Methodological Issues in the Neuropathology of Mental Illness

Joel E. Kleinman, Thomas M. Hyde, and Mary M. Herman

Although schizophrenia has been referred to as "the graveyard of neuropathology" (1), there has been a renaissance of postmortem studies in mental illness. This rebirth has been fueled by advances in neuroimaging that have allowed for in vivo neuropathology and new techniques in neuropathology such as immunocytochemistry, autoradiography, computerized neuronal morphometrics, and in situ hybridization. In point of fact, the "graveyard" quote was never meant to discourage neuropathological research, but to encourage the use of more advanced techniques. Most of this new research has involved schizophrenia and suicide, but drug addictions and specific affective disorders will probably follow suit. Although there have been numerous findings to date, many have unfortunately not been replicable or have failed to lead to significant advances. Therefore, a review of methodology in this area may prove useful.

One place to start this review is with a success story. One obvious success involves Parkinson's Disease. The discovery of reduced dopamine concentrations in the nigrostriatal pathway of patients with Parkinson's Disease (2) led to the L-Dopacarbidopa replacement treatment strategy (3), a modern neuroscience/neuropathology triumph. This success was based in large part by knowing where to look in the brain of Parkinson's Disease patients, a fact made more obvious by the loss of pigment in the substantia nigra (4). Advances in neurochemistry, such as assays to measure dopamine in fresh-frozen specimens, was also essential to this clinical research advance.

Neuropathological studies of mental illnesses have a number of issues that are not found in neurological diseases. A number of these problems are discussed in the first section, Neuropsychiatric Issues. The next two sections, Neuroanatomic Issues and Neuropathological Issues, cover issues that are probably relevant to all brain diseases.

The rate-limiting step in neuropathological studies of mental illnesses is the collecting of sufficient numbers of brain specimens to allow for a scientific study. There are at least five proven sources, which include medical examiners' offices, Veteran's Administration hospitals, state psychiatric hospitals, hospices, and brain banks. Each of these sources has advantages and disadvantages that must be considered.

Medical examiners' offices are the source of choice for the study of suicide. Most states have laws dictating that autopsies be performed by medical examiners on victims of suicide. This source of tissue offers the advantages of specimens from younger subjects who are also more likely to fall into the purview of the medical examiner. For the study of schizophrenia, this advantage may be offset by relatively less psychiatric history, which makes an accurate diagnosis more difficult. A second confound in medical examiner cases, substance abuse, can be mitigated by toxicological screening of urine, blood, or brain. The medical examiner's office provides two other major advantages: relatively short postmortem intervals and a source of normal controls from the same facility, which may reduce the effects of postmortem artifacts in the collection, resulting from factors such as ambient temperatures, which vary with the season of the year, delays in refrigeration, mode of death, and so on.

A second major source of brain specimens involves the Veteran's Administration hospitals. Along with medical examiners' offices this source is an excellent place to study alcoholism and other drug addictions. This source offers the advantages of relatively easy accessibility to medical records essential to an accurate diagnosis. Unfortunately, the large percentage of patients with dual diagnoses such as manic–depressive illness and alcoholism is a major problem in doing research in this system. Normal controls can also be obtained in these hospitals.

The state psychiatric hospital offers the most promise for prospective studies. Large numbers of schizophrenic patients who have not been placed in the community make prospective studies in the state and Veteran's Administration Hospitals feasible. The obvious advantages of this approach is that psychiatric diagnosis can be made with greater accuracy in living patients. The disadvantage of this approach is that the subjects collected in this fashion have the confound of advanced age, which may be associated with enlarged ventricles, brain atrophy, and dementia. Careful neuropathological examination for strokes and Alzheimer's Disease is a necessity in this type of study. A second major problem in this setting is that normal controls must be obtained from another source.

The hospice has been used with considerable success for the study of illnesses where death is imminent, such as from Alzheimer's Disease or other dementias. This approach is ideal for short postmortem intervals. It is of limited value for most traditional psychiatric illnesses since psychiatric patients rarely die in a hospice. Although nonpsychiatric disease controls can be obtained from this source, the patients usually have a severe debilitating systemic disease such as cancer or AIDS.

Lastly, for those researchers who do not have access to one of the aforementioned sources, brain banks at UCLA and Harvard are funded by the NIMH. Brain banks obtain tissues from any of the above sources in a standardized fashion. They have the disadvantages of obtaining tissues from great distances and multiple sources, limiting to some degree their abilities to satisfy the diverse needs of their users. The resourcefulness of their respective leaders, Drs. Wallace Tourtelotte and Edward Bird, have allowed the banks to assist numerous researchers despite their limited financial resources.

The second major problem after collection is diagnosis. This can be accomplished with prospective studies or after death using medical records, family interviews, and police reports. As mentioned previously, prospective studies are frequently confounded by age. Establishing a diagnosis after death can be difficult. The NIMH experience is that approximately a third of possible schizophrenic cases cannot be confirmed by existing records or interviews or can be confounded by alcoholism or substance abuse. Accurate psychiatric diagnosis of suicides may be even more difficult. Both of these seem simple next to determining an accurate history on substance abusers. For the most part, the latter are described by their toxicological screens. Attempts to better characterize their history by systematic use of segmental hair analysis is a possibility that has yet to become routine.

The need for proper controls is a third major issue. A good example of the problems may be a typical NIMH study. Schizophrenics collected through the medical examiner's office may involve half of the subjects dying by suicide. For this reason, a nonpsychotic suicide control group is used to control for manner of death. The confound of prior neuroleptic treatment can be met by using those schizophrenics who do not satisfy diagnostic criteria, or other neuroleptic treated patients (manic– depressive patients, psychotic depressed patients or the like). The two NIMH funded brain banks at Harvard and UCLA frequently use Alzheimer's and Huntington's Disease patients who have been treated with neuroleptics. Lastly, a group of normal controls is used as well. Each of these groups needs to be matched for age, gender, race, postmortem interval, and storage time. Controlling for socioeconomic status and intelligence is not feasible at this point in time. Tests for toxicology of urine and blood can be routinely performed, but routine brain toxicology has not been employed as of yet. Recently, testing for HIV has become routine.

A major impediment to postmortem research in mental illness has been medicolegal-ethical issues, which require family consent for brain tissue donations. Many mentally ill subjects do not have families available to donate brain tissue. Donor cards are rarely sufficient "evidence" for pathologists. Lastly, society has not been well educated to the public health benefits of neuropathological research. The need for other tissues (pituitary gland, pineal gland, adrenal gland, and the like) is also not appreciated by the public.

Despite these obstacles, this research area continues to grow. Two final methodological considerations are worth mentioning. If one collects enough tissue to match for all the relevant variables, a computerized inventory is a "must." Lastly, maintaining strict double-blind between the collectors and the biochemist or molecular biologist is an essential last step toward a good study.

The first consideration in neuroanatomical localization lies at the macroscopic or gross level. The central nervous system (CNS) is exceedingly complex, composed of innumerable cortical areas and subcortical nuclei. Random sampling of cerebral structures, like the proverbial hunt for a needle in a haystack, is unlikely to be productive. Research should be driven by hypotheses, and neuroanatomical investigations are especially in need of such an orientation. Specific regions of interest should be identified and selected on the basis of scientific evidence. This task is complicated by the subtle neuropathological basis of neuropsychiatric disorders. Brains from patients with well-recognized psychiatric disorders have been studied for more than 100 years, with comparatively little to show from this effort until very recently. The paucity of meaningful findings reflects less about the abilities of the researchers and more about the subtlety of the abnormalities. Gross inspection of the brain is a valuable screen for such comorbid conditions as tumor or infarct. In research, gross examination is useful in the assessment of subtle volumetric changes in large structures, such as the hippocampus (5).

Microscopic analyses of brain tissue probably will prove to be more valuable than macroscopic analyses in the investigations of psychiatric disorders. Recently, microscopic cytoarchitectural studies have led to a number of exciting findings in the brains of patients with schizophrenia. In the mesial temporal lobe, driven in part by data from in vivo neuroimaging studies (6), cytoarchitectural investigations have reported a subtle disarray in the entorhinal cortex (7, 8). Large structures in the brain often vary in neuronal type and arrangement regionally, in both the anterior–posterior dimension as well as the dorsal–ventral domains. This holds true for the sixlayered neocortex and the more primitive palleocortex. For example, the entorhinal cortex is characterized by clusters of neurons in layer two in its more rostral aspects, with a normal disappearance of these clusters caudally (9). It is incumbent to study the same level of entorhinal cortex in affected individuals, to make the appropriate comparison with normal controls.

Subcortical structures also have a great deal of often unrecognized anatomical complexity at the microscopic level. The patch-matrix pattern of the striatum reflects a heterogeneous distribution of neurons and neurotransmitters, and differing patterns of connectivity (10). The many nuclei of the thalamus, hypothalamus, and brainstem have been subdivided into subnuclei, differing in neuronal type, neurotransmitters, and patterns of connectivity. The nucleus of the solitary tract (NTS) in the brainstem illustrates these concepts. The rostral third receives gustatory input, and the posterior two-thirds receives input from a variety of chemoreceptors and mechanoreceptors associated with the viscera. Within the visceral NTS, there are ten distinct subnuclei (11). Input from one class of pulmonary receptors project primarily to just one, the ventrolateral subnucleus, whereas gastric afferents project to at least two different subnuclei (12, 13 ). Receptor distribution is also inhomogenous in this structure, with 5-HT3 receptors restricted largely to one subnucleus (14). Therefore, careful consideration of subtle neuropathological changes must take into account these levels of complexity.

In addition to anatomical heterogeneity in brain structures, there also is functional heterogeneity. For example, in many cortical and subcortical structures, there is a topographic organization that corresponds to specific parts of the body or of the visual fields. In the striatum, there is a complicated yet well-understood somatotopy, which should be considered in any investigations involving this structure (15). Therefore, investigations involving the striatum should involve anatomically identical regions across subjects.

Connectivity also plays a role. Neuronal networks are important in the generation and modulation of complicated behaviors. Neuropathological analyses must account for the primary and secondary changes in the neural network disrupted in the disorder. For example, pathological changes in the mesial temporal lobe could cause profound but secondary neurochemical changes in the frontal lobe. Secondary abnormalities may explain many of the clinical manifestations in a disorder, such as deficits in executive function in schizophrenia (16).

Neuropathology studies must also consider the normal cellular constituency of a structure under investigation. There can be a selective loss of one subset of neurons within a structure, with relative preservation of other neuronal subtypes. Within the striatum, there are multiple neuronal subtypes. In Huntington's Disease, there is a primary loss of Golgi type II neurons, with a relative sparing of larger neurons (17). Clarification of the pathophysiology of Huntington's Disease is now focusing on the neurobiological characteristics of the Golgi type II neurons. These principles may be applicable to psychiatric disorders as well.

When fresh brain specimens are obtained, the initial handling of the tissue will dictate how the aforementioned principles of anatomic inquiry can be applied. Blocks of tissue, taken in a preselected and uniform plane of section, allow consideration of many of the concepts outlined above, including issues of precise localization and cytoarchitectural configuration. Precise localization of pathology to cortical lamina, subcortical subnuclei, and neuronal subtype requires this type of tissue. Autoradiography for both receptors and in situ hybridization are best performed on tissue blocks that preserve the normal anatomy and landmarks. The immediate dissection of fresh tissue into smaller blocks can make detailed anatomical analysis difficult if not impossible. However, immediate dissection facilitates rapid neurochemical analyses resulting in, for example, the characterization of receptors using grind-and-bind techniques or measurement of neurotransmitter levels. Micropunch methodology can be applied to large tissue blocks, allowing more precise anatomical localization while rapidly accessing tissue for detailed neurochemical analyses.

Two other issues must be considered in human postmortem studies. The first issue is lateralization of function, which is important in cortical structures, and may be a consideration with subcortical structures as well. In right-handed individuals, regions within the left temporal and frontal lobes are the primary repositories of language function. Damage to analogous structures in the contralateral hemisphere produces a completely different, and much more subtle set of behavioral abnormalities. Lateralization of function is often tied to handedness. Unfortunately, handedness is rarely a consideration in many postmortem studies, and should be considered when studying structures whose function is lateralized. A second important issue is that of normal intersubject variability. There is a great deal of variation between individuals in brain size and configuration. Any conclusion of atrophy must consider normal variations. Variations in gyral patterns and sizes are another illustration of this principle. Lateralization of function, handedness, which is often linked to lateralization, and the normal amount of individual variability must be considered in any anatomical study of human brain tissue.

In summary, the anatomical complexity of the CNS must be recognized in the postmortem study of human brain tissue in behavioral disorders. Uniform selection of regions of interest, chosen on the basis of precise hypotheses, must then be studied using methods that take into account the macroscopic and microscopic characteristics of the structure. Consideration must be given to cytoarchitectural, laminar, and/or subnuclear divisions. Finally, lateralization of function, handedness, and normal intersubject variability cannot be ignored. Although these precepts are daunting in scope, rigorous application will improve the quality of human postmortem research into neuropsychiatric disorders, and increase the value of the findings from these investigations.

An often repeated phrase that compares neurology to real estate also holds true for neuropsychiatry: in the end, all that matters is location, location, location. Lesion localization, whether within an isolated site or more widely distributed throughout the cerebrum, holds the key to deciphering neuropsychiatric disorders.

The difficulties involved in this type of research make almost every brain desirable. These are some exceptions, however. These include respirator brains, decomposed brains, and some (but not all) gunshot wounds to the head. Although brains of HIV subjects may be used, they present obvious risks to neuropathology personnel and potential confounds for research. A common misconception involves the refusal to use brains after some arbitrary postmortem interval. To be certain, death and prolonged postmortem intervals are not good for the brain. However, time from death to refrigeration is probably a more meaningful variable than postmortem interval; however, this measure is more difficult to obtain and is rarely reported.

At the time of opening the thoracic cavity, cardiac blood is obtained for HIV, toxicology, or other testing. During removal of the brain, the calvarium is carefully sectioned to avoid saw blade marks on the cerebrum. The cerebral dura is reflected dorsally and the brain is carefully freed up, using a sharp scalpel for the cranial nerves and spinal nerve roots and a curved scissors for the tentorium cerebelli and vertebral arteries. A deep cut is made in the foramen magnum to obtain all of the medulla and as much as possible of the upper cervical cord. After the brain is completely freed, it is gently delivered from the skull, weighed, put in a plastic bag, and held on wet ice, which should surround the bag as much as possible. Traction must be avoided in order not to tear the cerebral peduncles or other brainstem structures. The pineal remains with the brain if the cerebral dura and proximal tentorium are kept attached to the brain. The pituitary is removed by carefully breaking the posterior clinoid processes and pulling them caudally. The gland is then dissected out with the tip of a sharp scalpel blade, leaving its capsule intact. The pituitary and additional segments of the cervical cord, if not obtained with the initial brain removal, are placed in a plastic bag on ice.

After returning to the laboratory, the pituitary and pineal are frozen separately before beginning to section the brain. If the pineal is not accessible at this point, it can be dissected out following the rostral midbrain section. Samples of dura and vessels from the base of the brain can also be frozen at this time. A flat section is made through the rostralmost midbrain, above the oculomotor nerves, to remove the brainstem and cerebellum, taking care not to damage the inferior medial temporal lobes (entorhinal cortex and hippocampus) in the process. After bissecting the cerebrum through the corpus callosum, the brain is frozen using 1- to 1.5-cm coronal sections. The portions of the brain that are not frozen immediately continue to be held in a plastic bag on wet ice. Where laterality is the principle research issue, bissection should be avoided if possible.

Throughout all procedures of brain removal and dissection, face masks, eye glasses or goggles, durable gloves (preferrably N-Dex Nitrile gloves, which are more resistant to accidental cuts), sleeve protectors and an apron should be worn. After the procedure is completed, all surfaces and instruments should be cleaned with a 10% solution of freshly prepared household bleach (Clorox) by submersion for a minimum of 20 min. This is done to protect personnel from potential infectious organisms such as hepatitis, tuberculosis, HIV, and other agents.

The coronal sections of the cerebrum are labelled left or right and sequentially, level one (being the most rostral) through the most caudal level. The coronal sections are arranged so that the second or third section begins at the rostralmost tip of the temporal pole. As sectioning proceeds, the neuropathologist examines each piece for pathological changes and takes sections for formalinfixation and histology from any questionable areas.

The cerebellum is separated from the brainstem by cutting through the cerebral peduncles and is then bissected into two pieces in the horizontal plane. A small section is made in both pieces, in order to examine the dorsal cerebellar vermis, still leaving both hemispheres attached in the midline. The brainstem is cut in sections perpendicular to its long axis, with at least two levels of midbrain, two of pons, three of medulla, and one or two of cervical cord. Thus tissue will be available for more than one study at each level. The levels can be indicated on the bag as to upper, middle, and lower levels. The brainstem should not be cut through the midline as this will destroy midline structures.

Although modern approaches have employed fresh-frozen tissue, there is still a place for fixed tissue. Formaldehyde-fixed specimens are essential for ruling out conditions such as cerebrovascular disease and Alzheimer's Disease. Moreover, this approach is especially useful for determining the size of brain structures. It is important when using this approach to control for storage time, because this will influence brain shrinkage. Formaldehyde should be changed regularly to maintain a 10% concentration, thus preventing tissue decomposition or growth of deep-seated bacteria because of lack of potency of the fixative. Lastly, electron microscopic studies for viral particles or the like may require glutaraldehyde as a primary fixative.

Formalin-fixed tissue can also be used for immunocytochemistry and in situ hybridization studies. Routine stains are hematoxylin-eosin and the Bielschowsky method for axons, adapted for paraffin sections.

Brain specimens can be frozen in a number of ways. Rapid freezing prevents ice artifacts, which are a major enemy of autoradiography and in situ hybridization studies. Rapid freezing can be accomplished by freezing 1- to 1.5-cm-thick coronal sections with a mixture of powdered dry ice and isopentane or between metal plates immersed in liquid nitrogen. Samples are then placed in airtight plastic bags and stored in -70°C freezers with either carbon dioxide or liquid nitrogen tanks as a backup.

Ruling out concurrent Alzheimer's Disease, cerebrovascular disease and tumors is essential, especially if the cases involve the elderly. Alzheimer's cases require sections from the frontal temporal/parietal, occipital cortex, and/or the amygdala/hippocampal region. The presence of cerebrovascular disease, hemorrhage, trauma, or tumors are screened by inspection and confirmed by appropriate sections from suspected areas. Inspection and sections of the locus coeruleus and substantia nigra are also necessary to rule out Parkinson's Disease and related disorders. A piece of dorsal cerebellar vermis is necessary to evaluate ethanol-induced atrophy, and sections of the cerebellar hemisphere with dentate nucleus and hippocampus and other cortical areas aid in the evaluation of anoxic/hypoxic alterations.

Fresh tissue is difficult to accurately dissect as the tissue is quite soft. Frozen tissue poses equally difficult problems with fogging from early thawing and the danger that thawing can lead to protein denaturation. Motor driven saws are especially problematic, because their heat may denature tissues in their immediate contact. Techniques such as autoradiography and in situ hybridization can be especially helpful in avoiding these problems as they lend themselves well to half or whole coronal 14- to 20-mm-thick sections that can be made from frozen coronal blocks by use of a cryostat. Both small structures, such as the locus coeruleus, and large structures, such as the cerebral cortex, are ideal for this approach.

Although the tasks may seem daunting, the rewards can be great. Neuropathological research is not for the faint of heart. It is not easy and it is not cheap. Any startup budget involves freezers ($10,000/freezer with backup and racks) and a cryostat ($10,000 to $70,000, depending on the model). Personnel ideally would include a neuroanatomist, a neuropathologist, a psychiatrist (preferrably two) and a minimum of two technicians. The potential, however, is great. When the infamous bank robber, Willie Sutton, was asked "why do you rob banks?" he responded, "that's where the money is." Such is the promise of neuropathological research for mental illness.

published 2000