Retinal Microperimetry: A New Tool for Identifying Patients With Type 2 Diabetes at Risk for Developing Alzheimer Disease

Accumulating evidence indicates that type 2 diabetes is associated with cognitive impairment and dementia, and numerous epidemiological studies have demonstrated that patients with type 2 diabetes have a significantly higher risk (twofold higher) of developing Alzheimer disease (AD) than do age-matched subjects without diabetes (1,2). This increased risk is maintained even after adjusting for vascular risk factors (3,4). In addition, patients with type 2 diabetes have an increase in mild cognitive impairment (MCI) in comparison with subjects without diabetes, which consists of cognitive impairment on standard tests but no impairment of activities of daily living and represents a transitional state between normal cognitive function and dementia. The annual conversion rate from MCI to dementia ranges between 10% and 30% in the general population (5–7).

The number of patients with type 2 diabetes and cognitive impairment or dementia is expected to increase because of the diabetes pandemic and the concomitant growth of aging populations worldwide (8). In this regard, severe cognitive impairment can be envisaged as a “new” long-term diabetic complication with dramatic consequences for affected subjects and their families, and with a significant impact on health care systems. Therefore, strategies are urgently needed to identify patients with diabetes who are at risk for dementia.

In clinical practice, no reported phenotypic indicators or reliable examinations are available to identify patients with type 2 diabetes with MCI. A diagnosis of MCI is based on complex neuropsychological tests (9), which makes their incorporation into current standards of care infeasible for the population with type 2 diabetes.

Recent evidence has shown that retinal neurodegeneration is an early event in the pathogenesis of diabetic retinopathy (DR) (10,11). The retina is ontogenically a brain-derived tissue, and it has been suggested that it may provide an easily accessible and noninvasive way to examine the pathology of the brain (12,13). Therefore, it seems reasonable to propose that the evaluation of retinal parameters related to neurodegeneration, such as retinal function, would be useful for identifying those patients with type 2 diabetes who have a high risk of developing AD.

Several methods can be used to measure retinal function. Among them, fundus-driven microperimetry has emerged as a simple, noninvasive, and rapid test that can be used in clinical practice (14,15). Microperimetry measures retinal sensitivity in terms of the minimum light intensity that patients can perceive when spots of light stimulate specific areas of the retina. Some recent studies have suggested that microperimetry is even more sensitive than multifocal electroretinography in detecting early functional changes of the retina (16).

On these bases, we hypothesized that microperimetry would be a useful and reliable screening test to assess retinal function and thereby identify patients with type 2 diabetes who are in the early stages of cognitive impairment. This simple and rapid test would permit one to select patients for whom further evaluation in a memory clinic would be cost-effective. Before this assumption can be accepted, however, it is necessary to examine whether a correlation actually exists between retinal sensitivity, as assessed by microperimetry, and brain neurodegeneration.

The main aims of this study were 1) to evaluate whether retinal sensitivity assessed by microperimetry correlates with parameters of brain imaging (MRI-based diffusion tensor imaging) and function (fludeoxyglucose-18 positron emission tomography [¹⁸FDG-PET]) commonly included in diagnostic procedures for patients with cognitive impairment, and 2) to determine whether retinal sensitivity assessed by microperimetry discriminates the different stages of cognitive impairment in patients with type 2 diabetes (absence of cognitive impairment, MCI, and AD).

A prospective, nested case-control study (1:1:1) was designed (identifier NCT02360527, www.clinicaltrials.gov). A total of 35 patients with MCI who met the inclusion criteria described below were selected from 214 consecutive patients with type 2 diabetes attending a memory clinic (Fundació ACE, Barcelona, Spain). Each patient with MCI was matched with a patient with type 2 diabetes with AD and a patient with type 2 diabetes with normal cognition by age (±4 years), diabetes duration (±4 years), and classic cardiovascular risk factors (i.e., hypertension and dyslipidemia). In addition, 20 subjects without diabetes were included as control subjects for each of the cognitive categories (normocognitive, MCI, and AD). The study was conducted according to the Declaration of Helsinki and was approved by the local ethics committee.

The main inclusion criteria were 1) age >65 years; 2) type 2 diabetes with a duration >5 years; 3) written informed consent, which included accepting participation in MRI measurements, ¹⁸FDG-PET, and a potential lumbar puncture; and 4) no apparent or mild nonproliferative DR, according to the International Clinical Diabetic Retinopathy Disease Severity Scale (17). Patients with more advanced DR were excluded because severe microvascular impairment could contribute to neurodegeneration, and the main aim of our study was to assess whether neurodegeneration of the brain and the retina runs in parallel, independent of the presence of overt microangiopathy.

The main exclusion criteria were 1) patients with other neurodegenerative diseases of the brain or retina (e.g., glaucoma) or cerebrovascular diseases (Fazekas scale score ≤1) (18) and 2) A1C >10% (86 mmol/mol). Patients with poor metabolic control (A1C >10%) were excluded because very high blood glucose levels could affect retinal function (19).

All patients underwent complete neuropsychological, neurological, and psychiatric evaluations as previously described (20), as well as biochemical analyses (including A1C and lipid profile) and APOE genotyping. Brain neurodegeneration was assessed by MRI (for structural changes) and ¹⁸FDG-PET (for functional changes). All MRI scans were performed with a 1.5-T MRI scanner (Magnetom Symphony; Siemens Medical Solutions, Erlangen, Germany). ¹⁸FDG-PET measurements (mean glucose metabolism uptake) focused on regions of interest that were chosen because they have frequently been reported as demonstrating hypometabolism in patients with AD.

Retinal sensitivity was evaluated by fundus-driven microperimetry (third-generation Macular Integrity Assessment) after pupillary dilation to a minimum of 4 mm. The standard Macular Integrity Assessment test covers a 10°-diameter area with 37 measurement points; a red 1°-radius circle was used as the fixation target. A four-level fixed strategy was used: Goldmann III size stimulus, background luminance of 4 asb and maximum luminance of 1,000 asb, with a 25-dB dynamic range. Notably, the microperimeter automatically compensates for eye movements during examination via a software module that tracks them. The characteristics of fixation (location and stability) were quantified and categorized according to the P1 and P2 parameters, the preferred retinal locus used to calculate the fixation stability, as follows: “stable” (P1 and P2 >75%), “relatively unstable” (P1 <75% and P2 >75%), or unstable (both P1 and P2 <75%). The device automatically calculates the reliability index, which assesses the accuracy of the test. The mean values for retinal sensitivity and reliability of the two eyes were used in data analyses.

Assuming that patients with type 2 diabetes present ∼30% of functional retinal abnormalities (21), and considering that this figure will rise to 60% in those patients with diabetes and cognitive impairment, a minimum of 25 patients were required in each group. This calculation was performed taking into account a two-sided risk level of 0.05 and a statistical power of 80%. Consequently, we considered it reasonable to extend the sample to include 35 subjects/group.

To assess differences between the groups, a χ² test was used for qualitative variables and ANOVA followed by a least significant difference (LSD) post hoc test was used for quantitative variables. To evaluate the correlation between retinal sensitivity and MRI and ¹⁸FDG-PET variables, a Spearman correlation test and regression analyses adjusted by age were performed. All P values were based on a two-sided test of statistical significance. Significance was accepted at P < 0.05. Bonferroni correction was used for multiple comparisons. Statistical analyses was performed with the SSPS statistical package.

The main clinical characteristics of patients with type 2 diabetes included in this case-control study are shown in Table 1. As expected based on the design, no differences were found among the groups in terms of age, diabetes duration, blood pressure, or lipid profile. In addition, no differences were found regarding blood glucose control (measured by A1C), BMI, sex, or family history of AD.

Retinal sensitivity was lower in patients with type 2 diabetes and AD than in patients with type 2 diabetes and MCI (P < 0.0001). In addition, patients with type 2 diabetes and MCI presented lower retinal sensitivity than age-matched patients with normal cognition (P = 0.027) (Table 2). Similar results were obtained in the control group without diabetes, but the differences between subjects with normal cognition and MCI did not reach statistical significance (Table 2). The mean of the reliability index of the microperimetry test was >90% in all groups, thus indicating that the sensitivity measurements were accurate. Notably, a correlation (r = −0.50; P < 0.0001) was found between retinal sensitivity and the Alzheimer’s Disease Assessment Scale cognitive subscale (ADAS-Cog) in patients with type 2 diabetes with cognitive impairment (Fig. 1).

A significant reduction in the volumes of the whole brain, gray matter, hippocampus, and mean cortex was observed in patients with AD compared with patients with MCI (Table 3). Regarding ¹⁸FDG-PET, a significant reduction in whole-brain glucose (bilateral composite) uptake and a reduction in angular gyrus, cingulum, and temporal glucose uptake were detected in patients with AD compared with patients with MCI (Table 3).

Retinal sensitivity assessed by microperimetry was correlated with the MRI parameters commonly assessed in patients with cognitive impairment. Therefore, retinal sensitivity was directly correlated with total gray matter volume (r = 0.38; P < 0.01), mean cortex thickness (r = 0.37; P < 0.01), and hippocampal volume (r = 0.35; P < 0.01) (Fig. 2A). In addition, retinal sensitivity correlated with whole-brain glucose uptake (r = 0.66; P < 0.001) and left angular (r = 0,46; P < 0.001), right angular (r = 0.64; P < 0.001), posterior cingulum (r = 0.48; P < 0.001), and temporal (left temporal: r = 0.55, P < 0.001; right temporal: r = 0.32, P = 0.02) glucose uptake on ¹⁸FDG-PET (Fig. 2B).

As expected, a correlation was detected between retinal sensitivity and age (r = −0.44; P < 0.001). However, all the aforementioned correlations between retinal sensitivity and both MRI and ¹⁸FDG-PET parameters were maintained at a significant level after adjusting for age.

Notably, when patients with type 2 diabetes and MCI were analyzed separately, the correlations between retinal sensitivity and ¹⁸FDG-PET parameters (whole-brain, temporal, and angular gyri glucose uptake) remained significant.

As far as we know, this is the first evidence that retinal sensitivity measured by microperimetry correlates with topography parameters of brain tissue loss (MRI) and brain hypometabolism (¹⁸FDG-PET) in patients with type 2 diabetes with cognitive impairment. We found that retinal sensitivity was lower in patients with type 2 diabetes and MCI than in patients with normal cognition. In addition, it was lower in patients with type 2 diabetes with AD than in those with MCI. Retinal sensitivity correlated with MRI and ¹⁸FDG-PET not only in AD but also in MCI, which can be considered a prodromal stage of AD. Notably, these correlations persisted at significant levels after adjusting for age.

Structural imaging based on MRI is an integral part of the clinical assessment of patients with suspected Alzheimer dementia. Brain atrophy is correlated with both tau protein deposition and neuropsychological deficits, and is a valid marker of AD and its progression (22). In our study we found a significant correlation between retinal sensitivity and total gray matter volume, cortical thickness, and hippocampal volume. In this regard, the degree of atrophy of medial temporal structures such as the hippocampus is now considered to be a valid diagnostic marker at the MCI stage (22). Therefore, our results suggest that retinal sensitivity measurements might be an adequate surrogate for MRI findings. In fact, the term “retinal sensitivity” is not appropriate because fundus-driven microperimetry assesses not only the functional status of the retina but also the entire visual system, and it is a dynamic test that requires short-term memory and adequate perceptual speed and executive function. Interestingly, type 2 diabetes affects all these domains (23,24), and therefore microprimetry could be a useful tool for testing all of them in an integrative manner.

Brain ¹⁸FDG-PET allows the in vivo study of cerebral glucose metabolism, reflecting neuronal and synaptic activity. While memory impairment due to AD is strictly linked to posterior cingulate and hippocampal hypometabolism, memory deficits observed during normal aging may reflect mainly a failure of encoding and the retrieval processes of episodic memory, which depend on frontal cortex integrity (25,26). In our study, specific correlations were found between retinal sensitivity and glucose uptake in the posterior cingulate and temporal gyri (which includes the hippocampus cingulum). These findings suggest that retinal sensitivity is a useful tool for identifying cognitive impairment related to AD, and not merely age-related neuropsychological deficits.

Several structural and functional measurements of the retina have been used to examine indirectly the events that take place in the brain during neurodegenerative processes. Among the structural methods, optical coherence tomography, a powerful tool in diagnosing and monitoring diabetic macular edema, has been proposed as a method for discriminating between subjects with AD, subjects with MCI, and healthy control subjects (27). In this regard, several meta-analyses have found a relationship between cognitive impairment and a reduction in retinal nerve fiber layer thickness (28–30). However, the methodologies and the results were not homogeneous, and the inclusion of patients with diabetes was not taken into account in the analyses of the results. This is an important point because although retinal neurodegeneration is a hallmark of the early stages of DR, glial activation or gliosis may precede the onset of neuronal death (11), thereby obfuscating the detection of retinal nerve fiber layer thinning. In addition, vascular leakage that occurs in the early stages of DR could also contribute to retinal thickness, thereby also acting as a confounding factor. New imaging methods such as optical coherence tomography angiography or high-resolution technologies such as the adaptive optics scanning laser ophthalmoscope could overcome these limiting factors, but specific studies of this issue are still required.

Regarding functional methods, multifocal electroretinography remains the gold standard and permits us to explore retinal function in terms of electrical response in different areas of the retina. This is, however, a cumbersome and time-consuming examination in which corneal electrodes are necessary, and therefore it is generally limited to research. By contrast, microperimetry is a simple test that requires <5 min to evaluate macular function. Microperimetry allows for exact topographic correlation between fundus details and light sensitivity (differential light sensitivity or retinal threshold). The principle of microperimetry rests on the possibility of seeing—in real time—the retina under examination (by infrared light) and to project a defined light stimulus over one selected location (15). Microperimetry examination is independent of fixation and any other eye movement (15); therefore microperimetry can be recommended for aging patients with cognitive impairment.

The American Diabetes Association recommends individualizing diabetes treatment, taking into account the patient’s cognitive capacity (31). Several neuropsychological questionnaires have been proposed to screen for cognitive decline in the population with type 2 diabetes (32). However, the number of patients whose cognitive function needs to be evaluated by a general practitioner or endocrinologist/diabetologist is potentially enormous, and a more simple and cost-effective case-finding strategy is needed to detect undiagnosed cognitive impairment. Our findings reveal that retinal microperimetry can be considered an effective and reliable tool for discriminating patients with AD from those with MCI. In addition, we found that retinal sensitivity measured by microperimetry was already lower in patients with diabetes with MCI than in age-matched subjects with diabetes with normal cognition. This is a relevant result because the identification of patients with prodromal stages of AD, such as MCI, is precisely the target of campaigns for the early detection of AD. In addition, a relationship between retinal sensitivity measured by microperimetry and the different stages of cognitive impairment (MCI and AD) was also found in control subjects without diabetes. In contrast with the results obtained in patients with type 2 diabetes, however, no significant differences were found between subjects with MCI and those with normal cognition. This could be attributed to the higher impact of neurodegeneration in the population with type 2 diabetes because of the presence of underlying mechanisms such as insulin resistance, inflammation, oxidative stress, and advanced glycosylation end product accumulation, which also play a key role in the pathogenesis of AD (33,34). However, further studies with large sample sizes are needed to examine whether microperimetry could also be useful a screening test for identifying cognitive impairment in the general population. Nevertheless, because patients with diabetes are prone to developing AD, they represent a high-risk group that should be prioritized in any program based on a case-finding strategy in the setting of cognitive impairment. In fact, identifying patients with diabetes in the early stages of cognitive impairment is important because unrecognized cognitive dysfunction can affect treatment adherence and diabetes self-management, thus resulting in poor glycemic control, more frequent severe hypoglycemic episodes, and more hospital admissions (35,36). For all these reasons, the diagnosis of cognitive impairment not only is recommended but also allows more personalized treatment for patients with type 2 diabetes.

In addition, retinal microperimetry is not influenced by noncognitive functions such as mood or depressive disorders, which could influence the results of neuropsychological tests. This is an important advantage given that the prevalence of depression is twofold higher in patients with type 2 diabetes than in the general population worldwide (37); depression has recently been reported in 27.5% of patients with type 2 diabetes in the Mediterranean population (38). This prevalence could be even higher in older adults with diabetes because depressive symptoms may be overlooked (39).

In this regard, our findings point to microperimetry as a tool to be incorporated into clinical practice and in particular into any program based on a case-finding strategy in the setting of cognitive impairment. In addition, retinal sensitivity assessed by microperimetry could be used to monitor the neuroprotective effectiveness of either antidiabetes drugs or new treatments for preventing the development of AD. Note that we found a high correlation between cognitive impairment assessed by ADAS-Cog and retinal sensitivity. ADAS-Cog is the most widely used general cognitive measure in clinical trials of AD and has been used as an outcome measure for trials of interventions in people with MCI (40). The ADAS-Cog consists of 11 parts and takes approximately 30 min to administer. Microperimetry is very fast and, depending on appropriate cost-effectiveness studies, might be envisaged as a good candidate to be incorporated into current methods for monitoring the effect of neuroprotective drugs.

Our study has several limitations. First, the inclusion of some patients with only mild nonproliferative DR, A1C <10% (86 mmol/mol), and no cerebrovascular disease does not allow us to extrapolate the results to the entire population with type 2 diabetes. Therefore, a study with less strict selection criteria and a larger sample size seems warranted. Second, the multiple statistical comparisons could have had some influence on our results, but the design of the study, the P values obtained, and the clinical coherence of the findings makes this potential limiting factor very unlikely. Finally, although the main aim of this study was to assess the relationship between retinal sensitivity measured by microperimetry and brain imaging in patients with type 2 diabetes with MCI or AD, the obtained results should lead us to create a database of normative microperimetry results that includes both aging healthy control subjects and subjects with type 2 diabetes without cognitive impairment.

In summary, our results suggest that retinal sensitivity assessed by microperimetry is related to brain neurodegeneration and could be a useful biomarker for patients with type 2 diabetes who are at risk of developing AD.

Funding. This study was supported by a grant from the European Foundation for the Study of Diabetes (EFSD/Lilly-Mental Health and Diabetes Programme).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. A.C. and O.S.-S. obtained study data, designed the study, analyzed data, and wrote the manuscript. C.H., M.B., and R.S. designed the study, obtained funds, and critically reviewed the manuscript. G.A., S.D., Á.S., Ó.S., and I.H. acquired, analyzed, and interpreted data and critically reviewed the manuscript. All authors approved the final manuscript. R.S. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.