Abstract
Purpose:
We evaluated the association of hyperreflective foci (HF) observed in early and intermediate age-related macular degeneration (AMD) with known AMD risk alleles.
Methods:
In this pilot case-control study, HF were defined as lesions with reflectivity equal or higher than the retinal pigment epithelium band in spectral domain optical coherence tomography (SDOCT). Hyperreflective foci in the outer nuclear layer and photoreceptor complex were evaluated in 518 individuals with early and intermediate AMD. Definite presence of HF was defined as at least 10 HF in all SDOCT scans. Genotyping was performed for 22 single nucleotide polymorphisms (SNPs). Associations between AMD severity stages, HF, and SNPs were determined by logistic regression analyses.
Results:
Hyperreflective foci (n ≥ 10) were significantly associated with AMD severity and the association was strongest with intermediate AMD (odds ratio [OR], 8.45; P = 1.092*10−8). Independently, HF showed associations with ARMS2 rs104909/HtRA1 rs11200638 (OR, 1.64; P = 0.017), CFH rs1061170 (OR, 1.70; P = 0.011), and APOE4/TOMM40 rs2075650 (OR, 2.26; P = 0.005) variants. Within the group of intermediate AMD, associations were similar (ARMS2 rs104909/HtRA1 rs11200638 OR, 1.79, P = 0.010; CFH rs1061170 OR, 1.77, P = 0.013; APOE4/TOMM40 rs2075650 OR, 1.98; P = 0.034) and showed additional trending associations with VEGFA rs943080 variant (OR, 0.59; P = 0.024). After Bonferroni-correction for 22 SNPs, none of the associations was statistically significant (P ≤ 0.0023).
Conclusions:
The presence of HF is related to AMD severity. Despite limited power of this pilot study, our results suggest an association of HF with polymorphisms in ARMS2/HTRA1, CFH, APOE4/TOMM40, and VEGFA genes which could be triggered by modification of the extracellular matrix, altered complement system or lipid metabolism.
Age-related macular degeneration (AMD) is a multifactorial neurodegenerative disease with a broad spectrum of phenotypic varieties. Early forms are characterized by the presence of drusen and pigmentary changes identified with fundus photographs (FP) while choroidal neovascularization (CNV) and central geographic atrophy (GA) represent late forms.
1 In recent years, AMD diagnosis was facilitated by the wide use of noninvasive high-resolution spectral domain optical coherence tomography (SDOCT) and even different pathologic morphologic features including hyperreflective foci (HF) could be identified.
2–7
Hyperreflective foci are defined as “discrete, well-circumscribed lesions with equal or greater reflectivity than the RPE band.”
3 Hyperreflective foci have been observed in the outer nuclear layer (ONL), in proximity to the drusen and RPE-atrophic areas.
2,5 In the longitudinal Age-Related Eye Disease 2 (AREDS2) study, HF proliferation and migration were associated with the development of GA.
8
Hyperreflective foci were associated with hyperpigmentations observed on FP
3,5 and showed dynamic changes during anti-VEGF therapy in neovascular AMD. They may present in vivo inflammatory components of the disease and, therefore, may be used as a new clinical biomarker.
4,6,9 In histologic analyses, intraretinal HF represented cholesterol crystal precipitations of phagocytic origin, which also may originate from microglia.
10
In AMD, heritability accounts for approximately 71% of the condition.
11 In recent years, >30 AMD risk loci have been identified and many of the genes at these loci encode components of the complement system, lipid metabolism, and extracellular matrix biology.
12,13 To our knowledge, it has not yet been investigated whether genetic polymorphisms are associated with the presence of HF, which may shed light on their etiology.
The purpose of this study was to analyze associations of HF with AMD severity and known AMD-related genetic risk polymorphisms.
All calculations were performed using SPSS software version 21.0 (IBM Software and Systems, Armonk, NY, USA). Associations between HF (≥10 vs. <10) and SNPs (0, no major allele; 1, major allele; 2, homozygous major allele; additive model) were calculated separately by logistic regression analyses after adjustment for age, sex, site (University of Cologne/University of Nijmegen), and AMD-severity (early/intermediate). Additional subanalysis were conducted for the intermediate AMD cases after adjustments. After Bonferroni correction for 22 SNPs, P values ≤0.0023 were accepted as significant. All SNPs were tested for Hardy-Weinberg equilibrium (HWE) and SNPs outside the equilibrium (P < 0.05) were not analyzed.
Mean age was 74.92 ± 10.80 years for subjects with definitive presence of HF and 72.27 ± 8.17 years for subjects with less than 10 HF. Our cohort was predominated by female sex (total number of females, 325 [62.74%]) but the distribution of sex in the groups HF ≥ 10 and HF < 10 was similar (number of females 42 [66.67%] vs. 283 [62.20%]).
Multivariate regression analyses of age, sex, and AMD severity stages with HF (n ≥ 10 vs. n < 10) were performed. Hyperreflective foci ≥ 10 were significantly associated with age (P = 0.022, odds ratio [OR] = 1.04, 95% confidence interval [CI], 1.01–1.07), and intermediate AMD (OR, 8.45; P = 1.092*10−8; 95% CI, 4.07–17.58).
In the present study, we investigated and quantified HF in the ONL and in the photoreceptor complex of early and intermediate AMD patients and found a strong association between HF and intermediate AMD in our cohort. Furthermore, the data suggested an association of genetic variants in the genes ARMS2/HTRA1, CFH, APOE, and VEGFA genes with the presence of HF.
Hyperreflective foci are distinctive SDOCT features which are observed repeatedly in association with photoreceptor layer thinning
2 and RPE atrophy.
20 Hyperreflective foci are considered a risk factor for disease progression to GA.
8 Hyperreflective foci phenotype seems to be tightly connected to large drusen considering its significant association with intermediate AMD and drusen are not only hallmark of AMD, but also are considered as biomarkers of local immune-mediated inflammation.
21,22
In this study, we found a trend that the HF phenotype was associated with
CFH rs1061170. Complement factor H (CFH) is a key regulator of the alternative pathway of the complement system and genetic changes can lead to dysregulation of the complement cascade and trigger drusen formation with subsequent accumulation of macrophages to Bruch's membrane (BrM).
23,24 Thus, it is possible that HF phenotype is a result of the proinflammatory responses associated with polymorphism in
CFH.
Furthermore, the study also suggests an AMD-severity independent association of HF phenotype with
ARMS2/HTRA1 genes. There is debate on whether the
ARMS2, HTRA1, or both genes are associated with AMD and the functions of their gene products are not fully understood. However, the ARMS2 and HTRA1 proteins have been reported to interact with extracellular matrix (ECM) proteins and with proteins involved in remodeling of BrM.
25,26 Alteration of BrM can lead to a disruption of signaling pathways between RPE and BrM
25 and aberrant expression of ECM proteins can influence immune cell activation.
27 Earlier studies proposed HF as migrating RPE cells
3,28 and recent histologic findings of a donor eye with vitelliform dystrophy described the intraretinal HF as a complex of lipofuscin granules, melanolipofuscin granules and melanosomes.
28 In contrast, histologic findings of AMD-donor eyes suggested that HF in AMD are very likely not solitary RPE granules alone but may consist of lipid droplets accumulated within phagocytes, such as microglia and macrophages.
10 This latter finding is supported by the observation of HF in other retinal diseases with neuroinflammatory components, such as diabetic maculopathy
29,30 or retinitis pigmentosa.
31
Microglial cells are the primary resident immune cells located in the inner retina involved in the pathogenesis of AMD, diabetic retinopathy, or retinitis pigmentosa.
16,17,32,33 Once activated, microglia can migrate to the outer retina and to the subretinal space
14–16,18,34 and induce structural and functional alterations of RPE.
17 In human retinas with GA, activated microglia were found in the ONL where they phagocytize cell debris.
16 In mouse and rat studies, VEGF blockade has been shown to inhibit/reduce microglia activation.
35,36 Moreover, decrease in HF quantity also was observed after anti-VEGF treatment in AMD and diabetic maculopathy patients.
4,6,9,30 Interestingly, in this study, HF also might be associated with rs943080 variant at the
VEGFA gene in intermediate AMD patients, which has been reported previously to have influence on treatment response.
37 Besides,
VEGFA variant rs943080 is known to be inversely associated with the presence of AMD.
12
Although HF also were associated with a variant of
APOE4/TOMM40 (translocase of outer mitochondrial membrane) in our study, the numbers of the cases regarding
APOE4/TOMM40 variants in our study are very restricted. Nevertheless, this gene also is known to be associated with neurodegenerative Alzheimer disease.
38 Human apolipoprotein E (ApoE) not only is essential for lipid transport and metabolism, but also is involved in neuronal repair, remodeling and degeneration.
38 The presumed nature of HF to consist partly from lipid droplets
10 could be related to an association with
ApoE variants. It is known that RPE and microglia can express ApoE.
39,40 In mouse model,
APOE4 genotype has been shown to be associated with increased microglia activation
41 and an ApoE-dependent microglial migration was demonstrated.
42 It is assumed that there is an autoregulatory feedback between microglia activation and ApoE expression for neuroprotective purpose.
40 It is possible that altered lipid metabolism may enhance microglial migration or activation, with induced formation of HF. However, to evaluate the association of
APOE4/TOMM40 variants with HF presence, studies with larger cohorts are needed.
Our study has several limitations. The phenotype-genotype associations were evaluated for 518 subjects but the findings should be confirmed in additional cohorts. Hyperreflective foci quantification also is challenging and not yet standardized. However, the inclusion of two graders reduced the possibility of subjective judgment of the phenotypes. We limited the HF grading to scans from ONL to RPE to increase accuracy. In the future, the distribution of HF in additional retinal layers and also advanced AMD could be evaluated.
Our study has a limited number of patients that showed HF (n = 63), which limited multiple testing criteria. Nevertheless, this is a pilot study to evaluate the association of HF with AMD risk alleles. After Bonferroni correction for 22 SNPs, none of the associations was statistically significant as our sample size was not sufficient for the analyses. Thus, our results are suggestive and should be validated in an independent cohort.
In summary, we found that HF are strongly related to AMD severity in patients with early and intermediate AMD. Furthermore, we suggested an association of ARMS/HTRA1, CFH rs1061170 with the HF-phenotype found in outer retinal layers and in the photoreceptor complex. Our findings supported the notion that modification of BrM or an altered complement activation triggered by ARMS2/HTRA1 and CFH polymorphisms, or altered lipid metabolism triggered by APOE may have a role on the formation of HF. Furthermore VEGFA polymorphism seems to be a protective factor for HF. Assuming that HF are accumulated inflammatory components, such as activated microglia, further histopathologic studies are needed to understand the relationship between these polymorphisms, inflammatory cells, and the HF phenotype.
Supported by the Deutsche Forschungsgemeinschaft (DFG-FOR2240).
Disclosure: L. Altay, None; P. Scholz, None; T. Schick, None; M. Felsch, None; C.B. Hoyng, None; A.I. den Hollander, None; T. Langmann, None; S. Fauser, Novartis (C), Roche (C), Bayer (C), Quantel (C)