Retinal microvascular alterations reflect cerebral small vessel disease burden in frontotemporal dementia: a multimodal OCTA–MRI Study Authors: Yufei Chen, William Robert Kwapong, Xu Chen, Yuan Gao, Min Chu, Xuxiang Zhang, Liyong Wu Published: 2025 ABSTRACT AbstractBackgroundFrontotemporal dementia (FTD) is a neurodegenerative disorder with a suspected vascular component. However, the relationship between retinal and cerebral microvascular alterations in FTD remains underexplored. This study investigates retinal microvasculature and its association with cerebral small vessel disease (SVD) markers in FTD using optical coherence tomography angiography (OCTA) and MRI.MethodsIn this cross-sectional study, 33 patients with clinically diagnosed FTD and 43 cognitively unimpaired controls (CU) underwent multimodal imaging, including OCTA and 3.0T MRI. Retinal vessel density and choriocapillaris perfusion were quantified using OCTA, while cerebral SVD markers, such as white matter hyperintensities (WMH) and perivascular spaces (PVS), were evaluated through both visual and automated methods. The correlation between retinal and cerebral microvascular changes was assessed using statistical models adjusting for potential confounders.ResultsFTD patients demonstrated significantly reduced retinal and choriocapillaris perfusion compared to CU (all p < 0.001). Notably, decreased superficial vascular complex (SVC) density correlated with higher WMH and PVS volumes. Choriocapillaris perfusion showed the highest diagnostic accuracy (AUC = 0.993) for distinguishing FTD from CU. Furthermore, OCTA metrics, particularly reduced SVC density, were significantly associated with increased SVD burden, indicating a link between retinal microvascular changes and cerebral small vessel pathology in FTD.DiscussionOur findings support a retina–brain microvascular axis in FTD, where retinal microvascular alterations may reflect cerebral SVD burden. These results highlight the potential of OCTA as a noninvasive biomarker for monitoring vascular contributions to frontotemporal neurodegeneration. The study provides a framework that using OCTA-derived metrics may help improve early diagnosis, disease stratification, and monitoring of vascular involvement in FTD.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13195-025-01945-x.Keywords: Frontotemporal dementia, Retinal microvascular alterations, OCTA, Small vessel disease, Choriocapillaris, White matter hyperintensities, Perivascular spaces Backrgound Frontotemporal dementia (FTD) represents a heterogeneous neurodegenerative disorder characterized by a wide spectrum of clinical manifestations, including progressive decline in behavior, executive function, or language due to degeneration of the frontal and temporal lobes [1, 2]. Although traditionally regarded as a primary neurodegenerative disorder, accumulating evidence suggests that vascular impairment may also contribute to the disease expression and progression [3]. Several neuroimaging studies have identified markers of cerebral small vessel disease (SVD) in FTD, implying a potential vascular-neurodegenerative interface [4–6]. Understanding this interplay is important as SVD may exacerbate neurodegeneration, ultimately leading to poorer cognitive performance in FTD. However, visualization of the cerebral microcirculation is limited. The retina, as an embryological extension of the brain, provides a non-invasive window into the cerebral microcirculation, positioning retinal imaging as a promising surrogate for the brain vascular pathology [7]. Retinal imaging studies using color fundus photography (CFP) have identified retinal microvascular impairment in neurodegenerative disorders such as Alzheimer’s disease (AD) and dementia compared to cognitively unimpaired individuals (CU) [7–10]. The recent introduction of optical coherence tomography angiography (OCTA) offers capillary-level visualization of the retinal microvasculature, providing quantitative vascular measures. Studies using OCTA have also identified retinal microvasculature changes in older adults, AD, and SVD, supporting its potential as a biomarker for neurovascular dysfunction [11, 12], and also linking retinal microvascular measures to SVD [13]. Yet, its application in FTD remains sparse. The few OCTA studies conducted in FTD cohorts have reported inconsistent findings – some describing reduced vessel density in the superficial capillary plexus (SCP), others noting preserved or even compensatory vascular patterns [14–16]. These discrepancies likely reflect methodological heterogeneity, including differences in OCTA devices, segmentation algorithms, and patient characterization across studies. Moreover, prior investigations have largely omitted parallel assessments of cerebral microvascular pathology, precluding direct evaluation of the relationship between retinal and cerebral small vessel changes in FTD. The present study addresses these gaps by systematically characterizing retinal microvascular alterations in a clinically well-defined FTD cohort using OCTA, and by examining their associations with MRI markers of SVD quantified through both standardized visual rating scales and automated volumetric methods. By integrating retinal and cerebral microvascular assessments, this work aims to elucidate whether retinal microvascular changes mirror intracranial small vessel pathology in FTD, thereby advancing understanding of the vascular contributions to frontotemporal neurodegeneration. In contrast to previous OCTA studies that primarily described retinal perfusion changes in isolation, our study adopts a multimodal, cross-system approach to evaluate the retina–brain microvascular axis in FTD. This integrative framework may help establish OCTA-derived biomarkers as accessible tools for detecting vascular pathology, improving disease stratification, and informing the development of targeted neurovascular interventions in FTD. Methods Figure 1 shows the schematic overview of our study design. This research followed the principles in the Declaration of Helsinki. The Ethics Committee and Institutional Review Board of Xuanwu Hospital, Capital Medical University in China, approved the clinical procedures of this study. Before enrolling in our study, participants were provided with written informed consent. Schematic illustration of our study design. FigureA shows the visual reading of cerebral small vessel disease (SVD). B shows the quantification of PVS and WMH volumes. I show a representative T1-weighted image. II shows the PVS segmentation result in the basal ganglia (red), deep white matter region masks (purple), which were obtained using the ShiVAi pipeline. III shows a representative axial slice from a T2 FLAIR image; IV. Corresponding WMH segmentation results obtained using WMH-SynthSeg. Periventricular WMH (marked in red) and deep WMH (marked in blue) were distinguished based on whether any voxel of a given WMH cluster was located within 3mm of the lateral ventricle. C shows the imaging and segmentation of OCTA metrics done by the OCTA machine Participants Between September 2024 and June 2025, participants were enrolled in this study at the Neurology Department of Xuanwu Hospital. Participants underwent ocular imaging via OCTA, cerebral MRI, and neuropsychological assessment. Demographic information, including age, sex, vascular risk factors (hypertension and diabetes mellitus), and educational years, was recorded. All participants underwent Mini-Mental State Examination (MMSE) and Montreal Cognitive Assessment (MoCA) as part of their cognitive assessment; these are brief dementia screening tools with a total score of 30, and a higher score indicates better cognition. We identified consecutive FTD patients following previous diagnostic criteria [17, 18]. The inclusion criteria for our FTD patients were as follows: (1) Had complete clinical information; (2) Normal appearance of optic nerve head and retina; (3) Able to cooperate, undergo, and complete MRI and OCTA imaging. Exclusion criteria were: (1) Diagnosed with forms of dementia such as AD, dementia with Lewy body disease, or vascular dementia; (2) History or presence of ocular disease that could affect ophthalmic vascular metrics, such as diabetic retinopathy, optic neuropathy, or severe cataract; (3) History or presence of concurrent neurological diseases such as stroke and psychiatric disorders (schizophrenia); (4) vascular disorders such as currently uncontrolled hypertension and diabetes mellitus; (5) History of cancer; (6) History of substance abuse. Cognitively unimpaired age-matched controls (CU) were enrolled from community-dwelling citizens through advertisements from residents who attended our hospital for a health examination. CU were defined as residents with no established neurological disorders, cognitively unimpaired in all cognitive domains, and did not show any abnormality on MR imaging. Lifestyle, cerebrovascular risk factors, and clinical information were recorded. Exclusion criteria were: (1) Diagnosed with forms of dementia such as AD, dementia with Lewy body disease, frontotemporal dementia, or vascular dementia; (2) History or presence of ocular disease that could affect ophthalmic vascular metrics, such as diabetic retinopathy, optic neuropathy, or severe cataract; (3) History or presence of concurrent neurological diseases such as stroke and psychiatric disorders (schizophrenia); (4) vascular disorders such as currently uncontrolled hypertension and diabetes mellitus; (5) History of cancer; (6) History of substance abuse. MR imaging A standardized protocol was conducted using a 3.0 T MRI system (SIGNA MR, GE Healthcare, WI, USA) in all participants, including 3D T1-weighted image, axial T2-weighted image, and axial T2 FLAIR image. For the present study, the high-resolution T1-weighted image were acquired using a magnetization-prepared rapid gradient-echo (MPRAGE) sequence with the following parameters: echo time (TE) = 2.19 ms, repetition time (TR) = 2000 ms, inversion time = 900 ms, field of view (FOV) = 240 mm × 240 mm, matrix size = 256 × 256, flip angle = 8°, and voxel size = 0.9 mm × 0.9375 mm × 0.9375 mm. Cerebral small vessel disease (SVD) markers such as lacunes, cerebral microbleeds, perivascular spaces, and WMH were rated according to the Standards for Reporting Vascular Changes on Neuroimaging consensus criteria [19]. WMH was rated on FLAIR images using the Fazekas scale [20]; WMH severity was rated (0–3) separately for periventricular (PWMH) and deep (DWMH) regions of the brain, with the sum of scores representing the total (TWMH) WMH burden. An ordinal score ranging from 0 to 4 was established to reflect the total burden of cerebral small vessel disease (SVD) as previously reported [21] (Fig. 1). A single rater (M.C.) who was blind to the clinical information of all participants evaluated MR images. A second rater (L.W.) evaluated a random sample of 20 participants to assess interrater agreement for WMH (kappa 0.85, p < 0.001). The quantification of white matter hyperintensities (WMHs) was performed through automated segmentation techniques. Axial T2 fluid-attenuated inversion recovery (FLAIR) imaging data were analyzed using the FreeSurfer WMH-SynthSeg algorithm for its consistent performance even in low-resolution MRI scans [22]. WMHs were categorized into two subtypes: periventricular and deep, with classification determined by the presence of at least one voxel within a WMH cluster positioned within a 3 mm proximity to the lateral ventricles [23] (Fig. 1). To ensure precision, all segmentations underwent expert review by a qualified neurologist (M.C.) to minimize classification inaccuracies. For cross-subject comparability, WMH volumes were normalized as a percentage of total intracranial volume (%ICV), thereby accounting for interindividual differences in brain size. Total intracranial volume was determined by summing the volumetric contributions of gray matter, white matter, cerebrospinal fluid, and WMH. PVS was also segmented using the ShiVAi pipeline (https://github.com/pboutinaud/SHiVAi) and classified according to the brain regions identified by the FreeSurfer SynthSeg algorithm [22]. PVS clusters segmented from T1-weighted images were categorized as basal ganglia PVS (BG PVS) and deep white matter PVS (WM PVS) as shown in Fig. 1. Data were expressed as a proportion of intracranial volume (%ICV) to account for differences in brain volume across subjects. OCTA imaging Retinal microvasculature was acquired by a swept-source optical coherence tomography angiography system (SS-OCTA, VG 200; SVision Imaging, Henan, China; version 3.1.352) between 8 am and 4 pm. The SS-OCTA contained a swept-source light source with a central wavelength of 1050 nm and a scan rate of 200,000 A-scans per second, as previously reported [11, 12, 24, 25, 26]. The tool was also equipped with an eye-tracking utility based on an integrated confocal scanning laser ophthalmoscope to reduce and eliminate eye-motion artifacts. The axial resolution, lateral, and scan depth were 5 μm, 13 μm, and 3 mm, respectively. Imaging was done in both left and right eyes without pupil dilation. OCTA fundus images were obtained with a raster scan protocol of 512 horizontal B-scans that covered an area of 6 × 6 mm2 centered on the fovea. The B-scans, which contained 512 A-scans each, were repeated 8 times and averaged. Angiograms with signal quality of 8 and above were selected for further analysis. The en face angiograms of the superficial vascular complex (SVC) and deep vascular complex (DVC) of the retina were generated by automatic segmentation to evaluate retinal perfusions. The segmentation between the SVC and DVC was set in the inner two-thirds and outer one-third interface of the ganglion cell-inner plexiform layer (GCIPL). Vascular density (%) was used to evaluate the retinal microvasculature. The SVP was identified as the microvasculature spanning from the base of the retinal nerve fiber layer (RNFL) to the junction between the inner plexiform layer (IPL) and the inner nuclear layer (INL) (Fig. 1). The ICP included the microvasculature between the IPL/INL junction and the junction between INL and the outer plexiform layer (OPL). The DCP comprised the microvasculature from the INL/OPL junction to 25 μm below the OPL. The choriocapillaris (CC) was defined as the microvasculature from the basal border of the retinal pigment epithelium-Bruch’s membrane complex to 20 μm beneath it [27]. The reproducibility of the OCTA metrics used in our study has been well documented in a previous study [28]. Angiograms used for processing followed the OSCAR-IB quality criteria [29] and the APOSTEL recommendation [30]. All angiograms were assessed by a trained, certified grader masked to participant characteristics (X.C.). Angiograms with a signal quality (SQ) of ≥ 7 were considered eligible for further analysis. Angiograms with motion artifacts, such as irregular microvascular patterns or blurred segmentation, were excluded from further analysis. Participants with retinopathy, such as age-related macular degeneration, severe cataracts, glaucoma, optic neuritis, and hemorrhages, were excluded. If a participant presented with any of these disorders in one eye, the other eye was used; if both eyes had the aforementioned disorders, the participant was excluded from the study. Statistical analyses The normality of the data was determined by visual inspection of the distribution and the Kolmogorov–Smirnov test. Consecutive variables with normal distribution were expressed as mean ± standard deviation (SD), and skewed distribution was expressed as median and interquartile range. Categorical variables are presented as frequencies and percentages (%). Frequency matching was used for CU selection according to the proportions in FTD to balance the distribution of sex and age between groups. The Fisher’s exact test, t-test, or Kruskal–Wallis tests were used to compare clinical characteristics between FTD and CU. A multiple linear regression model with a generalized estimating equation (GEE) was used to compare OCTA metrics between CU and FTD while adjusting for risk factors. The covariates were set as age, sex, vascular risk factors (hypertension and diabetes mellitus), educational years, and the working correlation matrix was set as exchangeable, considering the correlation of the 2 eyes. Correction for multiple comparisons (6 OCTA metrics) was performed using the Bonferroni method with a significance level set at 0.05/6 ~ 0.0083. The area under the receiver operating characteristic (AUROC) was calculated to determine the diagnostic capability of the OCTA metrics in FTD compared to CU; an AUROC of 1.0 represents perfect discrimination, while 0.5 indicates random discrimination. A multiple linear regression model with GEE was used to assess the association between OCTA metrics and neuroimaging metrics (visual and quantitative metrics) in FTD while adjusting for risk factors (age, sex, hypertension, diabetes mellitus) and intereye dependencies. Analyses were also conducted separately for the left and right eyes using general linear models to ensure the transparency of the results. Results were reported as regression coefficients [β, per standard deviation (SD) increase/decrease] with corresponding 95% CIs. In addition to the primary analyses, partial Spearman correlation was used to validate the association between OCTA and neuroimaging metrics in FTD. Results were reported as regression coefficients (r) with corresponding 95% CIs. Statistical analysis and plotting were conducted using R version 4.4.3 (gee package for GEE, mediation package for mediation analysis). A p-value of < 0.05 was considered significant. Results This study included 33 FTD and 43 CU were included in our data analyses. Supplementary Fig. 1 shows the flow chart of our study participants. Age, sex, vascular risk factors (hypertension and diabetes mellitus), and educational years were comparable between the two groups. Notably, FTD patients showed significantly lower MMSE and MoCA scores relative to CU. Of the 33 FTD, 27 (81.82%) were behavioral variant FTD (bvFTD), 5 (15.15%) were semantic dementia (SD), and 1 (3.03%) was progressive non-fluent aphasia (PNFA). Table 1 shows the demographics and clinical information of our study participants (FTD vs. CU). Table 1Baseline characteristics of our study populationCUN = 43FTDN = 33 p Demographics Age, years59.47 ± 5.5360.94 ± 6.160.281 Males, n (%)14 (32.56%)17 (53.12%)0.121 Educational years12 [9, 15]12 [9, 15]0.418 Duration, years3 [2, 4] Disease phenotype Behavioral variant, n (%)27 (81.82%) Progressive non-fluent aphasia, n (%)1 (3.03%) Semantic dementia, n (%)5 (15.15%) Vascular risk factors Hypertension, n (%)9 (20.93%)6 (18.75%)1 Diabetes mellitus, n (%)3 (6.98%)3 (9.38%)1 Cognitive measures MMSE30 [29, 30]21 [13, 25]< 0.001 MoCA29 [28, 30]14 [10, 17]< 0.001 Visual MRI measures Periventricular WMH1 [0, 1]2 [1, 2]< 0.001 Deep WMH1 [0, 1]1 [0, 1]0.672 Total WMH2 [1, 2]3 [2, 3]< 0.001 SVD burden0 [0, 1]2 [1, 2]< 0.001 Quantitative MRI measures TIV, mm31481.35 ± 92.241411.31 ± 98.920.004 Total PVS volume, %ICV0.04 ± 0.020.06 ± 0.040.006 BG PVS volume, %ICV0.01 ± 0.010.01 ± 0.010.245 WM PVS volume0.03 ± 0.020.05 ± 0.03< 0.001 Total WMH volume, %ICV0.23 ± 0.200.90 ± 0.23< 0.001 Periventricular WMH volume, %ICV0.20 ± 0.160.87 ± 0.25< 0.001 Deep WMH volume, %ICV0.03 ± 0.060.74 ± 0.32< 0.001 *OCTA metrics SVC, %45.85 ± 3.8939.78 ± 4.28< 0.001 SVP, %50.67 ± 4.3942.94 ± 4.31< 0.001 ICP, %34.61 ± 3.7829.48 ± 4.09< 0.001 DCP%17.24 ± 5.3412.87 ± 4.15< 0.001 DVC, %55.88 ± 3.1750.15 ± 4.41< 0.001 CC, %69.76 ± 1.9161.10 ± 4.60< 0.001MMSE Mini-Mental State Examination, MoCA Montreal Cognitive Assessment, WMH White matter hyperintensities, TIV Total Intracranial Volume, SVD Cerebral small vessel disease, ICV Intracranial volume, PVS Perivascular space, SVC Superficial vascular complex, DVC Deep vascular complex, SVP Superficial vascular plexus, ICP Intermediate capillary plexus, DCP Deep capillary plexus, CC Choriocapillaris*Generalized estimating equation models were used to compare the OCTA metrics between FTD and CU while adjusting for age, sex, hypertension, diabetes mellitus, educational years, and intereye dependencies Baseline characteristics of our study population MMSE Mini-Mental State Examination, MoCA Montreal Cognitive Assessment, WMH White matter hyperintensities, TIV Total Intracranial Volume, SVD Cerebral small vessel disease, ICV Intracranial volume, PVS Perivascular space, SVC Superficial vascular complex, DVC Deep vascular complex, SVP Superficial vascular plexus, ICP Intermediate capillary plexus, DCP Deep capillary plexus, CC Choriocapillaris *Generalized estimating equation models were used to compare the OCTA metrics between FTD and CU while adjusting for age, sex, hypertension, diabetes mellitus, educational years, and intereye dependencies OCTA showed choriocapillaris as the best diagnostic indicator for FTD Table 1 shows the comparison of OCTA metrics between FTD and CU. Our analysis showed that FTD showed significantly reduced retinal and choriocapillaris densities relative to CU (all p < 0.001). We found that choriocapillaris perfusion showed the highest area under the curve (AUC: 0.993, Youden index = 0.960, Sensitivity = 98.3%, Specificity = 97.7%, 95% CI 0.981 to 1.00, p < 0.001), making it the most effective among all OCTA metrics for detecting microvascular changes in FTD and CU (Table 2 and Fig. 2). Table 2ROC analysis of OCTA metrics in differentiating FTD from CUOCTA metricsAUCYouden IndexSensitivitySpecificitySVC0.8700.6130.7760.837SVP0.9130.7450.9310.814ICP0.8310.5840.7590.826DCP0.7610.4170.6380.779DVC0.8890.6360.7760.860CC0.9930.9600.9830.977SVC Superficial vascular complex, DVC Deep vascular complex, SVP Superficial vascular plexus, ICP Intermediate capillary plexus, DCP Deep capillary plexus, CC Choriocapillaris ROC analysis of OCTA metrics in differentiating FTD from CU SVC Superficial vascular complex, DVC Deep vascular complex, SVP Superficial vascular plexus, ICP Intermediate capillary plexus, DCP Deep capillary plexus, CC Choriocapillaris ROC analysis of OCTA metrics to distinguish FTD from CU. Choriocapillaris (CC) showed the highest OCTA metric to differentiate FTD from CU. SVC: Superficial vascular complex; DVC: Deep vascular complex; SVP: Superficial vascular plexus; ICP: Intermediate capillary plexus; DCP: Deep capillary plexus; CC: Choriocapillaris Association between OCTA metrics and quantitative MRI metrics in FTD We found that reduced SVC density was associated with higher total PVS volume (ß = −1.753, 95% CI −3.300 to −0.206, p = 0.034) and WM PVS volume (ß = −2.074, 95% CI −3.585 to −0.563, p = 0.012) as shown in Fig. 3. We also showed that reduced SVC density was associated with increased PWMH volume in FTD (ß = −1.809, 95% CI −2.930 to −0.689, p = 0.003). Similarly, reduced SVP density was associated with higher total PVS volume (ß = −1.880, 95% CI −3.443 to −0.317, p = 0.025) and WM PVS volume (ß = −2.009, 95% CI −3.569 to −0.450, p = 0.017). Also, reduced SVP density was associated with increased PWMH volume in FTD (ß = −1.387, 95% CI −2.565 to −0.208, p = 0.026). Table 3 shows the association between OCTA metrics and quantitative MRI metrics in FTD via GEE. The relationship between PVS and WMH volumes with OCTA metrics was also found using Partial Spearman correlation, as shown in Supplementary Fig. 2. Association between OCTA metrics and quantitative MRI metrics in FTD via generalized estimating equation models. OU (oculus uterque: both eyes), OD (oculus dexter: right eye), and OS (oculus sinister: left eye) Table 3Association between OCTA metrics and quantitative SVD markers in FTD using generalized estimating equationsSVC, %ß (95% CI) p SVP, %ß (95% CI) p Total PVS volume, %ICV−1.753 (−3.300 - −0.206)0.034Total PVS volume, %ICV−1.880 (−3.443 - −0.317)0.025BG PVS volume, %ICV−0.698 (−2.685–1.290)0.497BG PVS volume, %ICV−1.414 (−3.391–0.563)0.172WM PVS volume, %ICV−2.074 (−3.585 - −0.563)0.012WM PVS volume, %ICV−2.009 (−3.569 - −0.450)0.017Total WMH volume, %ICV−1.695 (−2.821 - −0.568)0.005Total WMH volume, %ICV−1.271 (−2.452 - −0.090)0.041PWMH volume, %ICV−1.809 (−2.930 - −0.689)0.003PWMH volume, %ICV−1.387 (−2.565 - −0.208)0.026DWMH volume, %ICV1.016 (−0.233–2.265)0.118DWMH volume, %ICV0.748 (−0.523–2.019)0.255ICP, %ß (95% CI)pDCP, %ß (95% CI)pTotal PVS volume, %ICV−0.451 (−1.835–0.933)0.528Total PVS volume, %ICV0.630 (−0.728–1.988)0.371BG PVS volume, %ICV−1.005 (−2.633–0.622)0.236BG PVS volume, %ICV0.376 (−1.266–2.019)0.657WM PVS volume, %ICV−0.415 (−1.813–0.984)0.565WM PVS volume, %ICV0.510 (−0.868–1.887)0.474Total WMH volume, %ICV0.883 (−0.196–1.962)0.116Total WMH volume, %ICV1.027 (−0.045–2.099)0.067PWMH volume, %ICV0.690 (−0.410–1.790)0.226PWMH volume, %ICV0.832 (−0.263–1.927)0.144DWMH volume, %ICV1.033 (−0.082–2.147)0.076DWMH volume, %ICV0.584 (−0.563–1.731)0.324DVC, %ß (95% CI)pCC, %ß (95% CI)pTotal PVS volume, %ICV−0.516 (−2.030–0.998)0.509Total PVS volume, %ICV−0.418 (−1.936–1.100)0.594BG PVS volume, %ICV−0.706 (−2.514–1.102)0.450BG PVS volume, %ICV0.483 (−1.335–2.300)0.607WM PVS volume, %ICV−0.642 (−2.164–0.879)0.415WM PVS volume, %ICV−0.878 (−2.385–0.628)0.262Total WMH volume, %ICV−0.025 (−1.204–1.154)0.967Total WMH volume, %ICV−0.588 (−1.765–0.588)0.333PWMH volume, %ICV−0.209 (−1.395–0.976)0.731PWMH volume, %ICV−0.789 (−1.926–0.428)0.219DWMH volume, %ICV0.583 (−0.632–1.797)0.352DWMH volume, %ICV0.065 (−1.173–1.303)0.919Data were adjusted for age, sex, hypertension, diabetes mellitus, educational years, and intereye dependenciesWMH White matter hyperintensities, TIV Total Intracranial Volume, SVD Cerebral small vessel disease, ICV Intracranial volume, PVS Perivascular space, SVC Superficial vascular complex, DVC Deep vascular complex, SVP Superficial vascular plexus, ICP Intermediate capillary plexus, DCP Deep capillary plexus, CC Choriocapillaris Association between OCTA metrics and quantitative SVD markers in FTD using generalized estimating equations Data were adjusted for age, sex, hypertension, diabetes mellitus, educational years, and intereye dependencies WMH White matter hyperintensities, TIV Total Intracranial Volume, SVD Cerebral small vessel disease, ICV Intracranial volume, PVS Perivascular space, SVC Superficial vascular complex, DVC Deep vascular complex, SVP Superficial vascular plexus, ICP Intermediate capillary plexus, DCP Deep capillary plexus, CC Choriocapillaris Association between OCTA metrics and visual scoring of SVD markers in FTD Increased PWMH burden (assessed via Fazekas scale) was associated with reduced SVC (ß = −2.481, 95% CI −3.474 to −1.488, p < 0.001), SVP (ß = −1.359, 95% CI −2.512 to −0.207, p = 0.025) and CC (ß = −1.419, 95% CI −2.598 to −0.241, p = 0.022) densities in FTD as shown in Fig. 4. We also found that increased SVD burden in FTD was associated with reduced SVC (ß = −2.179, 95% CI −3.126 to −1.232, p < 0.001) and SVC (ß = −1.768, 95% CI −2.788 to −0.748, p = 0.001) densities. Table 4 shows the association between OCTA metrics and visual scoring of SVD markers in FTD via GEE. Supplementary Fig. 2 shows the association between OCTA metrics and visual scoring of SVD markers in FTD using Partial Spearman correlation. Association between OCTA metrics and visual MRI ratings in FTD via generalized estimating equation models. OU (oculus uterque: both eyes), OD (oculus dexter: right eye), and OS (oculus sinister: left eye) Table 4Association between OCTA metrics and visual scoring of SVD markers in FTD using generalized estimating equationsSVC, %ß (95% CI) p SVP, %ß (95% CI) p PWMH−2.481 (−3.474 - −1.488)< 0.001PWMH−1.359 (−2.512 - −0.207)0.025DWMH−0.388 (−1.594–0.818)0.531DWMH0.086 (−1.137–1.309)0.891TWMH−1.524 (−2.692 - −0.355)0.013TWMH−0.628 (−1.868–0.612)0.325SVD burden−2.179 (−3.126 - −1.232)< 0.001SVD burden−1.768 (−2.788 - −0.748)0.001ICP, %ß (95% CI)pDCP, %ß (95% CI)pPWMH−0.601 (−1.699–0.496)0.288PWMH−0.105 (−1.322–1.112)0.866DWMH1.327 (0.263–2.391)0.018DWMH1.198 (0.010–2.386)0.053TWMH0.616 (−0.520–1.752)0.293TWMH0.777 (−0.465–2.019)0.225SVD burden−0.359 (−1.387–0.669)0.497SVD burden0.281 (−0.850–1.411)0.629DVC, %ß (95% CI)pCC, %ß (95% CI)pPWMH−0.436 (−1.573–0.702)0.456PWMH−1.419 (−2.598 - −0.241) 0.022 DWMH1.008 (−0.117–2.132)0.085DWMH−0.438 (−1.686–0.810)0.495TWMH0.478 (−0.698–1.654)0.429TWMH−1.021 (−2.274–0.231)0.116SVD burden−0.642 (−1.692–0.409)0.237SVD burden−1.011 (−2.132–0.110)0.083Data were adjusted for age, sex, hypertension, diabetes mellitus, educational years, and intereye dependenciesSVC Superficial vascular complex, DVC Deep vascular complex, SVP Superficial vascular plexus, ICP Intermediate capillary plexus, DCP Deep capillary plexus, CC Choriocapillaris Association between OCTA metrics and visual scoring of SVD markers in FTD using generalized estimating equations Data were adjusted for age, sex, hypertension, diabetes mellitus, educational years, and intereye dependencies SVC Superficial vascular complex, DVC Deep vascular complex, SVP Superficial vascular plexus, ICP Intermediate capillary plexus, DCP Deep capillary plexus, CC Choriocapillaris Discussion In this study, we demonstrated that patients with FTD exhibit significant reductions in retinal and choriocapillaris perfusion compared to CU, with choriocapillaris density showing the highest diagnostic accuracy for distinguishing FTD. Importantly, we observed that diminished retinal microvascular density, particularly within the SVC, was associated with increased SVD, as reflected by elevated WMH and PVS volumes. These findings provide convergent evidence for a retina-brain microvascular axis in FTD, supporting the concept that retinal microvascular changes mirror SVD pathology. By integrating OCTA and MRI markers within a multimodal framework, our results extend prior work in AD and SVD cohorts to FTD, highlighting the role of microvascular dysfunction in the pathophysiology and heterogeneity of frontotemporal neurodegeneration. Our results expand on previous works by situating FTD within the broader context of vascular contributions to neurodegenerative disease. While vascular dysfunction is a well-established component of AD pathology, its relevance to FTD has only recently gained attention. Prior neuroimaging studies [3, 4] have reported associations between WMH burden and clinical severity in FTD, suggesting that microvascular impairment may exacerbate neurodegeneration or accelerate disease progression. The current findings provide structural evidence linking retinal and cerebral microvasculature, suggesting that retinal vascular compromise may reflect SVD in FTD. This aligns with emerging models proposing that neurovascular dysfunction, through mechanisms such as endothelial impairment, impaired perivascular clearance, and chronic hypoperfusion, may contribute to neurodegenerative cascades [1, 4, 31]. Notably, the choriocapillaris showed the greatest diagnostic potential for differentiating FTD from CU, consistent with prior OCTA research in AD cohorts reporting early and pronounced choriocapillaris rarefaction [11]. The choroidal vasculature supplies the outer retina and is highly sensitive to neurovascular changes [32]. Reduced choriocapillaris perfusion in FTD may therefore reflect a shared downstream of microvascular dysfunction affecting both cerebral and ocular tissues. Alternatively, these changes may also suggest impaired neurovascular coupling, given that the retina and brain share similar autoregulatory mechanisms [33, 34]. The observed relationship between choriocapillaris perfusion and WMH volume may further underscore the physiological link, suggesting microvascular changes in the eye may be a reflection of the microvascular changes in the brain. The association between reduced superficial retinal vascular density and increased WMH and PVS burden may provide novel insight into the pathological overlap between the retinal and cerebral microvasculature. WMH and PVS are established imaging markers of SVD, reflecting chronic hypoperfusion, blood-brain barrier breakdown, and impaired glymphatic clearance [19]. The parallel reductions in retinal perfusion may therefore serve as an accessible proxy for monitoring cerebral microangiopathy in FTD. Importantly, the retina’s embryologic continuity with the brain and the feasibility of repeated non-invasive OCTA imaging may support its translational potential as a biomarker platform for vascular contribution to neurodegeneration in FTD. From a clinical perspective, our findings highlight the potential of OCTA as a tool for early vascular phenotyping and stratification in FTD. Given the absence of disease-modifying therapies and the clinical heterogeneity of FTD syndromes, the identification of microvascular biomarkers could facilitate personalized therapeutic targeting and monitoring of neurovascular interventions. Moreover, as vascular risk factors such as hypertension and diabetes mellitus are modifiable, recognition of their impact on the retina–brain axis may open new avenues for prevention and management in FTD populations. Several methodological strengths reinforce the robustness of our findings. The study employed standardized, quality-controlled OCTA acquisition following the OSCAR-IB and APOSTEL guidelines, and advanced MRI-based quantification of WMH and PVS using validated segmentation algorithms. The inclusion of both visual and automated SVD metrics provides a comprehensive assessment of cerebral microvascular burden, while the use of generalized estimating equations accounts for intereye correlations, enhancing statistical validity. This study positions itself as a significant advancement in the understanding of the retina-brain microvascular axis in FTD, particularly through the integration of OCTA with MRI-based SVD markers. While previous OCTA studies in FTD have focused on isolated retinal vascular changes, our study is novel in its systematic comparison of retinal microvascular alterations (such as reduced vessel density and choriocapillaris perfusion) with comprehensive SVD markers, including WMH and PVS, obtained via high-resolution MRI. This multimodal approach offers a more nuanced view of how retinal changes reflect cerebral small vessel pathology, extending prior research on AD. By incorporating both visual and automated SVD assessments, we provide stronger evidence linking retinal microvascular dysfunction to cerebral vascular pathology, positioning OCTA as a promising non-invasive biomarker for early diagnosis and disease monitoring in FTD. However, certain limitations should be acknowledged. First, the cross-sectional design precludes causal inferences regarding the temporal relationship between retinal and cerebral microvascular changes. Longitudinal studies are needed to determine whether retinal alterations precede or parallel the progression of cerebral SVD in FTD. Second, although we controlled for major vascular risk factors, unmeasured systemic influences such as blood pressure variability or endothelial biomarkers could contribute to the observed associations. Third, the modest sample size, while comparable to prior OCTA studies in neurodegenerative cohorts, may limit generalizability. Future multicenter studies incorporating larger, pathologically confirmed FTD samples and harmonized imaging protocols will be critical to validate and extend these findings. Another limitation of this study is that axial length was not included in the examination procedure, which may influence the interpretation of retinal and choroidal vascular measurements. Future studies could consider incorporating axial length to further refine the understanding of retinal microvascular changes in FTD. The lack of detailed consideration of cardiovascular factors, such as inflammation and oxidative stress, which could significantly influence the retinal microvascular changes [35–37] observed in FTD. These factors are known to play a crucial role in the development of SVD and could partially explain the findings. Future research should incorporate comprehensive cardiovascular data to better understand the interplay between vascular health and neurodegenerative processes in FTD. As the FTD spectrum encompasses diverse molecular subtypes, exploration of genotype-specific vascular phenotypes (e.g., GRN, MAPT, C9orf72) could elucidate whether the observed microvascular changes vary across underlying pathologies. Furthermore, insufficient detailed clinical characterization of the FTD cohort, including specific disease severity measures, is another limitation in our study. The study also combined heterogeneous FTD presentations, such as behavioral variant FTD, semantic dementia, and progressive non-fluent aphasia, which could introduce variability in the results. While pooling these different phenotypes offers broader insights into the generalizability of retinal microvascular alterations in FTD, stratified analyses by disease subtype and severity would have provided a more nuanced understanding of how these vascular changes vary across the spectrum of FTD. Future studies should consider subgroup analyses and detailed clinical assessments to better account for this heterogeneity. In conclusion, this study provides convergent multimodal evidence that retinal microvascular alterations may mirror SVD in FTD. The integration of OCTA-derived retinal metrics with MRI-based SVD markers underscores the relevance of neurovascular dysfunction in the pathogenesis of FTD. These findings support the use of retinal imaging as a noninvasive biomarker for detecting and monitoring microvascular contributions to frontotemporal neurodegeneration, with implications for early diagnosis, disease stratification, and therapeutic development targeting vascular pathways. Supplementary Information Supplementary Material 1. Acknowledgements We would like to thank the Neurology and Ophthalmology staff for their help with the patients and the project. Authors’ contributions Study concept and design: WRK, YC, LW.Data Acquisition: WRK, YC, XC, XZ, LW, YG.Data Analysis and Interpretation: WRK, YC, MC, LWDrafting of the manuscript: WRK, YC, LWCritical review of the manuscript: WRK, YC, LWAll authors reviewed and approved this version of the manuscript. Funding Beijing Natural Science Foundation (7242067). National Natural Science Foundation of China (82271464). Beijing Hospitals Authority’s Ascent Plan, Code: DFL20240805. Data availability The dataset generated and analyzed for this study will be made available by the corresponding author upon reasonable request. Declarations Ethics approval and consent to participate The protocol of this study was approved by the Ethics Committee of Xuanwu Hospital, Capital Medical University (Ethics Number: 2025 011-001). All participants or their legal guardians provided written informed consent before enrolling in the study. Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Yufei Chen and William Robert Kwapong contributed equally to this work. Contributor Information Min Chu, Email: cmsddhr@sina.com. Xuxiang Zhang, Email: zhang_xuxiang@hotmail.com. Liyong Wu, Email: wmywly@hotmail.com. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement The dataset generated and analyzed for this study will be made available by the corresponding author upon reasonable request.