Twenty-first century brain banking. Processing brains for research: the Columbia University methods

The psychological and financial burdens of neurodegenerative diseases increasingly strain the familial and social framework of our societies. This trend is linked to the gradual lengthening of life expectancy and is therefore worsening [25]. Thus, ongoing efforts to lessen the burden of neurodegeneration must be improved both qualitatively and quantitatively. Many neurodegenerative diseases occur exclusively in humans. For this reason, the natural course of their pathogenesis can only be methodically and thoroughly investigated within the human brain. Therefore, the availability of carefully categorized postmortem human brains is crucial for stripping bare the complex cascade of the deleterious mechanisms leading, for example, to dementia or movement disorders. Thus, it is essential to promote brain donation for research [13, 45, 51].

The lack of generally accepted methods for processing human postmortem brains for research persists despite the huge advancements in brain banking [4, 7, 16‐18, 27, 28, 30, 37, 40, 43, 47, 49]. Thus, developing standardized methods and networking existing brain banks of academic centers are the prerequisites to meet the growing requirements of modern research. This constellation of well-coordinated efforts of brain donation, processing, storage, and research is crucial for alleviating the miseries of neurodegeneration.

Nevertheless, brain banking cannot be achieved at the cost of the teaching mission of academic institutions by routing brains away from residency programs, particularly when the autopsy rate is steadily decreasing [24, 39]. A compromise must be worked out whereby a brain can be utilizable for diagnosis, research, and teaching.

Above all, any human brain donated for research must be accurately categorized according to clinical and pathological data. The best diagnostic categorization possible must be secured and reached, at the same time maximizing the yield of samples for basic investigations. Thus, carrying out reliable, reproducible methods is all-important.

Electronically recording either the absence of changes or the type, distribution, and extent of changes in brains banked, is imperative to best meet the precise requirements of explicit investigations in a qualitative, timely manner. Electronically tracking the samples to monitor their destiny greatly enhances the capability of brain banking. Indeed, coupling the electronic tracking system with a database including the diagnostic characteristics of the stored blocks assists in sorting out a set of samples that best match the parameters of a tissue request. Additionally, electronic tracking of the methodically categorized samples is essential for swift retrieval of those that are suitable for a specific request among the ones kept in freezers.

The methods currently applied at the New York Brain Bank––Columbia University, New York––are meeting these challenges. They were developed along basic guidelines provided by pioneers of brain banking, including, among others, Drs. E. D. Bird, N. J. Cairns, F. F. Cruz-Sanchez, R. Faull, M. Graeber, C. M. Hulette, K. Jellinger, R. Ravid, and W. T. Tourtellotte. The late Dr. E. P. Richardson and Dr. E. T. Hedley-Whyte were central in combining the tasks of postmortem neuropathologic evaluation, teaching, and brain banking.

The methods reported here have been gradually constructed, improved, and implemented since 2001, thanks to strong support from Columbia University, the Taub Institute under the leadership of Dr. Michael Shelanski and Dr. Richard Mayeux, and the Hereditary Disease Foundation. These methods are adequate for achieving reliable diagnostic assessments, efficient teaching, and obtaining many well-characterized and optimally prepared brain samples that fulfill the requirements of modern research, especially on neurodegeneration.

A rigorous organization of freezer space, coupled to an electronic tracking system with its attached software, fosters efficient access for retrieval within minutes of any specific frozen sample among more than 150,000 currently in storage. In 2006, the monthly mean of samples disbursed to eligible neuroscientists worldwide was up to 500. The monthly mean of source brains for the corresponding disbursement of samples was 150. The time line between the receipt of a request and the actual disbursement of samples was five working days. This report describes how this achievement is feasible with emphasis on the processing of brains donated for research.

Briefly, modern methods of brain banking must make the following crucial aims attainable:To achieve these aims, the brains donated for research must be optimally processed as soon as possible after death. The prosector must have a good command of the neuroanatomy, neuropathology, and the protocol. After harvesting the cerebrospinal fluid (CSF), one-half of each brain is immersed in formalin for performing thorough neuropathologic evaluation, which is combined with the teaching task. The contralateral half is extensively dissected in the fresh state. Up to 150 or more fresh samples are harvested per half-brain, either as blocks or as pulverized aliquots of parenchyma. The anatomical origin of each sample is recorded using the map of Brodmann for the cortical samples [11]. The samples are frozen using liquid nitrogen vapor (LNV) at −160°C, and barcode labeled. Currently, three measurable parameters serve to assess the quality of the tissue banked: the pH value, the yield of total RNA per unit of tissue, and the extent of degradation of ribosomal RNA.

Thus, samples banked are: (1) LNV fresh frozen blocks (size range between 0.3 × 0.5 × 1.0 cm and 0.5 × 2.5 × 3.0 cm), which are especially suitable for studies requiring preservation of the cellular morphology and that of the cytoarchitecture of the area of interest; (2) aliquots of fresh-frozen pulverized brain parenchyma (1.0–1.5 ml), especially for studies focusing on biochemistry, protein studies, or molecular biology; (3) fresh-frozen CSF aliquots (1.0–1.5 ml); (4) fresh-frozen miscellaneous tissues (e.g., pineal gland, pituitary gland, choroid plexus, vessels); (5) a standardized series of 18 buffered formalin phosphate-fixed and paraffin-embedded blocks, and spinal cord; and (6) 10% buffered formalin phosphate-fixed remnants of brain and spinal cord that are left from the dissection.

This report focuses on integrated methods (referred to as “Protocol 1”) of processing brains donated for research, which are designed to: (1) reach accurate neuropathological diagnosis; (2) optimize the yield of samples; and (3) process fresh-frozen samples suitable for a wide range of modern investigations, using either traditional or the latest technologies. These methods are now applied successfully at the New York Brain Bank (NYBB) of Columbia University. Indeed, our novel methods optimize the yield of precious human brain tissue for research, secure the postmortem diagnostic procedure, and foster teaching opportunities for scientists in training, medical students, residents, fellows, and health professionals. Although the logistics of the electronic tracking system and combined software are essential operative assets of the NYBB, only key relevant points will be addressed in this report. Not addressed in this report are the safety precautions and the issues of confidentiality that must be applied in human tissue repository for research.

The laboratories of the NYBB are next to the autopsy suite of the New York Presbyterian Hospital. The operation of the NYBB is fully integrated with the functions of the Department of Pathology and Neuropathology of the New York Presbyterian Hospital. This integration fosters the teaching tasks of the department, and prevents the duplication of efforts in the histology laboratory for processing the tissue used for microscopic evaluation of the banked brains. Thus, each brain donated to the NYBB is an opportunity to establish definitive postmortem diagnoses, teach trainees, and harvest samples for research.

The operational space assigned to the NYBB consists of five and one-half rooms: (1) storage freezer room (32.0 m²), (2) laboratory (35.0 m²), (3) dissection room (16.6 m²), (4) storage room for formalin-fixed samples (3.9 m²), (5) office of the director (27.0 m²), and (6) one shared room for administration (8.0 m²). The total surface of these rooms is 122.5 m².

The personnel include a laboratory supervisor, an administrative assistant, a postdoctoral research scientist, and a board-certified neuropathologist, who is the director and who has as an appointment in the Department of Pathology. The director and the postdoctoral research scientist, both MDs, perform the dissection of the brains and are on call 24 h/7 days.

To prepare a brain according to Protocol 1, we use the following items:Accessories for slicing the brainstem and cerebral hemisphere:Details on specific items used will be provided with the proceedings implicating them.

Only professionals who are familiar with neuroanatomy, neuropathology, and have a good command of the protocol perform the dissection of the brain to minimize the probability of overlooking unexpected changes [31]. Usually, two individuals are involved with the dissection, which lasts between 60 and 90 min.

The whole fresh brain is examined, photographed, and weighed. Any identified abnormality is recorded. Then the CSF is obtained. To harvest the CSF, the whole, fresh brain is placed on the cutting board with the ventral aspect facing the prosector and with the frontal poles away from the prosector. A transverse cut is performed through the distal end of the infundibulum. Then, the distally truncated safety needle cap attached to a clean 5 ml syringe is inserted into the infundibulum. The CSF is drawn up into the syringe. To increase the yield of CSF, the occipital lobes are gently raised to drain the CSF toward the infundibulum. Then, aliquots of the CSF are generated using an appropriate number of barcoded vials, each containing about 1.5 ml. The olfactory bulb(s), pineal gland, and optic nerve are then harvested, frozen, and placed in separate, barcoded vials.

The selection of which half-brain is prepared for research depends on the numerical day on which the fresh brain is received for processing (e.g., even day, right half; odd day, left half). However, if a unilateral lesion is detected on gross examination of the fresh brain, the side involved is formalin fixed. In instances in which the extent of the degenerative changes is asymmetric, as in corticobasal degeneration (CBD) for example, the more severely atrophic side is kept for neuropathological evaluation.

In the absence of a focal lesion or symptoms indicative of a localized, concomitant pathologic process, it is assumed that the findings in one half-brain reflect the parenchymal state of the contralateral half-brain, which is verified while processing the fixed counterpart. If the cut surface of any freshly processed slice shows unexpected changes, that sample is immersed in formalin. Thus, it is important that a trained professional processes the fresh half-brain at this seminal stage of brain banking. One must guarantee that the samples made available to neuroscientists harbor the anomalies proper to the disease investigated, or are eligible as control.

Figure 3 summarizes the fate of the half brain prepared for research, and that of the contralateral half used to reach the best diagnostic categorization, as an integrated part of the teaching mission of an academic center.

Multiple sampling eases the reproducibility of experiments using blocks of tissue from the same brain by the same team, or among diverse teams of investigators, which is compellingly essential for validating results. Moreover, multiple sampling allows comparisons of data, which may derive from different Brodmann areas. This is a must for correlating findings obtained with functional neuroimaging methods, including tensor- or voxel-based morphometry. Multiple sampling expedites the disbursements of brain blocks or aliquots with accurate topographic and diagnostic distinctiveness.

Three series of blocks are harvested: (1) a series of 36 blocks obtained from 18 precisely selected areas of each half-brain, which is referred to as “standard brain blocks” (SBB). The 18 blocks to be frozen, which are obtained from the fresh half-brain (Fig. 4: representative fresh frozen blocks, top row, right), exactly match the complementary set of 18 blocks obtained from the fixed half-brain (Fig. 4: representative matching formalin fixed blocks, middle row, right), which is processed for microscopic examination (Fig. 4: bottom row, right), and obtained from the fresh half-brain only; (2) a series of blocks, which includes the remainder of the same areas (homotopic extra series) as the SBB series defined above, without, however, matched counterparts from the fixed half-brain (e.g., additional levels including the visual cortex); and (3) a series of blocks (heterotopic extra series) referred to as “additional brain blocks” (ABB) [e.g., a block including Brodmann areas (BA) 11, 47] obtained outside the areas dealt with above. In addition to these three series of blocks, pulverized brain aliquots, pulverized in liquid nitrogen, are harvested and stored frozen in vials (see below).

The anatomical origin (e.g., Brodmann area) of each cortical sample that is banked is one of the factors used to identify any specimens in storage [11]. Thus, the dissection of the fresh brain must be meticulous and anatomically as precise as possible. Therefore, a few technical details are provided. Because the demand for the mesencephalon is far greater than its availability, two levels are obtained (Fig. 5).

The rostral level includes the red nucleus with the pars reticulata > pars compacta. The caudal level includes the decussation of the superior cerebellar peduncle with the pars compacta > pars reticulata. To best obtain these two levels, the following technique (Amaya’s version) is applied (Fig. 6): the hemi brainstem and hemi cerebellum from the fresh half-brain are detached from the cerebral hemisphere by a transverse cut through the rostral tip of the mesencephalon along a straight line, tangent to the ventral aspect of the mammillary body, and to the dorsal aspect of the superior colliculus (this line is parallel to the line passing between the superior and inferior colliculi, and the dorsal edge of the 3rd cranial nerve). Next, the hemi cerebellum is detached from the hemi brainstem by sequentially sectioning the superior, middle, and inferior cerebellar peduncles.

The cerebral hemisphere is cut coronally along an axis perpendicular to a line that is tangential to both the ventral aspect of the inferior temporal gyrus and the occipital pole (Fig. 8d). Either the medial or the lateral aspect of the hemisphere rests on the movable plastic cutting board in obtaining coronal slices. The stability of the cerebral hemisphere is optimal when its medial aspect, which is flat, rests on the cutting board. This option is best for obtaining 0.5 cm-thick slices that are free of distortion. However, the discrete anatomical landmarks of the medial aspect of the hemisphere are not visible. Nonetheless, topographic accuracy of the standard blocks is secured because of the distinctive condition of the slices obtained (Fig. 10).

We identified 18 brain sites (see below) from which the harvested blocks can match up to 95% of tissue requests, as per our experience during the past 20 years. The parts of these 18 brain sites can be reliably represented in a series of carefully selected blocks, the harvesting of which we have standardized. Furthermore, the availability of an identical, but contralateral series is crucial for the microscopical assessment and diagnostic categorization of the brains that are banked. Thus, as mentioned, the blocks of this series are referred to as “Standardized Brain Blocks” (SBB); 18 are obtained from the fresh half-brain (Fig. 12), and 18 from the fixed half-brain. Pictures including the anatomical landmarks of the SBB are available online (http://​www.​newyorkbrainbank​.​org). The 18 standard brain blocks depicted are the reference blocks (as per their origin and nature) of the SBB series. Because brain tissue is extremely valuable, more blocks are harvested. These extra blocks are either homotopic or heterotopic with reference to the standard series (SBB, see below).

Homotopic extra blocks contain either the same BA or structure(s) as the one(s) included in one of the 18 reference blocks or SBB. For example, SBB1 contains BA9. Many blocks containing BA9 can be obtained. Thus, the additional blocks containing BA9 are homotopic to the reference block SBB1, and are identified as SBB with a denominator that distinguishes each of them (SBB1.2, SBB1.3, etc.), and from the original block or reference block (SBB1.1).

A heterotopic block is a block that does not share its origin with any block of the standard series (SBB) and is referred to as “Additional Brain Block” (ABB), which is further identified according to the BA to which it belongs.

The minimal size of either SBB or ABB is about 0.3 × 0.5 × 1.0 m. Their maximal size is about 0.5 × 2.5 × 3.0 cm.

The series of SBB and ABB blocks are removed sequentially from the fresh, coronal slices. To freeze the blocks with liquid nitrogen vapor (LNV) we use a pair of Teflon-coated aluminum plates (each plate measuring 1.0 × 8.0 × 10.0 cm) and the XLC140 vessel (Fig. 13).

To obtain LNV, we use an XLC140 vessel (Fig. 13). The walls of this vessel include a sponge that absorbs liquid nitrogen (LN) and gradually releases it as vapor or LNV within a centrally located cavity where the samples are being frozen. Liquid nitrogen vapor allows freezing samples at LN temperatures without LN contact. Because LNV as compared to LN exerts less stress on the tissue, and minimizes the interface during the freezing process, XLC140 vessel (or an equivalent vessel) is crucial for the processing of a brain for research, yielding blocks with minimal or even without freezing artifacts. At regular intervals (once/24 h), LN is added to the XLC140 vessel to maintain its low temperature (−160 to −180°C; the actual temperature of liquid nitrogen is −195°C). LN is stored in a special tank in the dissecting room. A backup tank of LN is kept available.

The pair of pre-cooled (−70 to −100°C) Teflon-coated aluminum plates is used to create a “sandwich”. The “sandwich” consists of a lower plate on which five or more fresh blocks are laid flat (Fig. 14). Then, on top of the blocks, the upper plate is gently apposed. Next, the “sandwich” is instantly transferred within the cavity of the XLC140 vessel. A lid is placed on top of the XLC140 vessels while the freezing takes place (about 5 min). Once frozen, the blocks are removed from the “sandwich”, and carefully arranged on the cold shoulder of the XLC140 vessel in such a way as to avoid mixing up the blocks. Next, each block and its specific barcode label are transferred into an individual plastic bag, which is then sealed (Fig. 3). The bagged samples are subsequently stored temporarily at −80°C in an operational freezer and then electronically tracked and dispersed in storage freezers (see below).

In addition to both the SBB and ABB series, samples of cerebral cortex, cerebral white matter, and cerebellum are obtained separately to provide aliquots of fresh frozen parenchyma. These aliquots are ideal for biochemistry, protein studies, or molecular biology investigations. They are not suitable for studies requiring the preservation of cytoarchitecture or cellular morphology. The process of freezing these cerebral or cerebellar cortical aliquots differs from the process of freezing the white matter aliquots.

We use “ultra-low freezers” (< −70°C) grouped into two categories (Fig. 15):

Each freezer is connected to: (1) the standard, hospital electric supply network; (2) the emergency generator dependent electrical supply network; (3) CO₂ cylinders; and (4) remote calling alarm system.

The “box-areas” are stored in such a way that samples of a single brain are dispersed in 3–6 separate storage freezers, depending on the number of samples obtained from the source brain. Thus, in the event of a failure of an individual freezer, despite the provisions in place, it is unlikely that the entire set of samples obtained from a single brain would be ruined.

Samples cannot be added or retrieved from any storage freezers without the supervision of the laboratory supervisor, except in emergency situations. For example, if a storage freezer fails, whoever happens to be in a position to perform the transfer of samples from the idle freezer to the backup operational freezer will perform the transfer. If such a transfer must be undertaken, the following precautions must be taken. While transferring the racks, one must make sure that the topographic status or distribution of these racks in the back-up freezer matches that of the disabled freezer. This precaution keeps valid the coordinates of the electronic tracking of the samples that must be transferred and stored temporarily in the backup operational freezer.

Antemortem events, such as anoxia or hypovolemia, affect the molecules in brains more profoundly than a prolonged postmortem interval at least up to 24 h [3, 15, 32, 34]. There is no significant change in the pH of the CSF or of brain samples within a postmortem interval of 24 h [3].

Measurement of tissue pH provides a means to screen a postmortem brain for RNA preservation, and helps match control versus diseased brain samples [5, 32, 40].

Qualitative assessment of ribosomal RNA, and determination of tissue pH are performed using two blocks (one for RNA and one for pH evaluations, each block measuring 0.5 × 0.5 × 1.0 cm). These blocks are harvested at the preoccipital notch (BA37), where the watershed territory of the anterior, middle, and posterior cerebral arteries is located. If a longstanding, terminal hypoxia or hypovolemia occurred, then, theoretically, related changes involving this site should be reliably representative for the rest of the brain. Harvesting of these two blocks is performed at the end of the brain processing to maximize the accuracy of the assessment of tissue quality.

Clinical and pathologic data determine the categorization of the brain for research. Thorough, standardized pathologic examination is performed to give the best survey of normal or abnormal brains. For the neuropathological evaluation, the contralateral half-brain is used, immersed in 2.0 l buffered, 10% formalin solution, at room temperature, over 7–10 days.

The gross and microscopic examination results are recorded in two standardized, neuropathology reports for each brain. One report is narrative or text-based as part of the patient’s chart. The second report is electronic, including semi-quantitative assessments of changes, to identify samples stored for research according to criteria such as the extent of degenerative changes.

For the diagnostic categorization of the brains, we use the published criteria listed on the NYBB web site (http://​www.​newyorkbrainbank​.​org). For example, we assign the diagnosis of Alzheimer disease when the criteria of “The Consortium to Establish a Registry for Alzheimer’s Disease” (CERAD) are met [33]. The likelihood that dementia is due to the Alzheimer changes (neuronal loss, presence of neurofibrillary tangles of Alzheimer and of neuritic plaques) is assessed according to the criteria proposed by “The National Institute on Aging, and Reagan Institute Working Group on Diagnostic Criteria for the Neuropathologic Assessment of Alzheimer’s Disease” [46]. Furthermore, we assign a Braak and Braak stage to reflect the extent of involvement of the neurofibrillary tangles of Alzheimer [8, 9] or that of the Lewy body-containing neurons [10]. A brain is assigned to the category of Alzheimer disease Lewy body variant (ADLBV) if there is documented dementia, neuronal loss with neuritic plaques, and neurofibrillary tangles that occur in numbers of diagnostic significance for AD, and with cortical and subcortical Lewy bodies [19, 20]. The subcortical areas with Lewy bodies include substantia innominata, amygdala, hypothalamus, substantia nigra pars compacta, nucleus coeruleus, and dorsal nucleus of vagus. These criteria are subject to changes contingent to authoritative guidelines.

To identify samples in storage that best match the specific requirements of an investigation, a link combines the file of requests (filled by the investigator) with the file including the characteristics of the samples in storage (filled by the neuropathologist). The request file lists a set of questions addressed to the investigators; familiarity with either the neuroanatomy or neuropathology, or both, may vary. The questions guide the requestors; the answers greatly contribute to the pre-selection of the pool of samples that may or may not be eligible for a particular study. In addition, the request file offers a set of pictures providing details about the neuroanatomy and the SBB with their structures or areas, which are labeled.

The quintessence of the categorization of these samples is the “distributive diagnosis”, which is assigned to each brain and, by extrapolation, to each of its derived samples. Furthermore, a “matching diagnosis” is assigned to each brain, which is a more general diagnostic subgroup, and thus less specific than the “distributive diagnosis”. The subgroup “matching diagnosis” allows an electronic-supported, pre-selection of brain samples that may be suitable for a given request. Both the clinical (CDx) and the neuropathologic diagnoses (NPDx) determine the distributive diagnosis (DDx) of any brain processed and made available for research. Thus, the summary formula is CDx + NPDx = DDx. For example, if the thorough macroscopic and microscopic examinations using conventional methods of brain evaluation do not reveal any specific diagnostic abnormality, the neuropathologic diagnosis assigned to this brain would be “Diagnosis: There is no diagnostic abnormality recognized”. Furthermore, if the clinical diagnosis is “myocardial infarct” and the individual were without any neurological or neuropsychiatric impairment, the distributive diagnosis would be “No diagnostic abnormality recognized” and the matching diagnosis would be “Normal control.” However, if the clinical diagnosis were “Schizophrenia”, then the distributive diagnosis would be “No diagnostic abnormality recognized––Schizophrenia”, and the matching diagnosis would be “Schizophrenia”. Thus, this brain would be suitable for research on schizophrenia. That, in schizophrenia, volumetric loss of the brain or enlarged ventricles compared to individuals without neurologic or psychiatric disorders occurs, or that changes involving the pyramidal cells of the entorhinal cortex may be detectable using conventional methods of neuropathological evaluation, although not specific, would further support the eligibility of this brain for studies on schizophrenia [21, 26].

Brains from individuals without neurologic or psychiatric disorders may or may not have focal lesions. When no lesion is found on thorough neuropathologic examination, the distributive diagnosis assigned is “No diagnostic abnormality recognized” and the matching diagnosis is “Normal control.” However, if asymptomatic infarcts, or sequelae of a subacute or old trauma, or acute, focal damage involving an otherwise unremarkable brain were found on examination, the distributive and matching diagnoses assigned would be “Possible control” and the samples of this brain would be flagged to highlight its limitation. Thus, “Possible controls” are clearly segregated from “Normal controls”. Samples from brains categorized as “Possible controls” are used in rare instances, for example, when tissues are requested to test or titer antibodies, or to adjust reagents, or to check techniques in progress. Furthermore, the distinction between “Normal control” and “Possible control” is made to optimize the validity of the interpretations of data these brains yield.

A list of request diagnoses (RDx) is included in a pull-down menu on the electronic request form, which is accessible at our web site (http://​www.​newyorkbrainbank​.​org). An investigator seeking a tissue can select the diagnosis of interest among those listed on the pull-down menu. If the one sought is not itemized, the investigators can specify the diagnosis they are interested in by using a dedicated field.

The software identifies samples in storage with characteristics that best match the variables of the request, as a link exists between the “distributive diagnosis” and the “matching diagnosis”. The samples with the best complement between the “request diagnosis” and the “distributive diagnosis” are automatically identified. Simultaneously, the coordinates of the site where the eligible samples are stored are provided, allowing their retrieval from the storage freezers within minutes.

As mentioned, we perform a detailed, neuropathologic evaluation to characterize the variables of samples made available for research. We then generate two standardized reports based on the results of both macroscopic and microscopic examinations of the brain prepared for research. One report is text-based for the patient’s clinicians and chart. The second is electronic and figure-based to assess the extent of the severity of any detectable changes. The resulting data are used to identify electronically, for example, a block of cerebral cortex (BA9) of a woman with mild dementia, which contains less than three neuritic plaques per 100× microscopic power. The software will provide a list of the matching samples for BA9, including the freezer coordinates to locate these samples of interest. Once they have been disbursed, a code indicates who performed the disbursement and to whom, where, and when the samples were sent. In addition, upon retrieval of stored samples (currently up to 500/month), the software records the space made available for newly harvested samples, with a consequential 45% increased efficiency of the freezer cabinets.