Effects of prolonged fixation on vascular biomarkers in postmortem human brains

Introduction

Emerging evidence suggests that the development of neurodegenerative diseases (NDD), including Alzheimer’s disease (AD), may be preceded and exacerbated by vascular changes. Cerebrovascular disease (CVD) has been identified as an important contributor to age-related dementias, frequently co-existing alongside neurodegenerative pathologies and impacting disease onset and progression rate.^1–3 Markers of CVD, namely white matter hyperintensities (WMH), infarcts, and microbleeds, can be detected using structural magnetic resonance imaging (MRI).^4,5 However, structural MRI lacks the sensitivity and specificity required to reliably characterize the underlying microscopic pathologies associated with CVD. These imaging findings can instead be validated through postmortem histological evaluation of the affected tissues, which remains the gold standard for identifying the cellular and molecular alterations associated with CVD lesions.^6–8

Histology procedures such as immunohistochemistry (IHC) and histochemistry (HC) are commonly used for cellular assessments (i.e. number, location, or morphology) of cerebral tissues, confirming the presence and severity of various diseases-related pathological processes.^9–11 IHC and HC staining quality highly depends on tissue fixation quality^12,13 and can therefore be affected by differences in fixative chemicals used,^14,15 fixation length, ¹⁶ or postmortem interval (PMI). ¹⁷ Since postmortem tissue requests are typically submitted to brain banks after sufficient numbers of cases meeting specific inclusion criteria have accumulated, relatively few studies are able to consistently analyze samples shortly after donation. Consequently, most tissues are fixed in 10% neutral buffered formalin (NBF) for months, years, or even decades before being used for research,^18,19 which results in substantial variability in fixation duration both within and between studies, highlighting the need for a thorough assessment of its impact on downstream results.

NBF preserves tissue integrity by hardening the tissue to prevent decomposition, degradation, and shrinkage.^18,20,21 However, it also induces crosslinking of proteins and nucleic acids within tissues,^22,23 which in turn, can mask antigenic epitopes and reduce immunoreactivity in IHC.^13,24 To circumvent this drawback, antigen retrieval approaches are widely used to reverse the NBF-induced crosslinking in human brain tissues with prolonged fixation before IHC staining.^25–28 Prior studies have shown that the effect of prolonged fixation on IHC and/or HC staining in human brains fixed for periods ranging from weeks to years can be either negative or unaltered.^29–32 A recent study from our team showed that longer fixation times were associated with reduced staining intensity of the neuronal nuclei marker NeuN, and the ionized calcium-binding adapter molecule 1 (Iba1) as a microglia marker, while other markers (i.e. glial fibrillary acidic protein (GFAP)) and HC staining (i.e. hematoxylin and eosin (H&E)) intensity remained stable across fixation durations. ¹⁶ Furthermore, it has been previously suggested that gray matter (GM) and white matter (WM) regions could be differentially affected by the diffusion of chemical fixatives, as shown using MRI, ³³ however, this effect on HC and IHC studies has not been investigated. Taken together, these data suggest that prolonged fixation can produce differential effects on the quality of histological staining and neuropathological quantification and assessment. Since IHC and HC staining can be influenced by formaldehyde fixation but are necessary for identification of neuropathology and underlying mechanisms in CVD, it is essential to determine whether prolonged fixation alters the expression and detectability of biomarkers widely used to diagnose and study CVD.

Therefore, the present study aimed to determine whether prolonged fixation in NBF affects the staining intensity of antigens related to the microvasculature, the blood–brain barrier (BBB), and other CVD or NDD biomarkers in postmortem human brains, specifically within WM and GM regions.

Material and methods

Postmortem human brain samples

Brain samples were obtained from the Douglas Brain Bank (DBB; https://douglasbrainbank.ca). Consistent with the DBB protocol, hemispheres were separated by a sagittal cut in the middle of the brain, brainstem, and cerebellum upon receipt. ³⁴ One unsliced hemisphere (right or left, in alternation) was fixed in 10% NBF. For this study, 20 prefrontal cortex (PFC) blocks of the exact same anatomical region from 1 cm-thick coronal sections were extracted. Specimens fixed for 1, 5, 10, 15, and 20 years (n = 4/group, in total, n = 20) were requested. All specimens were consistently processed and fixed in the same solution (i.e. 10% NBF), with the sole difference being the duration of fixation.

All brains were recruited from the DBB after obtaining consent by next-of-kin in collaboration with the Quebec Coroner’s Office. All experimental procedures were approved by Human Research Ethics Committees of McGill University and Douglas Mental Health University Institute in accordance with the 1964 Declaration of Helsinki. The cases were recruited from the suicide brain bank, in which individuals with evidence of drugs or psychotropic medications or documented history of neurological disorders or head injury were excluded from the study. However, since no brain tissue was available for the 10-year group in that cohort, cases for this group were recruited from the aging and dementia brain bank. These brains underwent neuropathological evaluation prior to this study based on standard NIA-AA criteria, ³⁵ confirming their diagnosis of AD, presence of CVD and absence of other co-pathologies, which are reported in Table 1. Table 1 also presents the characteristics of the samples included in this study.

OPEN IN VIEWER

Table 1.

Demographics of the specimens used in this study.

Group	ID	Sex	Age	PMI (h)	Diagnosis
1 Year	1	M	50	54.50	NA
	2	F	41	51.50	NA
	3	M	19	46.00	NA
	4	M	33	70.50	NA
	All	3:1	Mean = 35.8 (SD = 13.1)	Mean = 55.6 (SD = 10.5)	NA
5 Years	5	M	53	49.50	NA
	6	M	59	47.50	NA
	7	M	55	51.00	NA
	8	M	59	46.42	NA
	All	4:0	Mean = 56.5 (SD = 3.0)	Mean = 48.6 (SD = 2.0)	NA
10 Years/DX	9	M	75	15.88	AD + CVD
	10	M	91	72.68	AD + CVD
	11	M	80	42.58	PART + CVD
	12	F	84	10.72	AD + LBD + CVD
	All	3:1	Mean = 82.5 (SD = 6.8)	Mean = 35.5 (SD = 28.5)	NDDs
15 Years	13	F	49	59.50	NA
	14	M	81	98.75	NA
	15	M	30	51.50	NA
	16	M	36	55.25	NA
	All	3:1	Mean = 49 (SD = 22.8)	Mean = 66.3 (SD = 21.9)	NA
20 Years	17	F	25	20.00	NA
	18	M	22	24.00	NA
	19	F	55	36.00	NA
	20	F	44	29.50	NA
	All	1:3	Mean = 36.5 (SD = 15.7)	Mean = 27.4 (SD = 6.9)	NA

F: female; M: male; NA: normal aging; AD: Alzheimer’s disease; DX: diagnosis; CVD: cerebrovascular disease; LBD: Lewy body disease; PART: primary age-related tauopathy.

Results

Mean age at time of death for the 1-, 5-, 10-, 15-, and 20-year groups were 35.7 (range 19–50), 56.5 (range 53–59), 82.5 (range 75–91), 49 (range 30–81), and 36.5 (range 22–55) years, respectively. The specimens included in the 10-year group were significantly older than all other groups (p < 0.05) and had postmortem evidence of CVD and NDD (Table 1). Male to female ratios were 3:1, 4:0, 3:1, 3:1, and 1:3, respectively. PMIs were not significantly different across groups.

Immunohistochemical staining

Table 2 summarizes the results of the mixed effect models used to assess the differences between mean staining intensities of the prolonged fixation groups (5, 10, 15, and 20 years) against the 1-year group, adjusted for age, sex, and PMI. Presented values reflect the t statistics of contrasting the groups of interest against the reference 1-year group, and stars indicate the significance of the results following FDR correction. Figure 1 summarizes the main findings of the study. Median intensity values for each marker in the five groups are reported in Supplemental Table 2. Figure 2 shows high magnification of the targeted biomarkers in the 1-year group.

OPEN IN VIEWER

Table 2.

Differences in mean staining intensities of the prolonged fixed groups compared against the 1-year group.

Marker	Tissue	5 Years	15 Years	20 Years	10 Years/DX	Age	Sex (M)	PMI
CD68	GM	−0.56	−0.62	−2.43	−0.09	0.20	−0.95	−0.78
CD68	WM	−0.28	−1.06	−1.68	0.42	0.03	−0.32	0.24
PLP	GM	2.29	1.50	0.45	3.70	−3.55	−1.62	2.41
PLP	WM	0.14	0.18	1.62	0.43	−0.10	−0.20	1.22
Ferritin	GM	−3.16	−5.91***	−4.03**	−2.24	−0.22	0.39	0.17
Ferritin	WM	−4.68**	−6.23***	−3.86**	−2.17	−1.01	0.96	−0.37
Claudin-5	GM	−0.66	0.50	−1.09	−0.15	0.49	0.33	−1.29
Claudin-5	WM	1.15	1.29	−0.03	2.45	−1.34	−0.22	0.36
Vimentin	GM	−2.42	−6.85***	−6.94***	−1.68	0.29	−1.25	−0.37
Vimentin	WM	−2.01	−7.48***	−10.21***	−2.31	−1.03	−2.00	−0.35
Collagen-IV	GM	−1.09	−6.88***	−3.85**	−1.56	0.55	0.73	0.82
Collagen-IV	WM	1.60	−2.74*	−1.39	0.48	−1.52	0.01	1.82
Trichrome Masson	GM	−0.26	−1.04	0.46	−0.11	0.66	0.32	0.29
Trichrome Masson	WM	−0.45	−0.56	1.54	0.64	1.09	0.18	0.63
Bielschowsky	GM	−0.37	−1.66	−3.15*	−1.48	0.66	0.71	−0.92
Bielschowsky	WM	−0.46	−1.70	−1.59	−0.83	1.09	0.66	0.20

GM; gray matter; WM: white matter; DX: diagnosis, reflecting the fact that this group had diagnoses of cerebrovascular and neurological disorders; PMI: postmortem interval.

Presented values reflect the t statistics and p values of the mixed effects model described in the methods section, contrasting the groups of interest (i.e. 5, 10, 15, and 20 years) against the reference 1-year group, accounting for age, sex, and PMI as covariates.

Bolded cell numbers are significant after a FDR correction for each column.

*

p < 0.05. **p < 0.005. ***p < 0.0005 (FDR corrected).

OPEN IN VIEWER

Figure 1.

Mean staining intensity of the five groups in the gray and white matter of the eight staining types: (a, b) activated microglia (CD68), (c, d) myelin proteolipid protein (PLP), (e, f) ferritin, (g, h) claudin-5, (i, j) vimentin, (k, l) collagen IV, (m, n) trichrome Masson staining, and (o, p) Bielschowsky staining. Mean staining intensity was measured based on RGB intensity values or a threshold intensity extraction using ImageJ.

OPEN IN VIEWER

Figure 2.

Photomicrographs (60×) of all the targeted markers in the gray matter of a 1-year specimen (ID-1): (a) CD68 showing microglia fine processes, (b) PLP showing labeled myelin fibers, (c) ferritin showing macrophages that phagocyted iron, and iron deposits in the neuropil, (d) claudin-5 showing blood vessel walls, (e) vimentin showing blood vessel walls and astrocytes bound to the blood–brain barrier, (f) collagen-IV showing blood vessel walls, (g) Masson’s trichrome staining showing fibers, red blood cells, and collagen, and (h) Bielschowsky silver staining showing axons and bundles. Scale bar = 20 µm.

CD68 labeled activated microglia in the five groups (Figures 2(a) and 3). In the GM and WM regions, the staining intensities decreased with fixation time and were higher in the 10-year/DX group, although these differences were not statistically significant (Figure 1(a) and (b), Table 2, and Supplemental Table 2). PLP stained myelin fibers in the GM and WM of the five groups (Figures 2(b) and 3). In GM regions, the gray values were not different across the five fixation groups (Figure 1(c), Table 2, and Supplemental Table 2). WM staining intensity of PLP had an increasing trend with prolonged fixation (Figure 1(d)), indicating that the fibers were less differentiated from the background (i.e. lower staining quality), but this was not statistically significant (Table 2 and Supplemental Table 2).

OPEN IN VIEWER

Figure 3.

Distribution of CD68-, PLP-, and ferritin-immunoreactive cells in the gray and white matter (20×) of the PFC sections of human brains fixed for 1, 5, 10, 15, and 20 years. Top two rows show activated microglia profiles. The two middle rows show myelin fibers profiles. The two bottom rows show macrophages and iron deposits. Scale bars = 100 µm.

Ferritin marker labeled dark deposits of iron and microglia-shaped cells in the five groups (Figures 2(c) and 3). In the GM, the 1-year group staining intensity was significantly higher than in the 5-, 10-, 15-, and 20-year groups (all p < 0.05; Figure 1(e) and Supplemental Table 2). However, the FDR corrected results in the 5 years did not reach the significance threshold (Table 2). In WM regions, the 1-year group staining intensity was significantly higher than in the 5-, 10-, 15-, and 20-year groups (all p < 0.05; Figure 1(f), Table 2, and Supplemental Table 2).

Figure 4 shows blood vessels positively stained for claudin-5 in GM and WM regions of the five fixation groups, and Figure 2(d) shows the blood vessels at higher magnification. We found no significant differences in the staining intensity of the GM ROIs (Figure 1(g) and Supplemental Table 2). The 10-year group showed higher staining intensities compared to the 1-year group in WM regions (Figure 1(h) and Supplemental Table 2), but this did not retain significance after a FDR correction (Table 2).

OPEN IN VIEWER

Figure 4.

Distribution of claudin-5, vimentin-, and collagen-IV-immunoreactive cells in the gray and white matter (20×) of the PFC sections of human brains fixed for 1, 5, 10, 15, and 20 years. Top two rows show endothelial junctions of the blood vessels (claudin-5). The middle two rows also show the cytoskeleton of the blood–brain barrier (vimentin), while the two bottom rows show the vascular media of the blood vessels (collagen-IV). Scale bars = 100 µm.

Vimentin marker stained blood vessels in the first four groups, while staining was completely absent in the 20-year group. Vimentin also labeled astrocytes-shaped cells in the 1-year group (Figures 2(e) and 4). In the GM, we found that the 1-year group showed significantly higher staining intensity than the 5-, 15-, and 20-year groups (all p < 0.05; Figure 1(i) and Supplemental Table 2). However, the difference with the 5-year group did not retain significance after a FDR correction (Table 2). In the WM ROIs, median staining intensity values of the 1-year group were significantly higher than in the 15-, 20-, and 10-year groups (all p < 0.05; Figure 1(j) and Supplemental Table 2). However, the 10-year group did not retain significance after a FDR correction (Table 2).

Collagen-IV stained blood vessels as well as glial cell bodies in the five groups (Figures 2(f) and 4). In the GM, staining intensity of the 1-year group was significantly higher than in the 15- and 20-year groups (p < 0.01; Figure 1(k), Table 2, and Supplemental Table 2). In the WM, staining intensity of the 1-year group was significantly higher than in the 15-year group (p < 0.001; Figure 1(l), Table 2, and Supplemental Table 2).

Histochemical staining

MTS stained neuropils in purple, cell nuclei in dark blue, and blood cells in red (Figures 2(g) and 5). We found no significant differences in the staining intensity of the GM nor the WM regions in brain sections fixed for 1, 5, 10, 15, or 20 years (Figure 1(m) and (n), Table 2, and Supplemental Table 2). However, we observed a trend where the increased staining intensity with prolonged fixation or in brain specimens with a NDD diagnosis, reflecting a lower differentiation between the labeled cells and fibers and the background (i.e. lower staining quality).

OPEN IN VIEWER

Figure 5.

Distribution of trichrome Masson and Bielschowsky silver staining in the gray and white matter (20×) of the PFC sections of human brains fixed for 1, 5, 10, 15, and 20 years. The top three rows show fibers in purple, blood cells in red and collagen in blue. High-magnification images (60×) are shown in the third row. The red arrow head shows red blood cells, while the green arrow head shows collagen in blue. The three bottom rows show the axons and tangles in brown on a yellow background. High-magnification images (60×) are shown in the last row. In the 10-year group, the red arrow head shows a neurofibrillary tangle, and the green arrow head shows a neuritic plaque. Scale bars = 100 µm.

BSS labeled axons and fibers throughout the GM and WM regions, where thick axonal fiber bundles were stained brown while the background of neuropils was stained yellow (Figures 2(h) and 5). Three out of four cases of the 10-year group presented high levels of AD neuropathologic changes, senile plaques, and neurofibrillary tangles in the GM when stained with BSS. In the GM, the 1-year showed significantly higher RGB values than in the 20-year group (p < 0.01; Figure 1(o), Table 2, and Supplemental Table 2). We found no significant difference in the WM ROIs in the BSS staining intensity with prolonged fixation (Figure 1(p), Table 2, and Supplemental Table 2).

Discussion

In the present study, we systematically evaluated the impact of prolonged fixation on antibodies targeting various biomarkers. We found that prolonged fixation resulted in a reduction of the IHC staining intensity for ferritin, vimentin, and collagen-IV, as well as HC staining intensity reduction for BSS. Given that when looking at the micrographs, the cells can still be observed even if they are lightly stained in long-term fixed samples, this should not affect cell density estimations. Furthermore, we did not find significant associations between increased fixation and the IHC staining intensity of CD68, PLP, and claudin-5, or the HC staining intensity of MTS, indicating that their staining intensity remains mostly stable during fixation. Taken together, these data indicate that prolonged fixation induces differential effects on commonly used biomarkers for CVD.

Due to the protein and RNA crosslinking caused by NBF, ³⁸ it is generally believed that prolonged fixation with NBF produces IHC antigenicity reduction in human brain samples. ¹³ Prior studies on human brains showed a negative correlation between NeuN intensity and fixation times ranging from immediate up to 3 years ³² or with fixation durations spanning 24 h, 4 months, and 10 years. ³⁰ A mice study also supported a similar decline in IHC staining by prolonged fixation. ³⁹ Similarly, NeuN levels were reduced in the swine brains fixed for 2 months, which completely disappeared after 3 months of fixation. ⁴⁰ Consistent with these studies, we previously showed a negative correlation between IHC staining intensity of NeuN and Iba1 and fixation years on human brain sections. ¹⁶ We therefore expected that fixation duration could impact other markers that are commonly used in neuropathology or for studying cerebrovascular disease burden. However, our prior study showed that prolonged fixation enhanced the IHC staining of the astrocyte marker GFAP in human brains. ¹⁶ While NBF-induced protein crosslinking is often considered the main cause for the negative effect of fixation on IHC staining by masking out epitopes,^13,41 crosslinking can also be responsible for increased staining intensity and/or antigenicity detection. For example, more GFAP epitopes could be exposed by breaking protein crosslinking during antigen retrieval. ¹⁶ While it is still unclear why some antigens are differently affected by fixation, taken together, these findings suggest that prolonged fixation may have differential effects on IHC staining quality for some but not all target proteins.

Since our prior study showed negative correlations between Iba1 staining intensity and fixation, ¹⁶ we expected that another microglia marker (CD68) would also be impacted by prolonged fixation. CD68 is a widely used marker for identifying and visualizing microglia, particularly when they are in an activated state, as macrophages,^42,43 and has been used extensively to identify inflammatory processes in neurodegenerative diseases, in particular AD.^42,44–46 Furthermore, microglia are known markers of neuroinflammation since their activation is implicated in CVD lesions, especially in WMH. ⁴⁷ A prior qualitative study showed that CD68 immunoreactive activated microglia were observed in the frontal lobes of human brains fixed from <1 to 20 years. While this is consistent with our results, they did not report quantitative evidence on the correlation between CD68 IHC staining intensity and the fixation length. ²⁹ Another study used CD68 IHC staining to assess Aβ immunized AD cases versus non-immunized AD cases. ⁴⁸ However, the potential impact of the fixation delay in the quantitative analysis of the pathology associated with this biomarker was not examined. Even though not statistically significant, our data shows a declining trend in staining intensity with fixation years, suggesting that while CD68 markers may be used in long-term fixed samples, fixation times should be considered in the analyses.

PLP is known as the human proteolipid protein, a biomarker for oligodendrocytic myelin, a major structural component of myelin, helping to compact and stabilize the myelin sheath.^49–51 PLP IHC is widely applied for the neuropathological evaluation of demyelinating disorders, including multiple sclerosis^49,50 and Pelizaeus–Merzbacher disease. ⁵² Furthermore, in the WM of AD brains, PLP decreases when the levels of amyloid-β increase, ⁵³ suggesting that abnormalities of myelin and oligodendrocytes are linked to AD pathology. ⁵⁴ However, our 10-year/DX group did not show any significant decrease of PLP staining intensity in the WM. Overall, our results showed no significant changes in the staining intensity of PLP in brains fixed for decades. Therefore, given that our previous study in the same specimens showed a reduction in the staining intensity of Luxol fast blue which also stains myelin sheath, ¹⁶ we recommend the use of PLP instead of Luxol fast blue in studies examining long-term fixed tissues.

Our ferritin marker stained iron deposits as well as microglia, since macrophages are known to perform phagocytosis of iron when neuroinflammation occurs in NDDs.^55,56 Ferritin is a widely used biomarker to assess the extent of CVD lesions in stroke ⁵⁷ and cerebral amyloid angiopathy,^58–60 as well as normal iron accumulation in aging ⁶¹ or in NDDs.^62–64 IHC using ferritin or other iron markers are also essential to validate ex vivo MRI or quantitative susceptibility maps.^65,66 However, to our knowledge, our study is the first to quantitatively evaluate the impact of fixation on the immunoreactivity of ferritin antibodies in postmortem human brains. Our results showed a significantly reduced staining intensity when brains were fixed for long periods. This should be considered in ex vivo studies that examine iron deposits using histopathology, although current studies mainly use Perl’s stain.^8,58,67–69

Claudin-5 is an essential protein for sealing the intercellular space between adjacent endothelial cells in the BBB (i.e. a biomarker for tight-junctions), primarily used to assess the integrity of the BBB and its potential disruption ⁷⁰ which is involved in aging and AD. ⁷¹ It is also investigated as a biomarker for various neurological and psychiatric disorders. ⁷² We found that claudin-5 staining intensity remained stable during fixation, suggesting that it is a reliable marker for BBB assessment.

Vimentin is a marker for intermediate filament III, present in ependymal cells as well as in the cytoskeleton of astrocytes.^73,74 It is also used as a marker for endothelial cells, particularly those lining blood vessels, ⁷⁵ and is used as a marker of BBB disruptions, ⁷⁶ reflecting its significance for CVD lesions studies. Furthermore, vimentin is closely associated with AD due to its role in Aβ aggregation and deposition.^77,78 Our current study showed that vimentin not only labeled blood vessels, but also activated astrocytes, components of the BBB, which is consistent with previous studies investigating vimentin co-localization with other astrocytes markers such as GFAP. ⁷⁹ The latter study used postmortem human brain blocks of 1 cm³ that were fixed immediately in NBF after brain extraction. They showed very high staining quality of astrocytes and blood vessels, ⁷⁹ while other studies showed no changes in vimentin staining intensity in dog brains fixed for 7 weeks.^26–31 This is also in agreement with our current study, where the 1-year group showed very high staining intensity of both cell-types, but that there is a significant decline in vimentin staining when brains are fixed for longer durations.

Collagen is a ubiquitous protein of the extracellular matrix of various connective tissues. ⁸⁰ Collagen-IV is a crucial component of blood vessel walls, contributing to their structural integrity and function, ⁸¹ and its presence and turnover are important indicators of vascular health and BBB integrity, with imbalances potentially leading to various conditions, ⁸² and NDDs.^83–86 Our current findings also suggest that glial cell bodies are stained by collagen-IV markers alongside blood vessels, representing the major components of the BBB, reflecting its significance as a marker of CVD. However, glial cells may have been labeled by non-specific binding of the antibody, and further studies are needed to examine the specificity of this marker. Previous studies have also investigated collagen as a potential marker for forensic evaluation, since the cellular matrix resists degradation during longer PMIs compared to the cellular components. ⁸⁰ However, we showed that staining intensity of collagen is reduced in brains fixed for decades. These discrepancies in collagen IHC quality may be due to (i) differences between collagen-I and collagen-IV studies; 9ii) tissue types (brain vs other forensic tissues); (iii) collagen inside blood vessel walls versus the extracellular matrix; and (iv) the type of antigen retrieval that was processed. These data suggest that further studies should include fixation length as a covariate when investigating collagen fibers in the brain.

BSS is a well-known stain used for differentiating between NDDs through selective staining of neuritic plaques, neurofibrillary tangles, and their axonal prolongations with silver deposits, thereby visualizing the two characteristic pathological hallmarks of AD.^10,87–96 Overall, BSS staining intensity decreased with fixation length reaching 20 years, since it is a silver revelation technique in which the tissue background may be affected by interaction with chemicals. ¹⁵ This suggests that, even though HC methods are not antigenicity-dependent (and therefore, not crosslinking dependent) compared to IHC methods, they can still be affected by prolonged chemical fixation, which should be taken into account when using these stains in future studies.

Finally, MTS is a common histopathological stain used to identify collagen and fibrin accumulation, ⁹⁷ as well as the extracellular matrix in brain tissue for BBB health and perivascular spaces (common markers of CVD lesions). ⁹⁸ MTS also stains myelin fibers within the extracellular matrix, allowing for the identification of CVD lesions, including WMHs.^8,99–101 It is commonly used to visualize stenosis and amyloid-β and tau accumulation within blood vessels, which are characteristic of cerebral amyloid angiopathy in NDD patients or animal models. ⁹⁰ Here, we found no significant differences in the staining intensity of MTS in groups fixed for decades, reflecting its robustness for use in multiple studies that employ brains fixed for years in brain banks.

At the time of our tissue request, our brain bank did not have any healthy-aging control PFC blocks available for the 10-year group. Instead of excluding this time frame from our study, we requested tissues with a diagnosis of CVD/NDDs for this group. This is a limitation also considering that the characteristics of the 10-year group were different from the others, especially since they were significantly older. However, including the 10-year/DX samples also allowed us to examine whether these markers could reliably detect the expected pathologies in a population of interest (Supplemental Figure 1 shows presence of p-tau and amyloid-β). For example, CD68 showed a wide range of staining intensities, from low values (3.68 in GM and 9.93 in WM) due to the decrease in staining intensity after 10 years of fixation, to higher values (20.46 in GM and 40.21 in WM), likely due to the increase of microglial activation resulting from neuroinflammation and neurodegeneration.^44–46 The highest values were obtained in Specimen 12, which was diagnosed with co-pathologies of AD, LBD, and CVD, suggesting increased microglial activation when more pathologies are involved, resulting in higher neuroinflammation. ⁴⁷ Claudin-5 also showed significantly higher intensity values in the WM regions than the other groups. This may be because claudin-5 is a tight junction protein involved in the BBB and may be dysregulated in AD.^70,71 We expected to find differences in BSS since it is used as a marker of neuritic plaques and neurofibrillary tangles, especially in AD.^10,87–96 Indeed, the presence of some neuritic plaques and neurofibrillary tangles were found in three out of four cases of the 10-year group. However, all of these differences in the 10-year/DX group did not reach significance after a FDR correction, potentially because this group was older, which results in a high collinearity between age and group variables, while our small sample size can limit our ability to detect significant differences.

Certain limitations should be noted. The sample size used to evaluate fixation-related changes was limited, reducing the statistical power of our analyses. Our current study was adequately powered only for large fixation-related effects (e.g. vimentin, collagen IV, and ferritin). In contrast, markers with moderate or more variable effects required substantially larger sample sizes. For example, for CD68, the observed effect size was Cohen’s f = 0.60 in GM and 0.66 in WM, corresponding to an estimated requirement of ~10 and 9 specimens/group, respectively, while PLP (Cohen’s f ~0.4) required ~20 specimens/group (see Supplemental Table 3). Taken together, the lack of significant results may reflect limited power and should be interpreted cautiously. Future studies with larger samples are required to further validate the reported effect sizes and differences. However, this limitation may be partly offset by inclusion of a broad range of fixation times, which reflects the diversity of samples typically available from brain banks, where tissue blocks are often obtained from brains preserved in formaldehyde for many years, and in some cases, for decades. Furthermore, we used the 1-year fixation group as reference, while 1 year itself is a relatively long fixation period. Ideally, a fresh-tissue baseline would have been included; however, this was not feasible within the constraints of our brain bank protocol, as hemispheres are typically fixed for several weeks before becoming available for research. Consequently, one limitation of our study is that fixation-induced changes may already have been present in the 1-year group, potentially resulting in an underestimation of the true magnitude of these effects. Furthermore, we did not assess the same cases longitudinally, which would have been ideal for evaluating fixation changes. However, this was not feasible since tissue availability from the same specimens is limited and longitudinal sampling for 20 years is not realistic. Nevertheless, we consistently selected the same region in all cases and matched the cases as closely as possible for age and sex. Since age and sex had no statistically significant effects on CVD markers and that these can be detected both in normal aging and in NDDs, we believe the impact of these covariates on our findings was minimal.

Another limitation of our study is that we did not optimize the protocols for each group (e.g. antibody concentration or DAB incubation could have been increased in the groups fixed for longer durations). The antigen retrieval protocol could also have been suboptimal for some antigens or fixation groups of interest, leading to some of the diminution of staining intensity observed. Although the HIAR chosen in this study showed sufficient results in the 1-year group for all antigens (i.e. suggesting HIAR was an adequate antigen retrieval), adapting the protocol or using stronger antigen retrieval techniques such as proteinase K or formic acid might lead to better results in the older groups. However, our goal was to assess the impact of the fixation length on the IHC staining quality, and modifications of the protocols between the fixation groups would have confounded comparisons.

Furthermore, while we focused on CVD markers, future studies should include additional neurodegenerative markers, such as amyloid-β, phosphorylated tau, TDP-43, or α-synuclein, especially since they are widely used for neuropathology assessment of NDDs and dementia.^{78,90,102–105} These markers were previously assessed according to fixation ranging from 0 to 14 years, where these studies also suggested differential effects of fixation lengths on different markers. ¹⁰⁶ These markers would therefore be of interest in future studies.

Finally, while we only assessed immunohistochemistry with DAB-enhanced chromogen, it would also be important to examine the markers through immunofluorescence staining in future studies. Indeed, a subset of markers (i.e. vimentin, collagen, and claudin-5) were stained through double immunofluorescence for the 1- and 5-year group (Supplemental Figure 2) in our sample. However, as autofluorescence is a widely known artifact in aldehyde-fixed tissue (especially for prolonged-fixed samples),^107,108 we did not perform immunofluorescence staining in all specimens, as even the 5-year group could reflect the higher intensity of fluorescent background. These protocols should be optimized in future studies to quench autofluorescence using different methods, such as photobleaching or Sudan black B, for example, Oliveira et al. ¹⁰⁸ and Wang et al. ¹⁰⁹

In this study, IHC targeting six biomarkers and two HC protocols widely used in studying NDDs were examined in postmortem human brains fixed for decades. Our findings suggest that prolonged fixation can produce negative effects on the IHC or HC staining intensities for human brain sections, depending on the biomarkers. Considering the effects of fixation on the IHC and HC staining quality is therefore necessary to enable accurate interpretation of subsequent experimental investigations. Studies using postmortem human brains should take fixation times into account as a covariate or use specimens fixed for the same durations for analyses, especially when measuring staining intensities that may decrease with prolonged fixation.

Supplemental Material