Co-Prevalence of Alzheimer’s Disease and Age-Related Macular Degeneration Established by Histopathologic Diagnosis

INTRODUCTION

Age-related macular degeneration (AMD) and Alzheimer’s disease (AD) are prevalent degenerative diseases of the aging that occur in older adults [1, 2]. AMD affects more than 1.75 million Americans while AD impacts nearly 5.1 million Americans over the age of 65 [3, 4]. Although AMD is the leading cause of central vision loss in people over the age of 65, the exact pathobiology of AMD is still under investigation [5].

Several studies have suggested a shared pathophysiologic mechanism in AMD and AD. An imbalance of iron homeostasis that contributes to protein misfolding and accumulation can occur before the onset of neurodegeneration in both AMD and AD [6 –9]. Drusen and amyloid plaques share certain histological characteristics including presence of amyloid-β, apolipoprotein, and other inflammatory markers [10 –12]. Dysfunction of proteosomal and lysosomal clearance pathways has been shown to be involved in the pathogenesis of AMD and AD via drusen and amyloid plaque accumulation [13, 14].

AD and AMD also share several environmental risk factors including smoking, hypertension, hyperlipidemia, obesity, and arteriosclerosis [15 –19]. Many epidemiological studies and a recent meta-analysis of clinical studies have suggested an association between AMD and AD or cognitive impairment [20 –27]. AD patients can also develop visual dysfunction, which may be similar to the dysfunction experienced by AMD patients, such as poor color vision and decreased retinal function on electroretinograms [28 –30]. Therefore, there are substantial grounds for exploring a prevalence link between AMD and AD.

Despite evidence for a potential association between AD and AMD, whether the pathogenesis of these two diseases is truly related has remained unclear [31 –34]. Epidemiologic studies aimed at interpreting the co-prevalence of these diseases have confronted significant diagnostic challenges and are difficult to compare since they have used varying clinical criteria and tests to establish the diagnoses of AMD and AD. Moreover, many patients with pathologic brain changes of early AD or with mild retinal changes associated with AMD may not seek care and may not be diagnosed with the disease if their symptoms are minimal [35]. Thus, population surveys may underestimate AD and AMD diagnoses and, as a result, skew measures of their co-prevalence. Since the gold standard for diagnosis of AD is based on brain autopsy, there is need for histopathologic diagnosis of AD to determine whether there is a true association between AD and AMD.

The primary objective of this study was to determine whether AD is more prevalent in patients with AMD compared to those without AMD using a database of autopsy cases with robust histopathologic diagnoses and staging determined by established grading systems [36, 37]. The secondary objective was to evaluate whether there was any relationship between the stage of AMD and likelihood of AD. Due to the documented aforementioned similarities in pathophysiology, we hypothesized that the prevalence of AD would be increased in AMD patients compared to patients without AMD and that increasing severity of AMD would be associated with greater likelihood of AD.

MATERIALS AND METHODS

This was a cross-sectional study of patients that underwent autopsy at the Duke University Medical Center between 1997 and 2017. This study was approved by the Institutional Review Board of Duke University Hospital and it was conducted in accordance with the principles of the Declaration of Helsinki.

RESULTS

157 autopsy eye and brain specimens were included in this study. According to Sarks staging, 91 ocular specimens were diagnosed with AMD and 66 with no AMD. Then, B & B staging was applied to classify the corresponding brain specimens into 105 AD specimens and 52 non-AD specimens. Table 1 reports the demographic and clinical information of the AMD and non-AMD specimens.

The Braak and Braak criteria was applied to further sub-categorize the severity of AD. Figure 1 displays the distributions of specimens by Sarks and B&B staging. The proportion of non-AD dementia was not significantly different between subjects with AMD (N = 54, 59.3%) and without AMD (N = 39, 59.1%) (p = 0.98). Table 2 shows the prevalence of AD stratified by severity of AMD. The overall prevalence of AD in the AMD group was 62.6% versus 72.7% in the non-AMD group, and there was no significant difference in the likelihood of AD between AMD and non-AMD patients (OR 0.63, p = 0.19). Similarly, there was no significant association of AD with intermediate (OR 0.62, p = 0.20) or severe AMD (OR 0.67, p = 0.42).

OPEN IN VIEWER

Fig. 1

Diagram illustrating the Sarks stages of age-related macular degeneration (AMD): A) Sarks I: normal. B) Sarks II: few small drusen. C) Sarks III: thin continuous sub-RPE deposits. D) Sarks IV: thick sub-RPE deposits overlying degenerating choriocapillaries. E) Sarks V: geographic atrophy. F) Sarks V: choroidal neovascularization. G) Sarks VI: disciform scar. NFL, nerve fiber layer; GCL, ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer; ONL, outer nuclear layer; PR, photoreceptors; RPE, retinal pigment epithelium; BM, Bruch’s membrane; chor, choroid. Reproduced with permission from Malek G, Lad EM (2014) Cell Mol Life Sci 71, 4617-4636.

OPEN IN VIEWER

Fig. 2

Bubble plot demonstrating % (n) in each category of Alzheimer’s disease (AD) and age-related macular degeneration (AMD) by severity. Size of black circle corresponds to number of cases in each category.

In order to further analyze the data, univariable logistic regression was performed to determine the association between various clinical and demographic variables and AMD (Table 3). In univariable analysis, female sex and race were significantly different between those with and without AMD (p < 0.05). AD, glaucoma, atherosclerosis, diabetes mellitus, hypertension, and depression were not significantly associated with AMD (all p > 0.05). A multivariable logistic regression model demonstrated that there was no significant positive association between AMD and AD after controlling for sex and race, or after controlling for all covariates (Table 4).

DISCUSSION

Studies examining the relationship between ocular disease and AD are ongoing due to the importance of this investigation. Both diseases impose a high degree of burden for both physicians and patients, and there is significant loss to follow-up in the AMD population [43, 44]. This can be compounded in patients with co-morbid neurocognitive impairment requiring significant caregiver support related to frequent and time-consuming visits. Further exploration of the link between AMD and AD may lead to a better understanding of the overlap in pathophysiology between these two diseases and of relevant targets useful for future clinical trials.

Due to the many barriers in enrolling and studying patients with AD and AMD, characterizing the relationship between these conditions in the same patient population has been challenging [35]. Recruitment and accurate examination of moderate to severely cognitively impaired or visually impaired patients can be limited, and many patients may have asymptomatic AMD and have not received a formal diagnosis. A major limitation of all prior work investigating the association of AD and AMD has been the reliance on clinical criteria for AD diagnosis. Despite standardized clinical criteria for AD diagnosis (National Institute of Aging-Alzheimer Association Criteria), and ongoing research on AD biomarkers, the gold standard for AD diagnosis remains the neurohistopathology analysis [45], whose sensitivity and specificity for a clinical diagnosis of “probable AD” is superior over clinical diagnosis. A clinical diagnosis of AD is difficult and requires access to specialists, as well as numerous imaging and laboratory tests [46]. In addition, the AD population in previous clinical studies had a high likelihood of being biased toward milder disease [20 –26], as patients with advanced dementia are less likely to receive coordinated care [32 , 48]. Because the postmortem diagnosis is the gold standard for AD, the findings from the current work are more accurate than prior epidemiologic studies that relied on non-specific clinical diagnoses [20–26 , 31–34]. Overall, our findings from an institutional autopsy database fail to support a positive association between AD and AMD, even when AMD is categorized by severity.

The association between AMD with AD has been a subject of debate in the literature. Several studies have suggested a weak association between AMD and generalized cognitive decline [20 , 49]. However, other work focusing on AD has found inconsistent results. Investigations based in Taiwan and Australia reported significant associations between AMD and AD after adjustment for several confounders with multivariable analysis, even though the AD sample size was only 22 in the Australian study [22 , 50]. In two large prospective Taiwanese studies, AMD and AD were significantly associated after controlling for age, monthly income, geographic location, urbanization level, hyperlipidemia, diabetes, hypertension, stroke, ischemic heart disease, and cataract status (p < 0.001) [22, 50]. In the retrospective Australian study, multivariable analysis was used to control for age, smoking, hypertension, high and low-density lipoproteins, history of cataract surgery, and APOE ɛ4 carrier status (p < 0.001). Most recently, Lee and colleagues prospectively studied the risk of developing AD and found a significant association with AMD after controlling for age, sex, education, APOE genotype, and smoking (HR 1.50, p < 0.001) [51]. Although these epidemiological studies suggested a possible association between AD and AMD, multiple studies have cast doubt on the link between these conditions after controlling for potential confounders. In the Rotterdam Study, Klaver and coworkers found that among patients with later stage AMD there was an increased risk of incident AD, but this association was no longer significant after controlling for atherosclerosis and smoking [52]. Another study similarly found an initial association between AD and AMD that was also not present in the multivariable analysis [53]. A population-based study by Baker et al. found an association between low digit symbol substation test score and early AMD, which became more robust when controlling for multiple potential confounders [31]. However, the authors did not note any relationship between low Mini-Mental State Exam scores, dementia, or AD with early AMD [31]. Keenan and colleagues did not detect an increased risk of developing AD following the diagnosis of AMD or of being diagnosed with AMD following AD [32]. Finally, a recent meta-analysis identified 21 studies that evaluated the association of AD dementia with AMD (as outcome) and of AMD with dementia/AD (as outcome) and found that patients with dementia or AD were at risk for AMD, especially late AMD. AMD was also significantly associated with increased risk of AD and with poorer cognitive function [27].

We used a large database of ocular and brain specimens with histopathologic diagnoses so that our diagnoses would be accurate. We did not identify a significant or positive association between AD and AMD, even when considering severity of AMD. We also did not find any significant difference in the prevalence of non-AD dementia between patients with and without AMD. A number of other groups have also evaluated histopathology findings. Blanks and coworkers have previously noted a substantial loss of ganglion cells in both the central and peripheral retina in autopsy eye specimens of patients with tissue-proven AD [54, 55]; however, unlike our study, AMD diagnoses were not specifically captured. In a small study of 17 eye specimens with AD, Williams et al. found evidence of macular degeneration in only two of the eyes with AD [56]. the Sarks stage was not determined for these two cases. Loffler and colleagues demonstrated increased anti-amyloid precursor protein immunoreactivity in ganglion cells and in the retinal pigment epithelium of eyes with AMD, but did not report on the proportion of subjects with AD [57]. Other studies have also examined post-mortem eyes of patients with and without AMD for evidence of AD, but they have not identified consistent AD markers [58 –60].

There are several possible explanations for our findings. Studies have often demonstrated a common neuronal degeneration mechanism for AMD and AD, but the temporal relationship is unclear. Brain degeneration may result in retinal degeneration via an unrelated sequence of events, and it is not known whether amyloid-β deposition is related to the neurodegenerative changes seen in the retina of AMD patients [15, 34]. Also, investigations of the role of APOE as a susceptibility gene have generated variable conclusions. Early studies reflected a possible protective effect with APOE4 allele expression and increased risk with APOE2, while more recent examinations have both supported [63 –65] and contradicted [33, 66] these population-based associations. Others suggest that APOE4 only delays diagnosis of AMD [67] or that an interaction with smoking status can significantly increase associated risk [63, 66].

Although our study did not detect a significant association between histopathologic diagnoses of AMD and AD, there is a growing body of evidence that examination of both the central and peripheral retina, may reveal other changes that are characteristic of AD [68 –70]. For example, a significantly greater burden of retinal amyloid-β₄₂ plaques (p = 0.0063) has been documented in the mid-peripheral retina of AD patients compared to matched controls [69]. Ultra-widefield imaging has also demonstrated an association between AD and both peripheral hard drusen and peripheral vascular changes [70]. Thus, several studies are underway to evaluate whether noninvasive retinal imaging can detect novel biomarkers of AD such as drusen deposits, vascular changes and other retinal pigment epithelium RPE changes [68].

Our study has several limitations. This is a single-institution study of eye specimens and corresponding brain specimens from patients over age 75 that underwent postmortem autopsy. Patients that did not undergo autopsy were not captured. The analysis was cross-sectional, so the temporal relationship between AMD and the development of AD was not evaluated. Finally, additional potential confounders, such as genotype, smoking, or other environmental factors that may mediate or modify the effect of the relationship between AMD and AD were not available for this analysis. Our study only controlled for demographic and clinical confounders that were recorded in the medical record and were statistically significant (p < 0.05) in the univariable analysis. This is a common approach taken in model-building, especially in smaller datasets. However, even if all variables were included in the multivariable model, no positive association was identified between AMD and AD. Despite these limitations, the current study has several important strengths. Unlike previous epidemiologic work that relied on surrogate markers or chart review for AMD and AD diagnoses, our study employed validated histopathologic criteria to establish these two diagnoses and provide robust staging of disease severity. Thus, patients with histopathologic evidence of disease but without clinical symptoms or clinical diagnoses are correctly classified, which minimizes bias. Further, we controlled for potential confounders and examined any potential effect modification based on severity of disease.

In conclusion, this cross-sectional study sought to examine the histopathological prevalence of AD in patients with and without AMD using autopsy specimens. We found no evidence of a positive association between AMD and AD, even when severity of AMD was considered. Based on this evidence, future studies should not focus on the shared characteristics between these aging degenerative diseases but instead be directed towards further characterizing the distinct pathophysiology of AMD and AD such that novel therapeutic approaches for these debilitating aging diseases can be developed.