Cerebral Microbleeds in the Elderly

Materials and Methods

The investigation consisted of 2 studies of brain tissue from elderly subjects. Initial survey study consisted of 12 subjects who underwent autopsy between January 2006 and December 2008 with no clinical diagnosis of cerebrovascular disease and were selected consecutively by the Department of Pathology, Harbor-UCLA Medical Center. Portions of subcortical white matter from frontal, parietal, occipital, and temporal lobes were sampled. Selected portions of brains were first fixed in 10% buffered formalin solution for 2 weeks, and then cut serially in 1.0 cm slices. To detect hemosiderin/iron, slices were immersion-stained using Prussian blue method. Slices (4 μm thick) were selected for paraffin embedding, sectioning (20 to 40 per brain), and hematoxylin and eosin staining.

To detect pericytes, macrophages, and amyloid deposition, detection kits based on immunohistochemistry were used according to the manufacturers’ protocols. Specifically, to detect pericytes of cerebral capillaries, a mouse monoclonal antibody against human α-smooth muscle actin (Sigma-Aldrich) was used. To detect macrophages, mouse monoclonal antibody against human CD68 (Dako North America) was used, and mouse monoclonal antibody against human β-amyloid epitope 6E10 (Covance) was used for amyloid detection. Antigen retrieval was performed by using 10 mmol/L sodium citrate (pH 6.0) at 95°C for 15 minutes. Slides were first incubated with primary antibodies at 1:2000 dilution for 30 minutes at room temperature. Slides were then incubated with a biotinylated secondary antibody. Staining was performed by using Cell and Tissue Staining Kit (R&D System) according to the manufacturer’s instruction. The slides were counterstained with hematoxylin and photographed by using a light microscope. Negative control samples were exposed to a secondary antibody with a similar IgG isotype (Cell Signaling) to the primary antibody.

For electron microscopy, tissues positive for Prussian blue staining of capillaries were selected and first fixed in 2.5% glutaraldehyde and emersion-fixed in osmium tetroxide, and then embedded in plastic. These sections were examined by using an electron microscope (Hitachi 600) according to the manufacturer’s instruction.

Follow-up study focused on basal ganglia and cortical tissue, material obtained from the University of California Irvine Alzheimer Disease Research Center, and from The 90+ Study. Blocks of lenticular nucleus from 21 subjects (who had died during the eighth, ninth, tenth, and eleventh decades of life) taken at the level of the mammillary bodies from brains that had been fixed for 2 weeks in 4% paraformaldehyde were embedded in paraffin and sectioned at 8 μm. In addition to hematoxylin and eosin and Prussian blue stains, immunostains using mouse antihuman β-amyloid protein diluted 1:10 000 (6E10; Covance), rabbit antihuman α-synuclein diluted 1:3000 (Chemicon), mouse antihuman CD68 (clone KP1) diluted 1:400 (Dako), mouse antihuman smooth muscle actin (clone 1A4) diluted 1:100 (Dako), and rabbit antihuman tau diluted 1:3000 (Dako) were performed using 3-amino-9-ethylcarbazole substrate chromogen (Dako). Braak staging was performed as previously described.⁹ Briefly, neurofibrillary tangles were assessed semiquantitatively (0 to +++) within 6 cerebral neocortical regions along with CA1 of the hippocampus, subiculum, entorhinal–transentorhinal region, and amygdala; tangle severity was scored on a scale of I to VI. Microbleeds were quantified (for follow-up study only) within the microscope field encompassed by a 2× objective; the mid putamen at the level of the mammillary bodies was examined, and the number of capillary and noncapillary channels bounded by ≥1 hemosiderin deposits were counted. From these counts, the number of vessels per square centimeter (capillary density) was calculated. Unpaired Student t tests and Pearson correlation were used to further evaluate microbleed scores; P<0.05 was considered significant.

Results

In the initial study, the age of subjects ranged from 71 to 92 years and mean age was 79.3 years; 3 were male and 9 were female. Microbleeds were found in 2 subjects and were located in subcortical white matter of occipital lobe and in putamen. In these subjects, iron was present in capillary wall at the ablumenal endothelial surface at the site of location of capillary pericytes (Figure 1). Electron microscopy of these microbleeds demonstrated iron in pericytes immediately adjacent to endothelial tight junction (Figure 2). Vascular β-amyloid deposition was not encountered at the sites of iron deposition. Nine of the 12 subjects had history of hypertension, including the 2 with microbleeds (Table).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Occipital lobe, subcortical white matter. A, Capillary with hemosiderin; hematoxylin and eosin and Prussian blue (original magnification ×520). B and C, Capillary with hemosiderin; hematoxylin and eosin and Prussian blue (original magnification ×520). D, The α-smooth muscle actin immunostain of capillary. Immunoreactivity is adjacent to ablumenal surface of endothelium, where pericytes are located (original magnification ×520).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. Electron microscopy of capillary in subcortical white matter, stained for iron (Prussian blue). Pericyte, attached to capillary wall and surrounded by basement membrane, contains dense deposits of iron. Tight junction of endothelial cell (red arrows) is immediately adjacent to pericyte. Also shown are basal lamina (green arrow), endothelium (blue arrow), hemosiderin in pericyte (white asterisk), red blood cell (yellow #), and blood vessel lumen (red arrowhead). Original magnification ×15 000.

In the follow-up study, age of subjects ranged from 77 to 105 years and mean age was 93.4 years; 4 were male and 17 were female. Postmortem interval ranged from 3 to 21 hours (mean, 6.3 hours). Thirteen subjects had history of hypertension and 6 were without hypertension history. In 2 subjects, hypertension history was uncertain; 1 of these subjects was not hypertensive at final reading. Microbleeds were seen within putamen in 20 of 21 subjects, including all subjects without history of hypertension. Iron was observed predominantly within macrophages adjacent to small arteries, arterioles, and particularly capillaries (Figure 3); it was also seen to be deposited free in the tissue. In most instances, it was distributed widely within the putamen. One subject (subject 31) had evidence of cerebral amyloid angiopathy involving vessels remote from the microbleeds. For the remainder, β-amyloid protein deposition was observed within diffuse and neuritic plaques, but not within the walls of involved capillaries or arterioles. The vast majority of vessels involved in these microbleeds were capillaries, and microbleed score (capillary density) is listed in the Table. There was no significant difference in microbleed score for males vs females and for subjects with and without hypertension, and there was no significant association between age and microbleed score and between microbleed score and postmortem interval (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3. Pericapillary deposition of hemosiderin (brown) within macrophage (red) in putamen, using CD68 immunostain. Original magnification ×600.

Discussion

We found frequent evidence of cerebral microbleeds in the aging brain, often with capillary involvement. The majority of the microbleeds occurred in putamen. The microbleeds occurred in vessels without amyloid deposition and microbleeds occurred in presence and in absence of hypertension. Moreover, blood–brain barrier pericytes in addition to brain macrophages appeared to have a role in microbleeds.

Iron uptake into brain is complex and may occur as a consequence of hemorrhage and by receptor-mediated endocytosis of transferrin-bound iron.^10,11 The latter occurs at endothelial cells of the blood–brain barrier and results in tissue iron distribution principally in oligodendrocytes.^10,11 This is a well-described and age-related process that does not, however, include phagocytosis or inflammation.^10–13

The neurovascular unit of the blood–brain barrier consists of endothelial cells in close approximation to pericytes, separated only by basement membrane.^14–16 In addition, macrophages are known to be found adjacent to neurovascular unit, either in residence or as cells migrating to that site.¹⁷ It is therefore not surprising that both cell types appear to have a role in microbleeds. Erythrophagocytosis is a well-described feature of macrophages,¹⁸ and Prussian blue stains of hemosiderin iron show punctuate staining similar to what was observed in the current study¹⁹ (Figure 1). Previous work has shown macrophage involvement in cerebral microbleeds.^4,6 Pericytes are also known to have phagocytic function,^14,20 and erythrophagocytosis has been observed in systemic pericytes.^21,22

Red blood cells may pass through endothelial junctions in systemic capillaries, resulting in petechial hemorrhages.²³ This is common in thrombocytopenia, along with other hemorrhagic diatheses. It is noteworthy that brain hemorrhage is rare in thrombocytopenia,²⁴ and this is likely attributable to structural and functional properties of the blood–brain barrier that can prevent local hemorrhage. Presence of tight endothelial junctions in the neurovascular unit is one likely component of the brain’s armamentarium against hemorrhage. Opening of the endothelial junctions to allow passage of red blood cells might represent age-related changes in barrier function; the latter has been described in cerebral white matter disease of aging with downregulation of blood–brain barrier efflux transporter p-glycoprotein.²⁵

Pericytes are known to be preferentially located adjacent to tight junctions of the blood–brain barrier and have been localized adjacent to histamine-induced gaps between endothelial cells.¹⁴ This location is ideally suited for a “gatekeeper” function of pericytes, in which these cells scavenge and phagocytose red cells that are able to pass the barrier between adjacent endothelial cells. In this scenario, macrophages adjacent to the neurovascular unit would be able to act as secondary scavengers for those red blood cells that are able to bypass pericytes and enter brain parenchyma (Figure 4).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4. Proposed mechanism for extravasation of red blood cells from brain capillaries. Opening of tight junction allows for pericyte erythrophagocytosis, with adjacent macrophage available as alternative or secondary site for phagocytosis.

Our study is not designed to provide pathological correlations to MRI findings of cerebral microbleeds. This represents a limitation of the present study because our findings indicate microbleeds primarily in basal ganglia location, whereas MRI studies show microbleeds more frequently in lobar site compared to deep brain regions.¹ MRI is known to overestimate the size of microbleeds (the “blooming effect”), with MRI diameter on average >150% of pathological lesions;⁴ nevertheless, MRI may not be sensitive to the relatively small microbleed findings of the current study. Other limitations of our study include absence of younger brains and predominance of female subjects. These represent potential future directions of research, along with cognitive function to which pathological features may be correlated. Moreover, it should be noted that the 2 studies described herein were not conducted identically. The initial study focused primarily on subcortical white matter, whereas the second study focused on cortex and basal ganglia.

In conclusion, cerebral microbleeds appear to be common in aging brain, occur at the capillary level, and are particularly prevalent in the putamen. Both pericytes and macrophages at the neurovascular unit seem to play a role modulating microhemorrhage. Microbleeds occurred both in the presence and absence of hypertension, and amyloid deposition was not found at site of microbleeds. This suggests that amyloid and hypertension may not necessarily be required for cerebral microbleeds in the aging brain. Further pathological investigations may determine the relationship between cerebral microbleeds as described herein and vasculopathic amyloid deposition and hypertension.