Alzheimer’s disease (AD) is a neurodegenerative disorder and one of the leading causes of dementia worldwide. It has huge implications for the quality of life of affected individuals, as it interferes with cognitive abilities, reasoning skills, language, and attention and ultimately alters their behavior and personality. In severe cases, it can lead to death. The pathophysiological mechanism underlying AD involves the accumulation of extracellular Aβ plaques and intracellular τ-proteins in cortical and limbic cerebral areas [3]. The definitive confirmation of AD can only be achieved postmortem by identifying neurofibrillary tangles and Aβ plaques in these parts of the brain [4].

Although OCTA is a relatively new imaging modality and thus its application in dementia is still being explored, the general consensus seems to be that patients with AD show an increase in the foveal avascular zone (FAZ), as well as a decrease in the superficial parafoveal and in the whole-vessel density (VD) [5]. More specifically, in one of the earliest studies on this topic, Jiang et al. showed that AD patients had significantly reduced VD in both the superficial (SCP) and deep capillary plexus (DCP) compared to controls [6]. These results have been replicated in subsequent studies [7,8]. The assessment of the FAZ has also shown some promising results as a potential biomarker for AD, with research suggesting an enlargement of this area in affected individuals [7,9]. The proposed pathological mechanism which might explain these findings is the accumulation of Aβ plaques in the retina in a manner similar to processes affecting the brain. More specifically, it seems that the retinal hypoxia is a combination of the binding of the vascular endothelial growth factor (VEGF) to Aβ and its confinement in the plaques, as well as the deposition of Aβ proteins in the internal vessel walls, causing vascular occlusion; reduced blood; and, ultimately, retinal hypoxia [7].

Given that AD represents a stage of established dementia, meaning that by the time of diagnosis the patient’s health has already been severely affected, researchers have also focused on the prodromal phase of the disease, termed mild cognitive impairment (MCI). Zhang et al. found a significantly reduced parafoveal SCP VD in an MCI group compared to controls [8]. Additionally, according to a recent prospective study, there is a marked loss of VD in the DCP in MCI patients. This is in accordance with the findings of Jiang et al., who found a lower VD in the superonasal quadrant of the DCP [6]. In a study by Querques et al., a quantitative analysis of the retinal vessels of patients with MCI and AD was performed using a dynamic vessel analyzer and OCTA [16]. Although they noted a decrease in the reaction amplitude and the arterial dilation in the MCI group, the OCTA values did not show any significant variability between the groups. In fact, in a monozygotic twins preclinical AD study, researchers reported a higher VD in all areas in the Aβ-positive group, with no differences seen in the FAZ area [17]. Regarding RNFL thickness, most studies agree that this is significantly reduced in the MCI group when compared to controls [18,19,20]. Furthermore, a few studies have noted significant differences in the retinal thickness between AD and MCI [21,22,23].

Choroidal thickness also seems to be a valuable biomarker for AD, as several studies have managed to demonstrate significant thinning in these patients compared to controls [25]. The pathophysiological mechanisms responsible for these findings are not completely clear yet. However, it seems that the accumulation of Aβ in the chοroid leads to inflammatory responses; complement activation; and, ultimately, to choroidal vascular angiopathy, in the same way as it occurs in AD brains [2].

The use of artificial intelligence and deep learning is constantly revolutionizing medicine and our understanding of certain diseases. A retinal OCTA segmentation database (ROSE) combined with an OCTA network has already been introduced and can provide the fractal dimension analysis of the retinal vasculature [26]. Despite being a novel method, the results appear to be promising, as researchers have already been able to detect significant changes between healthy controls and AD patients.

Vascular cognitive impairment and dementia (VCID), or vascular dementia (VaD), represents the second most common cause of cognitive impairment and a diagnostic challenge, in part due to the overlap with other dementia syndromes, including AD. The diagnosis of VaD, which is largely clinical, encompasses executive, visuospatial and/or memory dysfunction, among other cognitive aspects. There is a strong association with vasculopathy, including hypertension, hyperlipidemia and diabetes, which, through different mechanisms, ultimately lead to brain ischemia and degeneration [27]. White matter lesions or hyperintensities (WML/WMH), lacunar infarcts, microinfarcts, cerebral microbleeds and hemorrhages are the MRI hallmarks of this condition [28].

The retina is considered an extension of the cerebral tissue, and the study of its vasculature may reflect brain pathology. Therefore, OCTA is ideally suited to non-intrusively visualizing the functional microvascular changes which can be expected to be present in VaD. Indeed, it has been demonstrated that the flow density of the inner retinal layers might be a useful biomarker in differentiating vascular from degenerative dementia, as it correlates with the Fazekas scale but not with the presence of pathologic (Aβ, τ) proteins in the cerebrospinal fluid (CSF) [29,30]. A systematic review by Zhang et al. also revealed an association between WMHs and lower VD values, complementing the findings of Wang et al., who found the VD of the SCP to correlate with both WMHs and cognitive scores in patients with cerebral small-vessel disease (CSVD) [31,32]. Research in patients with subcortical vascular cognitive impairment (SVCI), a subtype of VaD, has shown lower capillary density (CD) values in the temporal RPC plexus as compared to healthy subjects. The researchers demonstrated that the CD values of the temporal and superior RPC quadrants were lower in the SVCI subgroup than in AD patients, a finding compatible with the potential use of this OCTA parameter as a differentiating biomarker. Furthermore, they noted a negative correlation between the RPC density and the CSVD score, an inherently important aspect of cognitive decline in VaD [28,33]. In a different study, patients with CSVD had lower VD values in both their temporal macular SCP and RPC plexuses versus healthy controls [32]. Of note, another recent study has shown VD to be significantly reduced in the DCP and RPC of healthy (i.e., cognitively normal) subjects with higher Fazekas scores as well, which highlights the potential of OCTA as an early biomarker [34]. The hypothesis that retinal microcirculation parameters might serve as biomarkers in VaD is further supported by the finding that lower vessel skeleton density (VSD)—a measure of perfused retina in OCTA images—is correlated with worse clinical (lower visuospatial and executive cognitive functions) and anatomical findings in patients with small-vessel disease (SVD) [29]. Specifically, visuospatial and executive cognitive functions and MRI findings related to cerebral perfusion and reactivity were found to be negatively affected. Cognitive function also appears to correlate with SCP density [32]. In one case report of a patient diagnosed with post-stroke VaD, a thinning of the choroid and electrophysiology consistent with outer retinal dysfunction was observed, which is argued to be in contrast to the inner retinal pathology reportedly associated with AD [35]. The underlying pathophysiological mechanism seems to be the expansion of the cerebral hypoperfusion to the choroidal circulation. However, additional research is required to supplement our knowledge of this case.

As VaD is inextricably linked to cerebrovascular disease (CVD), hereditary forms of CVD have fittingly been studied in the search for relevant retinal biomarkers. Cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL) has indeed been found to be associated with lower VD in the DCP, affected RPC plexus and even choroidal thinning [30]. Similarly, another study suggests that the OCTA parameters of macular VD may be used as early biomarkers for Fabry disease, as these predict myocardial changes which might precede cognitive impairment [36].

An ambitious ongoing cohort study by Clancy et al. is performing an extensive examination of factors which affect the CSVD burden in patients with mild stroke. Among other tests, patients are being subjected to OCTA. The results of this study will hopefully further elucidate the relationship between retinal imaging and the risk of developing VaD, as well as solidifying its diagnosis [37].