Retinal examination modalities in the early detection of Alzheimer's disease: Seeing brain through the eye

Retinal electrophysiological changes in AD

Visual impairment, such as dysfunction or deterioration in visual acuity, contrast sensitivity, color, depth, and motion perception, is common in AD. Electrophysiological examinations can objectively reveal a decline in these features. Commonly used methods include electroretinogram (ERG), visual evoked potential (VEP), etc. Pattern-ERG (p-ERG) parameters change significantly in the early stages of AD and can distinguish mild cognitive impairment (MCI) due to AD from MCI due to vascular etiology.[3] The latency of flash visual evoked potentials-P2 (FVEP-P2) is also significantly delayed in AD and MCI patients compared with that in healthy controls.[4]

Retinal structral changes in AD

Structural changes in retinal tissue can be detected via OCT, which can provide high-resolution two-dimensional cross-sectional imaging and three-dimensional volumetric measurements of the retina. It is a diagnostic technique widely used in ophthalmology and has recently been used for the analysis of retinal changes in neurodegenerative diseases including amyotrophic lateral sclerosis (ALS), Parkinson disease (PD), and AD. The mean total macular thickness in the peripapillary retinal nerve fiber layer (pRNFL), the inner plexiform layer (INL), and the outer nuclear layer (ONL) was found reduced in ALS patients, and there is evidence that pRNFL is diminished during PD progression.[5] Currently, spectral domain OCT and swept source OCT are commonly used for retinal structure detection in AD patients.

Though still controversial, many studies have shown that the thinning of pRNFL and total macular in OCT are promising indicators for AD and MCI and are correlated with the cognitive function.[6] Significant correlations were found between retinal measures and brain volume, neuropsychological test scores, Aβ or tau levels in the CSF, and Aβ or tau deposition detected by PET imaging.[7, 8, 9] The average RNFL thickness may reflect early AD pathology, as indicated by CSF Aβ⁴²/Tau level.[8] The inner perifovea retinal thickness was found to be related to the hippocampal and entorhinal cortex volume in structural MR scan in our previous study.[10] A Rotterdam study found that thinning of the ganglion cell layer–inner plexiform layer (GC–INL) was associated with the prevalence of dementia, while a thinner baseline RNFL was associated with incident dementia.[11] Moreover, MCI patients who eventually converted to AD had significantly thinner baseline RNFL and GC-IPL than stable MCI,[12] suggesting the potential predictive value of the RNFL and GC-IPL thickness for AD progression. However, retinal structure characteristics in AD patients overlap with those in other neurodegenerative diseases mentioned above, indicating the relatively low specificity of OCT.

Retinal vascular changes in AD

Retinal vascular dysfunction has been reported in AD patients, including narrowing retinal vessel diameter, reduction of venular fractal dimensions, increased tortuosity, decreased blood flow, microvascular changes, changes in choroidal thickness, etc.[13, 14, 15, 16] There are currently many methods available for retinal vascular parameters measurement, but a unified conclusion has not yet been reached.

In some studies, OCT-A showed increased size of the foveal avascular zone and reduced density of superficial capillary plexus in AD patients,[14, 15] while another research reported no significant difference in foveal avascular zone, but significantly reduced density of superficial capillary plexus in AD patients compared to the control group.[17] Quantitative retinal vasculature analysis with retinal fundus photography showed a reduction of venular fractal dimensions in cognitive impairment (CI) individuals.[16] rHSI showed higher retinal vein tortuosity and larger diameter of arteries near the optic head in CI patients with Aβ-positive PET.

Ad-related pathology in the retina

Several emerging retinal imaging technologies such as curcumin imaging, imaging of cellular apoptotic changes, and rHSI are also being explored which might provide additional value in retinal AD-related pathology detection. Among them, rHSI is the most applicable tool.

As an inexpensive, noninvasive retinal imaging modality, rHSI can detect the light scattering changes to reflect the structure and chemical composition of each retinal quadrant. Soluble Aβ oligomers form the most characteristic optical signal of AD.[18] In a study involving 19 patients with CI, whose MMSE scores ranged from 10 to 26, covering mild, moderate, and severe AD, rHSI was found to have higher sensitivity in early AD (MMSE 22–26), which was not affected by comorbid ophthalmologic pathologies such as glaucoma and cataract.[19] Significant difference was found between retinal reflectance spectra of Aβ-positive MCI patients and age-matched Aβ-negative controls. rHSI feature scores are also significantly correlated with cerebral Aβ load, suggesting that rHSI can predict the Aβ load in the brain.[20] rHSI could be a potential early screening tool for AD, but only a few studies have reported its preliminary application in small samples of AD patients. Based on the current evidence, more validation studies covering different stages of AD patients are warranted.

In summary, retina is a promising window for visualizing cerebral pathology, especially in patients with AD biomarkers. With retinal imaging technology being increasingly incorporated into general practices of ophthalmology, it becomes more available in clinical practice. Besides, retinal imaging avoids invasive procedures and is far less expensive than PET imaging. Features such as low cost, noninvasiveness, and easy accessibility make retina imaging suitable for large-scale population screening for AD.

Despite its beneficial potentials, there are still some limitations that need to be addressed. Some measures, for example, retinal nerve fiber thinning and vascular changes, might not be specific for AD, which are highly variable in the normal population and are also affected by other age-related neurodegenerative diseases. In addition, since the prospective studies covering the whole spectrum of the disease were limited, the sequence of the pathological events including retinal degeneration, Aβ/Tau-related change, and cognitive deterioration remains to be elucidated. Studies to date suggest a range of retinal imaging technologies can be used for the clinical research of AD. However, it is time for the research focus to shift from establishment of the method to validation in the real-world clinical practice. It is crucial to assess the sensitivity, specificity, diagnostic accuracy, and cut-off values before the retina-related measurement can be utilized as a routine clinical examination. In search of new retinal biomarkers for AD, it is also extremely important to test its diagnostic power in comparison with the gold standard biomarkers such as Aβ-PET, tau-PET, CSF Aβ⁴², p-Tau, etc. Clinical study design involving parallel comparisons of retinal measures and other AD biomarkers is also required for future research. Retinal biomarkers, as shown in a few cohort studies to date, undergo significant changes as the AD progresses. To confirm the current findings and validate their prognostic value, further studies with longitudinal designs and large sample sizes are needed. Moreover, as single imaging methods only reflect one specific retinal pathology, comprehensive measures integrating multiple imaging modalities are warranted to further improve its diagnostic accuracy in the times to come.