We used a convenience sample of 42 human brains (NBF: N = 12; SSS: N = 13; AFS: N = 17) from our body donation program (Université du Québec à Trois-Rivières, Canada) as well as the Douglas-Bell Canada Brain Bank (Douglas Mental Health University Institute, McGill University, Montréal, Canada). Prior to their death, the donors to the anatomy laboratory provided consent on body donation and sharing of medical information for teaching and/or research purposes. The study was approved by the University’s Ethics Subcommittee on Anatomical Teaching and Research. Donors from the Brain Bank were obtained in collaboration with the Quebec Coroner’s Office and with informed consent from the nearest relative, following the guidelines of the Douglas Research Center Ethic Committee.

The median age at time of death of the donors was 76 years old (range: 25 to 90). The male to female ratio was 3:1. The case number, age, sex, cause of death, post-mortem delay to the fixation treatment (i.e., post-mortem interval) and delay until the histology processing of the tissue, are reported in Table 1.

All specimens were fixed with one of three solutions (chemical components are shown in Table 2). NBF-fixed brains were extracted fresh and then immersed in 4 liters of NBF following the brain bank technique and kept in formaldehyde until use. For SSS and AFS, the bodies were perfused with 25 L of one of the solutions through the common carotid artery using a pump (41 kPa) (Duotronic III, Hygeco International Products) which is the common embalming technique performed in gross anatomy laboratories (Maranzano et al., 2020). All the specimens were kept refrigerated until use. The brains were extracted, and 1 cm³ blocks (neocortex for gray matter and subcortical white matter) were taken, choosing an area that was macroscopically of good quality (no superficial tears, good color and texture) and not damaged during the extraction and arachnoid removal procedure.

Blocks were rinsed in 0.1 M PBS and immersed in 30% sucrose for 2 to 5 days (until the floating block sinks, which is the sign of appropriate sucrose diffusion into the tissue) for cryoprotection. Blocks were then frozen on dry ice before cutting 40 μm sections on a cryostat (Leica CM1950) at -19°C, that were either 1-mounted on 2% gelatin-subbed slides for the histochemical stains (Cresyl violet, Luxol Fast blue, Prussian blue and Bielchowsky’s), or 2-placed in well plate with PBS for subsequent free-floating IHC procedures. After IHC or histochemical staining (see the sections below), the sections were dehydrated in 5-min baths of increasing grades of ethanol (70, 95, 100%) and coverslipped using Eukitt.

Free-floating sections were rinsed three times for 5 minutes in 0.1 M PBS. They were then incubated in a 20% aqueous methanol solution with 0.5% H₂O₂ for 5 minutes to quench endogenous peroxidase. After another round of rinses (3x5minutes in 0.1 M PBS), sections were incubated in blocking solution for 2 hours (3% Normal Donkey Serum; 0.5% Bovine Serum Albumin; 0.3% Triton X-100 in 0.1 M PBS), followed by the primary antibodies’ (Table 3) incubation at 4°C in the same blocking solution overnight. Sections were rinsed again (3x5minutes in 0.1 M PBS) before a two-hour incubation in donkey anti-rabbit biotinylated secondary antibody (1:500) at room temperature in the same blocking solution. After another rinse, sections were incubated in an avidin-biotin complex (ABC) kit for 30 minutes in the dark (VECTASTAIN^® Elite^® ABC-HRP Kit, Vector Laboratories, catalog#VECTPK6100). Sections were then rinsed for 5 minutes in TRIS-buffered saline (TBS) (0.05 M, 0.9% NaCl) before the final incubation in 0.07% diaminobenzidine (DAB) (Sigma, #D5905-50TAB) with 0.024% H₂O₂ in TBS for 10 minutes.

All specimens were also processed for four histological stains, commonly used in brain histopathology, namely 1-Cresyl Violet (endoplasmic reticulum of the cell nuclei, Nissl bodies), 2-Luxol Fast Blue (myelin), 3-Prussian blue (iron stain with a neutral red counterstain for Nissl bodies), and 4-Bielchowsky’s (neurofibrils and axons). Prior to the staining, sections were defatted in Xylene for 30 min and hydrated in increasing alcohols −100% Ethanol 2 × 1 min; 3–95% Ethanol 2 × 1 min; 70% Ethanol 2 × 1 min; H₂O_d 1 min. After the staining (see protocols below), a dehydration process was applied using the same baths, but in a reverse order.

The sections were assessed for multiple variables using a brightfield microscope (Olympus BX51W1) controlled by Neurolucida software (MBF Bioscience). We scored the IHC and histochemical stains variables to create qualitative categories that were compared for the three groups (fixatives) by looking through all the sections that were available for each specific staining. All variables were scored by a rater having 5-years experience in histology procedures and microscopy analysis (EMF). Figure 1 shows scoring of the criteria for different IHC variables (tissue quality and antigenicity preservation) that were assessed, while Figure 2 shows scoring of the criteria used for the four histochemical stains (see Supplementary Table S1 for all detailed scores of each specimen).

All variables mentioned above were also assessed in relation to two confounding variables: the post-mortem interval and the histology delay, to determine whether they might have affected the variables of interest. Raw data is shown in Table 1. Also, see Supplementary Table S1 for all detailed scores of each specimen, and Supplementary Figures S1, S2 for scatter dot plots of all the variables of interest in relation to the confounding variables.

All categorical variables described above were compared across the three fixative groups using Chi-square tests. All statistical analyses were adjusted for multiple comparisons using Bonferroni correction and were performed using SPSS statistics software (28.0.0 version).

To assess for confounding variables, we also used Chi-square tests to relate the variables of interest (section 2.6) to each confounding variable (section 2.7), separately in each fixative group (NBF; SSS and AFS).

We found no statistically significant differences between treatment groups in the ease of manipulation scores, but we observed more cases fixed with SSS that scored poor for this criterion, while brains fixed with NBF never scored very poor for ease of manipulation. Also, a higher proportion of cases fixed with NBF and AFS scored good for this criterion (Figure 3).

Tissue quality was assessed according to two criteria, namely the uniformity of the neuropil and cellular shape. We found no significant difference in the frequency of occurrence of fissured neuropil across the three groups (Figure 4A). However, we found a significant difference in the frequency of scores for the cellular shape, between the three groups (p < 0.001). Regular cell contours were always observed in the NBF-fixed brains (p < 0.001), while irregular (shrunken and shriveled) cell contours were always found in SSS-fixed brains (p < 0.001) (Figure 4B).

The distribution of labeled cells of the four tested antigens was used to assess antigenicity preservation. A homogeneous distribution of NeuN labeling was significantly more frequent in AFS-fixed brains (p = 0.0003), whereas significantly higher occurrence of isolated labeled cells (p = 0.0019) and no homogeneous NeuN labeling (p = 0.0019) were observed in SSS fixed brains. NeuN antigenicity was preserved in most NBF-fixed specimens, except for three cases in which no labeled neurons were observed (p = 0.005; significant before a Bonferroni correction) (Figure 5A). Homogeneous distribution of GFAP labeling was observed in only one NBF-fixed specimen, and no labeled astrocytes were observed in three cases (p = 0.005; significant before a Bonferroni correction) (Figure 5B). Patchy Iba1 labeling occurred significantly more frequently (p = 0.0037) whereas homogeneous labeling of this marker occurred less frequently in NBF-fixed brains (p = 0.005; significant before a Bonferroni correction). Homogeneous distribution of Iba1 labeling was significantly more frequent, whereas patchy labeling was significantly less frequent in AFS-fixed brains (p < 0.001) (Figure 5C). Finally, PLP labeling was more frequently uniform (homogeneous or cell patches) in all AFS-fixed brains (more cases of cell patches: p = 0.046 before a Bonferroni correction), Conversely PLP labeling was absent in three SSS-fixed cases and two NBF-fixed cases (Figure 5D).

In cases where our antibodies produced no labeling or only isolated labeled cells, we repeated the IHC procedures following the application of an antigen retrieval protocol. Antigenicity was recovered and produced a homogeneous labeling for all four markers in SSS and AFS-fixed brains (Figure 6). Nonetheless, NeuN and GFAP antigenicity was not recovered in three NBF fixed cases.

First, we assessed quality of the Nissl-stain. We found a significantly higher frequency of cases showing high cellular contrast in NBF-fixed specimens (strong labeling of neurons in all the cases) (p = 0.0037), and less cases with pale neurons (p = 0.016; significant before a Bonferroni correction only). The brains fixed with the other two solutions sometimes produced pale neurons staining (SSS: N = 6, p = 0.045; significant before a Bonferroni correction only, and AFS: N = 5), or even no staining at all (SSS: N = 1; AFS: N = 2) (Figure 7A).

Second, we assessed the quality of myelinated fibers labeling with LFB. NBF-fixed brains showed a higher frequency of cases in which stained fibers were absent (score = 0) (p = 0.016; significant before a Bonferroni correction only) and never homogeneous distribution (p = 0.036; significant before a Bonferroni correction only). Fiber labeling with LFB in SSS and AFS-fixed brains was quite inconsistent as they showed randomly distributed scores (Figure 7B).

Cell to background contrast was high following a Nuclear fast red (Nissl) counterstain to the Prussian blue staining (Figure 7C), regardless of the fixative. High cellular contrast occurred at a higher frequency in NBF specimens (all cases showed dark neurons: score = 2) (p = 0.016; significant before a Bonferroni correction only). Also, despite that not all the specimens showed ferric iron aggregation (depending on the pathology or age of the patient), blue spots indicating iron deposits were present in 10 NBF, 6 SSS and 8 AFS-fixed brains demonstrating the likelihood of detecting these deposits when either of these fixation protocols are applied (Figure 8, inserts in B, F, J).

Finally, we found a higher frequency of cases showing homogeneous labeling of silver impregnation of axonal fibers with the Bielchowsky procedure in the NBF-fixed brains (p = 0.0093; significant before a Bonferroni correction only) (Figure 7D). However, a higher occurrence of dark or intermediate background staining, was observed in the SSS and AFS compared to the NBF fixed brains, but this difference was not statistically significant (Figure 7E).

We were expecting that the SSS and AFS-fixed brains would be difficult to manipulate (softer tissue), according to our previous study in mice (Frigon et al., 2022). Accordingly, recent study in rat brains (Martins-Costa et al., 2022), and human anatomical studies describe these fixatives as soft-embalming solutions, which could lead to differences in the texture of the brains (Hopwood et al., 1989; Eltoum et al., 2001; Nicholson et al., 2005; Hammer et al., 2012; Hayashi et al., 2014; Balta et al., 2015; Tomalty et al., 2019; Rahman et al., 2021). The human tissue samples were relatively stiff and well-fixed macroscopically, and we found mostly that sections from AFS fixed specimens were easy to manipulate, and slightly more difficult in cases fixed with SSS. Nevertheless, SSS-fixed human sections were relatively easier to manipulate (stiffer) than the mouse sections of brains fixed with the same solutions (Frigon et al., 2022). This could be related to their different capacity to absorb the fixative (diffusion), related to their differences in shape, size, gyrification, and vascularization circuits across species (Ventura-Antunes et al., 2013).

Integrity of the neuropil was compromised in the three groups in which we found a higher occurrence of fissures. Conversely, a more uniform neuropil was observed in NBF-fixed mouse brain sections (Frigon et al., 2022). This could be explained by the quick perfusion fixation of mice and timely histological processing soon after fixation compared to the lengthy months to years delay before histological processing of the brain bank NBF-fixed human specimens which may enhance chemical artefacts on the tissue. Regarding the broken neuropils, the high concentration of isopropyl alcohol or ethanol in SSS- and AFS-fixed brains, likely dehydrates tissues significantly (Viktorov and Proshin, 2003), and may cause fissures in the neuropil by shrinking, by changing the osmolarity of the intracellular vs. extracellular spaces. This may also explain the irregular aspect of the cells in the SSS and AFS-fixed brains, compared to the NBF-fixed brains that only show regular cell contours.

We found that NeuN and GFAP antigenicity was not well preserved in all cases, especially in NBF-fixed brains, where no cells were observed in three specimens. This was expected of the NBF-fixed brain bank specimens, which are preserved in a solution with a higher concentration of formaldehyde often for years before histology processing (i.e., longer histology delays). The cross-linking, due to the over-fixation by formaldehyde immersion (Arnold et al., 1996; Stumptner et al., 2019) likely affected the conformation of the neuronal nuclei and the glial fibrillary acidic protein more profoundly than the other two analyzed antigens, due to a higher interaction between their specific tertiary conformations and the chemicals, which is in line with the qualitative description of formaldehyde and its chemical reactions (Fox et al., 1985; Puchtler and Meloan, 1985; Helander, 1994; Kiernan, 2000; Thavarajah et al., 2012; Hade et al., 2024). Moreover, an antigen retrieval protocol recovered the antigenicity for Iba1 and PLP in all cases in a homogeneous way, whereas NeuN and GFAP did not regain antigenicity in some NBF-fixed brains.

Antigenicity was generally best preserved by the AFS fixative. SSS fixation produced mostly labeling of isolated cells or patches. However, an antigen retrieval protocol recovered all the antigens in tissue fixed with the two solutions used in anatomy laboratories, either SSS or AFS. Therefore, sufficient quality IHC procedures are possible following the application of an antigen retrieval protocol. These observations differ from what we reported in our previous study in mice, in which the antigenicity of the four main cell populations was preserved and homogeneously distributed across all the types of fixation (Frigon et al., 2022), even when no antigen retrieval protocol was applied prior to the IHC procedures. This also supports the differences in mice and human brain tissue preservation properties.

The clear labeling of cells with Cresyl violet and the Nuclear red counterstain of the Prussian blue procedure, demonstrates that the brains fixed with solutions from anatomy laboratories could be used for cytoarchitectonic analysis (Amunts and Zilles, 2015; Amunts et al., 2020; García-Cabezas et al., 2020), brain morphology (Zachlod et al., 2022), neurodegenerative disease (Bobinski et al., 2000), and MRI validation (Hoffmann et al., 2011; Magnain et al., 2014; Alkemade et al., 2020). Furthermore, Prussian blue (Perl’s stain) was feasible in specimens fixed with the three solution and helped detecting ferric iron deposits.

Luxol fast blue fiber staining was not homogeneous in NBF fixed brain sections. In some NBF fixed cases, labeling luxol fast blue failed to stain myelinated fibers altogether. Similarly, luxol fast blue staining of myelinated fibers was not consistent in both SSS and AFS-fixed brains suggesting that IHC (i.e., PLP in our case) might be a better approach to routinely label the myelinated axons. Indeed, IHC (i.e., MBP, MAG, PLP antigens) are widely used in neuropathology (Stüber et al., 2014; Kuhlmann et al., 2017; Adiele and Adiele, 2019; Vakrakou et al., 2022; Beach et al., 2023).

Bielchowsky silver impregnation of fibers, was best achieved in NBF-fixed brains, followed by SSS and AFS in which more cases with heterogeneous staining and intermediate density backgrounds were observed. This may be related to the ethanol and/or amount of alcohols in their chemical compositions. An old study on silver impregnation technique showed that lipid extraction with organic solvents such as ethanol can lead to unsatisfactory impregnation and poor staining (Wolman, 1958). However, we found good cases and high-quality stains following all fixation procedures, suggesting that the fixatives used in anatomy labs may be reliable for neuroscientific research.

It is known that molecular conformations change at different paces after death, depending on the molecules being analyzed and the environmental conditions that affect the bodies (e.g., high temperatures). The changes happen within minutes for some cases, and within the first few hours of PMI for others. We therefore expected better histological results in brains with shorter PMI (Spokes, 1979; Palmer et al., 1988; Ravid et al., 1992; Hynd et al., 2003; Beach et al., 2015) and with a short HD, avoiding excessive post-mortem tissue decay and cross-linking from over-fixation (Arnold et al., 1996; Schulz et al., 2011; Stumptner et al., 2019). However, in our convenience sample, we found no evidence that neither lengthy PMI nor HD significantly affected the tissue; some good quality histology was obtained even in the cases with longer delays. This is in line with the recent work of Hade and their team, which reports good quality histology results with PMIs up to 37 h (Hade et al., 2024). It is worth noting that some individual cases suggested a slight influence of the PMI in the NBF-fixed brains, on the ease of manipulation, and antigenicity preservation of NeuN. These NBF brain bank tissue samples had significantly greater PMI (median = 45 h) than the anatomy laboratory specimens (median = 22 h) (p < 0.001), so it is conceivable that brains fixed with longer delays in anatomy labs could show the same patterns than brain bank specimens. However, we could not test this, given that our laboratory generally refuses bodies with PMIs of 48 h or more. Finally, we found significantly more homogeneous GFAP distribution in cases with a shorter HD and PMI in AFS-fixed brains, suggesting that this antigen might be more affected by these confounding variables, as reported by a recent study (Hade et al., 2024). Other results were not impacted in AFS-fixed brains, in which all histological procedures were successful.

Our results suggest that overall PMI and HD do not seem to impact the tissue quality. However, we saw a bit of an impact in some specific variables, such as ease of manipulation, NeuN and GFAP antigenicity. So, we consider that the prioritization of shorter delays should be preferred when feasible. Future studies increasing the number of specimens will elucidate more precisely the impact of longer delays on specific antigens.

First, working with human post-mortem tissue involves dealing with confounding variables, such as the post-mortem delay and the histology delay. The brains could also be affected by other variables such as the age, sex, presence of comorbidities and fixative perfusion quality (i.e., embalming of the full body). The post-mortem and histology delays have been assessed as confounding variables in each fixative group separately, and we found that they did not significantly impact our results. However, we acknowledge that the very different PMI and HD across fixative groups may affect the variables of interest, so further investigations across solutions in balanced samples with similar ranges of PMI and HD are warranted. Additionally, although our specimens were from a convenience sample, they mostly included normal-aging brains with absence of neuronal or vascular comorbidities, with a random distribution of sex. Also, all the brain specimens were removed from adequately fixed whole bodies that were used for regular dissection courses, translating an overall adequate perfusion procedure. However, a more detailed assessment of the perfusion of the fixative during the embalming (e.g., quantification of thrombi removed from the carotid arteries and jugular veins) could be used as an additional variable of interest in future studies. Second, using a convenience sample ensured that we obtained tissues from different brain lobes. However, we confirm that we used neocortex for all the specimens. Third, free-floating tissue sections stained with IHC procedures might show heterogeneous labeling, since the staining depends on the level of antibody penetration and the tissue background. Analyses might therefore be affected by the protocol itself, which was also evaluated as a variable of interest in our previous study in mice brains (Frigon et al., 2022). Therefore, the protocols were already tested and validated for this study, so we excluded the antibody penetration and tissue background quality as variables here. Fourth, we are aware that other antigens of interest could have been tested, but we limited our study to the antigens that were previously validated in our study in mice (Frigon et al., 2022). Fifth, we acknowledge that our study was conducted in a qualitative manner. However, quantitative assessment of the four cell types would be necessary to determine whether results obtained across solutions would be valid to assess specific pathological changes of the brains (e.g., loss of neurons in an area due to neurodegeneration in a brain fixed with NBF, vs. a brain fixed with SSS or AFS). Finally, the statistical power might have been limited by our relatively small sample size, especially when dealing with confounding variables. However, availability of human brain tissue is limited, and our study included all the specimens that were available at the time of the assessments. Future studies with larger sample sizes are warranted to further our understanding of the impact of confounding factors.