Relationship between the distribution of intra-retinal hyper-reflective foci and the progression of intermediate age-related macular degeneration

Intra-retinal hyper-reflective foci (IHRF) on optical coherence tomography (OCT) have been increasingly implicated as one of the most important predictors for progression to late-stage age-related macular degeneration (AMD), commonly defined as atrophy and/or macular neovascularization (MNV) [1‐11]. First reported by Khanifer et al. in 2008 [12], these OCT biomarkers have been defined as discrete, > 3-pixel size, well-circumscribed hyper-reflective lesions within the neurosensory retina, having a reflectivity at least comparable to the retinal pigment epithelium (RPE) band [13, 14].

Migration of stressed RPE cells into the retina as well as inflammatory cells has been proposed as potential originating sources of these lesions [15]. Microglial cell proliferation and subsequent trans-retinal migration from the outer retina inwards during the course of inflammation in various disease processes including AMD have been proposed as a mechanism by which IHRF may arise [16, 17]. Other investigators have proposed RPE as a more plausible source of origin of these foci, especially those eyes that are destined to develop late AMD [18‐20]. Previous reports have also shown that migration of RPE cells towards the deep capillary plexus (DCP) is a critical initial event in the development of intraretinal neovascularization (type 3 MNV) over time [19, 20]. Furthermore, a recent study by Cao et al. [21] conceptualized the origin of IHRF from “sloughed retinal pigment epithelial (RPE) cells” and proposed a progression sequence based on the available literature, OCT evidence, and immune-histochemical analysis of donor eyes [19‐24].

Given the apparent prognostic and pathophysiologic importance of IHRF [4], understanding their prevalence, distribution, and evolution of these lesions over time would appear to be of importance. The Amish Eye Study provides a useful resource for the study of AMD biomarkers over time as multimodal imaging data was collected prospectively at baseline and 2 years using a standardized protocol, and the Amish represent a relatively homogenous population with regard to both environmental and genetic factors which minimizes confounding factors [25‐27].

Thus, in this study, we analyze the extent and distribution of IHRF over 2 years in eyes with intermediate AMD (iAMD) [28, 29] from elderly Amish individuals, and evaluate the association of these IHRF parameters with the development of late AMD [30, 31].

In this retrospective longitudinal, Institutional Review Board-approved (University of Pennsylvania, University of Miami, Case Western Reserve University, and University of California—Los Angeles) study, we analyzed en face spectral-domain OCT images at baseline and 24 months later. The study was performed in accordance with the Health Insurance Portability and Accountability Act and adhered to the tenets of the Declaration of Helsinki.

The mean age ± standard deviation [SD] (range) of the 71 patients with iAMD at baseline was 72.27 ± 10.52 (51–97) years, 36 (50.7%) were female. Of these, 52 eyes (43.3%) of 42 patients had the evidence of both iAMD and IHRF, which were included in the final analysis [mean age ± SD (range) = 74.17 ± 9.37 (52–97) years, 24 were female]. IHRF were present bilaterally in 10 patients (20 eyes), and unilaterally in 32 patients (16 in the right eye and 16 in the left eye). Twenty-three eyes (19.0 %) showed progression to late AMD at the 2-year follow-up period. Figure 2 illustrates an example of a drusen with overlying IHRF which evolves to a region of atrophy by month 24. Additionally, 28 eyes with intermediate AMD which did not have IHRF at baseline developed IHRF over the 2-year period. Individuals who progressed were significantly older (76.30 ± 6.01 years), compared to the subjects who did not progress to late AMD (65.19 ± 10.14 years); P <0.001.

Quantitative analysis of this IHRF subgroup revealed a total of 243 IHRF lesions (average 4.6 per eye) at baseline, and 604 (11.6 per eye) [increase by ×2.48] at the 2-year follow-up. In those eyes that showed progression to late AMD, however, the IHRF count increased from 121 to 340 [increase by ×2.86]. Figure 3 shows the distribution of IHRF in various retinal slabs at baseline and at 2 years, in eyes that progressed to late AMD and in those that did not.

With regard to IHRF distribution, the IHRF count (average ± standard deviation, SD) at baseline by slab was as follows (Table 1): slab 1 = 58 (0.04 ± 0.35); slab 2 = 132 (0.1 ± 0.88); slab 3 = 49 (0.04 ± 0.49); slab 4 = 4 (0.0 ± 0.07); and slab 5 = 0. At the 2-year follow-up, the counts were as follows: slab 1 = 194 (0.14 ± 1.35); slab 2 = 287 (0.21 ± 1.77); slab 3 = 100 (0.07 ± 0.79); slab 4 = 23 (0.02 ± 0.26); and slab 5 = 0. There was a statistically significant increase in IHRF counts in slabs 1 (P=0.001), 2 (P=0.001), 3 (P<0.001), and 4 (P=0.02) over 2 years, though at both baseline and year 2, slab 4 had significantly fewer IHRF compared to outer slabs located closer to the RPE. The unweighted k value for intergrader repeatability was 0.90 (95% CI: 0.80–0.95; P <0.001) for IHRF quantification.

In this study, we evaluated the extent and distribution of IHRF over time and their impact on the risk for progression from intermediate AMD to late AMD over 2 years. We observed that IHRF are more extensive in the outer retinal layers compared to the inner retina and that IHRF more than double in number over a 2-year period. We also observed that that the risk of progression to late AMD over 2 years is significantly increased with the number of IHRF in slab 1 (OR = 2.13, 95% CI = 1.40–3.25; P=<0.001) and 2 (OR = 1.46, 95% CI = 1.14–1.86; P=0.002) spanning from roughly the inner border of ONL to the outer most portion of the neurosensory retina (Table 2). Also, the distribution of IHRF in each slab bears a significant risk of progression from iAMD to late AMD (OR = 1.21, 95% CI = 1.07–1.38, P = 0.003). Furthermore, IHRF distribution in the same slabs [slab 1 (OR = 1.74, 95% CI = 1.10–2.75; P=0.02) and 2 (OR = 1.38, 95% CI = 1.02–1.87; P=0.04)] appeared to be associated with the greatest risk and were the only independent predictors of progression. Similar trend of risk association was observed with the change in IHRF in each retinal slab over 2 years (Table 3). These findings highlight the fact that not only the presence but also the distribution of IHRF is important in terms of the risk for progression to late AMD.

The lack of an association between IHRF in one of the innermost slabs (slab 4) and progression to late AMD may be a finding of particular interest. On the one hand, if one assumes that these slab 4 IHRF are migrated RPE cells and further assumes that a greater distance between these foci and the RPE implies a greater duration of migration time, one might anticipate that such an eye may be more prone to late AMD since IHRF would have been present in the eye for a longer time [11, 21]. On the other hand, this finding may suggest that IHRF that are located in the inner retina may be distinct and of a different source of origin than IHRF in the outer retina (slabs 1–3) [19, 21‐23]. For example, IHRF in the outer retinal layers may be more likely to be distressed RPE cells that have migrated into the retina [35, 36]. We have observed that the choriocapillaris is more severely impaired in regions in which there are IHRF and have hypothesized that this relative ischemia may trigger these cells to migrate into the retina to seek nourishment from the deep capillary plexus (DCP) of the retina [19, 37‐39]. Furthermore, we have hypothesized that these RPE cells stimulate the DCP to produce type 3 MNV, which is typically preceded by the presence of IHRF [38, 39]. If the intended target of migrating RPE cells is the retinal DCP, there may not be significant drive for these cells to migrate further internally and may explain why the major risk for progression to advanced AMD is conferred by IHRF in the outer retina [19]. In contrast, IHRF located in the inner retina (slab 4 in our case) may be inflammatory cells, microglia, or other cells that have arrived from the retinal circulation rather than having come from the RPE layer [15‐17]. Testing these hypotheses, however, will likely require prospective studies with higher-resolution novel imaging devices (e.g., adaptive optics combined with fluorescence lifetime imaging) which are not currently available. It is notable that IHRF were never observed in the innermost slab (slab 5) which is primarily composed of the ganglion cell layer and nerve fiber layer. It is uncertain what properties of these innermost layers preclude the development of migration of IHRF to this location.

Although the number of IHRF in the inner retina at baseline was not a significant predictor of progression, an increase in IHRF over 2 years was associated with higher risk for progression to late AMD over 2 years on univariate analysis. While this was not an independent predictor on multivariate analysis, this does suggest that increasing IHRF in the inner retina may still be a biomarker of disease progression. It should be noted that increasing IHRF in the outer retina was also associated with progression to late AMD, and an increase in the outer most slabs (slabs 1 and 2) was both independent risk factors for progression.

Similar to previous studies, our eye level analysis indicated that the presence of any IHRF was associated with a higher risk for progression to late AMD [7, 11, 13, 21]. Because of this increased risk, subjects with intermediate AMD and IHRF have been suggested as potential candidates for early intervention trials of potential dry AMD therapies. Recently, post hoc analyses have shown that both pegcetocoplan (FDA approved) and avacincaptad can reduce the conversion from incomplete RPE and outer retinal atrophy (iRORA) to complete (cRORA) [40]. Thus, there has been interest in determining whether treatment could slow eyes with IHRF from progressing to iRORA and/or cRORA [40‐43]. The rate of these conversions within 2 years, however, may be relatively low, which may make these biomarkers unfeasible for use as clinical trial endpoints. Of note, our study demonstrated that IHRF can dramatically increase in number over a 2-year period. As increasing numbers of IHRF are associated with a greater risk for progression, preventing this increase with potential therapeutic agents may provide an alternative and more clinically feasible endpoint.

Our study has some limitations which must be considered when assessing our findings. First, although the OCT data was collected as part of prospective cohort study, the analysis in this report was not pre-specified. Second, while we quantified the number of IHRF, other morphometric parameters such as the shape and size of these lesions were not considered. Third, IHRF were quantified in somewhat arbitrary slabs that were based on a percentage of the retinal thickness and not in specific retinal layers. Although we considered segmenting the various retinal layers and quantifying the IHRF in each layer, the thickness of the layers varies throughout the macula, and important characteristics like reflectivity of certain layers like Henle’s layer may be impacted by underlying pathology such as drusen. This can decrease the confidence of graders to assess the accuracy of segmentation and further IHRF analysis. Furthermore, the advantage of only requiring segmentation of the RPE and ILM boundaries is that these structures can be evaluated more reliably, and segmentation errors can be readily identified and corrected or excluded. Fourth, while the Amish population offers some advantages with regard to relatively uniform genetic and environmental (e.g., no smoking, similar diet) background, our observations in the Amish may not necessarily extrapolate to the general population. Finally, although we counted a large number of IHRF, the number of eyes was relatively small. Our study also has several strengths including a standardized OCT protocol and the use of certified graders with demonstrated high grading reproducibility.

In summary, the extent, distribution, and progression of IHRF appear to be associated with the risk for progression to late AMD. Once IHRF are apparent, their numbers can dramatically increase, even over a 2-year period. IHRF are most prevalent in the outer retinal layers and these outer IHRF seem to be associated with the greatest risk for progression to late AMD. The cellular origin of IHRF at different levels of the retina warrants further study.