovascular impairment in DR cannot be adequately represented by a single parameter due to its complex, multifaceted nature. In our study, we employed UMAP to integrate vascular and neuronal parameters into a two-dimensional space. By evaluating the multifaceted disease in a multidimensional manner, we propose an integrated severity scale for neurovascular impairment in DRD. This approach allows for a more comprehensive assessment of the neurovascular unit, capturing the interplay between capillary nonperfusion and retinal structural changes. We demonstrated a continuous severity scale of neurovascular degeneration in DR. This scale could provide insights into the progression pathways of the disease, though longitudinal studies are needed to confirm this hypothesis. Furthermore, novel biological techniques, such as single-cell transcriptome and proteome analyses, could elucidate the pathological interactions between vascular cells, neurons, glia, and the extracellular matrix in the neurovascular unit. 43 , 44 These findings may provide useful information for the development of novel therapeutic strategies for DRD. There are several limitations to this study. The criteria for enrollment in this single-center study may result in selection bias. In particular, we excluded eyes with center-involving DME, so the clinical characteristics of neurovascular degeneration in DME remain to be elucidated. We do not have a complete solution for artifacts in OCTA images, which might affect the vascular parameters. 45 The quantification of retinal parameters was performed using specific devices and image processing methods, and future multicenter studies employing alternative methodologies are needed to confirm the generalizability of these findings. Additionally, while we used UMAP and the k-means method for dimensionality reduction and clustering, other algorithms may reveal different features. 46 A comprehensive assessment of visual function is also necessary to further explore the clinical relevance in DRD. In conclusion, we demonstrated the clinical characteristics of the neurovascular unit in DR eyes without macular edema. This study contributes to a deeper understanding of the pathogenesis of DRD. Declarations Competing Interests The authors declare no competing interests. Author Contribution M.Y. and T.M. contributed to the conceptualization, study design, and manuscript drafting. All authors contributed to data acquisition. T.M. and A.T. supervised the study. All authors reviewed and approved the final version of the manuscript. Acknowledgements This work was supported by a Grant-in-Aid for Scientific Research from the Japan Society for the Promotion of Science (Grant Number: 23K09004). The funding organization had no role in the design or conduct of this study. Data Availability The raw data are provided by the corresponding author upon reasonable request. References Antonetti, D. A., Klein, R. & Gardner, T. Mechanisms of disease: diabetic retinopathy. N Engl. J. Med. 366 , 1227–1239 (2012). Aiello, L. P. et al. Vascular endothelial growth factor in ocular fluid of patients with diabetic retinopathy and other retinal disorders. N Engl. J. Med. 331 , 1480–1487 (1994). Barber, A. J. et al. Neural apoptosis in the retina during experimental and human diabetes: early onset and effect of insulin. J. Clin. Invest. 102 , 783–791 (1998). Kim, K., Kim, E. S. & Yu, S. Y. Longitudinal relationship between retinal diabetic neurodegeneration and progression of diabetic retinopathy in patients with type 2 diabetes. Am. J. Ophthalmol. 196 , 165–172 (2018). Cheung, C. M. G. et al. Diabetic macular ischaemia–a new therapeutic target? Prog Retin Eye Res. 89 , 101033 (2022). Sun, J. K. et al. Updating the staging system for diabetic retinal disease. Ophthalmology 128 , 490–493 (2021). Levine, S. R., Sapieha, P., Dutta, S., Sun, J. K. & Gardner, T. W. It is time for a moonshot to find cures for diabetic retinal disease. Prog Retin Eye Res. 90 , 101051 (2022). Barber, A. J., Antonetti, D. A. & Gardner, T. W. Altered expression of retinal occludin and glial fibrillary acidic protein in experimental diabetes. Invest. Ophthalmol. Vis. Sci. 41 , 3561–3568 (2000). Wilkinson, C. P. et al. Proposed international clinical diabetic retinopathy and diabetic macular edema disease severity scales. Ophthalmology 110 , 1677–1682 (2003). Lo, E. H., Dalkara, T. & Moskowitz, M. A. Neurological diseases: mechanisms, challenges and opportunities in stroke. Nat. Rev. Neurosci. 4 , 399–415 (2003). Sohn, E. H. et al. Retinal neurodegeneration may precede microvascular changes characteristic of diabetic retinopathy in diabetes mellitus. Proc. Natl. Acad. Sci. U.S.A. 113, E2655–E2664 (2016). Simó, R., Stitt, A. W. & Gardner, T. W. Neurodegeneration in diabetic retinopathy: does it really matter? Diabetologia 61, 1902–1912 (2018). Abràmoff, M. D. et al. Approach for a clinically useful comprehensive classification of vascular and neural aspects of diabetic retinal disease. Invest. Ophthalmol. Vis. Sci. 59 , 519–527 (2018). Tang, Z. et al. Assessment of retinal neurodegeneration with spectral-domain optical coherence tomography: a systematic review and meta-analysis. Eye 35 , 1317–1325 (2021). Jia, Y. et al. Split-spectrum amplitude-decorrelation angiography with optical coherence tomography. Opt. Express . 20 , 4710–4725 (2012). Staurenghi, G., Sadda, S., Chakravarthy, U. & Spaide, R. F. International Nomenclature for Optical Coherence Tomography (IN•OCT) Panel. Proposed lexicon for anatomic landmarks in normal posterior segment spectral-domain optical coherence tomography: the IN•OCT consensus. Ophthalmology 121 , 1572–1578 (2014). Spaide, R. F., Klancnik, J. M. & Cooney, M. J. Retinal vascular layers imaged by fluorescein angiography and optical coherence tomography angiography. JAMA Ophthalmol. 133 , 45–50 (2015). Chalam, K. V. et al. Retinal thickness in people with diabetes and minimal or no diabetic retinopathy: Heidelberg spectralis optical coherence tomography. Invest. Ophthalmol. Vis. Sci. 53 , 8154–8161 (2012). Murakami, T. & Yoshimura, N. Structural changes in individual retinal layers in diabetic macular edema. J. Diabetes Res. 920713 (2013). (2013). Sun, J. K. et al. Disorganization of the retinal inner layers as a predictor of visual acuity in eyes with center-involved diabetic macular edema. JAMA Ophthalmol. 132 , 1309–1316 (2014). Murakami, T. et al. Optical coherence tomographic reflectivity of photoreceptors beneath cystoid spaces in diabetic macular edema. Invest. Ophthalmol. Vis. Sci. 53 , 1506–1511 (2012). Early Treatment Diabetic Retinopathy Study Research Group. Classification of diabetic retinopathy from fluorescein angiograms: ETDRS Report Number 11. Ophthalmology 98, 807–822. (1991). Samara, W. A. et al. Quantification of diabetic macular ischemia using optical coherence tomography angiography and its relationship with visual acuity. Ophthalmology 124 , 235–244 (2017). Kaizu, Y. et al. Optical coherence tomography angiography reveals spatial bias of macular capillary dropout in diabetic retinopathy. Invest. Ophthalmol. Vis. Sci. 58 , 4889 (2017). Tsai, A. S. H. et al. Diabetic macular ischemia: correlation of retinal vasculature changes by optical coherence tomography angiography and functional deficit. Retina 40 , 2184–2190 (2020). Terada, N. et al. Clinical relevance of parafoveal intercapillary spaces and foveal avascular zone in diabetic retinopathy without macular edema. Invest. Ophthalmol. Vis. Sci. 63 , 4 (2022). Yang, D. et al. Assessment of parafoveal diabetic macular ischemia on optical coherence tomography angiography images to predict diabetic retinal disease progression and visual acuity deterioration. JAMA Ophthalmol. 141 , 641–649 (2023). Yoshida, M. et al. Severity scale of diabetic macular ischemia based on the distribution of capillary nonperfusion in OCT angiography. Ophthalmol. Sci. 5 , 100603 (2025). Sim, D. A. et al. Quantitative analysis of diabetic macular ischemia using optical coherence tomography. Invest. Ophthalmol. Vis. Sci. 55 , 417–423 (2014). Scarinci, F., Jampol, L. M., Linsenmeier, R. A. & Fawzi, A. A. Association of diabetic macular nonperfusion with outer retinal disruption on optical coherence tomography. JAMA Ophthalmol. 133 , 1036–1044 (2015). Dodo, Y. et al. Diabetic neuroglial changes in the superficial and deep nonperfused areas on optical coherence tomography angiography. Invest. Ophthalmol. Vis. Sci. 58 , 5870–5879 (2017). Tsai, W. S. et al. Characterization of the structural and functional alteration in eyes with diabetic macular ischemia. Ophthalmol. Retina . 7 , 142–152 (2023). Sampani, K. et al. Association of retinal oximetry with peripheral diabetic retinopathy lesions and nonperfusion on ultra-widefield angiography. Ophthalmol. Sci. 5 , 100686 (2024). McInnes, L., Healy, J. & Melville, J. Uniform manifold approximation and projection for dimension reduction. arXiv Preprint . arXiv , 180203426 (2018). Becht, E. et al. Dimensionality reduction for visualizing single-cell data using UMAP. Nat Biotechnol (2018). Coyner, A. S. et al. Synthetic medical images for robust, privacy-preserving training of artificial intelligence. Ophthalmol. Sci. 2 , 100126 (2022). Yoshida, M. et al. Inference of capillary nonperfusion progression on widefield OCT angiography in diabetic retinopathy. Invest. Ophthalmol. Vis. Sci. 64 , 24 (2023). Wu, Y. et al. Developing a continuous severity scale for macular telangiectasia type 2 using deep learning and implications for disease grading. Ophthalmology 131 , 219–226 (2024). Canny, J. A computational approach to edge detection. IEEE Trans. Pattern Anal. Mach. Intell. 8 , 679–698 (1986). Early Treatment Diabetic Retinopathy Study Research Group. Photocoagulation for diabetic macular edema: Early Treatment Diabetic Retinopathy Study Report Number 1. Arch. Ophthalmol. 103 , 1796–1806 (1985). Snodderly, D. M., Weinbaus, R. S. & Choi, J. C. Neural–vascular relationships in central retina of macaque monkeys (Macaca fascicularis). J. Neurosci. 12 , 1169–1193 (1992). Mizutani, M., Gerhardinger, C. & Lorenzi, M. Muller cell changes in human diabetic retinopathy. Diabetes 47 , 445–449 (1998). Duh, E. J., Sun, J. K. & Stitt, A. W. Diabetic retinopathy: current understanding, mechanisms, and treatment strategies. JCI Insight . 2 , e93751 (2017). Van Hove, I. et al. Single-cell transcriptome analysis of the Akimba mouse retina reveals cell-type-specific insights into the pathobiology of diabetic retinopathy. Diabetologia 63 , 2235–2248 (2020). Spaide, R. F., Fujimoto, J. G. & Waheed, N. K. Image artifacts in optical coherence tomography angiography. Retina 35 , 2163–2180 (2015). Van Der Maaten, L. & Hinton, G. Visualizing data using t-SNE. J. Mach. Learn. Res. 9 , 2579–2605 (2008). Additional Declarations No competing interests reported. Supplementary Files SupplementaryFigures.pdf SupplementaryTables.pdf Cite Share Download PDF Status: Published Journal Publication published 11 Dec, 2025 Read the published version in Scientific Reports → Version 1 posted Editorial decision: Revision requested 22 Jul, 2025 Reviews received at journal 20 Jul, 2025 Reviews received at journal 17 Jul, 2025 Reviewers agreed at journal 15 Jul, 2025 Reviewers agreed at journal 15 Jul, 2025 Reviewers invited by journal 15 Jul, 2025 Editor assigned by journal 15 Jul, 2025 Editor invited by journal 08 Jul, 2025 Submission checks completed at journal 02 Jul, 2025 First submitted to journal 02 Jul, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7000479","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":486999135,"identity":"bee44cc8-d362-4dc9-8049-4e31c35a48ac","order_by":0,"name":"Miyo Yoshida","email":"","orcid":"","institution":"Kyoto University Graduate School of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Miyo","middleName":"","lastName":"Yoshida","suffix":""},{"id":486999136,"identity":"65eb5caf-6ea8-4c62-ad72-d2a72f45acc4","order_by":1,"name":"Tomoaki Murakami","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA9klEQVRIiWNgGAWjYPACGwMGCQYGxgYkIWbcqsFSaXAtEsRqOYypBSfQnZF/8NGNivPGBrd7DzDObLOp42fgMWD4UcPAbo5Di9mNZGbjnDO3zQzunEtg3NiWJiHZwGPA2HOMgdmyAacWNunctts2BjdyDBgfth2WMLj/xoCBt4GB2eAAXi3nkLQcANryl7CWA2ZgLRuhWpjx2nLmsTHQL8nGkjfyEg7OOJcmObOBreCwzDEJ3H45nvjwcU6FnWHfjdyDD3vKbPj5GZg3PnxTY5OMK8SQAA/DAUY2CBPoJIlkA2K0MDD8QXDtiNAyCkbBKBgFIwMAAOvNV45kFUVqAAAAAElFTkSuQmCC","orcid":"","institution":"Kyoto University Graduate School of Medicine","correspondingAuthor":true,"prefix":"","firstName":"Tomoaki","middleName":"","lastName":"Murakami","suffix":""},{"id":486999137,"identity":"439500ec-8ddd-459f-a603-fcb9baf31c13","order_by":2,"name":"Kenji Ishihara","email":"","orcid":"","institution":"Kyoto University Graduate School of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Kenji","middleName":"","lastName":"Ishihara","suffix":""},{"id":486999138,"identity":"fb2f5309-b040-475f-8e01-2b87be6ccdc1","order_by":3,"name":"Yuki Mori","email":"","orcid":"","institution":"Kyoto University Graduate School of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Yuki","middleName":"","lastName":"Mori","suffix":""},{"id":486999139,"identity":"96b2db5d-07fc-47fc-99a3-a71009760468","order_by":4,"name":"Akitaka Tsujikawa","email":"","orcid":"","institution":"Kyoto University Graduate School of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Akitaka","middleName":"","lastName":"Tsujikawa","suffix":""}],"badges":[],"createdAt":"2025-06-29 03:38:12","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7000479/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7000479/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41598-025-31862-w","type":"published","date":"2025-12-11T15:59:00+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":87272982,"identity":"e8669293-cd29-4fbe-a6e2-c8745755b471","added_by":"auto","created_at":"2025-07-22 08:34:15","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":841867,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMapping all 327 eyes with diabetic retinopathy onto the two-dimensional Uniform Manifold Approximation and Projection (UMAP) space based on capillary nonperfusion and retinal thicknesses.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eEach eye is represented by a dot in the UMAP space, and four representative cases are shown. The upper left, upper middle, upper right, and lower panels in each eye correspond to the superficial optical coherence tomography angiography (OCTA), deep OCTA, two-dimensional optical coherence tomography (OCT) map, and sectional OCT images, respectively.\u003c/p\u003e","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-7000479/v1/1ed8c0f86aefb70008e80195.png"},{"id":87272985,"identity":"403fbc8b-9b7b-43ee-a27b-fab31e819645","added_by":"auto","created_at":"2025-07-22 08:34:15","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":661875,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eHeatmap of each parameter on the two-dimensional Uniform Manifold Approximation and Projection (UMAP) space.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNonperfusion square (NPS) counts in the superficial (A) and deep (B) layers, central subfield thickness (C), and logarithm of the minimum angle of resolution (logMAR; D) are represented by pseudocolor scales. (E) International diabetic retinopathy severity grades. (F) The three groups are identified by k-means clustering.\u003c/p\u003e","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-7000479/v1/384f87d7cdddd882d4e4c4f8.png"},{"id":98244743,"identity":"a75f8a70-e05c-4796-a017-95db8dc75ffb","added_by":"auto","created_at":"2025-12-15 16:14:54","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2360091,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7000479/v1/c48a719b-2983-4480-a22f-f52abaca550e.pdf"},{"id":87272289,"identity":"fb5ea5ef-2a6f-4610-9547-ba42da6ea8f7","added_by":"auto","created_at":"2025-07-22 08:26:15","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":5908946,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryFigures.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7000479/v1/155b82389874813129bf82f2.pdf"},{"id":87272278,"identity":"854dceef-1605-4b89-94d1-e33077f67b16","added_by":"auto","created_at":"2025-07-22 08:26:15","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":70959,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryTables.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7000479/v1/6c4ad7884b607c2e8c55acf6.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Integrative Severity Scale for Diabetic Macular Atrophy and Ischemia Using Structural OCT and OCT Angiography","fulltext":[{"header":"INTRODUCTION","content":"\u003cp\u003eDiabetic retinopathy (DR) is a leading cause of vision loss among working-age populations. Degenerative processes and inflammatory responses reciprocally promote DR pathogenesis, leading to vision impairment.\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e In particular, vascular endothelial growth factor (VEGF) exacerbates angiogenesis and vascular permeability, both of which are hallmarks of disease activity, and anti-VEGF therapies stabilize active vascular lesions.\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e In contrast, degenerative processes, such as neurodegeneration and capillary nonperfusion, still exacerbate vision loss and remain inadequately understood.\u003csup\u003e\u003cspan additionalcitationids=\"CR4\" citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\u003cp\u003eThe concept of diabetic retinal disease (DRD) encompasses the full spectrum of pathophysiological changes.\u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e,\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e While DR is diagnosed based on vascular lesions, several clinical and biological studies suggest that neuronal dysfunction may precede the development of these vascular abnormalities, supporting the impairment of the neurovascular unit in DRD.\u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e,\u003cspan additionalcitationids=\"CR9 CR10 CR11 CR12 CR13\" citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e Hyperglycemia appears to exacerbate pathological processes in both vascular and neuronal cells, potentially mediated by glial cells and the extracellular matrix. Despite these findings, the clinical characteristics of neurovascular degeneration in DR remain poorly characterized.\u003c/p\u003e\u003cp\u003eAdvances in imaging technologies, such as optical coherence tomography (OCT) and OCT angiography (OCTA), have enabled detailed assessments of both retinal structure and the three-dimensional retinal vasculature.\u003csup\u003e\u003cspan additionalcitationids=\"CR16\" citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e OCT provides objective measurements of retinal thickness, facilitating the diagnosis of center-involving diabetic macular edema (DME).\u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e In addition, OCT can be used to assess retinal degeneration in age-related macular degeneration and retinitis pigmentosa. OCT images offer a layer-by-layer view of retinal structural lesions.\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e,\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e,\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e,\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e In particular, photoreceptor integrity is represented by the external limiting membrane and ellipsoid zone (EZ).\u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e,\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e OCTA allows us to evaluate capillary nonperfusion and diabetic macular ischemia (DMI) objectively.\u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e,\u003cspan additionalcitationids=\"CR23 CR24 CR25 CR26 CR27\" citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e These multimodal imaging techniques have enhanced our understanding of the complex pathogenesis of DME, DMI, and neurodegeneration.\u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e,\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e,\u003cspan additionalcitationids=\"CR30 CR31 CR32\" citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\u003cp\u003eUniform manifold approximation and projection (UMAP) is a method for dimensionality reduction and visualization and has been used to analyze high-dimensional data in the biomedical fields.\u003csup\u003e\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e,\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e Since the curse of dimensionality does not allow us to recognize high-dimensional data fully, dimensionality reduction improves our understanding of such datasets. In particular, UMAP has been used for single-cell transcriptome analysis to explore the organization of cell clusters, and has been recently applied to characterize the patterns of clinical features with high-dimensional data in retinopathy of prematurity, macular telangiectasia, and DR.\u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e,\u003cspan additionalcitationids=\"CR37\" citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\u003cp\u003eIn this study, we investigate the simultaneous degeneration of the neurovascular unit in DR on structural OCT and OCTA images, based on statistical approaches and dimensionality reduction using UMAP.\u003csup\u003e\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e"},{"header":"METHODS","content":"\u003cp\u003e\u003cb\u003eParticipants\u003c/b\u003e\u003c/p\u003e\u003cp\u003eIn this prospective study, we enrolled patients with DR without macular edema, who were examined at the Department of Ophthalmology, Kyoto University Hospital. The study adhered to the principles of the Declaration of Helsinki and was approved by the Ethics Committee of the Kyoto University Graduate School and Faculty of Medicine. Informed written consent was obtained from all participants prior to their inclusion in the study.\u003c/p\u003e\u003cp\u003eWe included eyes with DR that had swept-source (SS-)OCTA and spectral-domain (SD-)OCT images of sufficient quality. The inclusion criteria were as follows: the presence of DR, the absence of center-involving DME, acquisition of 3 \u0026times; 3 mm SS-OCTA images centered on the fovea, acquisition of three-dimensional SD-OCT images centered on the fovea, acquisition of horizontal and vertical OCT sectional images through the fovea, and written informed consent. Exclusion criteria included axial length less than 22 mm or greater than 26 mm, severe media opacities, the presence of other ocular diseases that could affect visual acuity, including glaucoma, epiretinal membrane, and vitreomacular traction, or previous ocular treatments within 6 months prior to image acquisition, such as intraocular surgery, anti-VEGF injections, ocular steroids, or photocoagulation. Additional exclusion criteria included poor image quality (signal strength index of 7 or less), significant segmentation errors in the superficial slab of OCTA images, and inadequate quantification of retinal thickness in OCT images. If both eyes met the inclusion criteria, the right eye was selected for the study.\u003c/p\u003e\u003cp\u003e\u003cb\u003eImaging\u003c/b\u003e\u003c/p\u003e\u003cp\u003eBest-corrected decimal visual acuity (VA) was measured and then converted to the logarithm of the minimum angle of resolution (logMAR) for statistical analysis. A comprehensive ophthalmic examination was performed, and the severity of DR was classified according to the International Clinical Diabetic Retinopathy Severity Scales.\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e Axial length was measured using partial coherence interferometry (IOL Master, Carl Zeiss Meditec, Inc., Dublin, CA). SD-OCT images were acquired using the raster scan and cross-hair modes of the Spectralis OCT (Heidelberg Engineering, Heidelberg, Germany). The retinal thickness in five subfields of the Early Treatment Diabetic Retinopathy Study (ETDRS) grid (center [within 1 mm diameter], nasal, superior, temporal, and inferior parafoveal subfields [1\u0026ndash;3 mm diameter]) was quantified based on the two-dimensional OCT map.\u003c/p\u003e\u003cp\u003eEyes with a central subfield thickness (CST) greater than 320 \u0026micro;m for males or 305 \u0026micro;m for females were diagnosed with center-involving DME.\u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e According to previous studies, the CST in eyes with no or minimal DR is 270\u0026thinsp;\u0026plusmn;\u0026thinsp;24 \u0026micro;m, so we selected the mean minus one (246 \u0026micro;m) or two (222 \u0026micro;m) standard deviation as the threshold.\u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e In this study, we observed that eyes with a CST\u0026thinsp;\u0026lt;\u0026thinsp;246 \u0026micro;m were less frequent and had a high percentage of reduced VA (Supplementary Fig. \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). Therefore, we chose 246 \u0026micro;m as the threshold to define \u003cem\u003ediabetic macular atrophy\u003c/em\u003e. For qualitative analyses, the status of the EZ line was assessed on cross-sectional OCT images, as previously described.\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\u003cp\u003eSS-OCTA images were obtained within a 3 \u0026times; 3 mm region centered on the fovea using the Plex Elite 9000 system (Carl Zeiss Meditec, Inc., Dublin, CA). To standardize the orientation, images of the left eye were horizontally inverted to ensure the nasal side was consistently aligned. The nominal 3 \u0026times; 3 mm area was scanned with 300 \u0026times; 300 A-scans, which were then digitally converted to a 1024 \u0026times; 1024 pixel array for further analysis. Nonperfusion squares (NPSs) were defined as follows: 1) creation of en face images of the superficial and deep retinal layers, 2) vessel detection using edge detection algorithms, and 3) identification of squares (15 \u0026times; 15 pixels) without visible retinal vessel edges.\u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e The superficial and deep layers were defined from the inner limiting membrane to the boundary between the inner plexiform layer and inner nuclear layer, and from the inner border of the inner nuclear layer to 70 \u0026micro;m above the retinal pigment epithelium, respectively. Vessel edges were automatically detected using the Canny Edge Detector plugin in ImageJ software (NIH, \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://imagej.nih.gov/ij/)\u003c/span\u003e\u003cspan address=\"http://imagej.nih.gov/ij/)\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003csup\u003e39\u003c/sup\u003e The circular area with a 2.5 mm diameter centered on the fovea was divided into 15 \u0026times; 15 pixel squares, and those without visible vascular edges were defined as NPSs. We counted NPSs within five subfields of the ETDRS grid, including the central 1-mm area and the four parafoveal sectors (1\u0026ndash;2.5 mm).\u003c/p\u003e\u003cp\u003e\u003cb\u003eUniform manifold approximation and projection\u003c/b\u003e\u003c/p\u003e\u003cp\u003eWe investigated the impairment of the neurovascular unit in DRD based on capillary nonperfusion and retinal thicknesses using complex and high-dimensional datasets. We applied UMAP for dimensionality reduction and visualization and subsequently demonstrated the arrangement of eyes according to the similarity of nonperfusion areas and retinal thinning in a two-dimensional space.\u003c/p\u003e\u003cp\u003eBriefly, we employed four steps for dimensionality reduction and subsequent clustering: 1. quantification and normalization of 15 variables related to capillary nonperfusion and retinal thicknesses, 2. principal component analysis (PCA) for variance-based dimensionality reduction, 3. UMAP for mapping each eye onto a two-dimensional space based on similarity, 4. subsequent k-means clustering. We selected 15 parameters for each eye: the NPS counts in the superficial and deep layers for each of the five ETDRS subfields (a total of 10 parameters), and the mean retinal thickness in the five subfields (a total of 5 parameters). These parameters represent both vascular and structural aspects of the retina, allowing for an integrative analysis of neurovascular degeneration. We first applied PCA, retaining over 90% of the total variance, which resulted in seven principal components. We then employed UMAP to project eyes with these components into a two-dimensional space.\u003csup\u003e\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u003c/sup\u003e UMAP is designed to preserve the local and global structure of high-dimensional data, making it suitable for identifying patterns and relationships in complex datasets.\u003csup\u003e\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u003c/sup\u003e Each eye was thus represented as a point in the two-dimensional UMAP space, with proximity between points indicating similarity in neurovascular parameters.\u003c/p\u003e\u003cp\u003eAfter the optimal number of clusters was determined utilizing the elbow method, we performed k-means clustering on the two-dimensional UMAP representation to categorize eyes into groups with similar neurovascular characteristics. The clusters were referred to as \u003cem\u003eMild\u003c/em\u003e, \u003cem\u003eIntermediate\u003c/em\u003e, and \u003cem\u003eSevere\u003c/em\u003e based on the NPS counts and retinal thickness values.\u003c/p\u003e\u003cp\u003e\u003cb\u003eStatistics\u003c/b\u003e\u003c/p\u003e\u003cp\u003eAll values were expressed as the median (interquartile range). Statistical significance was set at \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05. Normality was assessed using the Shapiro-Wilk test. For continuous variables, the Mann-Whitney U-test or Kruskal-Wallis test with Bonferroni correction was used, while categorical variables were analyzed using the Fisher\u0026rsquo;s exact test or Chi-square test. Spearman's rank correlation coefficient was used to assess statistical associations. All statistical analyses were conducted using SPSS software (version 24; IBM).\u003c/p\u003e"},{"header":"RESULTS","content":"\u003cp\u003e\u003cb\u003eSeverity scale of neurovascular pathology based on UMAP\u003c/b\u003e\u003c/p\u003e\u003cp\u003eIn this cross-sectional study, we analyzed 327 eyes from 327 DR patients without macular edema. The patients\u0026rsquo; characteristics are shown in Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e. All DR eyes were mapped onto the two-dimensional UMAP space according to the similarity of neurovascular parameters (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). The arrangement of DR eyes based on DMI and retinal thinning in the macula suggested a continuous severity scale of neurovascular pathology (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA-C). Cases in the lower left corner had smaller amounts of NPSs in both layers and normal CST. Deep NPSs increased in eyes located in the upper areas of the UMAP space. Heatmaps demonstrated that eyes in the right regions of the UMAP space had higher NPS counts in both layers and simultaneously smaller CST values. Additionally, these eyes also exhibited poorer logMAR (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eD). Interestingly, CST decreased in a subgroup of eyes located in the center of the UMAP space, despite no increase in superficial NPS counts (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC). DR severity was also illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eE. NPS counts and mean retinal thicknesses for each subfield were shown in Supplementary Fig. \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e-S4.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eSubsequent clustering using the k-means method divided the 327 cases into three distinct clusters (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eF; Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e), representing discrete grades referred to as \u003cem\u003eMild\u003c/em\u003e, \u003cem\u003eIntermediate\u003c/em\u003e, and \u003cem\u003eSevere\u003c/em\u003e groups based on NPS counts. The number of superficial and deep NPSs differed between the three groups (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). The central subfield of the superficial layer showed similar trends. In the parafoveal sectors of the superficial layer, there was a slight difference in NPS amounts between the \u003cem\u003eMild\u003c/em\u003e and \u003cem\u003eIntermediate\u003c/em\u003e groups, whereas the \u003cem\u003eSevere\u003c/em\u003e group exhibited much higher NPS counts (Supplementary Table \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e). CST decreased in a stepwise manner from the \u003cem\u003eMild\u003c/em\u003e to the \u003cem\u003eSevere\u003c/em\u003e group (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). Eyes in the \u003cem\u003eSevere\u003c/em\u003e group had thinner retinal thickness in the parafoveal sectors compared to those in the \u003cem\u003eMild\u003c/em\u003e and \u003cem\u003eIntermediate\u003c/em\u003e groups. In contrast, there were no significant differences or only minimal differences between eyes in the \u003cem\u003eMild\u003c/em\u003e and \u003cem\u003eIntermediate\u003c/em\u003e groups (Supplementary Table \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eComparisons of clinical parameters among three groups based on uniform manifold approximation and projection and subsequent clustering.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"5\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eVariables\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMild\u003c/p\u003e\u003cp\u003e(n\u0026thinsp;=\u0026thinsp;131)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eIntermediate\u003c/p\u003e\u003cp\u003e(n\u0026thinsp;=\u0026thinsp;129)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eSevere\u003c/p\u003e\u003cp\u003e(n\u0026thinsp;=\u0026thinsp;67)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cem\u003eP\u003c/em\u003e-value\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAge (years)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e64 (51\u0026ndash;71)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e66 (55\u0026ndash;74)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e70 (59\u0026ndash;77.5) \u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.009\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSex (male/female)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e82/49\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e78/51\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e47/20\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.401\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHemoglobin A1c (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e7.8 (6.8\u0026ndash;9.0)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e7.6 (6.9\u0026ndash;8.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e7.1 (6.6\u0026ndash;8.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.122\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDuration of diabetes (years)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e15 (8.5\u0026ndash;23)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e17 (10\u0026ndash;24)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e14.5 (10\u0026ndash;22.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.730\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSystemic hypertension (present/absent)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e73/58\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e88/41\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e46/21\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.067\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDyslipidemia (present/absent)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e61/70\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e58/71\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e30/37\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.957\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLogMAR\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e-0.079\u003c/p\u003e\u003cp\u003e(-0.128\u0026ndash;0.000)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.000\u003c/p\u003e\u003cp\u003e(-0.079\u0026ndash;0.097)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.301\u003c/p\u003e\u003cp\u003e(0.097\u0026ndash;0.699)\u003csup\u003e*\u0026dagger;\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePhakia/pseudophakia\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e74/57\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e68/61\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e12/55\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e\u003cp\u003eInternational DR severity grade (eyes)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMild NPDR\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e11\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\" morerows=\"3\" rowspan=\"4\"\u003e\u003cp\u003e0.004\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eModerate NPDR\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e72\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e65\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e35\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSevere NPDR\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e13\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e7\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e2\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePDR\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e35\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e55\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e30\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDiabetic macular atrophy (present/absent)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e8/123\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e24/105\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e37/30\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCentral subfield thickness (\u0026micro;m)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e283 (268\u0026ndash;297)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e272 (254\u0026ndash;289)\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e240 (190\u0026ndash;265)\u003csup\u003e*\u0026dagger;\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNPS counts in the superficial layer\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e207 (167\u0026ndash;278)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e368 (284\u0026ndash;423)\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e871 (665\u0026ndash;1007)\u003csup\u003e*,\u0026dagger;\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNPS counts in the deep layer\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e630 (507\u0026ndash;713)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1000 (878\u0026ndash;1127)\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1384 (1230\u0026ndash;1644)\u003csup\u003e*\u0026dagger;\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePrior PRP (present/absent)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e50/81\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e68/61\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e63/4\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePrior STTA (present/absent)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e11/120\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e7/122\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e11/56\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.036\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePrior anti-VEGF injection (present/absent)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e15/116\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e14/115\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e9/58\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.864\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePrior vitrectomy (present/absent)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e14/117\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e16/113\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e20/47\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003ctfoot\u003e\u003ctr\u003e\u003ctd colspan=\"5\"\u003eData are shown as numbers or median (interquartile range).\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd colspan=\"5\"\u003eAbbreviations: DR\u0026thinsp;=\u0026thinsp;diabetic retinopathy; NPDR\u0026thinsp;=\u0026thinsp;nonproliferative diabetic retinopathy; PDR\u0026thinsp;=\u0026thinsp;proliferative diabetic retinopathy; PRP\u0026thinsp;=\u0026thinsp;panretinal photocoagulation; STTA\u0026thinsp;=\u0026thinsp;subTenon\u0026rsquo;s injection of triamcinolone acetonide. *\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05 vs. Mild; \u003csup\u003e\u0026dagger;\u003c/sup\u003e\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.01 vs. Moderate.\u003c/td\u003e\u003c/tr\u003e\u003c/tfoot\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eClinical characteristics in each cluster\u003c/b\u003e\u003c/p\u003e\u003cp\u003eIn the \u003cem\u003eSevere\u003c/em\u003e group, CST was negatively correlated with NPS counts in the superficial layer, but not in the deep layer (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). LogMAR showed a positive correlation with NPS counts in the superficial layer and a negative correlation with CST, while no significant association was observed between logMAR and NPS counts in the deep layer (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eAssociations between neurovascular parameters within each group based on uniform manifold approximation and projection and subsequent clustering.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"4\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMild\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eIntermediate\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eSevere\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e\u003cp\u003eAssociation with central subfield thickness\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNPS counts in the superficial layer\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eρ\u003c/em\u003e=-0.111\u003c/p\u003e\u003cp\u003e\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.206\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cem\u003eρ\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.068\u003c/p\u003e\u003cp\u003e\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.441\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cem\u003eρ\u003c/em\u003e=-0.252\u003c/p\u003e\u003cp\u003e\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.039\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNPS counts in the deep layer\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eρ\u003c/em\u003e=-0.054\u003c/p\u003e\u003cp\u003e\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.537\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cem\u003eρ\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.004\u003c/p\u003e\u003cp\u003e\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.962\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cem\u003eρ\u003c/em\u003e=-0.235\u003c/p\u003e\u003cp\u003e\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.056\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e\u003cp\u003eAssociation with logMAR\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNPS counts in the superficial layer\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eρ\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.285\u003c/p\u003e\u003cp\u003e\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cem\u003eρ\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.286\u003c/p\u003e\u003cp\u003e\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cem\u003eρ\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.435\u003c/p\u003e\u003cp\u003e\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNPS counts in the deep layer\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eρ\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.136\u003c/p\u003e\u003cp\u003e\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.121\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cem\u003eρ\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.219\u003c/p\u003e\u003cp\u003e\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.013\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cem\u003eρ\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.159\u003c/p\u003e\u003cp\u003e\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.200\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCentral subfield thickness\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eρ\u003c/em\u003e=-0.005\u003c/p\u003e\u003cp\u003e\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.953\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cem\u003eρ\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.094\u003c/p\u003e\u003cp\u003e\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.290\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cem\u003eρ\u003c/em\u003e=-0.351\u003c/p\u003e\u003cp\u003e\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.004\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003ctfoot\u003e\u003ctr\u003e\u003ctd colspan=\"4\"\u003eData are shown as numbers or median (interquartile range).\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd colspan=\"4\"\u003eAbbreviations: logMAR\u0026thinsp;=\u0026thinsp;logarithm of the minimum angle of resolution; NPS\u0026thinsp;=\u0026thinsp;nonperfusion square.\u003c/td\u003e\u003c/tr\u003e\u003c/tfoot\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eIn the \u003cem\u003eMild\u003c/em\u003e and \u003cem\u003eIntermediate\u003c/em\u003e groups, logMAR was positively correlated with NPS counts in the superficial layer, but not with CST (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). In the \u003cem\u003eintermediate\u003c/em\u003e group, there was a modest association between NPS counts in the deep layer and logMAR.\u003c/p\u003e\u003cp\u003e\u003cb\u003eTwo subclasses of\u003c/b\u003e \u003cb\u003ediabetic macular atrophy\u003c/b\u003e\u003c/p\u003e\u003cp\u003eEyes with a thinner CST were mapped to two distinct areas in the UMAP space: the center and the right region of the UMAP space (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC). The \u003cem\u003eSevere\u003c/em\u003e group (55.2%) had a higher percentage of eyes with \u003cem\u003ediabetic macular atrophy\u003c/em\u003e (CST\u0026thinsp;\u0026lt;\u0026thinsp;246 \u0026micro;m) than the \u003cem\u003eMild\u003c/em\u003e (6.1%) and \u003cem\u003eIntermediate\u003c/em\u003e (18.6%) groups (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). Based on these results, we proposed the existence of two subgroups of \u003cem\u003ediabetic macular atrophy\u003c/em\u003e and compared several parameters between these two groups (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). Eyes in the center of the UMAP space belonged to the \u003cem\u003eMild\u003c/em\u003e and \u003cem\u003eIntermediate\u003c/em\u003e groups and had fewer NPS counts and a continuous EZ line more frequently (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). In contrast, eyes mapped to the right areas were included in the \u003cem\u003eSevere\u003c/em\u003e group, exhibiting higher NPS counts, worse logMAR, and more frequent disruption of the EZ line. Alternative thresholding of CST (\u0026lt;\u0026thinsp;222 \u0026micro;m) showed similar results (Supplementary Table S3).\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eComparisons of clinical parameters between two subgroups of eyes with \u003cem\u003ediabetic macular atrophy\u003c/em\u003e.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"4\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMild\u0026thinsp;+\u0026thinsp;Intermediate\u003c/p\u003e\u003cp\u003e(n\u0026thinsp;=\u0026thinsp;32)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSevere\u003c/p\u003e\u003cp\u003e(n\u0026thinsp;=\u0026thinsp;37)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cem\u003eP\u003c/em\u003e-value\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNPS counts in the superficial layer\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e383 (326\u0026ndash;456)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e931 (696\u0026ndash;1051)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNPS counts in the deep layer\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e931 (721\u0026ndash;1120)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1480 (1232\u0026ndash;1767)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCentral subfield thickness (\u0026micro;m)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e229 (210\u0026ndash;237)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e192 (163\u0026ndash;216)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEZ line (continuous/not continuous)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e26/6\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e10/27\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003elogMAR\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.023 (-0.079\u0026ndash;0.111)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.398 (0.222\u0026ndash;0.699)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003ctfoot\u003e\u003ctr\u003e\u003ctd colspan=\"4\"\u003eData are shown as numbers or median (interquartile range).\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd colspan=\"4\"\u003eAbbreviations: CST\u0026thinsp;=\u0026thinsp;central subfield thickness; EZ\u0026thinsp;=\u0026thinsp;ellipsoid zone; logMAR\u0026thinsp;=\u0026thinsp;logarithm of the minimum angle of resolution; NPS\u0026thinsp;=\u0026thinsp;nonperfusion square.\u003c/td\u003e\u003c/tr\u003e\u003c/tfoot\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e"},{"header":"DISCUSSION","content":"\u003cp\u003eIn this study, we investigated capillary nonperfusion and retinal thicknesses in the macula of DR eyes without macular edema. Statistical analyses revealed modest associations between NPS amounts and CST, while dimensionality reduction using UMAP highlighted various pathological patterns in the neurovascular unit. Some eyes exhibited greater amounts of deep NPSs without a decrease in CST, while others showed CST reduction with few NPSs. Additionally, certain cases presented with high NPS counts in both layers, retinal thinning, and concomitant VA reduction. We hypothesize that integrative analyses of neurovascular structures provide a more comprehensive understanding of retinal pathogenesis affecting visual function in DRD, compared to assessments based solely on a single parameter, such as capillary nonperfusion or neurodegeneration.\u003c/p\u003e\u003cp\u003eOur study utilizing UMAP highlights the clinical features of DRD, in which both vascular and neuronal pathological mechanisms contribute to vision loss.\u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e Previously, it was believed that vascular hyperpermeability and capillary nonperfusion lead to neuronal dysfunction and subsequent visual impairment.\u003csup\u003e\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e,\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e\u003c/sup\u003e Investigations using structural OCT have demonstrated novel imaging biomarkers of retinal neurodegeneration, such as disorganization of retinal inner layers and EZ disruption, in eyes with DME.\u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e,\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e Recent statistical analyses have also reported associations between capillary nonperfusion and neuronal biomarkers in eyes with DMI.\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e,\u003cspan additionalcitationids=\"CR30 CR31\" citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e These findings raise a key question: which comes first, vascular lesions or neurodegeneration? Or what are the common mechanisms for neurovascular degeneration? Our results suggest novel clinical entities, e.g., diabetic macular atrophy, DMI alone, and both atrophy and ischemia, in DRD. Furthermore, the distribution of cases in the two-dimensional UMAP space may represent a continuous and integrated severity scale of neurovascular degeneration.\u003c/p\u003e\u003cp\u003eInterestingly, CST was negatively correlated with NPS counts in the superficial layer in eyes of the \u003cem\u003eSevere\u003c/em\u003e group. Photoreceptors, which are the main component of the central subfield, are nourished by the deep retinal capillaries and the choroid. We applied the default setting for the superficial layer to create en face OCTA images, which include all retinal capillary layers at the fovea. The capillaries in the superficial and deep layers are fused in the perifoveal capillary network, which is visualized in superficial en face OCTA images.\u003csup\u003e\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e\u003c/sup\u003e This suggests simultaneous loss of both vascular and neuronal components at the fovea, and is consistent with previous findings that photoreceptor damage is associated with deep capillary nonperfusion.\u003csup\u003e\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\u003cp\u003eIn this study, we defined eyes with CST less than 246 \u0026micro;m as \u003cem\u003ediabetic macular atrophy\u003c/em\u003e, which were divided into two subgroups. One subgroup was mapped to the center of the UMAP space, while the other was located in the right regions. These two groups exhibited differences in several neurovascular parameters and logMAR. Eyes mapped to the right regions of the UMAP showed significant capillary nonperfusion in both the superficial and deep layers as well as photoreceptor disruption. This suggests that detrimental effects may mutually promote the impairment of the neurovascular unit or that common mechanisms could lead to simultaneous degeneration. In contrast, we could not identify definite retinal layer loss in eyes mapped to the center of the UMAP. These eyes demonstrated proportional retinal thinning and an enlarged foveal pit, which led us to hypothesize that the loss or volume reduction of M\u0026uuml;ller cells contributes to retinal thinning. Several studies have documented glial fibrillary acidic protein expression in M\u0026uuml;ller cells of diabetic retinas,\u003csup\u003e\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e\u003c/sup\u003e but the pathogenesis in these eyes remains to be clarified. In addition, we selected the mean minus one standard deviation, because of the frequency and the percentages of VA reduction in each group of retinal thicknesses in our single center (Supplementary Fig. \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). Future multi-center studies should determine the better thresholding to define \u003cem\u003ediabetic macular atrophy\u003c/em\u003e.\u003c/p\u003e\u003cp\u003eRegarding the question of defining thresholds for each severity scale for NPS counts in the superficial layer or the deep layer and retinal thicknesses, our study takes a different approach. Traditionally, diseases have often been diagnosed using a single parameter, e.g., CST. However, neurovascular impairment in DR cannot be adequately represented by a single parameter due to its complex, multifaceted nature. In our study, we employed UMAP to integrate vascular and neuronal parameters into a two-dimensional space. By evaluating the multifaceted disease in a multidimensional manner, we propose an integrated severity scale for neurovascular impairment in DRD. This approach allows for a more comprehensive assessment of the neurovascular unit, capturing the interplay between capillary nonperfusion and retinal structural changes.\u003c/p\u003e\u003cp\u003eWe demonstrated a continuous severity scale of neurovascular degeneration in DR. This scale could provide insights into the progression pathways of the disease, though longitudinal studies are needed to confirm this hypothesis. Furthermore, novel biological techniques, such as single-cell transcriptome and proteome analyses, could elucidate the pathological interactions between vascular cells, neurons, glia, and the extracellular matrix in the neurovascular unit.\u003csup\u003e\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e,\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e\u003c/sup\u003e These findings may provide useful information for the development of novel therapeutic strategies for DRD.\u003c/p\u003e\u003cp\u003eThere are several limitations to this study. The criteria for enrollment in this single-center study may result in selection bias. In particular, we excluded eyes with center-involving DME, so the clinical characteristics of neurovascular degeneration in DME remain to be elucidated. We do not have a complete solution for artifacts in OCTA images, which might affect the vascular parameters.\u003csup\u003e\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e\u003c/sup\u003e The quantification of retinal parameters was performed using specific devices and image processing methods, and future multicenter studies employing alternative methodologies are needed to confirm the generalizability of these findings. Additionally, while we used UMAP and the k-means method for dimensionality reduction and clustering, other algorithms may reveal different features.\u003csup\u003e\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e\u003c/sup\u003e A comprehensive assessment of visual function is also necessary to further explore the clinical relevance in DRD.\u003c/p\u003e\u003cp\u003eIn conclusion, we demonstrated the clinical characteristics of the neurovascular unit in DR eyes without macular edema. This study contributes to a deeper understanding of the pathogenesis of DRD.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003ch2\u003eCompeting Interests\u003c/h2\u003e\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eM.Y. and T.M. contributed to the conceptualization, study design, and manuscript drafting. All authors contributed to data acquisition. T.M. and A.T. supervised the study. All authors reviewed and approved the final version of the manuscript.\u003c/p\u003e\u003ch2\u003eAcknowledgements\u003c/h2\u003e\u003cp\u003eThis work was supported by a Grant-in-Aid for Scientific Research from the Japan Society for the Promotion of Science (Grant Number: 23K09004). The funding organization had no role in the design or conduct of this study.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003e The raw data are provided by the corresponding author upon reasonable request.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eAntonetti, D. A., Klein, R. \u0026amp; Gardner, T. Mechanisms of disease: diabetic retinopathy. \u003cem\u003eN Engl. J. Med.\u003c/em\u003e \u003cb\u003e366\u003c/b\u003e, 1227\u0026ndash;1239 (2012).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAiello, L. P. et al. Vascular endothelial growth factor in ocular fluid of patients with diabetic retinopathy and other retinal disorders. \u003cem\u003eN Engl. J. Med.\u003c/em\u003e \u003cb\u003e331\u003c/b\u003e, 1480\u0026ndash;1487 (1994).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBarber, A. J. et al. Neural apoptosis in the retina during experimental and human diabetes: early onset and effect of insulin. \u003cem\u003eJ. Clin. Invest.\u003c/em\u003e \u003cb\u003e102\u003c/b\u003e, 783\u0026ndash;791 (1998).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKim, K., Kim, E. S. \u0026amp; Yu, S. Y. Longitudinal relationship between retinal diabetic neurodegeneration and progression of diabetic retinopathy in patients with type 2 diabetes. \u003cem\u003eAm. J. Ophthalmol.\u003c/em\u003e \u003cb\u003e196\u003c/b\u003e, 165\u0026ndash;172 (2018).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCheung, C. M. G. et al. Diabetic macular ischaemia\u0026ndash;a new therapeutic target? \u003cem\u003eProg Retin Eye Res.\u003c/em\u003e \u003cb\u003e89\u003c/b\u003e, 101033 (2022).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSun, J. K. et al. Updating the staging system for diabetic retinal disease. \u003cem\u003eOphthalmology\u003c/em\u003e \u003cb\u003e128\u003c/b\u003e, 490\u0026ndash;493 (2021).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLevine, S. R., Sapieha, P., Dutta, S., Sun, J. K. \u0026amp; Gardner, T. W. It is time for a moonshot to find cures for diabetic retinal disease. \u003cem\u003eProg Retin Eye Res.\u003c/em\u003e \u003cb\u003e90\u003c/b\u003e, 101051 (2022).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBarber, A. J., Antonetti, D. A. \u0026amp; Gardner, T. W. Altered expression of retinal occludin and glial fibrillary acidic protein in experimental diabetes. \u003cem\u003eInvest. Ophthalmol. Vis. Sci.\u003c/em\u003e \u003cb\u003e41\u003c/b\u003e, 3561\u0026ndash;3568 (2000).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWilkinson, C. P. et al. Proposed international clinical diabetic retinopathy and diabetic macular edema disease severity scales. \u003cem\u003eOphthalmology\u003c/em\u003e \u003cb\u003e110\u003c/b\u003e, 1677\u0026ndash;1682 (2003).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLo, E. H., Dalkara, T. \u0026amp; Moskowitz, M. A. Neurological diseases: mechanisms, challenges and opportunities in stroke. \u003cem\u003eNat. Rev. Neurosci.\u003c/em\u003e \u003cb\u003e4\u003c/b\u003e, 399\u0026ndash;415 (2003).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSohn, E. H. et al. Retinal neurodegeneration may precede microvascular changes characteristic of diabetic retinopathy in diabetes mellitus. \u003cem\u003eProc. Natl. Acad. Sci. U.S.A.\u003c/em\u003e 113, E2655\u0026ndash;E2664 (2016).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSim\u0026oacute;, R., Stitt, A. W. \u0026amp; Gardner, T. W. Neurodegeneration in diabetic retinopathy: does it really matter? \u003cem\u003eDiabetologia\u003c/em\u003e 61, 1902\u0026ndash;1912 (2018).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAbr\u0026agrave;moff, M. D. et al. Approach for a clinically useful comprehensive classification of vascular and neural aspects of diabetic retinal disease. \u003cem\u003eInvest. Ophthalmol. Vis. Sci.\u003c/em\u003e \u003cb\u003e59\u003c/b\u003e, 519\u0026ndash;527 (2018).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTang, Z. et al. Assessment of retinal neurodegeneration with spectral-domain optical coherence tomography: a systematic review and meta-analysis. \u003cem\u003eEye\u003c/em\u003e \u003cb\u003e35\u003c/b\u003e, 1317\u0026ndash;1325 (2021).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJia, Y. et al. Split-spectrum amplitude-decorrelation angiography with optical coherence tomography. \u003cem\u003eOpt. Express\u003c/em\u003e. \u003cb\u003e20\u003c/b\u003e, 4710\u0026ndash;4725 (2012).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eStaurenghi, G., Sadda, S., Chakravarthy, U. \u0026amp; Spaide, R. F. International Nomenclature for Optical Coherence Tomography (IN\u0026bull;OCT) Panel. Proposed lexicon for anatomic landmarks in normal posterior segment spectral-domain optical coherence tomography: the IN\u0026bull;OCT consensus. \u003cem\u003eOphthalmology\u003c/em\u003e \u003cb\u003e121\u003c/b\u003e, 1572\u0026ndash;1578 (2014).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSpaide, R. F., Klancnik, J. M. \u0026amp; Cooney, M. J. Retinal vascular layers imaged by fluorescein angiography and optical coherence tomography angiography. \u003cem\u003eJAMA Ophthalmol.\u003c/em\u003e \u003cb\u003e133\u003c/b\u003e, 45\u0026ndash;50 (2015).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChalam, K. V. et al. Retinal thickness in people with diabetes and minimal or no diabetic retinopathy: Heidelberg spectralis optical coherence tomography. \u003cem\u003eInvest. Ophthalmol. Vis. Sci.\u003c/em\u003e \u003cb\u003e53\u003c/b\u003e, 8154\u0026ndash;8161 (2012).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMurakami, T. \u0026amp; Yoshimura, N. Structural changes in individual retinal layers in diabetic macular edema. \u003cem\u003eJ. Diabetes Res.\u003c/em\u003e 920713 (2013). (2013).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSun, J. K. et al. Disorganization of the retinal inner layers as a predictor of visual acuity in eyes with center-involved diabetic macular edema. \u003cem\u003eJAMA Ophthalmol.\u003c/em\u003e \u003cb\u003e132\u003c/b\u003e, 1309\u0026ndash;1316 (2014).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMurakami, T. et al. Optical coherence tomographic reflectivity of photoreceptors beneath cystoid spaces in diabetic macular edema. \u003cem\u003eInvest. Ophthalmol. Vis. Sci.\u003c/em\u003e \u003cb\u003e53\u003c/b\u003e, 1506\u0026ndash;1511 (2012).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eEarly Treatment Diabetic Retinopathy Study Research Group. Classification of diabetic retinopathy from fluorescein angiograms: ETDRS Report Number 11. Ophthalmology 98, 807\u0026ndash;822. (1991).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSamara, W. A. et al. Quantification of diabetic macular ischemia using optical coherence tomography angiography and its relationship with visual acuity. \u003cem\u003eOphthalmology\u003c/em\u003e \u003cb\u003e124\u003c/b\u003e, 235\u0026ndash;244 (2017).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKaizu, Y. et al. Optical coherence tomography angiography reveals spatial bias of macular capillary dropout in diabetic retinopathy. \u003cem\u003eInvest. Ophthalmol. Vis. Sci.\u003c/em\u003e \u003cb\u003e58\u003c/b\u003e, 4889 (2017).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTsai, A. S. H. et al. Diabetic macular ischemia: correlation of retinal vasculature changes by optical coherence tomography angiography and functional deficit. \u003cem\u003eRetina\u003c/em\u003e \u003cb\u003e40\u003c/b\u003e, 2184\u0026ndash;2190 (2020).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTerada, N. et al. Clinical relevance of parafoveal intercapillary spaces and foveal avascular zone in diabetic retinopathy without macular edema. \u003cem\u003eInvest. Ophthalmol. Vis. Sci.\u003c/em\u003e \u003cb\u003e63\u003c/b\u003e, 4 (2022).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYang, D. et al. Assessment of parafoveal diabetic macular ischemia on optical coherence tomography angiography images to predict diabetic retinal disease progression and visual acuity deterioration. \u003cem\u003eJAMA Ophthalmol.\u003c/em\u003e \u003cb\u003e141\u003c/b\u003e, 641\u0026ndash;649 (2023).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYoshida, M. et al. Severity scale of diabetic macular ischemia based on the distribution of capillary nonperfusion in OCT angiography. \u003cem\u003eOphthalmol. Sci.\u003c/em\u003e \u003cb\u003e5\u003c/b\u003e, 100603 (2025).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSim, D. A. et al. Quantitative analysis of diabetic macular ischemia using optical coherence tomography. \u003cem\u003eInvest. Ophthalmol. Vis. Sci.\u003c/em\u003e \u003cb\u003e55\u003c/b\u003e, 417\u0026ndash;423 (2014).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eScarinci, F., Jampol, L. M., Linsenmeier, R. A. \u0026amp; Fawzi, A. A. Association of diabetic macular nonperfusion with outer retinal disruption on optical coherence tomography. \u003cem\u003eJAMA Ophthalmol.\u003c/em\u003e \u003cb\u003e133\u003c/b\u003e, 1036\u0026ndash;1044 (2015).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDodo, Y. et al. Diabetic neuroglial changes in the superficial and deep nonperfused areas on optical coherence tomography angiography. \u003cem\u003eInvest. Ophthalmol. Vis. Sci.\u003c/em\u003e \u003cb\u003e58\u003c/b\u003e, 5870\u0026ndash;5879 (2017).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTsai, W. S. et al. Characterization of the structural and functional alteration in eyes with diabetic macular ischemia. \u003cem\u003eOphthalmol. Retina\u003c/em\u003e. \u003cb\u003e7\u003c/b\u003e, 142\u0026ndash;152 (2023).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSampani, K. et al. Association of retinal oximetry with peripheral diabetic retinopathy lesions and nonperfusion on ultra-widefield angiography. \u003cem\u003eOphthalmol. Sci.\u003c/em\u003e \u003cb\u003e5\u003c/b\u003e, 100686 (2024).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMcInnes, L., Healy, J. \u0026amp; Melville, J. Uniform manifold approximation and projection for dimension reduction. \u003cem\u003earXiv Preprint\u003c/em\u003e. \u003cb\u003earXiv\u003c/b\u003e, 180203426 (2018).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBecht, E. et al. Dimensionality reduction for visualizing single-cell data using UMAP. \u003cem\u003eNat Biotechnol\u003c/em\u003e (2018).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCoyner, A. S. et al. Synthetic medical images for robust, privacy-preserving training of artificial intelligence. \u003cem\u003eOphthalmol. Sci.\u003c/em\u003e \u003cb\u003e2\u003c/b\u003e, 100126 (2022).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYoshida, M. et al. Inference of capillary nonperfusion progression on widefield OCT angiography in diabetic retinopathy. \u003cem\u003eInvest. Ophthalmol. Vis. Sci.\u003c/em\u003e \u003cb\u003e64\u003c/b\u003e, 24 (2023).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWu, Y. et al. Developing a continuous severity scale for macular telangiectasia type 2 using deep learning and implications for disease grading. \u003cem\u003eOphthalmology\u003c/em\u003e \u003cb\u003e131\u003c/b\u003e, 219\u0026ndash;226 (2024).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCanny, J. A computational approach to edge detection. \u003cem\u003eIEEE Trans. Pattern Anal. Mach. Intell.\u003c/em\u003e \u003cb\u003e8\u003c/b\u003e, 679\u0026ndash;698 (1986).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eEarly Treatment Diabetic Retinopathy Study Research Group. Photocoagulation for diabetic macular edema: Early Treatment Diabetic Retinopathy Study Report Number 1. \u003cem\u003eArch. Ophthalmol.\u003c/em\u003e \u003cb\u003e103\u003c/b\u003e, 1796\u0026ndash;1806 (1985).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSnodderly, D. M., Weinbaus, R. S. \u0026amp; Choi, J. C. Neural\u0026ndash;vascular relationships in central retina of macaque monkeys (Macaca fascicularis). \u003cem\u003eJ. Neurosci.\u003c/em\u003e \u003cb\u003e12\u003c/b\u003e, 1169\u0026ndash;1193 (1992).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMizutani, M., Gerhardinger, C. \u0026amp; Lorenzi, M. Muller cell changes in human diabetic retinopathy. \u003cem\u003eDiabetes\u003c/em\u003e \u003cb\u003e47\u003c/b\u003e, 445\u0026ndash;449 (1998).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDuh, E. J., Sun, J. K. \u0026amp; Stitt, A. W. Diabetic retinopathy: current understanding, mechanisms, and treatment strategies. \u003cem\u003eJCI Insight\u003c/em\u003e. \u003cb\u003e2\u003c/b\u003e, e93751 (2017).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eVan Hove, I. et al. Single-cell transcriptome analysis of the Akimba mouse retina reveals cell-type-specific insights into the pathobiology of diabetic retinopathy. \u003cem\u003eDiabetologia\u003c/em\u003e \u003cb\u003e63\u003c/b\u003e, 2235\u0026ndash;2248 (2020).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSpaide, R. F., Fujimoto, J. G. \u0026amp; Waheed, N. K. Image artifacts in optical coherence tomography angiography. \u003cem\u003eRetina\u003c/em\u003e \u003cb\u003e35\u003c/b\u003e, 2163\u0026ndash;2180 (2015).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eVan Der Maaten, L. \u0026amp; Hinton, G. Visualizing data using t-SNE. \u003cem\u003eJ. Mach. Learn. Res.\u003c/em\u003e \u003cb\u003e9\u003c/b\u003e, 2579\u0026ndash;2605 (2008).\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"diabetic macular atrophy, diabetic macular ischemia, diabetic retinopathy, neurodegeneration, uniform manifold approximation and projection","lastPublishedDoi":"10.21203/rs.3.rs-7000479/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7000479/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eWe investigated the characteristics of neurovascular degeneration in diabetic retinopathy (DR) using optical coherence tomography (OCT) and OCT angiography (OCTA). En-face 3 \u0026times; 3 mm OCTA images were obtained from 327 eyes of DR patients without macular edema. Nonperfusion squares (NPSs) were defined as 15\u0026times;15-pixel regions lacking vascular signals. Neurovascular parameters were extracted from five subfields of the Early Treatment Diabetic Retinopathy Study grid. High-dimensional data were embedded into a two-dimensional space using Uniform Manifold Approximation and Projection, and clustering revealed three distinct groups: \u003cem\u003eMild\u003c/em\u003e, \u003cem\u003eIntermediate\u003c/em\u003e, and \u003cem\u003eSevere\u003c/em\u003e. Eyes with central subfield thickness (CST)\u0026thinsp;\u0026lt;\u0026thinsp;246 \u0026micro;m were classified as having \u003cem\u003ediabetic macular atrophy\u003c/em\u003e. The \u003cem\u003eMild\u003c/em\u003e group exhibited lower NPS counts, while the \u003cem\u003eIntermediate\u003c/em\u003e group showed increased deep-layer ischemia. The \u003cem\u003eSevere\u003c/em\u003e group had the highest NPS counts and the lowest CST, with a significant negative correlation between CST and superficial NPS counts (\u003cem\u003eρ\u003c/em\u003e = \u0026minus;\u0026thinsp;0.252, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.039). Eyes with \u003cem\u003ediabetic macular atrophy\u003c/em\u003e in the \u003cem\u003eSevere\u003c/em\u003e group demonstrated higher NPS counts, worse visual acuity, and more frequent ellipsoid zone disruption (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001). These findings suggest a pathological relationship between macular ischemia and retinal atrophy, offering new insights into DR progression.\u003c/p\u003e","manuscriptTitle":"Integrative Severity Scale for Diabetic Macular Atrophy and Ischemia Using Structural OCT and OCT Angiography","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-07-22 08:26:10","doi":"10.21203/rs.3.rs-7000479/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-07-22T04:09:51+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-07-20T10:40:55+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-07-17T15:51:38+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"121988618746051481467815000059646424538","date":"2025-07-15T10:16:26+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"95968388431620911908972813613341582607","date":"2025-07-15T07:21:16+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-07-15T07:10:41+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-07-15T06:09:11+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-07-08T04:13:38+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-07-02T08:45:35+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2025-07-02T08:42:11+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"8383a606-3aa2-4b84-836a-1ee03cfb97a8","owner":[],"postedDate":"July 22nd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":51704958,"name":"Health sciences/Diseases"},{"id":51704959,"name":"Health sciences/Endocrinology"},{"id":51704960,"name":"Health sciences/Medical research"}],"tags":[],"updatedAt":"2025-12-15T16:08:36+00:00","versionOfRecord":{"articleIdentity":"rs-7000479","link":"https://doi.org/10.1038/s41598-025-31862-w","journal":{"identity":"scientific-reports","isVorOnly":false,"title":"Scientific Reports"},"publishedOn":"2025-12-11 15:59:00","publishedOnDateReadable":"December 11th, 2025"},"versionCreatedAt":"2025-07-22 08:26:10","video":"","vorDoi":"10.1038/s41598-025-31862-w","vorDoiUrl":"https://doi.org/10.1038/s41598-025-31862-w","workflowStages":[]},"version":"v1","identity":"rs-7000479","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7000479","identity":"rs-7000479","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}