17 Techniques in neuropathology
Neuropathology has classically been seen as something of a dark art by general histologists because of the tradition of using a large number of obscure and often capricious stains. However, with the ascendancy of molecular pathology, many of the more unreliable stains are being replaced by immunohistochemistry. As such, many of the preparations that have been described in previous editions of this volume have been omitted from this chapter as they are no longer in use. Nonetheless, a number of reliable and useful tinctorial and metal-based stains remain in common usage. These are described in this chapter together with some preparations that are still employed, albeit less frequently. Finally, in neuropathology as in most other areas of histological practice, hematoxylin and eosin (H&E) remains the most useful and commonly used stain, as it demonstrates most cell types well and with good detail.
The nervous system can be subdivided into the central, peripheral and autonomic nervous systems. This chapter will concern itself principally with the central nervous system and secondarily with the peripheral nervous systems. The principal components of the central nervous system are:
The neuron is an excitable cell that is responsible for processing and transmitting information. Neurons communicate with each other via intercellular interfaces called synapses. At the synapse, an electrical impulse in the presynaptic neuron causes it to release a chemical transmitter that diffuses across a narrow gap to influence the electrical activity of the postsynaptic target neuron. Neurons have several components (Fig. 17.1):
• The axon, an elongated fibrous process that transmits electrical impulses away from the soma to synapses with either other neurons or muscle fibers. This may be a meter or more in length in the case of the lower motor neurons that reside in the lower (lumbar) spinal cord and innervate muscles of the lower leg.
(Courtesy of Patrick Elliott of the Medical Illustration Department, Royal Hallamshire Hospital, Sheffield, UK.)
Two naturally occurring pigments may be observed to accumulate in the brain with age: lipofuscin and neuromelanin. Both are generally believed to represent cellular waste products. Lipofuscin is a yellow-brown, autoflourescent, granular substance composed of peroxidized protein and lipids. It is seen in larger neurons, such as the lower motor neurons of the spinal cord and pyramidal cells of the hippocampus in the context of Alzheimer pathology. Large amounts of lipofuscin-like pigment accumulate in the context of inherited neuronal ceroid lipofuscinoses, of which Batten’s disease is the most common (Goebel & Wisniewski 2004). Neuromelanin is most commonly seen in the cytoplasm of neurons of the substantia nigra and the locus ceruleus. It is the cause of the macroscopic pigmentation of these structures. Under the microscope, it is a dark brown, granular material that is believed to be the by-product of oxidative metabolism of catecholamines (Sulzer et al. 2008).
• Astrocytes, which have a number of functions. They maintain the extracellular ion and neurotransmitter balance. They are involved in repair and scarring responses to brain damage and also form part of the blood–brain barrier, which protects the brain from harmful blood-borne substances.
Neuropil is a term used to denote the feltwork of neuronal processes in which neuron cell bodies reside. Central nervous system tissue is classically subdivided into gray matter, which contains the majority of neuronal cell bodies and little myelin, and white matter, which is predominantly formed of myelinated axons and few neurons.
The meninges form three layers of protective covering over the brain. The outer layer, beneath the skull, is formed by the dura mater. It is a tough, fibrous membrane. The arachnoid mater is a more delicate, fibrillary covering that lies inside the dura mater and is more closely adherent to the brain surface, but it does not invaginate into the surface infoldings of the brain (or sulci). The pia mater is the most delicate covering. It is closely apposed to the brain surface, following its contours down into the depths of sulci.
Hematoxylin and eosin (H&E) preparations demonstrate most important features of neurons. However, Nissl preparations are also popular for examining the basic architecture of neural tissue and its components. These are often combined with the luxol fast blue myelin stain. Granules of Nissl substance are found in the cell body (Fig. 17.1) and correspond to rough endoplasmic reticulum. They are basophilic due to the associated nucleic acid (Palay & Palade 1955). Many basic dyes (e.g. neutral red, methylene blue, azur, pyronin, thionin, toluidine blue and cresyl fast violet) stain Nissl substance. Variation in the stain used, pH and degree of differentiation allow preparations to label either Nissl substance alone, or Nissl substance in combination with cell nuclei.
Motor neurons generally have very coarse (‘tigroid’) Nissl substance, and regions such as the anterior horns of the spinal cord, where these cells are abundant, are good tissues to use when learning these stains (Fig. 17.2). For paraffin-embedded sections of formalin-fixed tissue, the cresyl fast violet stain is reliable and relatively straightforward. As such it is by far the most commonly used Nissl preparation. Toluidine blue may also be used, whilst Einarson’s gallocyanin method, being more suited to alcohol-fixed tissue, is largely unused (Kellett 1963).
Figure 17.2 Anterior horn cells in spinal cord. Notice their large size and the prominent nucleolus. Paraffin section, stained with toluidine blue. Similar results can be obtained with cresyl fast violet.
Cresyl fast violet (Nissl) stain for paraffin sections
|Cresyl fast violet||0.5 g|
|Distilled water||100 ml|
|Glacial acetic acid||250 μl|
1. Neuronal cytoskeletal proteins. Neurofilaments (NF) are intermediate filaments specifically expressed by mature neurons. They are composed of protein subunits that are classified by molecular weight into NF-L, NF-M, and NF-H which may be variably phosphorylated (Gotow 2000). Antisera raised against different neurofilament proteins in different states of phosphorylation are available. NF-H, in particular, and NF-M, to a lesser extent, are normally unphosphorylated in the neuronal cell body, but become phosphorylated in the axons. Thus, antibodies to phosphorylated NF-H mark axons but not cell bodies in normal nervous system tissues. Antibodies to non-phosphorylated neurofilament will label neuronal somata (Trojanowski et al. 1986). Microtubule-associated protein 2 (MAP-2) is a protein involved with microtubule assembly and is expressed by neurons in dendrites and cell bodies (Maccioni & Cambiazo 1995; Shafit-Zagardo & Kalcheva 1998). It is therefore often used to as a marker of neuroepithelial differentiation (Wharton et al. 2002; Blumcke et al. 2004).
2. Cytoplasmic proteins. PGP9.5 and neuron-specific enolase (NSE) are strongly expressed in neurons and can be reliably labeled by commercially available antisera. Unfortunately, they are not specific for neuronal cells, making interpretation tricky. They are best used in the context of a broad antibody panel (Ghobrial & Ross 1986; Wilson et al. 1988).
3. Neuronal nuclear proteins. NeuN is a neuron-specific DNA binding protein, which starts to be expressed around the time of initiation of terminal differentiation of the neuron (Mullen et al. 1992). Antibodies to NeuN therefore label neuronal nuclei, and neuronal components of other tumors (Edgar & Rosenblum 2008). However, in the context of neuro-oncology, it lacks specificity, being expressed to a variable degree in a diverse range of primary brain tumors. Therefore, NeuN is best used as part of a panel of antibodies in the investigation of clear cell primary brain tumors, but is of limited utility for other tumors (Preusser et al. 2006).
4. Proteins associated with neurosecretory granules. Antisera to these proteins can be useful to establish neuronal and neuroendocrine differentiation (Koperek et al. 2004; Takei et al. 2007). Synaptophysin is a membrane glycoprotein component of presynaptic neurosecretory vesicles. The cell body of normal neurons is usually unstained by synaptophysin (Fig. 17.3), resulting in early claims that cell body labeling was a feature of neoplastic neuronal cells that differentiated them from native neurons (Miller et al. 1990). However, it is now evident that a population of normal neurons also show cell body labeling which detracts from the use of this feature as a diagnostic marker (Quinn 1998). Synaptophysin is a useful marker of neuroendocrine differentiation and so also stains cells in metastatic neuroendocrine tumors (Wiedenmann et al. 1987). Chromogranin A is a protein of the dense core matrix of neurosecretory granules, and antibodies to it can be used to identify cells containing dense core vesicles (Nolan et al. 1985). It is therefore used predominantly to elucidate neuroendocrine differentiation in tumors.
The annals of neurohistology describe several methods to demonstrate various special structures of the neuron, including axons (viable and degenerate), dendrites, synapses, dendritic trees and peripheral nerve endings. Many of these were block impregnation and free-floating frozen section methods and are rarely performed in time-constrained modern diagnostic laboratories. Immunohistochemistry has largely replaced the old silver preparations for the demonstration of axons as it is reliable and produces adequate results for diagnostic neuropathology with ease. However, for formal quantitation of axons, many still find that Palmgren’s method is superior to neurofilament immunohistochemistry (Chance et al. 1999). This technique has classically been used for staining axons of the peripheral nervous system. However, it is also an excellent preparation for staining central nervous system axons as well. Palmgren’s method uses potassium nitrate to suppress staining of reticulin. It is considered that with the Palmgren method, cresyl violet preparations and immunohistochemistry for neurofilament and MAP-2, there is no longer requirement for older silver preparations such as that of Bielschowsky (Bielschowsky 1902) and Marsland (Marsland et al. 1954).
Silver techniques, such as the Palmgren method, require great care and attention to detail such as clean glassware and pure distilled water for a successful outcome. Stock solutions should be well maintained and not more than a few months old (in some cases less than a week).
Modified Palmgren’s method for nerve fibers in paraffin-embedded material (Palmgren 1948)
5. Without rinsing, drain the slide and flood the section with reducer that has been heated to 40–45°C. Rock the slide gently and add fresh reducer. Leave for 1 minutes. A beaker placed on a hot plate is useful for this stage.
6. Wash in three changes of distilled water. Examine microscopically and, if necessary, repeat from step 4, reducing the time in the silver solution and decreasing the temperature of the reducer to 30°C.
The only proviso when using glycine is that it must be made up fresh prior to use; as it is only stable for approximately one week. However the Palmgren silver, once made up, is stable for several weeks.
The silver incubation time may need to be increased for tissues which have had a short formalin fixation time, but as a rule of thumb you should see a slight yellow tinge to the tissues when the optimum time has been reached. The reducer keeps for several months.
The original method stated that it had to stand for 24 hours before use, but this is not the case. The reducer will darken with time, changing from pale yellow to dark amber. It is important that at the reduction step the slides are gently agitated to ensure an even reduction of the tissue; if the sections are not dark enough, they can be rinsed in distilled water and steps 4–6 repeated but with a shorter time in the silver solution.
The hotter the reducer, the faster the reduction will take place and it may well be uneven, leading to suboptimal preparations. Sections can be toned using gold chloride prior to fixing, which is an optional step.
The original method used an intensifying step prior to fixing: this employs aniline. Some have found this to be of little value. The method was originally designed for use with paraffin sections. However, it can be applied to cryostat sections which have been pretreated with 20% chloral hydrate overnight prior to carrying out the Palmgren method.
Examination of axons in peripheral nerve in diagnostic neuropathology now relies largely on toluidine blue-stained semi-thin resin-embedded tissue. Capricious techniques such as Eager’s method for detecting degenerating axons are no longer in use (Eager et al. 1971).
The Golgi preparation and its variants are excellent for the visualization of the three-dimensional nature of the neuron and its dendritic processes. However, modern diagnostic neuropathology practice has no requirement for this. Golgi techniques are occasionally used in research (e.g. Garey 2010) although new antisera are increasingly allowing immunohistochemical indices of these aspects of cell morphology. It is suggested that the interested reader consider the Pugh and Rossi modification for use on paraffin-embedded tissue (Pugh & Rossi 1993) if a Golgi stain is to be attempted.
Myelin forms an electrically insulating sheath around axons. It is approximately 80% lipid and 20% protein and is formed from sheet-like processes of glial cells that are concentrically wrapped multiple times around the axon. This greatly improves the speed and efficiency of impulse conduction along the axon. Myelin is formed by oligodendrocytes in the central nervous system and Schwann cells in the peripheral nervous system. A single oligodendrocyte may myelinate multiple axons in its vicinity, whereas a single Schwann cell may only myelinate a single segment of a single axon. Loss of myelin (as is seen in multiple sclerosis in the central nervous system or Guillain-Barré in the peripheral nervous system) can be severely debilitating.
Modern tinctorial stains for myelin are simple and reliable and can be performed on formalin-fixed paraffin-processed tissue. Many can be combined with a Nissl stain to demonstrate neuronal localization. Older methods may give more even and consistent staining, but are considerably more time consuming and have fallen from use (Weigert 1904; Loyez 1910; Weil 1928). Both luxol fast blue and solochrome cyanine preparations are now favored.
Luxol fast blue is a copper phthalocyanine dye which is employed in myelin staining of paraffin-processed tissue (Kluver & Barrera 1953). This can be combined with cresyl violet or hematoxylin to outline cellular architecture (Fig. 17.4) or with periodic acid-Schiff (PAS) to demonstrate myelin degradation products in demyelinating disease. It can be used on central nervous system tissue only.
Luxol fast blue stain for myelin with cresyl violet counterstain (Kluver & Barrera 1953)
|Luxol fast blue||1 g|
|Methanol (absolute)||1000 ml|
|10% acetic acid||5 ml|
5. Differentiate in 0.1% lithium carbonate solution until the gray and white matter are distinguished. This may be more easily controlled by using 0.05% lithium carbonate followed by 70–95% alcohol instead.
9. Drain sections and transfer to 70% alcohol. Avoid placing the section in water at this stage as the cresyl violet staining loses some of its intensity. Gently agitate the sections; the cresyl violet dye will flood out. The 70% alcohol differentiates the cresyl violet stain. Optimally, the cresyl violet should be removed, leaving the cell bodies and Nissl clearly visible. Do not over-differentiate; the 70% alcohol will take out the cresyl violet and to a certain extent the luxol fast blue. The cresyl violet counterstain will deepen the color of the luxol fast blue stained myelin from turquoise to a deep blue.
a. If the section is over-differentiated with lithium carbonate/alcohol, the section can be restained with the luxol fast blue and then differentiated to obtain the optimum staining result. This may apply to tissues which have very low amounts of myelin (e.g. baby/neonatal brains). These tissues can be very challenging in achieving optimum staining. Unfortunately, once the cresyl violet counterstain has been applied the over-differentiation cannot be rectified.
Page’s solochrome cyanine technique for myelin in paraffin sections
|Solochrome cyanine RS||0.2 g|
|Distilled water||96 ml|
|10% iron alum||4 ml|
|Concentrated sulfuric acid||0.5 ml|
Immunohistochemistry for S-100 is useful in the diagnosis of tumors derived from Schwann cells both in the central nervous system and peripherally (Hirose et al. 1986; Winek et al. 1989). Many antisera are used as markers of myelination, the most useful being myelin basic protein and myelin associated glycoprotein (Itoyama et al. 1980a, 1980b; Ludwin & Sternberger 1984; Lindner et al. 2008). However, given the reliability and simplicity of the tinctorial stains, immunohistochemical myelin makers are largely unused in routine diagnostic neuropathology and remain the preserve of research laboratories.
Myelin loss may occur in a region of brain damaged by any of a number of processes such as trauma, neoplasia, multiple sclerosis or toxic insult. It may also occur secondary to the loss of axons emanating from a lesioned brain region. In modern practice, degeneration of myelinated tracts is most commonly demonstrated by showing loss of normal myelin staining by either luxol fast blue or solochrome cyanine preparations, or by showing a microglial reaction using CD68 immunohistochemistry (e.g., Ince et al. 2008). Historically, the Marchi technique (Swank & Davenport 1935) and neutral lipid stains have been used to detect early and late myelin degeneration products, respectively. However, the Marchi technique requires block staining or free-floating sections and lipid stains cannot be performed on paraffin-embedded material. Given these issues, the sensitivity of CD68 immunohistochemistry and the ease of the tinctorial preparations for normal myelin (see above), the Marchi and neutral lipid techniques have been rendered obsolete.
The term neuroglia refers to the supporting cells of the central nervous system and comprises ependymal cells, astrocytes, oligodendrocytes, and microglia. As is becoming a recurrent theme, immunohistochemistry is increasingly replacing tinctorial stains for their identification.
Ependymal cells are epithelioid and line the ventricles of the brain and the central canal of the spinal cord. They are easily located with conventional stains such as H&E and immunohistochemistry for GFAP, vimentin and S-100. Immunohistochemistry for epithelial membrane antigen (Uematsu et al. 1989; Hasselblatt & Paulus 2003) labels both normal and neoplastic ependymal cells, whilst cytokeratin markers are negative.
Astrocytes have multiple, fine processes and (in their reactive state) are ‘star-shaped’ (hence the name). On standard H&E sections, only the nucleus of resting astrocytes is distinct, as the cell body cannot be discerned from background neuropil. These nuclei are slightly larger with more open granular chromatin than that of the more compact oligodendrocyte. Modern neuropathology relies most heavily on GFAP (Fig. 17.5) immunohistochemistry for demonstration of astrocytes, although antibodies to S-100, αB-crystallin and glutamine synthetase may also be used. Metal-based and tinctorial methods such as Cajal’s gold sublimate, PTAH (Chan & Lowe 2002) and Holzer (1921) are no longer in use as these are variously more expensive, more technically demanding or less specific than their immunohistochemical equivalents.
Figure 17.5 Reactive astrocytes in white matter, stained by anti-GFAP immunohistochemistry technique with hematoxylin nuclear counterstain. Fine GFAP-containing processes form a felt-like mat in which the stellate cell bodies are evident.
Astrocytes are principally classified into protoplasmic and fibrous forms. These are similar in function. However, whereas protoplasmic astrocytes have shorter, thicker, highly branched processes and are generally found in the gray matter, fibrous astrocytes have longer, thinner, less-branched processes and usually reside in white matter. Astrocytic reactions in the cerebellum are characterized by Bergmann or radial astroglia which have processes that run radially from the Purkinje cell layer of the cortex to the pial surface.
In response to injury of the brain parenchyma, astrocytes react by increasing in size with a more prominent, eosinophilic cytoplasm. The nucleus moves from a central to a more eccentric position within the cell cytoplasm and processes become more prominent. Astrocytic gliosis is a response to permanent injury, whereby astrocytes proliferate to fill tissue defects with a fibrous glial scar.
In neuro-oncology, astrocytic differentiation is best demonstrated by GFAP immunohistochemistry. GFAP immunoreactivity is also seen in other tumors including ependymoma, oligodendroglial tumors and choroid plexus tumors (Eng & Rubinstein 1978; Velasco et al. 1980; Eng 1983; Doglioni et al. 1987). Astrocytic tumors also label with vimentin and S-100, but these are also seen in many other tumor types, rendering them of little use for differential diagnosis. Astrocytes occasionally show cross-reactivity as seen for the pan cytokeratin AE1/AE3 (Cosgrove et al. 1989). Therefore, it is expedient to use other cytokeratin markers such as CAM5.2 or MNF116 to exclude the diagnosis of epithelial cell tumors, such as metastatic adenocarcinoma.
The proliferation marker Ki-67 (MIB-1) is often used in the assessment of surgical neuropathology specimens as an aid to grading tumors and to help differentiate reactive from neoplastic astrocytic populations. The latter will tend to have a higher number of nuclei labeled with this marker.
An emerging and potentially more powerful tool for the diagnosis of diffuse oligodendroglial and astrocytic neoplasms is the use of antibodies to isocitrate dehydrogenase 1 (IDH1) carrying the R132H mutation. This is the most frequent mutation in diffuse gliomas (Hartmann et al. 2009). There is an emerging literature that appears to demonstrate that antisera to this mutant protein may be used to differentiate reactive gliosis from grade II and III astrocytomas (Camelo-Piragua et al. 2010; Capper et al. 2010) and oligodendrogliomas from lesions with similar morphologic appearances (Capper et al. 2011).