Article Text
Abstract
Background/aims Drusen, the pathognomonic lesion of age-related macular degeneration, are dynamic and undergo both growth and regression. Using histochemistry to localise esterified cholesterol (EC), we investigated small drusen to discover signs of dynamism.
Methods Flat mounts of Bruch's membrane were prepared from peripheral retinas of six donor eyes without chorioretinal pathology that were preserved within 6 h of death. Tissues were pretreated with ethanol to extract native unesterified cholesterol, incubated with cholesterol esterase and stained with filipin to bind unesterified cholesterol that was newly released by hydrolysis. Tissues were imaged with wide-field epifluorescence microscopy. Diameters were measured and internal substructures (shells, lakes) assessed using previous descriptors.
Results Of 676 drusen with mean diameter of 26.87 µm, 41.6% were stained homogeneously and 45.7% had lakes of pooled EC. Clusters of 2–7 drusen with similar staining patterns accounted for 25.3% of drusen. Increased EC content near the druse rim (shells) occurred in 10.5%.
Conclusions Over half of very small drusen at the edge of clinical detectability have evidence for internal remodelling, suggesting that both formative and removal events are present early in the druse lifecycle.
- Macula
- Pathology
- Retina
- Degeneration
Statistics from Altmetric.com
Introduction
Drusen, the pathognomonic lesions of age-related macular degeneration, are dynamic, exhibiting both growth and regression.1 Large population-based epidemiology studies indicate that eyes with many small drusen eventually exhibit larger and softer drusen with worse prognosis for progression.2 These findings imply expansion and merging at the level of individual lesions. Growth and clustering of drusen over months to several years have been documented by angiography, optical coherence tomography, retro-mode scanning laser ophthalmoscopy and fundus autofluorescence.3–5 Drusen can also regress, fading to spots of geographic atrophy.6 ,7 A druse at any given moment is balanced between formative and removal processes, which, if ongoing, simultaneously constitute remodelling and may be cyclic.5 While valid systems for experimental study of the drusen lifecycle are still being developed,8 ,9 histological review of human drusen can supply useful clues to these processes.
Lipid is the earliest discovered10 and largest single volumetric component of drusen, accounting for ≥40% of hard druse volume.11 This component is attributed to an accumulation of apolipoprotein (apo) B and E containing lipoproteins, rich in esterified cholesterol (EC), secreted basolaterally by the retinal pigment epithelium (RPE) into Bruch's membrane (BrM) for clearance.12 Previously, we, with others, described internal substructures within drusen defined by the relative amounts of esterified and unesterified cholesterol (EC and UC) among other components. One subregion type is a ∼15 µm diameter central core near BrM, originally defined by lectin-binding properties, which are also UC-rich, EC-poor and containing immunoreactivity for non-fibrillar amyloid.13–15 A second subregion type is thin shells at the surfaces of domed drusen that are strongly positive for EC and apo E and C-I immunoreactivity.13 ,16–18 A third subregion type was polygonal or comma-shaped EC-rich lakes delimited by phospholipid monolayers, which ultrastructurally corresponds to pooled neutral lipid in soft drusen.3 ,19–21 These regions were distinct from other druse subregions defined by the concentration of ß-amyloid and zinc, and from other particulate components such as melanin, lipofuscin and exosomes.22–25
The size of the smallest drusen detectable in the clinic varies with imaging technology, with base diameters of 30–50 µm reported.4 ,26–29 Histologically confirmed drusen <30 µm diameter are believed to be undetectable clinically.26 The purpose of our current study is to determine constraints to drusen remodelling processes by examining small drusen at or below the limits of clinical detectability, stained for the demonstration of EC. Whereas we used previously analysed randomly oriented sections of manually isolated and pelleted RPE-capped drusen,13 ,24 here we analysed RPE–BrM flat mounts30 to ensure consistency of viewing angle and to maximise sample size. Many drusen were found within groups, and all drusen in a group had the same staining characteristics.
Methods
All research adhered to the tenets of the Declaration of Helsinki, and Institutional Review Board approval was obtained prior to initiation of any studies. From the Alabama Eye Bank, we obtained eyes from six female Caucasian donors (64–86 years) with no grossly visible maculopathy, ≤6 h postmortem. Eyes were preserved by immersion in 4% paraformaldehyde in 0.1 M phosphate buffer for 24 h following removal of the anterior segment and then stored in 1% paraformaldehyde at 4°C until used. The BrM–choroid whole-mount preparation and EC staining procedure with filipin, a fluorescent polyene antibiotic specific for the 3-β-hydroxy group of sterols including UC, has been described.31 ,32 UC is a ubiquitous cell membrane component. For that reason, it is detectable in choroidal cells throughout a BrM–choroid whole-mount preventing unobstructed viewing of BrM.32 Thus, we turned our attention to EC only, which is the specific storage and transport form of cholesterol. Age-related Bruch's membrane deposits and drusen are rich in lipoprotein-derived EC and stain specifically with filipin with our reported EC staining protocol.32 ,33 In brief, after removal of the anterior segment and vitreous, the neurosensory retina was detached from the optic nerve head, and the RPE was brushed away. The BrM–choroid complex was separated from the sclera, and large choroidal vessels and connective tissues were removed with fine tweezers and art brushes. Tissue flatness was ensured by mounting on organosilane-pretreated coverslips with BrM apposed to the glass. After pretreatment with Proteinase K, UC was extracted by 70% ethanol. EC was hydrolysed by cholesterol esterase (15 µg/mL; Roche Diagnostics, Indianapolis IN) for 5 h at 37°C. UC released by hydrolysis was bound by filipin during a 2.5 h incubation. Coverslips were inverted and mounted on to glass microslides with Glycergel (DAKO, Glostrup, Denmark).
Specimens were examined using differential interference contrast microscopy (Eclipse 80i, Nikon Instruments Inc, Melville, New York, USA) and a 60× oil immersion plan apochromat objective (numerical aperture=1.4). Images were captured with a CCD camera (QImagingRetiga 4000R Fast) and the accompanying software (Qcapture V.2.8.1, QImaging, Burnaby, British Columbia, Canada). Images of the same areas were also taken using wide-field epifluorescence with UV filter cubes (UV-2A, Chroma Technology Corp., Rockingham, Vermont, USA). Intact drusen were evaluated for the presence of cores, lakes and shells,24 plus additional measures of confluence and completeness of staining (table 1). The longest diameters of whole drusen and internal cores, if present, were measured with ImageJ (V. 1.42g, NIH, USA). If drusen were merged, size was determined for each single druse within the confluent group. We performed statistics with Biostat 2009 Professional (V.5.8.4.3, AnalystSoft Inc). t test was performed to analyse drusen size of different subgroups.
Results
A total of 676 extramacular drusen were analysed. The mean diameter of these deposits was 26.87 µm ± 11.54 (range 7.23–93.77 µm). These drusen were less than half the diameter, on average, than those previously examined in filipin-stained, randomly oriented,13 a highly significant difference (p<0.00001; figure 1).
We found all staining patterns in all six eyes, and we regularly identified small clusters of drusen. In 283/676 drusen (41.9%), filipin fluorescence due to EC was intense and evenly distributed throughout the entire druse (figure 2H). In these drusen, other hyperfluorescent structures were undetectable due to insufficient contrast with the already highly fluorescent druse interior. In other drusen, EC-rich lakes were commonly observed. Of 305/676 drusen with lakes (45.1%), 59.0% (n=186) had numerous lakes (figure 2E) and 37.7% (n=119) had only scattered lakes (figure 2D). In contrast, shells were present in only 71/676 drusen (10.5% of total). A shell is normally quite thin. By focusing through a druse, in small drusen we could also distinguish quite well if the whole druse content is stained or only a small hyperfluorescent outer rim. A hypofluorescent EC-poor core, typically solitary, was found at the base of 86/676 drusen (12.7%), with a mean diameter of 8.5 µm ± 3.1 (figure 2C,D). However, we did find 15/676 drusen (2.2%) with two or three cores. The diameter of multicore drusen (33.8 µm ± 12.0) was similar to drusen with ≤1 core (26.1 µm±11.3; p=0.66). Evidence of drusen confluence was present in 171/676 (25.3%) (figure 2E–I). The median number of merged drusen in a group was 2, but some clusters contained as many as 7. It is noteworthy that all drusen in a merging group exhibited the same staining characteristics. Drusen with complete staining were on average 28.8 ± 12.4 µm diameter, whereas drusen with signs of substructure were 25.5 ± 10.7 µm (p<0.001).
Discussion
This is the largest number of drusen subject to detailed histological analysis. Our principal finding in very small drusen at the edge of clinical detectability is evidence for internal remodelling of EC in over half of observed lesions. Further, drusen with apparent internal remodelling were not larger than drusen with homogeneous interiors, suggesting that these events are present early in the druse lifecycle.
A strength of this investigation is that the filipin method applied to extracted and hydrolysed tissue is very specific for EC.34 A second strength is that the whole-mount technique allows visualisation of all drusen sizes, especially smaller ones, access to a large number of drusen, and the ability to focus through a whole. In contrast, in our previous studies using pellets of isolated drusen, which also included a large sample, drusen were dissociated from their natural surroundings and subject to mechanical damage, and smaller drusen may be overlooked during isolation. The limitations of this study are that we analysed only extramacular drusen, which are common in aged eyes, because they are biomechanically robust and easy to handle24 and we stained for EC only because staining for UC labels cell membranes throughout the thickness of the whole-mounted choroid.32
Although the range of mechanisms underlying druse remodelling remain to be determined, we can use the current data to evaluate extant hypotheses. Our biochemical model for Bruch's membrane neutral lipid accumulation posits a steady release of very low density lipoprotein-size apolipoprotein B,E-containing lipoprotein particles from RPE, either individually or in small groups within coated membrane-bounded bodies, into Bruch's membrane, as part of a programme of recycling lipids taken up for outer retinal nutrition.8 ,35–37 Thus, a homogeneous distribution of EC-containing particles may be considered the ground state onto which other physical processes such as pooling are imposed, as believed for atherosclerotic plaque.38 We previously suggested that UC-rich, EC-poor cores could be explained by neutral pH cholesterol esterase activity that converted EC to UC by hydrolysis in the extracellular compartment.13 Cellular sources of the putative esterase are the RPE-overlying drusen, choriocapillary endothelium and transiently appearing dendritic cells and macrophages, already known for their proximity to drusen and abundance in choroid.39 ,40 The professional antigen-presenting dendritic cells are particularly attractive because compelling images demonstrated their proboscis-like intrusion into small drusen.41 Those observations were interpreted by their discoverers as evidence for dendritic cells serving as a nucleation site for druse initiation. Because dendritic cells perform important surveillance functions in lipid-laden atherosclerotic intima,42–44 it is equally possible that these or other cells are also participating in clearing Bruch's membrane and drusen, as evidenced by core formation.
We not only found merging drusen but also could identify numerous small groups of drusen of similar staining characteristics. One possible mechanism is a patchy heterogeneity the RPE cells with differing gene and/or protein expression based on genetic cellular mosaicism.45 ,46 RPE mosaicism is generated on a genetic or epigenetic level during normal aging, resulting in a varying RPE phenotype or genotype.46 ,47 The RPE protein expression changes with age and might be influenced by local age-related alteration of associated structures such as Bruch's membrane.48
Prophylactic laser treatment of large, risk-conferring drusen was a modality predicated on mobilising or stimulating naturally occurring mechanisms for clearing sub-RPE lipid that sets the stage for choroidal neovascularisation.49 Although clinical trial results were inconclusive,50–52 the idea that endogenous mechanisms should be identified and characterised for possible clinical exploitation still has merit.53 It remains to be seen whether our observations in small drusen are applicable to the large drusen targeted by these therapies to date. In any case, these ideas may be eventually testable in disease-in-a-dish model systems that incorporate RPE cells and dendritic cells in coculture.8 ,9 ,54
References
Footnotes
-
Contributors All authors meet the BJO authorship and contributor criteria.
-
Funding (CAC) National Institutes of Health grant EY06109, EyeSight Foundation of Alabama, Research to Prevent Blindness, Inc, Macula Vision Research Foundation.
-
Competing interests None.
-
Ethics approval University of Alabama at Birmingham IRB.
-
Provenance and peer review Not commissioned; externally peer reviewed.