Retinal scans can provide an early sign of Alzheimer’s disease


While it has been said that the eyes are a window to the soul, a new study shows they could be a means for understanding diseases of the brain.

According to new research by scientists at the UCSF Weill Institute for Neurosciences, retinal scans can detect key changes in blood vessels that may provide an early sign of Alzheimer’s disease, while offering important insights into how one of the most common Alzheimer’s risk genes contributes to the disease.

“The most prevalent genetic risk for Alzheimer’s disease is a variant of the APOE gene, known as APOE4,” said lead author Fanny Elahi, MD, Ph.D. “We still don’t fully understand how this variant increases risk of brain degeneration, we just know that it does, and that this risk is modified by sex, race, and lifestyle.

“Our research provides new insights into how APOE4 impacts blood vessels and may provide a path forward for early detection of neurodegenerative disease.”

Studies in mice have explored the effect of APOE4 on capillaries in the brain. Elahi, an assistant professor of neurology and member of the UCSF Memory and Aging Center (MAC), has long suspected these tiny blood vessels may play a significant role in Alzheimer’s disease, since they deliver nutrients and oxygen, carry away waste, and police immune system responses through the protective shield known as the blood–brain barrier.

Damage to these blood vessels could cause a host of problems, she says, including the protein buildup and cognitive decline seen in individuals affected by Alzheimer’s disease.

Because the technology does not exist to visualize individual capillaries in living people’s brains, Elahi and colleagues turned to the eye. In the new study, which publishes May 15, 2021, in the journal Alzheimer’s and Dementia: Diagnosis, Assessment & Disease Monitoring, Elahi and her team have shown that APOE4-associated capillary changes can be detected in humans through an easy, comfortable eye scan.

As a light-penetrating tissue that shares biology with the brain, the retina, researchers believe, may help determine what APOE4 variants may be doing to similar capillaries inside the brain, even in those without dementia.

The team – which includes Ari Green, MD, a neuro-ophthalmologist, professor, and director of the UCSF Neurodiagnostic Center, and Amir Kashani, MD, Ph.D., associate professor of ophthalmology at the Johns Hopkins Wilmer Eye Institute – used an advanced retinal imaging technique known as optical coherence tomography angiography (OCTA) to peer into the eyes of aging people with and without APOE4 mutations to evaluate the smallest blood vessels at the back of the eye.

The team leveraged the well-characterized cohorts of people enrolled in ongoing studies of brain aging and neurodegenerative disease at the MAC. By adding OCTA scans to existing MRI and PET scan data, they gain comparative insights without putting volunteer participants through additional discomfort.

“That’s the beauty of this technique,” Elahi said. “It’s very easy, noninvasive and participant-friendly.”

Analyzing the retinal scans, the researchers found reduced capillary density in APOE4 carriers, an effect that increased with participant age. To test whether those scans accurately reflected what was happening in the brain, the team then compared the abnormalities seen in OCTA scans of retinal capillaries to measurements of brain perfusion, or the flow of blood through the brain, as measured via MRI. They found that people with higher retinal capillary density also had greater blood flow in the brain.

Finally, the team looked to participants with prior PET scans of beta-amyloid, the protein associated with Alzheimer’s disease, to see how their retinal capillary measurements related to the burden of amyloid plaques in the brain, which is the major focus of Alzheimer’s disease diagnosis, research and treatment to date.

They found that capillary density did not differ between groups with and without amyloid plaques, nor did it vary along with amyloid burden. According to Elahi, that independence suggests that capillary abnormalities are unlikely to be driven by amyloid pathology, or that their relation may at most be indirect.

“This is the first time that we have demonstrated in living, asymptomatic humans that the smallest blood vessels are affected in APOE4 gene carriers,” said Elahi.

That’s important, she added, because it suggests that the increased risk of brain degeneration and Alzheimer’s disease in APOE4 carriers may be through its effect on blood vessels.

Elahi and her colleagues plan to follow their study participants to better understand blood vessel dysfunction at a molecular level. That work could help detect the onset of Alzheimer’s disease before significant damage occurs to the brain and identify new vascular targets for early treatment.

“This is just the beginning,” Elahi said. “But the implications for early detection and possible intervention can be significant in combatting Alzheimer’s disease and other neurodegenerative disorders. It’s much harder to regenerate neurons than to stop their degeneration from happening in the first place. Similar to cancer, early detection can save lives.”


The neurosensory retina is a developmental outgrowth of the brain and is the only (CNS) tissue not shielded by bone (eg, with the exception of the bony orbital floor and wall). Because of the eye’s anatomy and transparency to light, the neurosensory retina is more accessible to non-invasive, repeated, high spatial resolution imaging.

Moreover, the retina shares structural and pathogenic pathways with the CNS including the cerebral microvasculature and neural cells. Retinal cells are arranged in multiple layers: three layers of nerve and glial cell bodies and three layers of synapses. Photoreceptors capture visual information and transmit it via interneurons to retinal ganglion cells (RGCs). Extending from the inner retina, these cells make direct synaptic connections with the CNS through the optic nerve and optic pathways.13

Retinal changes have been observed across the continuum of AD and in other proteinopathies and neurodegenerative disorders, clearly establishing the importance of eye pathology in these diseases.14 For example, the hallmark biomarkers of AD—amyloid beta (Aβ) plaques and neurofibrillary tangles (NFTs) composed of the protein tau—have been identified in the retina of individuals with AD15 and may either trigger, or be otherwise associated with, a pathological cascade.16,17

This cascade can lead to retinal degeneration, thinning of the retinal nerve fiber layer (RNFL), and other structural and functional changes observed in AD.18 Although there is conflicting literature, several studies have corroborated findings by Koronyo-Hamaoui et al., pointing to the existence of Aβ40 and Aβ42 alloforms, Aβ deposits, (p)tau, NFTs, and inflammation in the retina of AD patients.16,17,19–25 Aβ in the retina is also thought to cause microvascular changes similar to those observed in the cerebral microvasculature in AD.26,27

Indeed, a recent study demonstrated in retinal tissue from mild cognitive impairment (MCI) and AD patients the accumulation of Aβ40 and Aβ42 in blood vessels, especially capillaries, and that this was associated with early and progressive loss of retinal vascular platelet-derived growth factor receptor-beta and pericytes, which are key components of the blood-retina and blood-brain barriers.24

More than one research group has shown, in both rodent and human neuropathological studies, similarities between Aβ plaques, including vascular Aβ deposits, in the retina and those in the brain of individuals with AD, although some plaques in the retina are smaller16 than those in the brain14,15,17 and larger ones differ in shape from those in the brain.17,28

Moreover, the presence of amyloid pathology has been demonstrated also in melanopsin-containing retinal ganglion cells (mRGCs).17 These cells, projecting to the suprachiasmatic nucleus of the hypothalamus, are crucial for circadian photoentrainment and they are lost in AD, possibly contributing to the occurrence of circadian and sleep disturbances in AD.17

Still, in a cross-sectional study of cases with a clinical diagnosis of AD and neuropathological Braak staging ≥4, Den Haan et al. concluded that amyloid precursor protein and Aβ pathology in the retina differ from that seen in the cerebral cortex, and that no retinal differences were found between individuals with AD and those who did not have AD.

They did observe diffuse phosphorylated or fibrillary or paired helical filament tau in AD retina, particularly in the peripheral retina of individuals with AD. In addition, they did not observe Aβ plaques in the AD retinas.29 Contrariwise, Koronyo et al. have reported finding neurofibrillary tangles in human AD retina,16 and this is a topic requiring further exploration.


Different retinal imaging modalities provide different types of structural and functional information. An empirically derived combination of both structural and functional retinal measures may be required to understand how this complex neurovascular/glial system is affected by AD.30 Moreover, this is a rapidly evolving field, and the pace of engineering advances in these imaging modalities, like any other, out-paces needed advances in signal analyses and data interpretation. In addition, this provides a roadblock in moving retinal biomarkers from the discovery to the validation phase. In the sections below, we provide a cursory review of current work across several imaging modalities; see Alber et al., 2020, for a review and discussion of these technologies.14

Optical coherence tomography to detect structural changes in the retina
Optical coherence tomography (OCT) provides two-dimensional cross-sectional images and three-dimensional (3D) volumetric measurements of retinal anatomy, and it is widely used for assessing retinal abnormalities in ophthalmologic and neurologic diseases such as glaucoma and multiple sclerosis.31 OCT generates high-resolution images by measuring the time delay of reflected light. This type of imaging has high depth resolution and can image 2D and 3D volumes over a large field of view.

Using interference data from multiple rapid scans, the OCT scanner generates 3D maps of the retina, which enable measurement of the thickness of the RNFL, the layer composed of axons; The ganglion cell layer (GCL) comprising primarily the neuron somata; the inner and outer plexiform layers (IPL and OPL) composed of microglia; and the outer retina where photoreceptors are found. OCT, first introduced clinically in the mid-1990s, has evolved over the years as the laser technology improved.

An advanced technology called spectral-domain OCT (SD-OCT) was introduced in 2001.32 SD-OCT produces faster imaging and is now a commonly used technology in clinical settings,31 although advances with ultra-high-resolution and OCT-AO (adaptive optics) are promising too.

In a meta-analysis of 11 OCT studies in AD and MCI, Mahajan et al. concluded that AD was associated with thinning of the retina and loss of retinal ganglion cells, as well as decreased volume of the optic nerve. Similar but less-robust changes were seen in individuals with MCI.33,34 A meta-analysis of 25 OCT studies found that individuals with AD and MCI had RNFL thinning compared to those without AD or MCI,35 with glaucoma singled out as an important confounding factor to consider.

In another review of 33 cross-sectional studies that compared OCT measures from individuals with and without AD, Santos et al. also found RNFL thinning in those individuals with AD compared to cognitively unimpaired individuals, but noted that although there is normal age-related thinning of the RNFL, most studies failed to control for the participants’ age.36

Several studies have shown that age-related loss is important to consider in assessment of RNFL changes in any chronic disease.37 However, limitations to RNFL thickness as a marker include the lack of standardization in OCT studies, and this limits comparison across studies38 and it is not specific for AD.

Santos et al. evaluated, within individuals, longitudinal relationships between retinal thickness and cognitive functioning, as well as AD brain changes among adults with preclinical AD. Participants with preclinical AD were selected based on evidence of Aβ burden as determined by amyloid PET imaging and reduced cognitive functioning in response to a challenge with very low-dose of scopolamine hydrochloride, a muscarinic anticholinergic agent.

SD-OCT imaging was used to assess volume and thickness of all retinal neuronal layers and participants completed PET amyloid imaging at the beginning and end of the study period. Over a 27-month period, as compared with normal controls, participants in the preclinical stage of AD showed decreases in the macular RNFL, outer nuclear layer, and inner plexiform layer volume. Decreases in RNFL volume correlated with increased PET amyloid burden and reduced performance on a cognitive measure of audiovisual integration efficiency.36,39

OCT may also have predictive value according to two recent studies. Ko et al. demonstrated an association between lower RNFL thickness at baseline and cognitive decline over a 3-year period.40 Mutlu et al. reported an association between thinner RNFL at baseline and an increased risk of developing dementia.41

Imaging of retinal Aβ and other proteinopathies
The retinas of individuals with AD may exhibit specific pathological changes in addition to RNFL thinning. Some groups have identified Aβ plaques in the retina in living AD patients, using the food additive curcumin.15–17,19,21–25,29,42 Curcumin, which has fluorescent properties and binds to fibrillary Aβ, has been studied for its diagnostic potential in AD by enabling non-invasive retinal fluorescent imaging.15,16,43 The presence of amyloid plaques has been demonstrated in human retinas using curcumin and scanning laser ophthalmoscope (SLO).15,16,44

Non-invasive polarimetric imaging of retinal amyloid deposits (which also detects pure fibrillary Aβ) may provide a low-cost, non-invasive screening method for presymptomatic detection of AD Imaging, with polarized light correlating with fluorescent Aβ markers. Campbell et al. have shown that the number of Aβ deposits in human retinas ex vivo correlate with brain histopathology, even in early stages, and are now working to compare polarimetric markers with PET imaging of amyloid.45,46

Hyperspectral imaging of the retina may provide an alternative non-invasive method of measuring Aβ. The approach is based on the finding that Aβ has an influence on light scatter that varies with the wave-length of light. Hadoux et al. recently used retinal hyperspectral imaging in conjunction with machine-learning image analysis methods to distinguish between people with MCI and high brain Aβ burden on PET imaging from matched controls (Aβ PET negative).21

The technique was also used to distinguish transgenic AD mice (5xFAD), known to accumulate Aβ in the brain and retina, from age-matched wild-type control mice. More et al. used a similar imaging method to distinguish between people with clinically diagnosed AD and cognitive unimpaired participants.47

These retinal imaging modalities require replication and neuropathological validation to move from the biomarker discovery to the biomarker validation phase.

Vascular imaging of the retina
Because the retina shares developmental, structural, and pathogenic pathways with the CNS, its microvascular system is also very similar to, and contiguous with, the vascular supply of the CNS. Changes in the condition of the retinal microvessels thus may provide information about various cerebrovascular and neurodegenerative disorders including AD. Like the brain, the retina does not have traditional lymphatic vessels.

The drainage of interstitial fluid from the brain occurs along the basement membranes of capillaries and arteries as intramural periarterial drainage fails in cerebral amyloid angiopathy (CAA).48 It remains to be seen if a similar mechanism for clearance exists in the eye and whether the accumulation of amyloid in retinal vessels may be an early biomarker for cerebral CAA.

Emerging technologies provide a larger perspective that allows peripheral vasculature of the retinal microcirculation to be evaluated, without the need for angiography and contrast dyes.49 These technologies include OCT angiography (OCT-A) and dynamic vessel analyzer examination of the fundus.

For example, among individuals with AD and MCI, OCT-A imaging has revealed loss of retinal microvascular density in some studies50,51 although not in others.52 Retinal microvascular abnormalities and structural alterations have also been observed using OCT-A in people with elevated Aβ PET imaging or by CSF protein assays, indicative of preclinical AD.53

Although not explicitly discussed at the Retinal Imaging in AD (RIAD) meeting, there are several retinal vascular changes that have been examined in AD, including, but not limited to, fractal dimension, vessel caliber, vessel branching, blood flow velocity, and retinal oximetry.

These techniques have shown mixed results when comparing AD patients to healthy controls (see Alber et al., 2020 for a review), and standardization of methods is necessary to determine the context of use for these potential retinal vascular biomarkers.

Imaging neuroinflammation and metabolism in the retina
In retinal tissue of triple transgenic AD mice, astrocytes and microglia are found in association with Aβ in the RGC. Widefield autofluorescence (AF) in vivo imaging of the retina has been used in mice to demonstrate the presence of inflammatory microglial cells, which may reflect CNS inflammation.54 Adaptive optics SLO has also been used to examine leukocyte-endothelial interactions in the retinae of mice. Further analysis revealed that retinal microglia express an anti-inflammatory phenotype in presymptomatic stages of AD but adopt a pro-inflammatory phenotype as the disease progresses.20

Next-generation retinal imaging approaches
Changes in cellular metabolism are among the earliest signs of retinal disease55 and can be detected using AF imaging approaches, since intracellular coenzymes involved in mitochondrial function and energy metabolism are naturally autofluorescent.55,56 Fundus autofluorescence (FAF) imaging can be accomplished using several different imaging systems including the fundus camera, sSL orchestrator (sSLO), and ultrawide field (UWF) imaging devices.57

Fluorescent lifetime imaging ophthalmoscopy (FLIO) is one advanced retinal imaging method that measures the decay in fluorescence intensity over time for endogenous fluorophores.58 The decay time of the fluorescence signal depends on the specific fluorophores present and may be affected by many factors, including age.

In patients with AD, fluorescence lifetime parameters have been shown to correlate with cognition (Mini-Mental State Examination [MMSE] score) and CSF concentrations of Aβ42 and phosphorylated tau.59 In a pilot study of individuals identified to have AD-related brain pathology and therefore identified as preclinical AD, FLIO parameters correlated not only with CSF AD biomarkers but also with GCL-inner plexiform layer (IPL) thickness on OCT, suggesting its use as a simple, non-invasive diagnostic and assessment tool to identify individuals defined as having preclinical AD.60 Further studies in larger longitudinal cohorts are needed, however, to define the most important parameters for discrimination.

As well as novel imaging of amyloid deposits with polarized light mentioned earlier, another novel approach involves the detection of apoptotic retinal cells (DARCs). Neurodegenerative diseases including AD are characterized by loss of RGCs. Using fluorescently labeled Annexin, Cordeiro et al. have demonstrated the ability to visualize apoptosis of individual RGCs in experimental models of neurodegeneration and in humans with glaucoma.61

DARCs may be able to register early stages of RGC pathology, although it is not well established whether DARCs can detect changes in RNFL thickness reliably earlier than OCT62; however, recent published findings from the Phase 2 DARC trial demonstrate it to predict RNF loss by 18 months in glaucoma eyes.73

Furthermore, studies in Parkinson disease (PD) models suggest that these retinal changes occur before pathological signs of PD in the CNS are apparent, raising the possibility that this may represent a biomarker of early PD as well as a biomarker of treatment efficacy.63 The application of this technology for early detection of AD has not yet been well delineated, and further validation is needed.

An important caveat when examining these retinal imaging techniques is that their specificity to AD cannot be determined without within-subject, longitudinal studies including histology- and/or biomarker-confirmed diagnostic criteria. There are few studies that make these comparisons, most of which are cross-sectional in nature. Another potential confounder is the presence of multiple pathologies, including cardiovascular and cerebrovascular insults, which can affect retinal and cerebral pathology.

reference link:

More information: Fanny M. Elahi et al. Retinal imaging demonstrates reduced capillary density in clinically unimpaired APOE ε4 gene carriers, Alzheimer’s & Dementia: Diagnosis, Assessment & Disease Monitoring (2021). DOI: 10.1002/dad2.12181


Please enter your comment!
Please enter your name here

Questo sito usa Akismet per ridurre lo spam. Scopri come i tuoi dati vengono elaborati.