The BTRC has developed a unique brain banking model to recruit individuals affected by AUD along with neurologically normal controls (Figure 1).

Coordinated preservation and collection of human brains began in the late 19th century, with these practices evolving into the modern brain banks of today (21). Early post-mortem brain research was largely histological in nature but evolved into omics work, requiring high-quality fresh frozen tissues. This trend continues with the latest spatial platforms, particularly spatial omics, that require both high-quality molecular components and highly preserved cytoarchitecture. The ability to supply bank tissue that is fit for purpose across evolving technologies requires an acute awareness of researcher needs, in addition to flexibility, time, and proactive networking across international brain banks to create effective and standardised best practices. Ideally, the brain bank team or closely affiliated members should be active researchers who can predict trends, oversee tissue protocol development, and perform the beta-testing of new methodologies.

The spatial technologies on the market are varied (e.g., MERFISH, NanoString Geo, and 10x Visium) and growing rapidly. The intricacies inherent to these techniques and level of optimisation mean that banks and researchers may be better placed working collaboratively. One issue is that banks are tasked with prolonging tissue resources and ensuring that tissue is supplied in an equitable manner. It may be easier for banks to do these techniques on behalf of researchers as opposed to providing whole FFPE or frozen tissue blocks to researchers so these techniques can be firstly optimised and then used at the researcher’s discretion.

Another major issue with the spatial technologies is the trade-off between resolution and capture area. This increases the already large challenge of reproducibility between studies that aim to sample the exact same brain region. Currently, regions are visually identified and manually processed by the staff of each bank, leading to variation in the exact site sampled. A coordinate-based schema for directing dissection would be a major step forward in this regard, particularly if adopted widely. There are a number of initiatives in play to achieve this through three-dimensional (3D) reconstitution of 2D tissue slices (22), which often incorporate in vivo or ex vivo MRI (23–25). One of the first comprehensive attempts to create ‘cytoarchitectonically parcellated’ regions of interest, combining histological sections with in vivo MRI, was carried out in macaques (26). This utilised whole fixed brains, sectioned in celluloid moulds, Nissl-stained, and converted to grey-scale images for MRI registration via a multi-stage algorithm. As a labour-saving task, they repeated the registration, substituting photographic images of the sections taken before staining, and the results were similar. Given that the majority of banks hemi-sect a brain, photographic images of slices of both hemi-brains taken prior to storage may represent the most effective method of the 3D reconstruction (24). Gazula et al. (24) used surface scans (3D scanning) of the whole brain, rather than just relying on a probabilistic atlas. This method is agnostic to slice thickness and thus is suited to the hemi-brain approach, where the side to be frozen is sliced fresh, with variable thickness, relative to the more rigid, and thus finely sliceable, fixed hemi-brain.

The BTRC has ethics approval from the Human Research Ethics Committee (equivalent to an IRB) at the University of Sydney, which covers all stakeholders, including in-kind contributors, to the consenting, retrieval, and dissemination of tissue. The UoB consent form has been developed as ‘broad consent’, such that each donor agrees to allowing tissue to be provided to researchers internationally for unspecified research and that data arising from this research may be shared in public repositories. This has been an important point, as many researchers use post-mortem brain tissue from the BTRC or other banks under a waiver from their own IRB. It is now mandated by many funding bodies and journals that omics data be uploaded to data repositories and that effective data management should operate under findable, accessible, interoperable, and reusable (FAIR) principles (27). However, these same IRBs are then called upon to guarantee to data repositories that the work producing these data had been carried out in an ethical manner. They are increasingly deferring to brain banks to ensure that the work is in line with the original consent of the donor. This is problematic as two sources of genomic data can be proven by a third party to come from the same individual and may in some jurisdictions amount to de-identification. These third parties will increasingly bring artificial intelligence-assisted technologies to data mining, but the corollary is that fewer donors may consider donation if their anonymity, privacy, and sense of data ownership cannot be guaranteed (28). This may also differ between those affected by a disease and healthy volunteers (29).

In response to researcher uptake of spatial technologies, which requires both high-quality RNA and cytoarchitecture, the BTRC has refined and continues to refine the processing of donor tissue. The fixed hemisphere is now perfused-fixed in 15% neutral-buffered formalin, followed by immersion for up to 24 h; a reduction from 3 weeks (diffusion fixation) as previously reported (3). We are currently exploring changes to our procedures to freeze tissue to enable better cytoarchitecture in this tissue. Initial trials of rapid freezing methods and varying cryoprotectants were carried out in sheep necropsy brain tissue (20), with trials of freezing methods for human tissue ongoing.

As research technologies evolve and change in popularity, so do brain regions of interest. Post-mortem human brain is not easily accessible and has limitations; therefore, animal models and imaging studies have long led the way into brain research. As new technologies have allowed deeper analysis of animal tissue and human images, various new regions of interest have emerged as candidates for validation in human tissue. To keep up with this, the BTRC continuously adds regions to be dissected as standard blocks (Table 1).

Given the relative scarcity of human post-mortem brain tissue available for research, the BTRC is committed to exploring ways to maximise research output using minimal tissue. One such option is the creation of tissue microarrays (TMAs). Briefly, FFPE TMA blocks for one case are created containing tissue cores from 10s to 100s of parent blocks for multiple brain regions (30). This allows concurrent staining and comparison of many brain regions in one section. Multiplex immunofluorescence (mIF) is another method that can maximise outcomes whilst reducing reagent and tissue use. Combining TMAs with mIF is a particularly effective way to broadly sample multiple brain regions simultaneously to look at the regional impact of neurodegenerative diseases (30) or AUD (53). Furthermore, mIF can be combined with spatial transcriptomics technologies to provide unprecedented data on the juxtaposition of pathology and cellular subtypes. We have shown how more effective mIF can be used with 24-h FFPE tissue, but many important markers used in neuroscience can be robustly used in 3 weeks—or longer—of fixation times. mIF is constantly evolving but remains relatively expensive when using commercial technologies that use covalent amplification reactions to circumvent antibody specificity and fluorophore abundance issues. At the same time, the BTRC has shown that more traditional methods can be used with iterative strip-incubate cycles to approach the outputs achieved with newer technologies (Maskey et al., in this issue). Finally, digital pathology is increasingly important in human post-mortem tissue analyses. The BTRC recognises the need to use machine learning algorithms to speed up the quantification of cells and pathology and their spatial interactions whilst maintaining stereological principles of sufficient sampling (31). Current progress at the BTRC is best described as semi-quantitative (30) with ongoing work to use newer deep learning algorithms such as nnU-Net (32) to achieve full automation.

Historically, pH and RINs have been used as the standard measurements to assess macromolecule quality in brain samples. FFPE tissue would typically not be the first choice for RNA-based experiments. However, new sequencing-based spatial transcriptomics platforms are designed for FFPE tissue (33). In tissue fixed for 24 h, the integrity of RNA measured in terms of fragments exceeding 200 bp (DV200) can exceed 60%, even though the RIN is less than 2 (20), whereas standard blocks typically result in DV200 approximately 25% or lower. In comparison, frozen cryosections are typically approximately a RIN of 7, with DV200 of 80–90%, but, as discussed elsewhere, current freezing methods compromise cytoarchitecture and potentially allow transcripts to disseminate from their cell of origin.

The BTRC has typically recorded blood alcohol content (BAC) to confirm active drinking in coronally sourced AUD cases. Then using medical records, our team has estimated lifetime alcohol consumption, DSM IV and 5-based alcohol dependence, and AUDIT-C score for all donors. More recently, we have augmented our current drinking determinants by measuring phosphatidylethanol (PEth), a direct biomarker of alcohol use, in frozen brain tissue using a robust in-house liquid chromatogrpahy mass spectometry (LCMS) workflow (34). PEth is an effective indicator of alcohol use in the months prior to death and particularly useful when BAC is unavailable.

The BTRC has genotyped the majority of cases using the Axiom UK Biobank Genotype Array platform to further characterise and broaden the downstream analysis that can be performed using our tissue. This is a genome-wide platform with 805,000 single-nucleotide polymorphism, including common mutations known to result in disease (35, 36). Researchers can opt to receive whole-genome data, which allow for analyses such as GWAS to be performed, or QTL analysis when combined with existing transcriptomic or methylome data that have been generated using BTRC tissue (37). They also allow the BTRC to select a cohort for research based on common variants such as apolipoprotein ε2, ε3, and ε4.

As discussed above, funders and an increasing number of journals are asking that data on which published findings are based be made available to other researchers. This not only meets the FAIR principles (data that is findable, accessible, interoperable, and reusable) but also aligns with the basic tenet of scientific endeavours as being reproducible (38). As methodologies get more complex, they are creating a so-called ‘data deluge’ for both clinicians and researchers (39). It is unclear whether raw data or normalised data should be uploaded, given the expert bioinformatics skills required to re-analyse the former. Brain banks are theoretically in a strong position to host research data derived from their tissue, as they can ensure donor and region fidelity and explore outliers by investigating medical records and potentially performing meta-analyses. The BTRC has facilitated several recent meta-analyses of transcriptomic data but has stopped short of hosting raw data (45–47). Other issues of concern include maintaining de-identification inherent to the donor consent. One option that is used by large epidemiological studies such as the UK Biobank is to set up an independent data enclave, where both data curation and analysis are carried out in a secure environment, with the bank ensuring that access and results made public meet consent and privacy stipulations. Such a data coordination centre would be another avenue where a brain bank can directly facilitate research and, importantly, make AUD data available to non-content experts who may derive hitherto unknown relationships with other diseases or general pathomechanisms.

The BTRC has been funded by the NIAAA in some form for 30 years to support AUD research. Focusing on a stigmatised disease such as AUD has many challenges, but this is offset to some extent by its commonality. An even more challenging focus is rare brain diseases, particularly in a geographically large country such as Australia. For diseases that typically have a prevalence of less than one per million, it is almost impossible for a single bank to collect a cohort of reasonable study power. One option is to make more use of the more than 60 brain banks around the world and the estimated ~100,000 banked brains (40). There are many obstacles to creating a single physical collection, such as the majority of banks not accepting tissue from outside their own country. One option that the BTRC is pursuing is a virtual international brain bank that curates tissue from rare brain disease cases worldwide. This solves a problem for researchers by establishing whether sufficient cases exist for their work, although it does not solve the likelihood of multiple tissue transfer agreements and courier fees. Moving forward, a virtual brain bank could act as a nidus for greater internationalisation in brain banking, where individual banks could be the custodians for a single rare disease collection, making dissemination of tissue to researchers simpler and cheaper.

There is much interest in finding non-invasive, clinical brain biomarkers. Plasma-based biomarkers are well established for cardiovascular and liver disorders but in their infancy for brain disease. A brain bank could expand its collection to include DNA, plasma, and cell lines. There are already existing centralised repositories for DNA, serum, and peripheral blood mononuclear cells (PBMCs) for disorders such as Alzheimer’s disease (AD), although these have not been coordinated with the subsequent donation of brain tissue from the same individuals. Recently, one such facility expanded to incorporate a biomarker assay laboratory to run AD-specific assays such as Aβ 40, Aβ 42, and P-tau 217 but also more general assays of brain damage, including Glial Fibrillary Acidic Protein (GFAP) and Neurofilament Light (NFL) (48). Given the increasing certainty that plasma assays do reflect conditions behind the blood–brain barrier (41), the BTRC could expand its collection and monitor AUD patients for potential brain damage whilst also simultaneously providing serum and cell lines to researchers. This would allow the BTRC to facilitate AUD research whilst the patient is still alive and may even encourage subsequent brain donation. It has the added benefit of contributing to the clinical characterisation of future donors.

Expanding a post-mortem tissue bank to one housing clinically derived samples would require substantive philanthropic government support. In Australia, federal research infrastructure funding is given based on maximum beneficiaries. Therefore, brain banks should consider greater interoperability with the wider biobanking community. One option would be a multi-organ bank that stores blood and cells from potential donors within the patient’s lifetime as above but also sets the foundations for deriving hypothesised systemic manifestations of brain diseases, including addictions and vice versa (42). For example, the combination of a brain, heart, liver, and cancer tissue biorepository planned for the University of Sydney would be useful for the exploration of multi-organ dysfunction due to chronic alcohol consumption (49). There are also interesting connections between somatic mutations in cancer and germline mutations in neurodegenerative diseases (50), whilst brain-specific diseases such as multiple sclerosis may represent the CNS manifestation of past, systemic viral infection that may have also affected other organs (51).

A related idea is the rapid tissue retrieval of multiple organs from a post-mortem donor. Outside of brain banking, the majority of other organ researchers can rely on surgical biopsy tissue. The use of post-mortem tissue is less common, but then the donation of deceased tissue as a legacy option is largely unknown to Australians (10). A multi-organ, multi-disease focus would allow banks to make a much more compelling case for national infrastructure funds. Finally, donated tissue is only as good as the clinical information held on that individual individual. There are currently studies, such as the UK Biobank, that are longitudinally creating the most characterised individuals in research history. There needs to be more effort to bank tissues from these individuals. In NSW, the 45 and Up study has a similar focus, and some biobanking of participants has been carried out (52). A multi-organ tissue bank from these individuals could create a unique legacy for medical research.

Measuring sample utilisation and the outcomes arising from tissue use is vital in assessing the impact and value of the collection (43). Researchers are required to report publication and presentation outcomes to the BTRC annually. More recently, the BTRC has used an effective method for tracking and identifying publications where tissue or BTRC-derived datasets have been utilised. Briefly, a robust set of search terms was devised to run an active search of citation databases, Dimensions and Scopus, to alert the BTRC when a relevant publication is released. Since 2001, 776 papers, including preprints and papers published using a BTRC-derived dataset, have been published in peer-reviewed journals, with over 1,000 oral and poster presentations given at conferences. Further analysis of these publications has revealed 456 individual authors from 443 different institutions across 44 countries. The collection has 47,961 lifetime citations and a mean relative citation ratio of 2.03.

The regulatory challenges faced in Australia are both like and different from those across national borders (5). However, there are many reasons why the internationalisation of (post-mortem) brain banks would be useful (4, 40). Chiefly amongst them is that this small industry could punch further above its weight in terms of research outputs if protocols and data collection were standardised worldwide. Danner et al. described this as a more ‘collective initiative’ to meet researchers demands (4). One stretch objective for the industry is to centralise sufficiently sized (powered) collections to facilitate research into uncommon or rare brain diseases. The transfer of tissue, data, and funds across state or country borders are all current impediments to such initiatives, and many countries are grappling with the same general privacy issues as Australia. A useful starting place would be an international virtual brain bank or database that curates and catalogues all the brains available worldwide, e.g., Brain Bank Connect (44). This could easily morph into an international community of practice in brain banking, setting the bar for best practice, and potentially hosting data, but importantly providing a precedent for bankers to advocate locally for greater internationalisation of their tissue and associated data.