Article Text

Download PDFPDF

Bruch's membrane and choroidal macrophages in early and advanced age-related macular degeneration
  1. S Cherepanoff1,2,
  2. P McMenamin3,
  3. M C Gillies4,
  4. E Kettle5,
  5. S H Sarks6
  1. 1Anatomical Pathology, SEALS, Prince of Wales Hospital, Randwick, NSW, Australia
  2. 2Department of Pathology, School of Medical Sciences, University of New South Wales, Sydney, Australia
  3. 3Department of Anatomy & Human Biology and Centre for Ophthalmology and Visual Sciences/Lions Eye Institute, University of Western Australia, Crawley, WA, Australia
  4. 4Save Sight Institute, University of Sydney, Sydney, NSW, Australia
  5. 5Children's Medical Research Institute, Westmead, NSW, Australia
  6. 6Suite 4, 15 Parnell St, Strathfield, NSW, Australia
  1. Correspondence to Dr Svetlana Cherepanoff, Anatomical Pathology, SEALS, Prince of Wales Hospital, Level 4, Campus Centre, Barker St, Randwick NSW 2031 Australia; svetlana.cherepanoff{at}


Aim To determine the sub-macular Bruch's membrane (BrM) macrophage count and the choroidal and BrM macrophage immunophenotype in normal eyes and in eyes with early and advanced age-related macular degeneration (AMD).

Methods BrM macrophages were counted in 125 human eyes (normal, normal aged, early AMD and geographical atrophy), and CD68 and inducible nitric oxide synthase (iNOS) immunohistochemistry was performed on 16 human eyes (normal, normal aged, early AMD, geographical atrophy and disciform scarring). All eyes were examined clinically ante mortem. Results were correlated with histopathological features, including basal laminar deposit and membranous debris, and with clinical fundus appearance.

Results CD68+ macrophages were found in the choroid of normal human eyes, and did not express iNOS. Expression of iNOS by choroidal macrophages (as well as endothelial cells and pericytes) was associated with: (1) recruitment of macrophages to BrM in early AMD eyes with soft drusen or thick continuous basal laminar deposit, corresponding to clinically detectable soft drusen or pigment changes; and (2) active disciform scarring. iNOS expression was absent in BrM macrophages, suggesting immunomodulatory differences between the choroid and BrM. The highest BrM macrophage counts were found in eyes with subclinical choroidal neovascularisation.

Conclusion The presence of extracellular deposits (soft drusen and thick continuous basal laminar deposit) is associated with macrophage recruitment to BrM and alteration in the immunophenotype of resident choroidal macrophages.

  • AMD
  • macrophage
  • BLamD
  • membranous debris
  • CNV

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Macrophages have been implicated in the pathogenesis of age-related macular degeneration (AMD) since their discovery within drusen1–3 and choroidal neovascular membranes.4 Significantly, macrophages were found adjacent to Bruch's membrane (BrM) breaks, through which subclinical new vessels grew from the choroid into the retina.1 5 6 Animal studies showed that acute, laser-induced breaks in BrM resulted in macrophage recruitment and choroidal neovascularisation (CNV),7 8 further supporting a pathological role for macrophages in AMD. More recent animal models suggest that the normal recruitment and turnover of resident choroidal macrophages may, in fact, prevent the development of AMD-like lesions.9 10

These apparently contradictory observations may be explained by the extraordinary heterogeneity of tissue macrophages. Networks of choroidal macrophages are found in normal mouse,11 rat12–15 and human16 eyes. These resident macrophages appear to be programmed for phagocytosis without eliciting inflammation. By contrast, ‘classically activated’ inflammatory macrophages express inducible nitric oxide synthase (iNOS),17 18 an enzyme that converts l-arginine to nitric oxide, a damaging reactive nitrogen species with cytotoxic19 and pro-angiogenic properties.21 22

We postulate that the progressive accumulation of the extracellular deposits that characterise early AMD—basal laminar deposit (BLamD) and membranous debris23—may alter the phenotype of local macrophages. We also postulate that macrophages associated with CNV may be phenotypically different to those associated with geographical atrophy (GA) or disciform scarring.

In the present study, the infiltration of BrM by macrophages expressing CD68 (a lysosomal marker specific to macrophages/monocytes24) and iNOS, was examined in 141 human eyes (representing normal, normal aged, early AMD, GA and disciform scarring), and correlated with clinical fundus appearance.



Archival slides of human eyes were selected from a large clinicopathological collection of more than 600. Eyes were processed for histopathology and graded (histopathological groups I–VI) according to the appearance of BLamD as previously described.6 All of the eyes were examined (best corrected visual acuity and direct funduscopy) and most were photographed (Zeiss 30° fundus camera; Carl Zeiss Meditech AG, Jena, Germany) during life.

For histopathology and BrM macrophage counts, the best foveal sections (picro-Mallory) of 125 eyes, representing histopathological groups I–V were examined with a Leitz Dialux microscope, using the PL APO 40× (NA-0.75) objective. Unstained paraffin sections of 16 eyes, belonging to histopathological groups I–VI, were used for CD68 and iNOS immunohistochemistry.

Histopathological definitions and grading: BLamD and membranous debris

BLamD is abnormal basement membrane-like material found between the plasma membrane of the retinal pigment epithelium (RPE) and its basement membrane. Its accumulation correlates with progressive RPE abnormalities25 and it is found in all eyes with early or advanced AMD.26–31 Early BLamD stains blue with the picro Mallory method, with faint striations, while late BLamD stains red.23 The presence of late BLamD corresponds to moderate to severe RPE abnormalities and clinically evident pigment changes.23 Eyes were divided into histopathological groups depending on the appearance of BLamD,6 as this reflects the stage of degeneration of the RPE. Briefly, eyes with no BLamD (group I) or those with only patchy early type BLamD (group II) were considered normal or aged. Eyes with early AMD had continuous BLamD, thin in group III and thick (greater than half the height of a normal RPE cell) in group IV. Late AMD eyes had geographic atrophy (group V) or disciform scarring (group VI).

Membranous debris is composed of membrane fragments32 33 and lipid,34 35 and is specific for AMD.29 It is first found in eyes with early AMD as basal linear deposit (BLinD), seen ultrastructurally as a thin layer external to the RPE basement membrane.27 29 32 Membranous debris can also accumulate as small focal collections internal to the RPE basement membrane, forming basal mounds23 (not clinically visible). Larger collections external to the RPE basement membrane become soft drusen29 (clinically visible if >30 μm). Due to the exclusive use of light microscopy in this study, the presence of BLinD was not determined. However, BLinD has been found in all eyes in which there is a thin continuous layer of BLamD.23 Thus, the amount of submacular membranous debris in increasing order is: (1) no basal mounds (presumed BLinD); (2) basal mounds only; (3) basal mounds and soft drusen.

Histopathological definitions: subclinical CNV

Subclinical CNV was defined as growth of fibrovascular tissue into the sub-RPE space through breaks in BrM in the macular region, not evidenced on funduscopy.

Clinical definitions

Clinical data from the last ophthalmological visit prior to death was used. Best corrected visual acuity was converted from Snellen to the logarithm of the minimum angle of resolution (logMAR) for statistical purposes. The fundus was graded as: (1) normal, which included up to five small (<63 mm) hard drusen; (2) hard drusen clusters (eight or more small hard drusen); (3) soft drusen (≥63 μm) without pigment abnormalities; (4) soft drusen (≥63 μm) with pigment abnormalities; and (5) pigment abnormalities alone.

BrM macrophage counts

Macrophages were identified by their abundant pale pink-staining cytoplasm, large oval or bean-shaped nuclei and fine chromatin. Macrophages within the zone bounded by the inner margin of the inner collagenous zone and the outer margin of the choroidal capillaries/intercapillary pillars were counted.

An eyepiece graticule (which equalled 280 μm in total with the 40× objective) was used to examine the foveal field and one graticule length on either side, giving a total of three fields (840 μm). In eyes with hard drusen, soft drusen or RPE changes, additional fields (up to five, all within the inner macula) were examined. Cell counts were divided by the number of fields examined, multiplied by three, and expressed as mean count per three 40× fields.


Sections were deparaffinised, rehydrated and pre-treated in 1% hydrogen peroxide to block endogenous peroxidase and 10% normal horse serum to block non-specific binding. Sections were incubated (1 h, 37°C) with primary monoclonal mouse anti-human antibodies (anti-iNOS (Transduction Laboratories BD Biosciences, California, USA); anti-CD68 (Zymed Invitrogen, Carlsbad, California, USA)). Pepsin digestion preceded blocking in the case of anti-CD68. After washing (Tris), sections were incubated (30 min, 37°C) with biotinylated anti-mouse IgG (Vector Laboratories, Burlingame, California, USA) and re-washed prior to incubation (30 min, 37°C) with ABC (Vector Laboratories). iNOS staining was visualised with 3,3-diaminobenzidine (DAB) (Sigma-Aldrich, St Louis, Missouri, USA) in nickel ammonium sulphate and peroxide solution (10 min, 37°C), producing a dark grey reaction product. CD68 staining was visualised with Vector SC Blue (Vector Laboratories), producing a dark blue reaction product. All sections were washed, dehydrated and coverslipped. Peripheral blood smears were used as positive controls.


Statistical analysis was performed using SPSS for Windows (v.15.0; SPSS Inc., Chicago, Illinois, USA). The Student t test was used to compare mean BrM macrophage counts and visual acuity measurements between histopathological groups. Contingency tables of histopathological group versus the presence of subclinical CNV were evaluated by the χ2 and Fisher's exact tests. A p value ≤0.05 was considered statistically significant.


Groups I and II: normal and normal aged eyes (none or patchy BLamD)

The fundus was normal in all group I and II eyes (18 eyes), with the exception of two group II eyes with pigment abnormalities in which hard drusen clusters were present (table 1). BrM macrophages were absent in group I and II eyes, except in the two group II eyes with pigment changes, in which occasional macrophages were found beneath the hard drusen. CD68 cells, with ill-defined cytoplasmic borders, were present in the choroid of all group I and II eyes, intermixed with choroidal melanocytes (figure 1C and D). Expression of iNOS was absent in BrM or the choroid, but was occasionally seen in the basal aspect of RPE cells (figure 1E).

Table 1

Clinicopathological features versus Bruch's membrane and choroidal macrophage findings in study eyes

Figure 1

Morphological and immunohistochemical features of normal (group I) and normal aged (group II) eyes. (A) A typical group I eye from a 73-year-old patient with a normal fundus. The retinal pigment epithelium (RPE) is normal, there is no basal laminar deposit (BLamD) and little Bruch's membrane (BrM) hyalinisation. (B) In a representative group II eye from a 72-year-old patient with a normal fundus, BrM hyalinisation extends down the intercapillary pillars to reach the base of the choroidal capillaries (black arrows), resulting in widened intercapillary spaces. The black arrows also define the area within which BrM macrophages were counted. Note a very fine layer of early BLamD just internal to the hyalinised BrM, which in group II eyes does not form a continuous submacular layer. In the place of choroidal capillaries, cells with small, round nuclei and scanty cytoplasm (white arrow) are occasionally seen. Rarely, small foci of fibrous tissue (asterisk) were found in the same location. (C) CD68 cells with ill-defined cytoplasmic borders were found in the choroid (among melanocytes) of group I and II eyes, as shown in this group I eye from a 62-year-old patient with a normal fundus. (D) A group I eye from the same patient as (A), with numerous CD68+ choroidal cells intermixed with melanocytes (black arrows). An intravascular monocytoid CD68+ cell is seen within a choroidal vessel (white arrow). (E) Occasional inducible nitric oxide synthase (iNOS) expression was found in the basal aspect of RPE cells (arrows) in group I and II eyes, as seen in this group II eye from a 75-year-old patient with a normal fundus. iNOS expression was not found in cells within BrM or the choroid in eyes from groups I and II. (A, B) Picro-Mallory stain. (C–E) CD68 and iNOS immunohistochemistry. The CD68 colour product is dark blue and the iNOS colour product is dark grey. Scale bars, about 100 μm. CC, choriocapillaris; ONL, outer nuclear layer.

Groups III and IV: early AMD (thin and thick continuous BLamD)

The fundus was normal in 15 (79%) group III eyes without basal mounds (presumed BLinD), and in six (40%) eyes with basal mounds. In group IV, all eyes had clinically evident pigment changes, with or without soft drusen (table 1). CD68 cells were again seen in the choroid of all, and basal RPE expression of iNOS was more frequently present (table 1). BrM macrophages were found in eyes with clinically evident early AMD, that is, pigment changes with or without soft drusen (except in three older eyes with normal fundi). These were noted occasionally replacing choroidal capillaries (figure 2A and C) or in the intercapillary pillars (figure 2B). Occasional iNOS+ cells with indistinct borders were found in the choroid (figure 2D). Choroidal vascular endothelial cells (figure 2E) and perivascular cells (figure 2F) were also found to express iNOS. Notably, no iNOS expression was seen in BrM macrophages.

Figure 2

Morphological and immunohistochemical features of eyes with early age-related macular degeneration (AMD) (groups III and IV). (A) A representative group III eye from an 88-year-old patient with pigment changes and soft drusen. Bruch's membrane (BrM) macrophages (black arrows) are seen in the place of a lost choroidal capillary, and possibly eroding an intercapillary pillar. There is mild RPE disorganisation and thin continuous basal laminar deposit (BLamD) of the early (pale blue staining) type (square brackets). A soft druse (asterisk) appears empty due to loss of lipid contents during processing. A pigment-laden cell (white arrow) is also found replacing a choroidal capillary. (B) A group IV eye from an 89-year-old patient with pigment changes only. The retinal pigment epithelium (RPE) appears more disorganised compared with the eye in (A) and there is thick continuous BLamD, with continuous segments of the late, red staining type (square brackets) type overlying the early type. A BrM macrophage is found apparently eroding an intercapillary pillar (arrow). (C) CD68 immunohistochemistry in a group IV eye from an 80-year-old patient with soft drusen and pigment changes, confirming the presence of BrM macrophages (arrows). Note again the network of CD68 cells in the choroid. (D) Inducible nitric oxide synthase (iNOS)+ cells in the choroid (white arrow) in the same eye as in (C), have similar morphology and localisation as CD68 cells. Basal expression of iNOS by the RPE cells (black arrows) was also found in group III and IV eyes. (E) Choroidal vascular endothelial cell iNOS expression (arrow) in a group IV eye from a 74-year-old patient with soft drusen and pigment changes. (F) iNOS expression by cells with ill-defined cytoplasmic borders found between choroidal melanocytes (white arrow) and near a choroidal vessel (arrows) of the same eye as in (C). (A, B) Picro-Mallory stain. (C–F) CD68 and iNOS immunohistochemistry. Scale bar, about 100 μm. CC, choriocapillaris.

Groups III and IV: BLamD and membranous debris versus BrM macrophages

Where a thin continuous layer of BLamD (group III) was present (figure 2A), BrM macrophage counts were highest in eyes with soft drusen (table 2). Where a thick continuous layer of BLamD was present (group IV), high BrM macrophage counts were recorded in all eyes (table 2) irrespective of the amount of membranous debris present (figure 2B).

Table 2

Bruch's Membrane (BrM) macrophages, membranous debris and subclinical choroidal neovascularisation (CNV)

BrM macrophages and subclinical CNV

Subclinical CNV was present in sixteen of the 125 eyes examined histologically. With the exception of two eyes from group V, all eyes belonged to groups III and IV. Macrophages, as identified histologically, were noted frequently eroding intercapillary pillars or BrM (figure 3A).

Figure 3

Morphological features of eyes with subclinical choroidal neovascularisation (CNV). (A) A group III eye from an 88-year-old patient with soft drusen and pigment changes. Thin continuous basal laminar deposit (BLamD) of the early type (asterisk) is overlain by mildly disorganised retinal pigment epithelium (RPE). Bruch's membrane (BrM) macrophages are seen eroding BrM (arrow) beneath a soft druse, with contents emptied by processing and overlain by persistent early type BLamD. (B) A group III eye from an 81-year-old patient with soft drusen only. Fibrovascular tissue is seen invading the sub-RPE space (asterisk) via a focal break in BrM (arrow). (A, B) Picro-Mallory stain. CC, choriocapillaris. Scale bar, about 100 μm.

In eyes with thin continuous BLamD (group III), subclinical CNV was only found in eyes with soft drusen (table 2), suggesting an association with the amount of membranous debris present. However, this relationship did not hold once continuous BLamD (group IV) was present (table 2).

Eyes with subclinical CNV had significantly higher BrM macrophage counts than eyes without subclinical CNV in groups III (p = 0.011; mean difference 13.32; 95% CI 4.51 to 22.13) and IV (p = 008; mean difference 6.94; 95% CI 1.95 to 11.93). Of the 23 paired eyes in the study, greater numbers of BrM macrophages were also found in the fellow eye of eyes with subclinical CNV (p = 0.002; mean difference 9.18±2.52; 95% CI 3.96 to 14.41) compared with the fellow eyes of eyes without subclinical CNV.

Group V: GA

Compared with groups I–IV, fewer CD68 cells were seen beneath areas of atrophy; however, they were distributed at the margins of areas of atrophy, adjacent to viable RPE cells (figure 4A). iNOS expression was not found in BrM or the choroid of these eyes (figure 4B).

Figure 4

CD68 and inducible nitric oxide synthase (iNOS) immunohistochemistry in eyes with geographical atrophy (group V) or disciform scarring (group VI). (A) CD68 cells were seen at the edges of areas of atrophy, aligned with the remnant retinal pigment epithelium (RPE) (arrow) in a group V eye from an 84-year-old man. Few CD68 cells were found in the choroid immediately beneath the atrophic region. (B) In the same eye in (A), iNOS+ cells were not found within Bruch's membrane (BrM) or choroid. (C) A group VI eye from an 80-year-old man with an ‘inactive’ flat disciform lesion without haemorrhage or exudates. Microscopically, subretinal fibrous tissue surrounding a large feeder vessel was present. Numerous plump CD68 cells are seen within the scar (arrows). (D) In the same eye in as (C), iNOS+ cells are absent. (E) A group VI eye from 76-year-old man with an exudative (‘active’) disciform lesion. A fibrovascular membrane consisting of numerous small vessels and cells, some pigmented, is seen. While CD68 cells (arrow) are less numerous in this scar compared to the eye in (C), plump CD68 cells are present in the choroid (arrowhead). (F) In the same eye as (E), iNOS expression is seen in choroidal perivascular cells (arrows) and endothelial cells. Scale bars, about 100 μm. CC, choriocapillaris.

Group VI: disciform scarring

While numerous CD68 cells were present within both the inactive (figure 4C) and active (figure 4E) scar, iNOS expression was only seen in the active lesion, within large perivascular cells in the choroid and choroidal endothelial cells in this region (figure 4D and F). These observations are consistent with a two-dimensional analysis of human CNV membranes, which found higher macrophage numbers in an eye with active CNV compared with an eye with inactive CNV.36


Observations that macrophages may both facilitate and prevent AMD-like lesions underscore the complex relationship between this heterogeneous group of cells and AMD pathogenesis. Unravelling this paradoxical functional relationship requires the phenotypic characterisation of both macrophages resident in the normal choroid, and those associated with AMD.

Although macrophages have been known for some time to be a feature of the normal choroid of mice and rats11–14 they are less characterised in the human eye.16 The current study confirms that while macrophages are present in the normal human choroid, they do not express iNOS, and are thus phenotypically unlike classically activated ‘inflammatory macrophages’.18 Local regulation of macrophage function, known to be important in other organs,18 is probably also important in the uveal tract. While macrophage recruitment is facilitated by RPE-derived macrophage chemoattractant protein-1 (CCL2),37 immunomodulatory factors such as complement factor H,38 transforming growth factor-β1,39 interleukin-1 receptor antagonist,40 pigment epithelium-derived factor, somatostatin and interleukin-1041 probably prevent recruited macrophages from developing an inflammatory phenotype.

In contrast to resident choroidal macrophages, frequent BrM macrophages are only seen in eyes with early AMD in which there is either: (1) thin continuous BLamD and large amounts of membranous debris (soft drusen); or (2) thick continuous (late type) BLamD and significant RPE abnormalities. BrM macrophages are therefore only present when there is clinical evidence of early AMD (soft drusen or pigment changes).

The presence of injured/apoptotic cells or accumulated extracellular debris is known to attract components of the innate immune system,42 which participate in opsonisation and chemotaxis.43 In early AMD, the complement components and acute-phase proteins deposited in drusen44 and BLamD45 may recruit macrophages to BrM and promote phagocytosis. Accumulation of oxidised cholesterol, a by-product of photoreceptor function, in drusen and BLinD may directly attract macrophages and promote vascular endothelial growth factor (VEGF) expression.46 Recruited macrophages may also contribute to, or alter, the biochemical composition of drusen or BLamD.47 Animal studies show that recruited macrophages may be derived from the vasculature or from the resident choroidal pool.11–14 In mice, ‘resident’ choroidal macrophages have a greater turnover or replenishment rate than the nearby macrophages of the neural retina, the microglia.15

Our results suggest the possible sequence of events in macrophage recruitment to BrM and CNV pathogenesis may be: (1) replacement of choroidal capillaries or intercapillary pillars; (2) erosion of BrM including intercapillary pillars; (3) facilitation of CNV by breakdown of the physical barrier provided by BrM and elaboration of factors promoting angiogenesis.

Expression of iNOS in normal human eyes at the basal aspect of RPE cells may represent a response to the relative ischaemia48 associated with normal, age-related choroidal capillary fallout.49 In eyes with BrM macrophages, iNOS was expressed by choroidal cells with indistinct cell borders and by choroidal perivascular and endothelial cells, but not by the BrM macrophages themselves, suggesting local immunomodulatory differences between BrM and the underlying choroid. While double labelling with CD68 was not preformed, the morphology and location of the iNOS+ cells in the choroid were highly reminiscent of (CD68+) resident choroidal macrophages seen in adjacent sections and of those recently demonstrated by Schroedl et al.16 Macrophage nitric oxide production has known cytotoxic and proangiogenic effects in the choroid and retina21 and is involved in pathological ocular neovascularisation.50 Both iNOS and VEGF expression have been noted in macrophages and endothelial cells in excised neovascular membranes.51 52

We postulate that in eyes with soft drusen or thick continuous BLamD, macrophage recruitment to BrM facilitates CNV by erosion of the structural barrier, while iNOS expression by choroidal macrophages (and possibly by choroidal pericytes and endothelial cells) promotes angiogenesis. This proposal is supported by the following observations: (1) high BrM macrophage counts are found in eyes with subclinical CNV; (2) choroidal iNOS expression is seen in active disciform scarring but not in inactive scarring or in GA and (3) the fellow eye of eyes with CNV have high BrM macrophage counts, consistent with the clinical observation that these eyes are at increased risk of developing CNV.53

In summary, resident choroidal macrophages are present in normal human eyes and do not express iNOS. The presence of soft drusen or continuous thick BLamD is associated with macrophage recruitment to BrM and iNOS expression in the choroid, suggesting altered resident choroidal macrophage phenotype and local immunomodulatory differences between BrM and the choroid. Understanding the role of macrophages in the pathogenesis of AMD will require better characterisation of these differences and improved understanding of factors regulating normal choroidal macrophage recruitment and turnover.


We thank Adeline Akkari for assistance with immunohistochemistry. SC was supported by a National Health and Medical Research Council (Canberra ACT, Australia) PhD Scholarship (No. 008103).



  • Funding National Health & Medical Research Council, Canberra, ACT, Australia (SC).

  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval This study was conducted with the approval of the University of NSW Human Research Ethics Committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Linked Articles

  • At a glance
    Harminder S Dua Arun D Singh