Hyperreflective Foci in Age-Related Macular Degeneration are Associated with Disease Severity and Functional Impairment

Abstract

Purpose:

To analyze presence of hyperreflective foci (HRF) across different age-related macular degeneration (AMD) severities and examine its correlation with other structural and functional AMD features.

Design:

Longitudinal, single-center, case-control study.

Participants:

One hundred and fifty-eight participants aged > 50 years old with varying AMD severities (including no AMD).

Methods:

Color fundus imaging was used to assess AMD severity and hyperpigmentation (PGM) presence. Subretinal drusenoid deposits (SDD) and HRF were detected on OCT volumes. The correlations of HRF with additional AMD features were evaluated using linear and logistic mixed-effects models. One study eye per participant underwent dark adaptation (DA) testing to measure rod intercept time (RIT) for structure function associations. Eyes were followed longitudinally and changes in AMD severity and RIT were measured relative to HRF presence.

Main Outcome Measures:

The primary outcome was presence of HRF, which was compared with presence of other AMD features and DA impairment.

Results:

One hundred and fifty-eight participants (median baseline age of 73.1 [interquartile range (IQR) = 66–79] years) contributing 1277 eye visits were included. Hyperreflective foci (HRF) were detected more frequently in higher AMD severities. Hyperreflective-foci presence was significantly associated with PGM presence (odds ratio 832.9, P < 0.001) and SDD presence (odds ratio 9.42, P = 0.017). Eyes with HRF demonstrated significantly longer DA (median 27.1 [IQR = 16–40] minutes) than those without HRF (13.5 [10–22] minutes) but less than eyes with SDD only (40 [28–40] minutes). Highest RIT values were found in eyes with both HRF and SDD (40.0 [40–40] minutes). Age and HRF explained a similar proportion of RIT variability as age and SDD. Eyes that developed HRF demonstrated baseline RITs closer to eyes with HRF at baseline, compared with eyes that never developed HRF (29.1 [16–40], 38.5 [22–40] versus 13.1 [10–22] minutes; Kruskal–Wallis P < 0.001).

Conclusions:

The progressively increased presence of HRF in higher AMD severities, and its correlation with previously associated AMD biomarkers, suggests HRF is an important OCT feature adding to the understanding of disease progression. Hyperreflective foci presence was associated with delays in DA, indicating HRF is a marker for visual cycle impairment.

Financial Disclosure(s):

Proprietary or commercial disclosure may be found after the references.

Age-related macular degeneration (AMD) is the leading cause of blindness in industrialized countries.^1,2 The development and progression of AMD, from aging changes to early, intermediate, and late AMD stages, is currently defined by the presence of drusen, pigmentary changes, and geographic atrophy or evidence of neovascularization based on fundus photography.³ Consensus groups, such as Classification of Atrophy Meeting and Consensus on Neovascular Age-Related Macular Degeneration Nomenclature Study Group, seek to establish an alternative corresponding OCT framework for defining criteria for late stage AMD, such as neovascularization, atrophy, and their precursors.^4–6

Although the historical natural history data based on color fundus photography (CFP) remain the gold-standard modality for AMD staging, additional disease-specific features have been identified using OCT.^7,8 One of these OCT features is hyperreflective foci (HRF), which are described as well-circumscribed lesions with a reflectivity equal to or greater than the retinal pigment epithelium (RPE) and can correspond to hyperpigmentation (PGM) on CFP.⁹ Impaired RPE-derived cells and/or lipid-filled cells have been linked to HRF on histology.^10,11 These lesions often occur over drusen, where it is hypothesized that reduced oxygenation drives the anterior migration of RPE cells, manifesting as HRF.¹² Another proposal, coexistent with the other, is that HRF in AMD are microglia, a commonly found immune cell in the inner retina, which migrates from the inner retina to the outer retina when activated in environments with degeneration.¹¹

Eyes containing HRF have been reported to be associated with functional consequences of impaired dark adaptation (DA).¹² However, little is known regarding the context and especially the AMD stages in which HRF occur. With an increased reliance on imaging modalities additional to CFP, a better understanding of how and where OCT-based features fit in our CFP-based AMD grading are needed. Furthermore, the correlation with other AMD features sharing anatomic or functional overlap, such as PGM and subretinal drusenoid deposits (SDDs), also known as reticular pseudodrusen, is not well explored. Both will contribute to a better understanding of the disease pathophysiology and are necessary to further characterize HRF as a biomarker in AMD.

Therefore, this current study investigated the presence of HRF across different stages of AMD and its relation to cooccurring SDD and PGM. We also studied the functional correlates of HRF with rod intercept time (RIT), a measure of DA, first examined through cross-sectional study and then in a longitudinal manner.

Methods

Study Population

The study population included participants of an ongoing longitudinal study of DA in AMD that consists of subjects aged ≥ 50 years with and without AMD. Participants were recruited from the eye clinic at the National Eye Institute, National Institutes of Health, Bethesda, MD, between May 2011 and March 2020. The study was approved by the institutional review board of the National Institutes of Health, and the tenets of the Declaration of Helsinki were followed. Although not a clinical trial, the study is registered on ClinicalTrials.gov (NCT01352975). All participants provided written informed consent after the nature and possible consequences of the study were explained. A secondary objective of the original DA study was to analyze functional characteristics relative to AMD structural risk factors as conducted in this analysis.

After examination, eyes were classified as study or fellow eye based on the presence of features of intermediate AMD and advanced AMD, respectively, as described previously and briefly summarized here.^13,14 If neither eye had large drusen (≥ 125 μm) or if large drusen were present bilaterally, either eye could be assigned as the study eye, at the investigator’s discretion. If large drusen were present in only one eye, that eye was designated as the study eye. Participants with advanced AMD in only one eye had the nonadvanced eye assigned as the study eye. The study eye also had to meet the additional criteria of a visual acuity ≥ 20/63. Only the study eye of each participant underwent DA testing.

The exclusion criteria were as follows: (1) advanced AMD in both eyes at baseline; (2) any other active ocular or macular disease (e.g., glaucoma, diabetic retinopathy, or Stargardt disease); (3) a condition preventing compliance with the study assessment; (4) cataract surgery within 3 months before the enrollment; (5) history of vitamin A deficiency; (6) high oral intake of vitamin A palmitate supplement (≥ 10,000 IU/day); and (7) active liver disease or a history of liver disease.

Participants underwent study visits at baseline and 12, 24, 36, 48, and 60 months. If there was a progression to late AMD in a participant’s eye, they were terminated from the study at that point and excluded from any further analysis.

Examination and Imaging

Participants underwent multimodal imaging (CFP, OCT, fundus autofluorescence [FAF], and infrared reflectance [IR]) in both eyes (study and fellow eyes) at annual visits. Both eyes (study and fellow) underwent a comprehensive, prespecified, multimodal imaging protocol The designated study eye only underwent functional testing using DA.

Color fundus photography was acquired with the TRC-50DX retinal camera (Topcon Medical Systems). Infrared reflectance and short-wavelength (488-nm excitation) FAF images and spectral–domain-OCT scans were acquired with the Heidelberg Spectralis (Heidelberg Engineering). The OCT macular cube scan consisted of 121 horizontal B scans spanning 30° horizontally and 25° vertically.

Reading Center Grading

To identify AMD morphology, including screening for the presence of HRF, SDD, and pigment, and for the assessment of AMD severity group (AMDSG), CFP, OCT, FAF, and IR images were analyzed by trained graders at the University of Wisconsin Fundus Photograph Reading Center. All grading on each individual image was conducted independent of other modalities (except SDD as described) and independent of other visit dates. Age-related macular degeneration severity grades were assigned to each eye at each study visit, based on the CFP, according to the Age-Related Eye Disease Study (AREDS) 9-Step Severity Scale (AREDS report #17).¹⁵

Common terminology was used to determine the presence or absence of any abnormality. Images were graded as either absent, questionable, present, cannot grade, or not applicable for all AMD features. Absent was given when the abnormality under consideration was not visible on the image. The grade of questionable was assigned when the grader had a 50% to 89% confidence that the abnormality is present. For an abnormality that had a ≥ 90% confidence, the grader assigned a present/definite grade. If required, the grader could add additional criteria (area/location/extent) to further classify present grades. Images received “cannot grade” if the image was of insufficient quality to confirm detections. “Not applicable” was assigned to images for insufficient information to answer the question. The reading center maintained an ongoing masked quality control program. Quality control included contemporaneous resampling methodology and was internally assessed by both the group as a whole and by particular evaluators. A 5% masked regrading of all imaging modalities was conducted for quality control.

Hyperreflective foci were determined to be present if they were present on OCT and met all of the following criteria: (1) hyperreflective dot overlying drusen or regressed drusen, (2) blocks signal underneath, and (3) location anywhere between drusen and internal limiting membrane. Additionally, the reading center’s grading of HRF intensity was qualitative, requiring lesions with similar intensity to the RPE, and there were no absolute size criteria. Qualitative size is conferred by the requirement that the lesion blocks signal underneath.

Subretinal drusenoid deposits were also assessed by the reading center for their presence categorized as definite, questionable, cannot grade, and not applicable. For a definite grade, SDD had to be present on OCT as well as an additional imaging modality (FAF, IR, or CFP). If SDD were not present on an additional imaging modality (i.e., if present on OCT only), a grade of questionable was assigned. On OCT, a cluster of SDD present on > 1 B scan was required. On IR and FAF, the description of SDD included the following: (1) hypo (dark) spots; (2) discrete clusters of dark dots that are similar in size; (3) minimum area of one half the disc diameter; and (4) pattern can be ribbon-like if confluent, reticular/giraffe-type pattern, a mix of discrete clusters and ribbon-like pattern and may have intervening hyper spots or areas. On CFP, for SDD to qualify as present, they first had to appear subretinal on OCT, then identified as yellowish in color and appearing as soft drusen arranged in an ill-defined network of broad interlacing regions.

Fundus photographs were used to evaluate the features and characteristics used previously in the grading of the AREDS and AREDS2 studies.^3,15 Specifically, areas of PGM on CFP were evaluated with spatial considerations on a 4-step scale (absent, questionable, definite < C-2, definite > C-2 where C-2 = 0.069 mm²), for locations in center and inner subfields using the standard ETDRS grid and AREDS drusen circles. At each visit, both participants’ eyes also received an AMD Severity Scale grade per AREDS report no. 17.¹⁵ Representative images are presented in Figure 1. For this analysis, a simplified, binarized version of the grading system was used, where questionable and definite detections for all fundus features were considered as positive detections given the stringent criteria of the definitions used in the grading. Eyes receiving reading center’s additional grades of not applicable or cannot grade were not included. Individual eye imaging visits were excluded if they did not receive complete multimodal imaging grading from the reading center. In addition, eyes with neovascular AMD (AMD severity 11) were not included in this analysis because they were not graded for additional feature presence by the reading center.

Figure 1.

Five participants’ color fundus photo (CFP) image and corresponding B scan on OCT. Each row represents a different phenotype and grading for hyperreflective foci (HRF), hyperpigmentation (PGM), and subretinal drusenoid deposits (SDD). A, Eye Age-related Macular Degeneration Severity grade (AMDSG) 3, − HRF/− PGM; B, AMDSG 6, + HRF/− PGM; C, AMDSG 5, − HRF/+ PGM; D, AMDSG 8, + HRF/+ PGM; (E) AMDSG 6, + HRF/+ SDD/ (− PGM).

DA Testing

All participants underwent DA testing in the study eye only at each visit using a prototype of the AdaptDx dark adaptometer (MacuLogix). The details have been described previously^13,14,16,17 but are briefly described here. In brief, the study eye was dilated, and the participant was asked to focus on a fixation light. A photoflash producing an 82% focal bleach centered at 5° on the inferior visual meridian was performed, and threshold measurements were made at the same location with a 1.7-degree diameter, 500-nm wavelength circular test spot, using a 3-down/1-up modified staircase threshold estimate procedure. The time it took for an eye’s visual sensitivity to recover to a threshold within the second component of rod-mediated DA (5 × 10⁻³ cd/m² [a decrease of 3 log units]) (RIT) was recorded. Eyes that were unable to reach the threshold before the time limit of 40 minutes were conservatively assigned a RIT value of 40 minutes.

Statistical Analysis

Statistical analysis was conducted through the statistics computing program RStudio (Version 1.2.5033) with the add-on packages lme4, plyr, dplyr, tidyr, and ggplot2.

Normally distributed data were summarized using the mean and standard deviation, whereas nonnormally distributed data were summarized using the median and interquartile range (IQR). Univariate and logistic mixed-effects models were developed to describe associations between the presence of HRF (dependent variable) and the presence of PGM, SDD, AMD stage, and age (independent variables). Mixed-effects models with the visit ID and eye ID as a random-effects term were used to account for the hierarchical nature of the data (i.e., visits nested in eye, eye nested in patient). Likelihood ratio test were applied to test for significance. Feature selection was performed using forward selection with evaluation of the models with all combinations of covariate effects, based on the Schwarz’s Bayesian Information Criterion.

P values of ≤ 0.05 were considered as statistically significant.

Analysis Plan

Because there are several analyses conducted, we will provide a brief summary of the analysis plan here.

First, the baseline characteristics and AMD severity (per AREDS report no. 17¹⁵) were reported on study and fellow eyes (excluding eyes with macular neovascularization⁵).

Next, all visits from study and fellow eyes were utilized in the analysis of structural associations of HRF, SDD, and PGM.

Visits from study eyes only were pooled for function-structure correlations. Rod intercept time, the measure of DA, was assessed, and associations with the presence of HRF and SDD were investigated in univariate and then multivariate analyses.

Last, eyes with longitudinal follow-up were used to document the progression of AMD severity according to their HRF status (presence, development, or absence). Study eyes also were also evaluated for the longitudinal changes in RIT according to their HRF status.

An outline of the cohort used for each analysis is presented in Figure S1 (available at www.ophthalmologyretina.org).

Results

Cohort Characteristics

A total 158 participants (281 eyes), with a median (IQR) age of 73.1 (66–79) years, contributing 1277 eye study visits were included in the analysis. Each participant contributed both eyes (excluding eyes with macular neovascularization⁵ or ungradable data) for structural analyses and one study eye for functional analyses and associations (see Fig S1). Median (IQR) follow-up time for subjects was 60 [24–60] months. Average AMD severity at baseline was determined to be 4.77 ± 2.77. Demographic characteristics are detailed in Table 1.

Table 1.

Demographic Characteristics of Patient Population at Baseline

AMD Severity Group in Study Eye	n (Study Eyes)	Age	n (Fellow Eyes)	Avg AMDSG in Fellow Eyes	Avg Follow-Up (Median, mos)
0	18	74.1	18	1.0	60
1	9	71.6	9	1.4	60
2	13	70.5	12	2.1	48
3	9	67.7	8	3.3	60
4	17	73.5	14	1.9	60
5	16	68.2	12	4.8	60
6	23	76.6	17	5.9	60
7	30	72.6	21	6.7	30
8	13	71.8	7	8.0	60
9	7	69.6	6	9.0	36
10	1	71.2	1	10.0	24

Eyes with neovascular disease excluded from analysis.

AMD = age-related macular degeneration; AMDSG = age-related macular degeneration severity group.

Association of HRF Across Different AMD Severity Groups

The absolute number of (eye) visits with a positive grade of HRF presence was tallied according to severity group, as demonstrated in Figure 2A. Hyperreflective foci were less prevalent in lower AMDSGs (Fig 2A). When analyzing relative prevalence (Fig 2B), the percentage of detections followed a near-linear trend with increasing severity. The highest percentage of HRF was found in AMDSG group 10 (a combined 84.3% of the total 32 visits in group 10).

Figure 2.

Number of eyes in all visits with hyperreflective foci (HRF) across all age-related macular degeneration (AMD) severity grades. A, Absolute number of eyes with HRF detected per AMD severity group, (B) HRF detected per severity group as a percentage of the number of eyes in corresponding severity group. AMDSG = Age-related Macular Degeneration Severity grade.

Cooccurrence of HRF and PGM

Of the 1277 eye visits, a total of 623 (48.8%) had PGM on CFP, whereas only 345 (28.1%) had HRF on OCT. To gain insight into the cooccurrence of AMD features in each eye assessed across different modalities, 4 mutually exclusive phenotypes were examined. Each eye, both study and fellow, was categorized in the following manner: eyes without either HRF or PGM (− HRF/− PGM), eyes with HRF but no PGM (+ HRF/− PGM), eyes without HRF but with PGM (− HRF/+ PGM), and eyes with both HRF and PGM (+ HRF/+ PGM). The majority of imaging grades, (n = 621, 48.6%), were graded as −HRF/− PGM. When PGM was present (+ PGM), there were approximately similar numbers of visits with HRF present (n = 312) and HRF absent (n = 311). A markedly smaller group (n = 33) exhibited HRF but no pigment (Table 2). Additionally, this analysis was repeated with baseline visit data only and produced comparable results, see Table S1 (available at www.ophthalmologyretina.org).

Table 2.

Total Number of Visits with AMD Features Present

	PGM		SDD
	−	+	−	+	Total
− HRF	621 (48.6%)	311 (24.4%)	834 (65.3%)	98 (7.7%)	932
+ HRF	33 (2.6%)	312 (24.4%)	228 (17.9%)	117 (9.2%)	345
Total	654	623	1062	215

Distribution of visits per AMD feature across all AMD severity groups. HRF and PGM considered independently of HRF and SDD.

− indicates visit with feature graded as absent, + indicates visit with feature graded as present.

AMD = age-related macular degeneration; HRF = hyperreflective foci; PGM = hyperpigmentation; SDD = subretinal drusenoid deposits.

In earlier stage AMD (AMDSG 0–3), 92.3% of visits appeared as − HRF/− PGM. The percentage of eyes that were −HRF/− PGM declined steadily with increasing AMD severity until reaching 0 visits in severity groups 8 and beyond. At severity group 7, 96.4% of visits had PGM, with approximately half (47.3%) of those also having HRF present. In the stages after severity group 7, even more eyes were graded as + HRF/+ PGM. The rarest phenotype of + HRF/− PGM (n = 33) peaked in severity group 6 while still only representing a small percentage (8.1%) of visits within that severity group (Fig 3A).

Figure 3.

Assessment of all visits across all age-related macular degeneration (AMD) severity groups according to presence of fundus features. Presence (+)/absence (−) of hyperreflective foci (HRF) was assessed in addition to presence/absence of (A) hyperpigmentation (PGM) and (B) subretinal drusenoid deposits (SDD). AMDSG = age-related macular degeneration severity grade.

Cooccurrence of HRF and SDD

Because SDD represent another OCT-based phenotype identified to confer increased risk of progression to late AMD¹⁸ and have demonstrated correlation with DA dysfunction, we investigated the cooccurrence of HRF and SDD. Hyperreflective foci were found in 345 eye visits (28.1%), and SDD were present in 215 eye visits (16.4%) across all AMD severities. Using a similar grouping to that described in the previous section, 4 types were analyzed across all severity groups, as seen in Table 2. The most common phenotype in our cohort was − HRF/− SDD, accounting for 834 out of the total 1277 eye visits (65.3%). More visits were found to have + HRF/− SDD (n = 228, 17.9%) than − HRF/+ SDD (n = 98, 7.7%). The double positive phenotype of + HRF/+ SDD consisted of a total of 117 visits (9.16%).

In the earlier stages of AMD (0e4), visits were graded mainly in the − HRF/− SDD phenotype (at 91.2%). With AMD severity beyond increasing AMDSG 4, the percentage of visits with HRF increased monotonically, whereas the percentage with SDD reached a relatively steady percentage of 15% to 45% in grades 7 to 10. In severity group 10, eye visits graded as + HRF/− SDD was the predominant type, accounting for 71.9% of visits (Fig 3B).

Association of DA with Presence of HRF and SDD

To investigate potential correlations between the OCT-based features HRF and SDD and DA function, analyses were performed on the population of study eyes (i.e., not fellow eyes), with each AMD feature’s association to RIT considered independently and in combination. Out of all study eye visits (n = 712, average severity grade at baseline = 4.75 ± 2.7) there were 221 (31.0%) with HRF present and 491 (69.0%) with HRF absent. Average RIT, as median (IQR), for eyes without HRF was 14.5 (10–25) minutes, in contrast to 40 (22–40) minutes for those with HRF present (Fig 4A). The RITs differed significantly between the 2 groups (Mann–Whitney U test = 22 879, P < 0.001 2-tailed).

Figure 4.

Rod intercept time (RIT) categorized by presence of age-related macular degeneration (AMD) morphology, (A) hyperreflective foci (HRF), (B) subretinal drusenoid deposits (SDD), and (C) combined presence of HRF and SDD sorted into mutually exclusive phenotypes. D, Percentage of eyes that reached the 40-min time limit during dark adaptation testing in the same groups used in (C). n = number of eyes across all visits.

In eyes without SDD, the average RIT was 15.1 (10–26) minutes and 40 (40–40) minutes in eyes with SDD (Fig 4B). Similarly, the RITs differed significantly between the 2 groups (Mann–Whitney U test = 9081, P < 0.001 2-tailed).

Eyes were categorized according to both their HRF and SDD grades (Fig 4C). Eyes with − HRF/− SDD had the widest range of RIT values, with an average RIT of 13.5 (10–22) minutes. The average RIT for eyes with + HRF/− SDD was 27.1 (16–40) mins, as compared with 40 (28–40) minutes for eyes with − HRF/+ SDD. In eyes with + HRF/+ SDD, the average RIT was 40 (40–40) minutes. There was a statistically significant difference between the RITs of the 4 different phenotypes (Kruskal–Wallis P < 0.001).

Given that there was a limit on the DA testing time, the results were analyzed according to the percentage of eyes reaching the test ceiling. Of the 445 study eye visits in the − HRF/− SDD category, only 4.0% reached the test ceiling. Eyes with + HRF/− SDD had 32.4% reach the test ceiling, in comparison with 54.3% of the − HRF/+ SDD eyes. The subset of eyes with + HRF/+ SDD had 87.1% of visits at the test ceiling (Fig 4D).

Because some of the eyes in the cohort were aged normal without AMD, this analysis was repeated with eyes with AMD severities 0 and 1 separated out as a control group. Comparing the control group with eyes with or without HRF and SDD, we found a very similar pattern. The control eyes demonstrated better DA and shorter RIT compared with eyes with AMD but without HRF or SDD. All other relationships maintained as can be seen in Figure S2 (available at www.ophthalmologyretina.org).

Analysis to Determine Relative Significance

Multivariable logistic regression was used to determine the relative significance of correlation with presence of HRF. This analysis identified PGM presence (odds ratio [OR], 832.9; 95% confidence interval, 92.38–7509.47) and SDD presence (OR, 9.4; 95% confidence interval, 1.50–59.21) each to be independently associated with the HRF presence (Table 3).

Table 3.

Multivariable Model of Hyperreflective Foci

Presence of Hyperreflective Foci
Predictors	ORs	CI	P
Intercept	0.00	0.00–0.00	< 0.001
Age	1.14	1.02–1.27	0.024
SDD (presence)	9.42	1.50–59.21	0.017
PGM (presence)	832.90	92.38–7509.47	< 0.001
SD (intercept)	94.08
SD (observations)	2.72
Random Effects
σ2	3.29
τ00 PATID	20.65
ICC	0.86
N PATID	156
Observations	712
Marginal R2/conditional R2	0.377/0.914

CI = confidence interval; ICC = intraclass correlation coefficient; OR = odds ratio; PATID = patient study identifier; PGM = hyperpigmentation; SD = standard deviation; SDD = subretinal drusenoid deposits.

Age-corrected univariate analyses with RIT as the dependent variable and HRF or SDD presence as the independent variable, respectively, are shown in Tables S2 and S3 (available at www.ophthalmologyretina.org). Although SDD presence showed a greater estimate (cf. deterioration of RIT when feature is present) of 4.10 minutes, compared with 2.43 minutes for HRF, the marginal R² was relatively similar (0.290 and 0.271, respectively). Marginal R² is a measure of how much variability of the dependent variable (here, RIT) is explained by the features (age + SDD versus age + HRF), and the similar results show that HRF are of comparable importance to explain altered DA (although leading to less impairment) as SDD.

Multivariable Analysis of DA Function

To assess the relative contributions of variables associated with RIT in univariate analyses and in the context of other variables that have been demonstrated to correlate with RIT, a multivariable model using forward selection of variables of age, color fundus AMD severity grade, pigment and OCT features of HRF and SDD was performed. The resulting model that best fit the RIT data included the variables of age, AMD severity, SDD, and HRF (marginal R² = 0.426, conditional R² = 0.917). The analysis demonstrates that the presence of HRF on OCT is significantly associated with RIT even while accounting for other relevant AMD features and grades (Table 4). The parameter estimates indicate that RIT increases more in the presence of HRF than on AMD severity grading alone but less than for the presence of SDD.

Table 4.

Multivariable Analysis of Dark Adaptation Function

	Rod Intercept Time
Explanatory Variables	Estimates (min)	CI	P
(Intercept)	−27.64	−36.07 to −19.21	< 0.001
Age (per y)	0.55	0.44–0.66	< 0.001
AMD severity (1)	2.68	−1.60 to 6.97	0.220
AMD severity (2)	5.74	2.57–8.91	< 0.001
AMD severity (3)	7.05	3.75–10.35	< 0.001
AMD severity (4)	7.00	3.75–10.25	< 0.001
AMD severity (5)	9.14	5.88–12.41	< 0.001
AMD severity (6)	11.63	8.35–14.90	< 0.001
AMD severity (7)	13.94	10.57–17.30	< 0.001
AMD severity (8)	13.29	9.69–16.89	< 0.001
AMD severity (9)	13.29	9.47–17.10	< 0.001
AMD severity (10)	12.25	7.55–16.95	< 0.001
SDDs (presence)	3.17	1.21–5.14	0.002
Hyperreflective Foci (presence)	2.05	0.89–3.21	0.001
SD (intercept)	8.14
SD (observations)	1.80
Random Effects
σ2	10.40
τ00 PATID	66.29
ICC	0.86
N PATID	156
Observations	712
Marginal R2/Conditional R2	0.448/0.925

AMD = age-related macular degeneration; CI = confidence interval; HRF = hyperreflective foci; ICC = intraclass correlation coefficient; PATID = patient study identifier; SDD = subretinal druseoid deposits.

Longitudinal Structural and Functional Changes relative to the Presence of HRF

The first and last visit imaging grades of 258 eyes with longitudinal follow-up (median [IQR] follow-up time = 60 [36–60] months) were compared and examined for the progression of AMDSG in relationship to HRF presence (Table 5). Eyes were divided into 4 types: eyes that did not have HRF present at either visit (constant −, 171 eyes) HRF present at both visits (constant +, 53 eyes), eyes with − HRF at baseline with + HRF by last visit (0 to 1, 24 eyes) and eyes with + HRF at baseline but with − HRF at the final visit (1 to 0, 10 eyes) The average starting AMDSG and change in AMDSG were analyzed and presented in Table 5.

Table 5.

Longitudinal Assessment of Eyes According to Presence of HRF

	Constant (−)	0–1	Constant (+)	1–0
N	171	24	53	10
Avg starting AMDSG	3.13	5.33	7.36	5.20
Change in AMDSG	0.585	0.750	0.509	1.90
n (study eyes)	91	13	36	5
RIT at first visits, median (IQR)	13.1 (10–22)	29.1 (16–40)	38.5 (22–40)	15.0 (14–21)
RIT at last visit, median (IQR)	17.1 (11–31)	37.1 (22–40)	40 (29–40)	19.7 (16–23)
Avg change in RIT	3.1 (0.95–8)	1.4 (0–6.3)	0 (0–2.55)	2.3 (0.6–7)
% eyes that reached test ceiling (40 min)	8.79	38.5	54.2	10.0

Distribution of visits per AMD feature across all AMDSG.

AMDSG = age-related macular degeneration severity group; Avg = average; HRF = hyperreflective foci; IQR = interquartile range; RIT = rod intercept time.

The change in RIT over follow-up was assessed in study eyes only (145 eyes), in relationship to the HRF status as seen in Table 3. The average RIT values at the first visit were 13.1 (10–22) for eyes with constant − HRF grades, 29.1 (16–40) for eyes that transition from − HRF at baseline to + HRF at final, 38.5 (22–40) for eyes that remain + HRF, and 15.0 (14–21) minutes for eyes with grading changing from + HRF at baseline to − HRF at the final. A Kruskal–Wallis test was used to determine the differences were statistically significant (P < 0.001). Interestingly, eyes with − HRF at baseline but that developed + HRF by final had baseline RIT testing that was more similar to eyes that had + HRF grades that were maintained over follow-up than to eyes that had − HRF and remained − HRF.

Discussion

This study interrogated the significance of HRF as an OCT feature relative to color image features and functional testing to understand AMD disease severity and progression. In the multidimensional analyses, we find that HRF are increasingly likely to occur with higher CFP-based AMD severity stages and to correlate with worse DA function.

The design of this study population allowed us to investigate the presence of HRF across all levels of AMD severity, as well as aged eyes without AMD. It confirmed that HRF tend not to be present in eyes without AMD or in lower AMD severity grades (< 5% in severity grades 0–3), but that their prevalence increases to 10% at AMD severity score 4 and increases monotonically across higher AMD stages, reaching 85% in eyes with geographic atrophy. The longitudinal analysis demonstrated that once an eye has HRF, it is very likely to continue to demonstrate HRF presence over years of follow-up; this shows that the finding is generally not a transient feature but rather, one that persists into advanced disease.

Hyperreflective foci have previously been found to be associated with increased risk of progression^19,20 and delayed DA (which itself is also associated with increased progression risk²¹). In the prospective AREDS2 Ancillary spectral domain-OCT Study, Christenbury et al.⁷ tracked the quantity and location of HRF from eyes with intermediate (category 3) AMD. In their study, HRF proliferated from baseline to follow-up and had a substantial distribution shift over time, from the RPE to the inner retinal layers. The presence of HRF, along with greater distribution, were predictive of geographic atrophy development.

The finding that eyes with HRF also mostly had PGM graded on CFP was expected from previous observations, including those by Ho et al.⁹ and a histologic study by Balaratnasingam et al.¹⁰ It has been debated that HRF represent proliferations of lipid-filled cells and/or anteriorly migrating impaired RPE cells.¹¹ To document intraretinal RPE pigment migration, Ho et al. ⁹ compared 3-dimensional OCT scans and CFP and found that all eyes with intraretinal RPE migration on OCT were seen also to have corresponding areas of pigment clumping on CFP, commonly occurring over areas of drusen.⁹ Laiginhas et al.²² explained these findings further by stating that PGM also correspond to regions of thickened RPE, in addition to having a connection with intraretinal HRF. Folgar et al.²³ previously reported a method of automated coregistration of PGM on CFP and HRF on OCT. Spatial correspondence of the 2 intermodal features revealed that the PGM and HRF were often colocalized, strongly indicative of representing the same lesion. However, in the current study, the finding that half of the eyes graded as having PGM present did not have HRF graded as present was unexpected, as many consider the resolution of OCT imaging makes it a more sensitive imaging modality than CFP.²⁴ In considering our data, there are several possible explanations for this discrepancy. First, it could be that the definition of HRF on OCT is more restrictive (i.e., has a higher threshold) than the assessment of pigment grading on color images. The grading of HRF was performed in a rigorous manner by the Wisconsin reading center but did require that the lesion be in a location overlying drusen or regressed drusen and also be large enough to block the signal underneath. Another explanation for the lower frequency of detections could be the spacing of B scans (60 μm),²⁵ leading to potential areas where the retina was not sufficiently sampled. Future studies are warranted that follow not only specific eyes and their status over time but follow specific HRF lesions across time, including spatial alignment and correlation between imaging modalities.

In this study, we were able to also investigate the functional correlations of HRF, as one eye of each study participant underwent DA testing, a promising functional biomarker of AMD severity.²⁶ Our findings demonstrated that, in univariate analyses, HRF presence was strongly associated with increased delays in DA. Previous research has also shown a striking correlation between SDD and rod dysfunction.²⁶ In a longitudinal study conducted by Chen et al.,¹⁴ study eyes with SDD had significantly higher rates of RIT prolongation than those without during follow-up.

Given that both HRF and SDD are observed on OCT imaging and are associated with hyperreflective alterations above the RPE,²⁷ a faster progression of AMD, and more dysfunction in DA,^12,13,28 and are both believed to result from an impaired RPE,²⁹ we investigated the cooccurrence of these features. We demonstrated that HRF have a significant association with both SDD and PGM in subjects with AMD. Additionally, 87% of eyes that had both SDD and HRF present reached the ceiling of the 40-minute DA test. However, we also establish that HRF show an independent association with delayed DA and are of comparable importance with SDD in explaining delayed RIT. Multivariable modeling demonstrated that HRF, in addition to age, AMD severity, and SDD presence, was significantly important to the performance of the model.

The longitudinal aspect of the study gave possible insights into the development of HRF. The range of average RIT values for eyes that developed HRF in between their first and last visits more closely resembled the range of RIT values for eyes that had an HRF at both visits than those that did not have an HRF at either visit. This could indicate that RIT impairment precedes HRF formation. This parallels previous studies of SDD development by Chen et al.,¹⁴ which found that eyes that developed SDD over the course of the study had initial DA testing that matched the SDD group even before having positive gradings. Additionally, Flamendorf et al.¹³ and Laíns et al.³⁰ previously reported that any abnormality present (HRF, SDD, drusen, etc.), not just within the DA testing spot but in the entire macula, had a significant association with rod dysfunction. Only 10 eyes in the longitudinal analysis were found to change from HRF present to absent. Investigating these cases, we attribute these findings to variability in imaging and grading that may have altered the capturing of a lesion on subsequent scans and also development of advanced disease, which in some cases obscured or changed the presence of the lesion.

Limitations of this study include the subjective nature of the grading involved with these feature assignments. Although the grading was performed in a rigorous manner, it relies of the subjective appreciation of features that can be subtle and time-consuming to evaluate. Ongoing work in the field seeks to develop automated grading for features relevant to AMD severity and risk of progression, including 2-dimensional imaging³¹ and OCT.^32–33 Moving to automated detection has the potential to provide a more comprehensive and quantitative evaluation. Furthermore, a limitation of our study design is that HRF were graded on a per-eye binary basis, and questions remain about the quantity and spatial distribution of the HRF over follow-up. Spatial correlation will enable further understanding of the pathophysiology of disease progression and functional cellular consequences.

Future studies are necessary to determine the precursors to the development of HRF (possibly as debris of impaired RPE/drusen) and their migration over time, paralleled by multimodal imaging and functional testing to determine the impact of HRF in different stages. The goal is to understand the implications that each biomarker has on function and the risk of progression. This would enable the development of models that will facilitate both the selection of patients for clinical trial as well outcomes that are more robust and sensitive.

Abbreviations and Acronyms:

AMD

age-related macular degeneration

AMDSG

AMD severity group

AREDS

Age-Related Eye Disease Study

CFP

color fundus photography

DA

dark adaptation

FAF

fundus autofluorescence

HRF

hyperreflective foci

IR

infrared reflectance

IQR

interquartile range

OR

odds ratio

PGM

hyperpigmentation

RPE

retinal pigment epithelium

RIT

rod intercept time

SDD

subretinal drusenoid deposit