STP Position Paper: Recommended Best Practices for Sampling,Processing,and Analysis of the Peripheral Nervous System (Nerves and Somatic and Autonomic Ganglia) during Nonclinical Toxicity Studies

Basic Philosophy

The working group concluded that a rigid “one-size-fits-all” approach to sampling, processing, and evaluating PNS tissues is inappropriate due to the variety of situations, modes of action (MOAs), molecular-initiating events, and potential target sites that might be encountered. Instead, the working group is of the unanimous opinion that the appropriate and achievable objective is to delineate a strategy for evaluating key PNS structures to differentiate common classes of neurotoxic lesions, but let the experiences and needs of individual institutions ⁴ drive selection of the specific battery of sampling, processing, and analytical methods undertaken to provide a suitable survey of the PNS. The methods used for implementing such institutional decisions should be articulated clearly in the study protocol and/or study report or alternatively documented in advance in an institutional standard operating procedure (SOP) or study-specific procedure. Such institutional decisions should be made using a “weight of evidence” (WOE) approach, where expanded sampling and evaluation of the PNS is considered only when evidence of PNS neurotoxicity is substantial enough to be an important factor in the final risk assessment. In general, WOE decisions incorporate such factors as the degree of PNS neurotoxicity versus toxicity to other target tissues (i.e., how sensitive is the PNS to the test article relative to other systems), the extent of PNS neurotoxicity that develops at relevant levels of exposure, and the variable sensitivities of different species, including humans, to the PNS neurotoxic effects of a given test article or compound class.

Scenarios for PNS Neurotoxicity

Four general situations during which PNS tissues may be sampled in the course of toxicity studies were considered (Table 1). For each, the recommended sampling strategy varies slightly based on the different locations in which the PNS is affected and the types of studies that are conducted. The first three situations involve general (Tier I or “screening”) toxicity studies, while the last scenario relates specifically to dedicated neurotoxicity (Tier II or “advanced”) assessments.

Situation 1 is a general toxicity study in which (1) no potential for PNS neurotoxicity was detected in data obtained during prior studies (in vivo, in vitro, and/or in silico) and (2) no in-life behavioral or neurological deficits are seen in the current study. This strategy represents a rational default approach when analyzing new molecular entities for which no or few prior in vivo toxicity studies have been done. Situation 2 is a general toxicity study in which in-life signs of peripheral neuropathy or other data reflect possible damage mainly to the somatic (motor and/or sensory) nerves and/or their associated ganglia. Situation 3 represents a general toxicity study in which in-life signs of peripheral neuropathy or other data suggest injury to autonomic nerves and/or ganglia, which collectively regulate involuntary, visceral homeostatic functions. For both situations 2 and 3, other data that might trigger the recommended expansion of the PNS analysis include known or presumed MOA and quantitative structure–activity relationship (QSARs) models for the test article, its metabolites, and/or related molecules. Situation 4 is the dedicated neurotoxicity study, which usually is required for test articles in which human epidemiological findings, experimental data from animal studies (in vivo and/or in vitro), and/or MOA or QSAR similarities to known neurotoxic agents indicate a high probability that PNS neurotoxicity may occur under likely exposure scenarios. Some agents may simultaneously impact the somatic and autonomic PNS and thus may necessitate increased sampling (combining situations 2 and 3) and evaluation to fully assess both arms of the PNS.

A side-by-side comparison of PNS specimens to collect and evaluate and baseline tissue sampling and processing recommendations for the four situations are given in Tables 2 and 3, respectively. The working group recommends that this information be used to define institutional SOPs or study-specific procedures that describe collection and processing for PNS tissues. These documents should be detailed but sufficiently flexible, so that the study director may adjust the study protocol as necessary to address the issues encountered during a particular situation.

The four situations noted above typically arise in nonclinical toxicity studies for adult animals, where PNS effects related to test article exposure occur with some frequency. No impediment exists with respect to applying these best practice recommendations to the PNS of developing animals. However, the working group notes that current practice in developmental neurotoxicity studies typically handles PNS sampling, processing, and analysis using the situation 1 recommendations for both the early (juvenile) and late (adult) time points. If other evidence suggests that a more wide-ranging examination is required to fully characterize the PNS findings, the more extensive PNS neuropathology examination typically is concentrated on the adult time point (Garman et al. 2016).

Best Practice Recommendations for All Four Situations

The PNS sampling strategy should be guided by observed in-life neurological signs or other information for the compound/target. The choice of which PNS samples to collect and whether or not special histology processing and/or investigative techniques need to be applied should be defined at the discretion of the institution for a given toxicity study using a WOE approach. For all situations, PNS structures (nerves, ganglia, and effector organs) typically should be collected bilaterally but may be processed and evaluated unilaterally. This strategy is acceptable since toxic lesions in the PNS are generally bilateral whereas background alterations secondary to nerve trauma (induced either in vivo or post mortem) usually are unilateral; collection of the samples bilaterally also allows the contralateral structure to be utilized to verify questionable findings (e.g., where test article–related neurotoxicity must be discriminated from artifacts related to handling or other PNS testing procedures) and as a substrate for additional special methods that might be needed to better characterize lesions (immunohistochemistry [IHC], teased nerve fiber preparations, etc.). Nerves and skeletal muscle (an effector organ) should be evaluated in both cross and longitudinal orientations. All PNS specimens from the treatment groups selected for initial evaluation (generally the high-dose and control animals) should be processed in the same time frame using methods to avoid systematic variation in processing conditions (“block” effects).

Where plastic embedding is undertaken, hard plastic resin is the recommended medium. Soft plastic (e.g., glycol methacrylate [GMA] or methyl methacrylate [MMA]) embedding offers no substantive improvement in resolution over paraffin embedding with respect to preserving myelin architecture, so soft plastic is not an acceptable substitute for hard plastic resin.

The recommended best practice for light microscopic evaluation is to undertake a tiered, semiquantitative analysis with foreknowledge of the study design. A subsequent masked (“blinded” or “coded”) analysis of PNS tissues with findings of concern may be conducted at the discretion of the study pathologist (or peer review pathologist) but usually is done only to aid in defining the dose response and/or establishing a “no observed adverse effect level” (NOAEL).

Best Practice Recommendations for Situation 1

For general toxicity studies with no known concern for PNS neurotoxicity (situation 1), the majority of the working group concurs that one large, mixed (i.e., sensorimotor) somatic nerve such as the sciatic nerve (or tibial nerve if the sciatic trunk has been traumatized), the autonomic ganglia within the walls of protocol-specified organs, the spinal cord (white matter tracts of the cervical and lumbar divisions), and one effector organ (e.g., the medial head of the gastrocnemius muscle) represent a suitable baseline PNS survey. At least one dorsal root ganglion (DRG ⁵ ) from the lumbar division either isolated or (in rodents only) processed in situ in the vertebral column segments is recommended for collection at necropsy and retention but need not be assessed unless nerve or spinal cord lesions require additional characterization. Standard processing—immersion fixation in conventional (i.e., methanol-containing) neutral buffered 10% formalin (NBF), paraffin embedding, and hematoxylin and eosin (H&E) staining—is generally acceptable. ⁶ For this situation, the decisions regarding whether or not to obtain, process, and evaluate additional samples (e.g., DRG and ganglia) as well as undertake special procedures (discussed in more detail below) to further investigate possible neurotoxic changes in the PNS are left to the discretion of the institution.

Best Practice Recommendations for Situation 2

For general toxicity studies where somatic PNS neurotoxicity is a concern under likely exposure scenarios (situation 2), three spinal nerves are evaluated. Typically, the sciatic nerve and two or more of the following nerves (most of which are distal branches of the sciatic nerve) are sampled: tibial, fibular (i.e., common peroneal), plantar, saphenous, sural, or (in rodents) caudal nerves. The sciatic, tibial, and fibular nerves in all species and the sural and caudal nerves in rodents are mixed sensorimotor structures; the saphenous, plantar, and (in nonhuman primate) sural nerves are sensory-predominant branches (sometimes referred to as “sensory-only” nerves). Nerve selection generally should be based on in-life findings. Institutions may consider including sensory-predominant nerves to facilitate data translation among species since these distal branches can be biopsied more easily for clinical monitoring in humans. The spinal cord (at minimum the cervical and lumbar divisions), at least four DRG (two or more associated with both the cervical and lumbar intumescences, which supply nerve fibers to the brachial and lumbosacral plexuses, respectively), the associated spinal nerve roots, and one trigeminal (cranial nerve [CN] V [Gasserian]) ganglion ⁷ with associated nerve fibers are recommended for collection and evaluation.

Conventional processing conditions (immersion fixation in buffered formalin, paraffin embedding, and H&E staining) are suitable for PNS tissues, with three exceptions. First, methanol-free formaldehyde (MFF ⁸ ) or medical-grade glutaraldehyde (MGG, typically 2.5%) rather than NBF ideally should be employed to minimize processing artifacts. Generally, MFF is used for routine histologic preparations, while MGG is employed for samples in which postfixation with osmium is used to highlight myelin organization. The working group recognizes that this first adjustment may not be feasible on short notice, especially if the in-life PNS-related signs develop late in the course of a large general toxicity study. Second, if further characterization of PNS changes observed in H&E-stained sections is necessary, acquisition of serial sections for at least one mixed nerve may be considered for special neurohistological staining to highlight axonal morphology (silver stain) and explore myelin integrity (myelin stain). Third, at least one nerve cross section (usually a mixed-function distal trunk like the tibial or fibular nerve or a sensory-predominant branch like the sural or plantar nerve) should be postfixed by immersion in MGG followed by osmium, processed into hard plastic resin, and then stained with toluidine blue for light microscopic evaluation; this specialized technique better stabilizes myelin during processing than does fixation with MFF or NBF alone. The last two adjustments should be feasible regardless of whether MFF or NBF is utilized as the initial fixative at necropsy.

Best Practice Recommendations for Situation 3

For general toxicity studies where autonomic PNS neurotoxicity is a concern at relevant levels of exposure (situation 3), elements of the parasympathetic, sympathetic, and enteric ⁹ PNS should be evaluated, including multiple autonomic ganglia and, at the discretion of the institution, selected autonomic nerves (e.g., vagus for parasympathetic division and sympathetic chain for sympathetic division). Common postganglionic parasympathetic ganglia that may be assessed are found in the walls of protocol-specified hollow organs, most commonly the intestines (i.e., the submucosal [Meissner’s] and myenteric [Auerbach’s] ganglia) and urinary bladder, but ideally at sites related to in-life findings. Typical sympathetic sites to consider include the cranial cervical, cervicothoracic, celiac, cranial mesenteric, and/or sympathetic chain ganglia and their attached nerve roots, as well as the sinoatrial and atrioventricular nodes of the heart; in practice, working group members have found that the sympathetic structures that can be isolated most reliably by trained technicians are the cranial cervical ganglia in small animals and cervicothoracic ganglia in larger species, while the next easiest site in all species is the cranial mesenteric ganglion. In addition to autonomic PNS nerves and ganglia, somatic PNS nerves and ganglia should be collected as described in situation 2, and the sciatic nerve and at least four DRG (two each for the cervical and lumbar divisions) should be examined as a baseline somatic PNS screen. All spinal cord divisions (cervical, thoracic, and lumbar) should be collected and evaluated. Routine processing (immersion fixation in NBF or ideally MFF, paraffin embedding, and H&E staining) is suitable for most autonomic PNS samples. Postfixation with MGG and osmium followed by hard plastic embedding may be useful for small-caliber autonomic trunks despite the lower degree of myelination in most autonomic nerves.

Best Practice Recommendations for Situation 4

For dedicated neurotoxicity studies where PNS neurotoxicity is likely or certain (situation 4), expanded sampling and evaluation includes at least three spinal nerves (sciatic, tibial, and fibular, saphenous, sural, plantar, or caudal), white matter tracts of all three spinal cord divisions, multiple DRG and their associated spinal nerve roots, and a trigeminal ganglion. At least six DRG should be examined (two or more DRG for each of the cervical, thoracic, and lumbar spinal cord divisions). In general, DRG should be removed from the vertebral column rather than processed and evaluated in situ to avoid soft tissue degradation associated with skeletal decalcification, but in small animals like rodents, in situ analysis following vertebral column decalcification is acceptable. Fixation is undertaken by whole-body perfusion with a methanol-free fixative (typically MFF or mixtures of MFF and MGG). Paraffin embedding is suitable for most nerves and ganglia, although at least two distal nerve cross sections (typically the tibial nerve and a more distal branch) should be postfixed in MGG and osmium and then embedded in hard plastic resin. Paraffin-embedded nerves should be stained with H&E and, if warranted, axonal and myelin stains, while plastic-embedded nerves are stained with toluidine blue. Ganglia usually are stained only with H&E, although silver and myelin stains may be beneficial. Other special neurohistological methods (see below) may be considered at the discretion of the institution to better characterize any neurotoxic lesions.

Adjustments to Recommended PNS Sampling, Processing, and Evaluation Practices

The working group recommendations for PNS sampling in situations 1–4 are designed to be applicable to cases where test articles have been delivered systemically (i.e., where all PNS tissues are liable to some degree of test article exposure) and thus may need to be modified for selected scenarios and/or unusual test articles. For example, even in situation 4 (where neurotoxicity is expected), less extensive PNS evaluation may be permissible if the pattern and severity of PNS lesions for the doses and/or the dosing regimen used in a study have been well defined in one or more previous studies; however, the working group recommends that all PNS tissues described in situation 4 still be collected and archived as wet tissue. Retained tissues that were fixed in MFF should be held in MFF until the PNS neuropathology is complete to avoid the potential for methanol-associated myelin artifacts, after which PNS samples may be transferred per institutional SOP to NBF for long-term archiving. Similarly, additional PNS samples (e.g., forelimb nerves) may have to be evaluated if clinical signs suggest that PNS damage has occurred at these sites. Finally, local delivery of a minimally diffusible test article ¹⁰ generally warrants increased collection and prioritized analysis of nerves near the administration site, while more distal PNS elements may be collected but retained as wet tissue without processing. Such modifications in sampling and evaluation may be made at the discretion of the institution. The rationale for such adjustments typically should be given in the study report.

Situation-specific PNS Sampling Strategies

Situation 1

In general toxicity studies where no neurotoxic potential is known or anticipated (situation 1), the minimal list of PNS tissues to be evaluated in all species is a readily accessible, large, spinal-origin somatic nerve, the autonomic ganglia within the walls of major viscera, the cervical and lumbar divisions of the spinal cord, and a skeletal muscle (i.e., richly innervated effector organ). This PNS sampling strategy is identical to that proposed in the STP best practices document for CNS sampling in nonclinical general toxicity studies (Bolon, Garman, et al. 2013) and reflects the current practice for general toxicity studies.

Nerves

The usual PNS sample for situation 1 is sciatic nerve. The rationale for selecting this nerve is that it contains both sensory and motor nerve fibers, which permits analysis of major peripheral sensorimotor structures in a single sample. The sciatic nerve is exposed by reflecting and/or removing the overlying skeletal muscle (Figure 1). Sciatic nerve samples commonly are acquired at a distal location (i.e., just proximal to where the tibial and fibular nerves branch, which occurs near the femorotibial joint [knee]), although collection at a more proximal site (i.e., midway between the vertebral column and knee) is a frequent alternative. The choice of sciatic nerve site to be sampled (proximal vs. distal) is left to the discretion of the institution.

OPEN IN VIEWER

Figure 1.

Locations for harvesting somatic nerves from the rodent hind limb. Nerves: a = sciatic; b = tibial; c = common peroneal (fibular); d = lateral sural; e = plantar, f = saphenous. Muscles: 1 = gluteus medius; 2 = biceps femoris; 3 = semitendinosus; 4 = quadriceps femoris; 5 = gastrocnemius lateralis; 6 = tibialis cranialis; 7 = rectus femoris; 8 = gastrocnemius medialis. Bones: F = femur (condyles); T = tibia. (Schematic diaphragm rendered by Mr. Tim Vojt, with confirmation of anatomic structures by Ms. Kate Anderson.)

A sciatic nerve branch, typically the tibial nerve (another trunk carrying both sensory and motor nerve fibers; Figure 1), may be evaluated instead of the sciatic nerve if likely artifactual changes might confound sciatic nerve analysis. A common scenario in which this substitution may be warranted is in nonhuman primates that have received intramuscular injections of ketamine in the region where the sciatic nerve is routinely collected. Chemical and mechanical trauma associated with such injections has been demonstrated to damage the nearby sciatic nerve trunk (Carrier and Donnelly 2014).

While sciatic nerve (or tibial nerve) commonly is the only PNS structure evaluated for situation 1, additional spinal-origin somatic nerves may be collected at necropsy at the discretion of the institution. Retaining other nerves in the archived wet tissues may prevent the need to repeat studies in the event that changes observed in the sciatic nerve necessitate evaluation of other portions of the PNS. A simple means for accomplishing this task in small species such as rodents is to retain an entire hind limb (after removing the skin) and the proximal tail (which contains the caudal nerves). In larger species, the distal nerve trunks should be removed at necropsy, although some institutions may prefer to save the distal limb (e.g., below the knee, with skin removed). Other nerves to consider for collection are listed below (under situations 2–4 and in Table 2). The choice of which additional nerves to harvest, if any, also should remain the decision of the institution.

Ganglia

A majority of working group members, with a few dissenters, recommend that DRG need not be evaluated routinely for situation 1 since no PNS neurotoxicity is known or anticipated. The working group does endorse collection and archiving of at least one lumbar DRG against the possibility that an explanation might need to be sought for lesions observed in the sciatic nerve. The rationale for this recommendation is that DRG, as well as the nerves they serve, lack effective neurovascular barriers (Olsson 1990; Abram et al. 2006; Sapunar et al. 2012) and thus may be exposed to test articles that are excluded from the CNS by the blood–brain barrier. Usually, the chosen DRG are associated with the spinal cord segments from which the sampled spinal nerve arises (i.e., the lumbar intumescence for the sciatic nerve and its branches; Table 4). A fast and simple means for retaining the DRG (and their associated spinal nerve roots) in the wet tissues is to harvest an extended portion (rodents) or region-specific segments (all species) of the vertebral column (after removing the overlying musculature and skin). The DRG may be processed and evaluated as isolated ganglia (all species) or in situ in decalcified vertebral column sections (rodents and other small species). Autonomic PNS ganglia to be assessed in situation 1 are limited to the parasympathetic ganglia already present within protocol-specified hollow viscera (e.g., intestines and urinary bladder). Specific sampling of additional autonomic ganglia is not needed.

OPEN IN VIEWER

Table 4.

Expected Spinal Cord Origins of Principal Forelimb and Hind Limb Mixed Nerve Trunks for Common Vertebrate Species.

Species	Limb	Spinal Cord Segmental Originsa of Major Nerves	Positionsa of Spinal Cord Segments for Major Nerves Inside Vertebral Column	Reference
Chicken	Fore	C13–C15	C12–C15	Jungherr 1969; Baumel 1975
	Hind	Syn3–Syn5 (L3–L5)	Syn2–Syn4 (L2–L4)	Jungherr 1969; Baumel 1975
Mouse	Fore	C6–C8	C5–T1	Kaufman and Bard 1999; Sengul and Watson 2012
	Hind	L3–L4 (±L5)b	L1–L4b	Rigaud et al. 2008; Sengul and Watson 2012
Rat	Fore	C6–C8	C5–C6	Greene 1935
	Hind	L4–L5b	L1–L2	Bolon et al. 2006; Rigaud et al. 2008
Dog	Fore	C6–C7	C5–T1	Ghoshal 1975a; Sharp et al. 1990; Fletcher 1993
	Hind	L4–S1	L3–L4	Ghoshal 1975a; Bailey et al. 1988; Fletcher, 1993
Pig	Fore	C7–T1 (±C6)	C6–C7	Ghoshal 1975b; Dellman and McClure 1975
	Hind	L6–L7 (±L5)	L5–L6	Ghoshal 1975b; Dellman and McClure 1975
Primate	Fore	C6–C8b	C5–C8b	Turnquist and Minugh-Purvis 2012; Watson et al. 2015
	Hind	L4–L5	L1–L2b	Turnquist and Minugh-Purvis 2012; Bolon, Garman et al. 2013

Note: C = cervical; L = lumbar; S = sacral; Syn = synsacral (representing fused lumbar and sacral vertebrae); T = thoracic.

^a Positions correspond to the locations of spinal cord motor neurons that supply the nerve roots which are carried in the main mixed somatic spinal nerve trunks that serve the proximal limbs (musculocutaneous and sciatic nerves, respectively, for forelimb and hind limb). Entries represent extrapolated “average” values if multiple sources are cited. For practical purposes, the major contributions to these two nerves are carried in the spinal nerve roots of largest diameter arising from the designated range of cervical and lumbar segments.

^b Denotes that the nerve fiber origins for somatic nerves as well as the positions of the relevant spinal cord segments within the vertebral canal vary with the strain (rodents) or species (nonhuman primates).

Effector organs

In situation 1, skeletal muscle typically is examined as a protocol-specified tissue. Reductions in myofiber diameter may serve as indirect evidence of PNS damage due to nerve fiber (i.e., motor axon) degeneration if direct evidence of myopathic injury is not seen. Although tongue is a common choice for histologic evaluation of skeletal muscle (as a means of assessing many myofibers in several orientations in a single section), other skeletal muscle groups can be collected along with their innervating nerves. Muscles commonly selected for sampling are composed mainly of type I (“slow twitch,” fatigue-resistant) fibers (e.g., diaphragm and soleus) and/or type II (“fast twitch,” glycolytic) fibers (e.g., biceps femoris, quadriceps femoris, and gastrocnemius; Schiaffino and Reggiani 2011; Gropp 2017). Some investigators substitute biceps brachii (if the forelimb appears to be affected as a specimen comprised mainly of type II myofibers).

The working group recommends the gastrocnemius as the default effector organ sample since it has a mixed (but mainly type II fiber) composition (Armstrong and Phelps 1984), is a common site of neurogenic atrophy in both humans (Spencer and Schaumburg 1977) and animals with peripheral neuropathy, and the size of the muscle can be assessed qualitatively during life by palpation. The medial head of the gastrocnemius is larger, making it a logical choice for sampling this muscle. The biceps femoris is a suitable alternative sample as it is also a common location for detecting neurogenic atrophy. The exact choice of muscle(s) to sample and evaluate should be left to the discretion of the institution.

Situation 2

In general toxicity studies where in-life clinical signs or other data (e.g., MOA and QSAR similar to known PNS toxicants) suggest the potential for somatic (sensorimotor) PNS effects (situation 2), the number of PNS specimens subjected to light microscopic analysis should be expanded. Specific neurological evidence warranting additional sampling of the somatic PNS includes local or generalized signs of paresis, paralysis, proprioceptive defects, or muscle atrophy (Table 1). Nonspecific clinical observations related to possible somatic nervous system dysfunction (e.g., abnormal movement, circling, difficulty walking, lameness of unknown origin, and generalized skeletal muscle weakness) may also trigger collection of additional PNS samples, at the discretion of the institution. The clinical signs listed here as possible evidence for PNS neurotoxicity are illustrative, not inclusive, and generally would be considered to warrant a more detailed PNS neuropathology evaluation only if present as chronic and slowly reversible or irreversible adverse effects that clearly cannot be attributed to other systemic toxicities (anemia, dehydration, emaciation, emesis, neurotoxicity related to CNS targets, etc.).

Nerves

Multiple mixed (sensorimotor) spinal nerves are sampled bilaterally during the initial tissue analysis (Figure 1; Spencer and Schaumburg 1977). As noted in situation 1, sciatic nerve samples usually are collected at a distal location (i.e., near the femorotibial joint) due to the increased sensitivity of distal sensory nerve fibers to neurotoxic agents (Martyn and Hughes 1997; Azhary et al. 2010). More proximal sampling may be elected by some institutions for characterizing myelin changes since the proximal sciatic nerve is populated by larger Schwann cells, covering longer axonal lengths, such that the loss of individual cells is more obvious relative to depletion of the smaller Schwann cells in more distal nerves (Friede and Bischhausen 1982; Krinke 2011). The choice of sciatic nerve site to be sampled (proximal or distal, or both) is left to the discretion of the institution.

In addition to the sciatic nerve, the choice of other nerves to collect may be dictated by the spectrum of neurological signs observed in-life or may conform to a predefined battery specified in an institutional SOP. Typically, distal nerve branches are preferred for evaluation since they usually contain a high proportion of sensory axons, and clinical cases of peripheral neuropathy often present as altered sensation (Martyn and Hughes 1997; Azhary et al. 2010). Furthermore, hind limb nerves rather than forelimb nerves usually are sampled in toxicity testing because the longer nerve fibers that serve the hind limb usually are affected first during peripheral neuropathies (Krinke 2011). That said, forelimb nerve branches also should be harvested if the in-life neurological signs suggest that forelimb function has been affected. Evaluation of nerves near the administration site may be prioritized in instances where a locally delivered test article has limited systemic bioavailability.

At least three spinal-origin nerves (usually sciatic nerve and two of its branches) are recommended for evaluation, but the decision regarding which nerves to assess should be guided by the clinical signs and left to the discretion of the institution. The tibial (all species), fibular (all species), and/or sural (rodents; Peyronnard et al. 1986) nerves are common choices as they are mixed sensorimotor tributaries of the sciatic nerve. In rodents, the caudal nerve (a mixed nerve that extends the entire length of the tail) may also be considered for evaluation as electrophysiological testing (e.g., nerve conduction velocity) combined with light microscopic examination of this nerve affords an opportunity to correlate structural and functional findings related to PNS neurotoxicity (Schaumburg et al. 2010). Some working group members have found that aldehyde fixation of the proximal to middle tail (via intravascular perfusion or immersion) allows for later harvest and analysis of caudal nerve. In general, nerves are evaluated unilaterally (in which case nerves that are to be examined for a given animal typically are harvested from the same side), but bilateral evaluation may be considered at the discretion of the institution or if necessitated when iatrogenic nerve damage is likely due to in-life trauma (e.g., intramuscular injection sites).

Collection of dedicated sensory-predominant or motor-only nerves generally is not necessary for safety assessment since the approach to microscopic evaluation is similar for both mixed- and single-modality nerves. However, if observed clinical signs are indicative of a sensory neuropathy (which is the most common presentation of peripheral polyneuropathy in humans and animals), the working group recommends that at least one PNS specimen be a sensory-predominant nerve. Readily accessible sites include the plantar (usually the lateral branch in dogs [Ghoshal 1975a], but the medial branch in rodents [Sant’Anna et al. 2016] and pigs [Ghoshal 1975b]), saphenous (dogs: Braund et al. 1980, and rodents: LaMotte, Kapadia, and Shapiro 1991), and sural (primates [including humans]; Butt et al. 2014) nerves. Evaluation typically may be undertaken using H&E supplemented if necessary by special neurohistological procedures (see below). However, effective examination of unmyelinated (Group C) sensory nerve fibers may necessitate transmission electron microscopy (TEM) because these elements have diameters less than 1 µm, which is near the effective resolution of conventional bright-field microscopes. The only motor-specific nerves in all species are the ventral spinal nerve roots, which may be assessed individually or in sections that also include the sensory-only dorsal spinal nerve root and its associated DRG. For this purpose, serial DRG sections may be necessary to ensure that the desired nerve root is examined as their morphologic features are identical. The choice regarding whether or not to sample sensory-predominant and/or motor-only nerves should be left to the institution.

CNs other than CN V typically are sampled and analyzed only if in-life neurological signs suggest that their function has been compromised (reviewed in Bolon and O’Brien 2011) or if they can be seen within other protocol-specified tissues. For instance, cross sections of CN I (olfactory nerve, an unmyelinated sensory nerve) and major branches of CN V (trigeminal nerve, a mixed somatic nerve) may be readily observed in situ in standard nasal cross sections evaluated in inhalation toxicity studies (done most commonly for rodents; Young 1981). The working group agrees that the decision regarding whether or not to include CNs in the PNS neuropathology analysis should be left to the discretion of the institution. The optic nerve (or CN II), while routinely included in the list of protocol-specified tissues for GLP-type general toxicity studies, develops as an evagination arising from the forebrain and is myelinated by oligodendrocytes and not Schwann cells (Butt et al. 2004; Garman 2011b) and so is not a part of the PNS. ¹¹

Ganglia

If evidence of a somatic peripheral neuropathy is observed, at least two DRG should be evaluated for both the cervical and lumbar divisions of the spinal cord (i.e., at least four total DRG). The DRG need not be paired (i.e., represent bilateral samples from the same spinal cord segment). The inclusion of cervical DRG is recommended as evaluation of these samples represents the sole assessment of the forelimb PNS in standard nonclinical toxicity studies. The best practice for GLP-type nonclinical studies is to remove DRG from the vertebral column (Figure 2) when the spinal cord is extracted (Bolon, Garman, et al. 2013) to preclude the induction of handling artifacts associated with vertebral decalcification needed for in situ examination. However, an acceptable practice in rodents and other small species is to assess DRG in situ (Bolon, Garman, et al. 2013) to avoid trauma produced during their removal. Because soft tissue gathered when seeking DRG sometimes represents connective tissue or fat, more than two DRG should be harvested to ensure that at least two DRG from each specified spinal cord division actually are available for histologic evaluation. Even more ganglia may need to be collected and examined when the test article is delivered directly nearby (e.g., epidural or intrathecal injection) or when clinical signs suggest that nerves arising from a particular spinal cord segment or segments have been affected. The DRG typically are chosen from those associated with the origins of the brachial plexus (i.e., origin of the brachial nerve) and lumbosacral plexus (i.e., origin of the sciatic nerve) because axons emanating from these ganglia are some of the longest (and thus among the most susceptible) in the body. The locations of DRG serving the brachial and sciatic nerves vary by species and sometimes strain (Table 4).

OPEN IN VIEWER

Figure 2.

Diagrams show how to approach and localize dorsal root ganglia (DRG) associated with the origin of the sciatic nerve. Left panel: note that caudal segments of lumbar spinal cord (L4–L7) are displaced cranially relative to the DRG and vertebrae of the same number. Regional anatomy is based on the dog vertebral column. Right panel: DRG are best approached via removing the vertebral arches (at the location of the dotted lines at 2 o’clock and 10 o’clock) using bone-cutting rongeurs. (Images crafted by Mr. Tim Vojt.)

In addition to DRG, the trigeminal ganglion (i.e., the sensory ganglion of CN V) should be collected for evaluation. Ganglia of the autonomic PNS are assessed when seen in situ within other protocol-specified organs (e.g., intramural parasympathetic ganglia in the intestines and urinary bladder). Similar to situation 1, additional autonomic ganglia need not be sampled specifically for this scenario.

Effector organs

Skeletal muscle from sites other than the tongue should be examined from two or more distinct muscles. The specific sampling location(s) may be left to institutional preference and the parameters of the study design (e.g., muscle near sites of locally delivered test articles should also be sampled). The working group recommends that gastrocnemius serve as the default choice for one of the two muscle specimens.

Muscle weights acquired at necropsy may provide an indirect but quantitative means of discriminating peripheral neuropathic effects but should be employed at the discretion of the institution. Weights typically are acquired from isolated biceps brachii, biceps femoris, gastrocnemius, and/or quadriceps femoris, which can be easily identified and collected in a consistent fashion (Greene 1935; Vleggeert-Lankamp 2007; Magette 2012). When a muscle weight will be included as an end point, the working group recommends the gastrocnemius for weighing since peripheral neuropathies usually occur first in portions of longer axons (which in the hind limb are most distant from their supporting neuronal cell bodies; Krinke 2011). Care is required in interpreting the relevance of muscle weights as meaningful conclusions may be impossible in situations where animals were in moribund condition or impacted by other toxicities (e.g., if weights were gathered from samples taken near the site of local test article administration or from long-term recumbent animals) or where historical control data are limited or nonexistent. In such cases, muscle weight measurements have little impact on the risk assessment for PNS toxicity.

Where present within muscle sections, muscle spindles (i.e., sensory end organs) and intramuscular nerves should be assessed. The choice regarding how to record test article–related findings observed in these structures should be left to the institution.

Central (CNS) motor and sensory tracts

An adequate evaluation of somatic PNS neurotoxicity requires that the motor and sensory pathways in the spinal cord also be evaluated. At minimum, this analysis should incorporate examination of longitudinal (or oblique) and transverse sections through the cranial cervical division (particularly C₁, which is near the termini of the cuneate and gracile fascicles, which are the key sensory tracts in the white matter of the dorsal funiculus) and lumbar divisions. The rationale for this recommendation is that many fiber tracts in the spinal cord white matter carry axons from PNS neurons and sensory receptors; because spinal cord has a minimal capacity to regenerate, damage to PNS-derived axons often is more apparent in the spinal cord. Spinal cord evidence indicative of PNS injury is especially prominent in the relatively large dorsal funiculus of the cranial cervical region, which harbors all the sensory axons arising directly from DRG serving the fore and hind limbs. The actual level at which samples are obtained is dictated by the test species (Bolon, Garman, et al. 2013; Watson et al. 2015) and in rodents may be influenced by the strain (Rigaud et al. 2008). The position of the lumbar intumescence within the vertebral canal varies modestly among rodent strains and nonhuman primate species. Therefore, a practical approach to correctly sampling the mouse and monkey lumbar spinal cord segments from which sciatic nerve-targeted axons arise is to harvest the spinal cord that is housed within the relevant lumbar vertebrae (Table 4).

Situation 3

In general toxicity studies where autonomic PNS neurotoxicity is suspected (situation 3), expanded sampling of autonomic PNS structures is necessary. Evidence warranting more extensive autonomic PNS sampling includes nonspecific signs of visceral dysfunction including abnormalities in gastrointestinal motility, heart rhythms, micturition (urinary retention or incontinence), ocular responsiveness (mydriasis and miosis), salivation, or vascular tone (Mathias 2003; Table 1). A WOE approach is especially important in deciding whether or not to engage in expanded sampling and analysis of the autonomic PNS. In general, isolated and/or transient signs of visceral distress (e.g., affecting one or two autonomic functions) usually reflect signs of toxicity to extraneural organs rather than to the autonomic PNS and thus would not serve as a trigger for increased autonomic PNS sampling. Instead, expanded autonomic PNS collection is recommended if a generalized autonomic dysregulation is suggested by multiple anomalous signs originating in the autonomic CNS or PNS, especially if adverse effects persist over a long period and are only slowly reversible or irreversible.

When autonomic PNS neurotoxicity is suspected, care should be taken to properly define the extent to which the histopathologic evaluation of the nervous system should be increased. Sometimes both autonomic divisions (i.e., parasympathetic and sympathetic) may be affected at once, which would warrant careful sampling to obtain representative specimens of both divisions. In addition, autonomic neuropathies may also be accompanied by somatic neuropathies, which may warrant expanded sampling of the somatic PNS (as defined for situation 2 above) as well. As noted above, the final PNS sampling strategy should be driven by the constellation of PNS-related in-life neurological signs.

Nerves

Although the number of autonomic nerves conducive for ready sampling is limited, multiple autonomic nerves may be considered for assessment during the initial analysis for situation 3. Autonomic PNS sampling may include parasympathetic (e.g., CN X [vagus]) and/or sympathetic (e.g., sympathetic chain branches) nerves (Figures 3 and 4). Somatic nerve sampling often mirrors that described above for situation 2 (Figure 1). The decision regarding which autonomic nerves (if any) and somatic nerves to examine remains with the institution.

OPEN IN VIEWER

Figure 3.

Location of key autonomic ganglia in the cervical region. Two cervical ganglia, the cranial (superior) cervical ganglion (C, a part of the sympathetic division) and the caudal vagal (nodose) ganglion (X, a visceral afferent [i.e., sensory] element), may be isolated adjacent to the trachea and thyroid glands in the vicinity of the bifurcation of the carotid artery. In histologic sections, the two ganglia can be distinguished readily because the sympathetic neurons in the cranial cervical ganglion are much smaller than the sensory neurons in the caudal vagal ganglion. Samples: left column = adult rat (provided courtesy of Dr. Magalie Boucher, by permission); right column = adult Beagle dog showing collection of both the cranial cervical ganglion (C) and caudal vagal ganglion (X) in the same histologic section and high-power depictions of these ganglia at the same magnification demonstrating their distinct features. Processing (right image): immersion fixation in neutral buffered 10% formalin, paraffin embedding, sectioning at 4 μm, hematoxylin and eosin staining.

OPEN IN VIEWER

Figure 4.

Location of key autonomic ganglia and nerves in the thoracic and abdominal regions. In the thorax (upper) of a cadaver dog, the major autonomic (sympathetic) structures are the sympathetic trunk (1), rami communicantes (arrows) extending from the spinal nerve roots, cervicothoracic ganglion (2), vagus nerve (3), and vagosympathetic trunk (4). Other labeled structures are the spinal nerve ventral branches that form the brachial plexus (asterisks), esophagus (5), external jugular vein (6), left common carotid artery (7), costocervical trunk (8), left subclavian artery (9), and left cranial lobe of the lung (10). The position of the middle cervical ganglion (another sympathetic structure) beneath the costocervical trunk is outlined by a dashed oval. (Image republished from Image 1 of the “Carnivore Dissection, Lab 12” portal, Clarkson et al. 2015, as informed by content in the “Canine Autonomic Nervous System” portal, Fletcher and Brown 2010, with permission of Dr. Thomas Fletcher.) In the abdomen (lower) of a cadaver cat, the primary autonomic (sympathetic) structures are the celiac ganglion (8) and cranial mesenteric ganglion (10). Other labeled structures include the diaphragm (1), aortic hiatus (2), aorta (3), lumbar artery (4), major splanchnic nerve (5), celiac artery (6), caudal phrenic artery (7), celiac plexus (9), cranial mesenteric plexus (11), cranial mesenteric artery (12), minor splanchnic nerve (13), adrenal gland (14), cranial abdominal artery (15) with its diaphragmatic branch (15’), stomach (16), liver (caudate lobe, 17), and kidney (18). (Image republished from https://en.wikipedia.org/wiki/Celiac_ganglia; last accessed March 1, 2018.) The working group has observed that autonomic ganglia in unfixed carcasses are paler (light tan or white) and rarely appear brown or yellow as they do here in fixed tissues.

Ganglia

Where feasible, ganglionic sampling should be based on in-life findings (i.e., visceral dysfunction). However, in most cases, intramural autonomic (parasympathetic) ganglia in protocol-specified hollow organs (e.g., intestines, urinary bladder) represent the most common autonomic ganglia available for evaluation in situation 3. If intramural ganglia are missing from initial sections, then preparation of additional tissue sections from protocol-specified viscera should be considered.

In addition, several sympathetic ganglia may be considered for analysis. The working group recommends that at least two sympathetic ganglia be evaluated if an autonomic PNS neuropathy is suspected. Frequently sampled sites include the cranial (superior) cervical ganglion (Figure 3), cervicothoracic ganglion (Figure 4), cranial (superior) mesenteric ganglion (Figure 4), and celiac ganglion (Figure 4); if present in heart sections, the sinoatrial and atrioventricular nodes may also be assessed. Working group members have found that the cranial cervical ganglia in small animals and cervicothoracic ganglia in larger species, and the cranial mesenteric ganglia in all species, are the most readily harvested sympathetic ganglia. The caudal vagal (nodose) ganglion—which is a sensory (visceral afferent) portion of CN X—is easily confused with the cranial cervical ganglion since both are located in proximity to the bifurcation of the carotid artery, although the two ganglia can be differentiated readily in tissue sections (Figure 3). Somatic sensory PNS ganglia, such as multiple DRG (cervical and lumbar) and a trigeminal (CN V) ganglion, should also be considered for sampling.

Effector organs

In most toxicity studies, the list of protocol-specified tissues will include multiple effector organs that are innervated by the autonomic PNS (e.g., glands, heart, and hollow organs with abundant smooth muscle like the digestive tract and urinary bladder).

Lesions of the autonomic PNS have been linked on occasion to structural changes in some effector organs. For example, systemic administration of ganglioplegic drugs (i.e., “ganglionic blockers,” which inhibit transmission between preganglionic and postganglionic autonomic neurons in both the parasympathetic and sympathetic systems) can induce sperm granulomas in the epididymis of rats (Bhathal, Gerkens, and Mashford 1974). However, sperm granulomas are a common incidental background finding in this species, so their presence should not be interpreted as confirmation that a test article produces autonomic dysfunction in the absence of additional evidence to support this conclusion.

Central (CNS) autonomic centers

Preganglionic neurons for autonomic nerves reside in various brain nuclei (parasympathetic role) and the lateral (intermediate) column of the thoracic ± rostral lumbar spinal cord (sympathetic role). The hypothalamus serves many significant autonomic tasks. The most important autonomic structure in this region is the paraventricular nucleus (PVN) of the hypothalamus, which contains neuroendocrine cells that innervate the median eminence and pituitary gland (Ulrich-Lai and Herman 2009). In rodent brains trimmed according to current STP best practices for CNS sampling (Bolon, Garman, et al. 2013), the PVN should be present in level 3. CNs III, VII, IX, and X carry both somatic motor and parasympathetic nerve fibers; the parasympathetic components innervate involuntary functions of multiple muscles and glands. Locations of these brainstem parasympathetic nuclei reside outside the seven levels recommended for assessment under current STP best practices for CNS sampling (Bolon, Garman, et al. 2013) and instead will need to be localized using a species-specific neuroanatomy atlas (Paxinos, Huang, and Toga 2000; Paxinos and Franklin 2001; Paxinos and Watson 2007; Palazzi 2011) if in-life signs warrant their assessment. The lateral column of the sacral spinal cord also contains preganglionic autonomic neurons. Dogma for the past century has classed these sacral neurons as parasympathetic, but recent functional and molecular data indicate that these neurons may actually regulate sympathetic functions in pelvic viscera (Espinosa-Medina et al. 2016). These CNS sites may be considered for sampling and evaluation at the discretion of the institution if the potential for an autonomic neuropathy is present.

Situation-specific Fixation Options for PNS

Strategies for Trimming PNS Samples

Great care should be exercised when handling nerves and ganglia (even when fixed) as even subtle manipulation may cause artifactual changes. Tissue trimming of the PNS includes one or more nerve trunks and skeletal muscle (an effector organ) in all four situations as well as DRG (including spinal nerve roots) and/or autonomic ganglia for situations 2–4. Nerves and skeletal muscle should be trimmed to permit analysis of fibers in both cross (transverse) and longitudinal orientations. Particular attention should be given to evaluating skeletal muscle in cross sections because the morphological features of myofibers affected by PNS lesions (e.g., “fiber group atrophy” from denervation) are assessed most readily in this orientation. Myofibers in the diaphragm and tongue are arranged in crisscrossing patterns that preclude most fibers from being viewed in truly longitudinal and cross orientations, thereby adding to the challenge of detecting “fiber group atrophy.”

A properly prepared nerve cross section (Figure 5) allows for an assessment of the density and numbers of myelinated axons and to a lesser extent those of unmyelinated axons (Raimondo et al. 2009). The cross section also allows for an evaluation of myelin integrity (including discrimination between demyelination and remyelination) and may reveal Schwann cell changes not readily seen in longitudinal sections. The longitudinal section provides a means for demonstrating axonal or myelin damage spanning several internodes (Figure 5) and may, due to the length of nerve examined, allow for a better assessment of associated changes such as inflammatory reactions. Longitudinal nerve samples should be approximately 1 cm long if feasible (Bolon, Garman, et al. 2013) to ensure that sufficient numbers of nerve fibers will be visible over extended lengths. Spinal nerve roots may be isolated if necessary for evaluation (after embedding in hard plastic resin) but generally are embedded along with their associated DRG, typically in longitudinal orientation (Figure 6).

OPEN IN VIEWER

Figure 5.

Nerves should be available for histopathologic analysis in both cross (top row) and longitudinal (bottom row) orientations. The cross (transverse) view allows comparison of the numbers and densities of myelinated nerve fibers (large-caliber, pale blue axons bounded by thick, dark myelin sheaths) and possibly unmyelinated fibers (small-caliber axons with minimal myelin [often found in small clusters]), although such fine discrimination is only possible in specially prepared nerves exhibiting high contrast between axons (pale) and myelin sheaths (dark; upper left panel) and not in routinely processed sections (upper right panel) where contrast is modest and extensive clear space exists between nerve fiber bundles as a very common processing artifact. The longitudinal plane permits axonal and myelin integrity to be assessed over extended distances. Samples: sciatic nerve from normal (i.e., control) adult rat. Processing: left column = whole-body perfusion fixation with 4% glutaraldehyde, postfixation in 1% osmium tetroxide, hard plastic resin (epon) embedding, sectioning at 1 μm, toluidine blue staining; right column = immersion fixation in neutral buffered 10% formalin, no glutaraldehyde or osmium postfixation, paraffin embedding, sectioning at 4 μm, hematoxylin and eosin staining. (Images of hard plastic-embedded nerves [left column] were provided courtesy of Dr. William Valentine, by permission.)

OPEN IN VIEWER

Figure 6.

Placement of multiple ganglia in a single cassette ensures that sufficient tissue is available for histopathologic evaluation. The grouping may be ganglia from all spinal cord divisions (as shown here) or alternatively grouping as a single spinal cord division (e.g., cervical, thoracic, or lumbar). Sample: dorsal root ganglia and spinal nerve roots (arrows) from adult control rat. Processing: immersion fixation in neutral buffered 10% formalin, paraffin embedding, sectioning at 4 μm, hematoxylin and eosin staining.

For situation 3, isolated sympathetic ganglia should be processed in a fashion similar to other ganglia.

Situation-specific PNS Embedding Strategies

Embedding of PNS tissues is a critical factor in determining the data quality derived from the evaluation of PNS tissues. Paraffin allows detection of primary degenerative and infiltrative processes and therefore is a suitable embedding medium for PNS samples in general toxicity studies where PNS neurotoxicity is not a concern (situation 1). Paraffin is also used for most specimens in general toxicity studies where PNS neurotoxicity is suspected (situations 2 and 3) as well as in dedicated neurotoxicity studies where neurotoxicity (CNS and/or PNS) is likely or certain (situation 4) due to its low cost, ready availability, and suitability of such tissues for special pathology procedures (IHC, etc.). One neurotoxicity testing guideline states that “[p]lastic embedding is required for tissue samples from the peripheral nervous system” (US EPA 1998a, p. 6). The intent of this recommendation is to improve discrimination of fine cellular detail in myelinated and unmyelinated fibers. Use of plastic embedding media permits acquisition of thinner sections, thus providing improved resolution of cellular features (especially myelin sheath architecture).

Plastic embedding is expensive and labor-intensive. In studies where it is deemed that plastic embedding is warranted, the working group advises the following practical adaptation of regulatory guidance requiring plastic embedding for the PNS.

The working group recommends plastic embedding for at least one (situations 2 and 3) or two (situations 4) nerve cross sections (Figure 5), which are scenarios in which a concern exists that a test article may elicit PNS neurotoxicity. Indeed, for situation 3, nerve fibers (and especially the myelin sheaths) of autonomic nerves often are so small in diameter that plastic sections of osmicated nerves may be essential for light microscopic assessment. In such cases, PNS specimens slated for plastic embedding should have been postfixed in MGG and osmium. Cross sections of these nerve samples permit ready evaluation of the features and diameters for both axons and complete nerve fibers (i.e., axons plus myelin). Plastic embedding of longitudinal nerve sections is used less often as osmium deposition in myelin may obscure features in superimposed PNS nerve fibers due to overlap of the metal-impregnated myelin sheaths. However, plastic-embedded longitudinal nerve sections may be useful for evaluating nodes of Ranvier.

Several working group members suggest that laboratories and sponsoring institutions be encouraged to consider adjusting their PNS processing procedures for situation 1 to incorporate routine preparation of one osmicated, plastic-embedded nerve cross section as a means of attaining ideal morphologic preservation for PNS samples. However, the majority of the working group accepts that this modification may not be practical for the many general toxicity studies where no concern exists that the test article has induced PNS neurotoxicity and thus should be used in situation 1 at the discretion of the institution.

Plastic embedding for nerve samples usually employs one of two variants: “hard plastic” (hydrophobic) resins such as araldite, epon, or Spurr’s, or combinations thereof (e.g., epon-araldite); or “soft plastic” (hydrophilic) materials like GMA and MMA. Section thicknesses that are reproducibly attainable for PNS using hard plastic (<1 μm, Figure 5) and soft plastic (2 μm, Figure 7) are considerably reduced relative to that which is readily achievable for paraffin (4–6 μm, Figures 5 and 7). Soft plastics are more expensive than paraffin but are less costly and easier to process and section than are hard plastics. However, hard plastics can be used with osmicated PNS samples while soft plastics are not compatible with osmium; thus, myelin lamellae are only imperfectly conserved in soft plastic sections, which negates the original reason why plastic embedding of PNS tissues was required (EPA 1998a). The working group is of the unanimous opinion that soft plastic embedding media offer very little if any improvement in cytological resolution over paraffin embedding for nonosmicated nerve samples (Figure 7) and that soft plastic offers substantially inferior tissue preservation relative to hard plastic combined with osmication (Figure 5). Accordingly, the working group recommends hard plastic resin (of osmicated samples) as the best practice for plastic embedding of PNS and further advises that the use of soft plastic is not a suitable alternative for hard plastic embedding of the PNS. Methodological details for hard plastic embedding are found in the manufacturer’s instructions available with commercially available kits.

OPEN IN VIEWER

Figure 7.

Soft plastic (left panel) is not a suitable medium to comply with regulatory guidelines that mandate plastic embedding of nerves, as soft plastic provides no significant improvement in resolution relative to conventional paraffin embedding (right panel). Samples: sciatic nerve from normal (i.e., control) adult rat. Processing: left column = whole-body perfusion fixation in neutral buffered 10% formalin, no osmication, soft plastic (glycol methacrylate) embedding, sectioning at 2 μm, hematoxylin and eosin (H&E) staining; right column = immersion fixation in neutral buffered 10% formalin, no osmication, paraffin embedding, sectioning at 4 μm, H&E staining.

Osmium-impregnated nerves may be embedded in paraffin (Bolon, Butt, et al. 2013) or frozen sections (Strich 1968). The preservation and visualization of myelin is enhanced in osmicated, paraffin-embedded, and frozen nerve sections in comparison to nonosmicated, paraffin-embedded, and frozen nerve sections, respectively, but remains inferior to osmicated, hard plastic resin-embedded sections. Therefore, the working group recommends that paraffin embedding of osmicated tissues be avoided as a substitute for hard plastic resin embedding.

For dedicated neurotoxicity studies (situation 4), the working group recommends that nerves and DRG should be embedded in individual blocks (with or without other tissues), so that lesions may be tracked to identifiable PNS sites. Alternatively, some institutions place many DRG from all spinal cord divisions in one cassette (Figure 6) or group DRG from specific spinal cord segments into separate cassettes. The working group recommends that sampling more DRG is preferable, even if the exact locations of each individual DRG are not tracked. Therefore, the choice of embedding scheme (individual vs. multiple samples in one block) is left to the discretion of the institution.

For both paraffin- and plastic-embedded specimens, the working group recommends that all PNS tissues from the treatment groups selected for initial evaluation (e.g., high-dose and control animals) should be processed in the same time frame to avoid any systematic variation (e.g., block effects) in such technical factors as the lengths of time spent in fixative or dehydrating solutions. If a quantitative end point has been built into the study design, tissues from all study groups for which the end point might need to be collected should be processed into blocks at the beginning of the study, even if sectioning of the blocks for intermediate dose groups will be delayed. This strategy will greatly reduce the likelihood that variations in processing will impact the quantitative data. A key means of standardizing the effects of fixation and processing across treatment groups is to include cassettes from animals in the different cohorts within each processing “run,” so that handling-related artifacts are balanced by mixing samples from all dose groups.

Situation-specific PNS Staining Strategies

Staining of PNS tissues employs H&E for paraffin sections and toluidine blue for hard plastic resin sections (Table 3). The working group recommends these two stains as a suitable initial screen for PNS specimens in toxicity studies. When preparing PNS tissues, an important consideration is that delayed processing of some treatment groups (as opposed to immediate processing of all groups into blocks) may result in altered tinctorial intensity in H&E-stained nerve sections, which might confound any quantitative or post hoc masked (blinded) histopathologic evaluations. For studies in which PNS neurotoxicity represents a possible concern (situations 2–4), other neurohistological methods may be undertaken in nonosmicated, paraffin-embedded nerves at the discretion of the institution to further characterize or clarify the interpretation for any PNS findings discerned during the initial analysis. Stain quality for such special procedures varies depending on many factors including section thickness, technician experience, and regularity with which the procedure is performed.

When needed, the working group recommends stains for axons and myelin in paraffin sections as the most useful special methods for further evaluating PNS findings related to test article exposure. Silver stains (e.g., Bielschowsky’s [Figure 8], Bodian’s, or Holmes) or IHC methods to detect neurofilament proteins are helpful to highlight neurofilament-rich structures, including axons and cytoplasmic organelles in neuronal cell bodies, and for demonstrating damaged axons in axonopathy (e.g., fragmentation) or neuroaxonal dystrophy (e.g., axonal spheroids). Myelin-staining methods used in PNS tissues include Luxol fast blue (Figure 8) and Sudan black, which are especially beneficial for intact myelin, IHC to detect myelin basic protein, and the Marchi stain, which often is used to reveal demyelination (Strich 1968). The reasons for recommending these procedures are that axons and myelin are the two key components of PNS structures, and thus, many laboratories routinely perform these stains.

OPEN IN VIEWER

Figure 8.

Special methods used to highlight nerve fibers include silver and myelin stains. Upper panel: Bielschowsky’s silver stain demonstrates axons and neuronal cytoplasm as dark profiles against a pale background. Arrows indicate swollen axons. Lower panel: Luxol fast blue stain reveals myelin as intact dark blue sheaths. A single degenerating axon is revealed at the bottom of the image as a series of vacuoles containing fragmented debris. The myriad tiny, clear vacuoles in the myelin sheaths represent a processing artifact. Processing: immersion fixation in neutral buffered 10% formalin, paraffin embedding, sectioning at 4 μm.

Special neurohistological methods used to showcase neurotoxic damage in the CNS typically are not utilized when evaluating PNS neurotoxicity. Routine IHC methods to demonstrate glial fibrillary acidic protein (GFAP, upregulated in reactive astrocytes and in some satellite glial cells) and ionized calcium-binding adaptor molecule 1 (Iba1, expressed by microglia and macrophages) may be used in DRG to detect satellite glial cells and activated macrophages, respectively (Patro, Nagayach, and Patro 2010; Ton et al. 2013). However, such techniques typically are deployed in the research setting rather than in toxicity testing. In ganglia, glutamine synthetase (a preferred marker for satellite glial cells; Miller, Richards, and Kriebel 2002; Schaeffer et al. 2010) and CD68 (a macrophage marker; Jimenez-Andrade et al. 2006) may be used with or in place of anti-GFAP and anti-Iba1 procedures; leukocyte biomarkers (e.g., anti-CD3 for T-lymphocytes) may be helpful in differentiating inflammation from increased satellite cell numbers. Fluoro-Jade, a fluorescent stain used to detect necrotic neuronal cell bodies in the CNS (Schmued and Hopkins 2000; Schmued et al. 2005), may be attempted in the PNS to detect degenerating neuronal cell bodies in ganglia (Marmiroli et al. 2009). In the experience of working group members, Fluoro-Jade has not been consistently helpful in discriminating dying ganglionic neuronal cell bodies in DRG and autonomic ganglia (Krinke et al. 2001). The “red dead” morphology observed in necrotic neurons in the brain (Garman 2011b; Bolon, Butt, et al. 2013) typically is not observed in ganglia. Where a marker besides H&E is required to confirm that ganglionic neuronal cell bodies are dying or dead, the working group recommends anti-caspase-3 IHC (Schmeichel, Schmelzer, and Low 2003), terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL: Oliveira et al. 1997; Lee et al. 2017), or TEM.

While the use of soft plastic is not recommended for PNS tissues, archival samples embedded in this medium may be stained routinely with H&E. Other procedures that may be undertaken in soft plastic-embedded PNS tissues include silver stains for axons and histochemical stains (e.g., Sudan black; Cerri and Sasso-Cerri 2003) or IHC methods (Mueller et al. 2000) to reveal myelin lipids. The experience of working group members is that special techniques are applied to soft plastic-embedded PNS specimens mainly in the research setting.

Hard plastic-embedded nerve cross sections usually are osmicated during processing and then stained with toluidine blue (Figure 5). The concentration of toluidine blue used for this purpose varies among laboratories but typically is set at 1% (1:100; Hancock et al. 2005). Paraphenylenediamine may also be employed to highlight lipid-rich cell membranes in osmicated, hard plastic resin-embedded sections (Shirai et al. 2016) including myelin sheaths (Sadun, Smith, and Kenyon 1983).

Special histochemical procedures may be applied to differentiate various myofiber types in skeletal muscles (Armstrong and Phelps 1984; Kremzier 1984; Staron et al. 1999; Kammoun et al. 2014; Gropp 2017). These methods generally are not used to evaluate muscle samples in situations where PNS neurotoxicity is a concern as muscle lesions due to PNS damage (i.e., fiber group atrophy) may be seen easily by H&E.

Special Procedures for Evaluating PNS Neurotoxicity

If warranted, additional techniques may be undertaken to better characterize PNS lesions. Examples include TEM (Peters, Palay, and Webster 1991), morphometry (Diemer 1982; Kristiansen and Nyengaard 2012; Butt et al. 2014), stereology (Hyman, Gomez, and Irizarry 1998; Butt et al. 2014), teased fiber preparations (Krinke, Vidotto, and Weber 2000; Hancock et al. 2005), and quantification of intraepidermal nerve fiber ending density (IENFD; Lauria, Cornblath, et al. 2005; Lauria, Lombardi, et al. 2005; Myers and Peltier 2013; Mangus et al. 2016), motor end plates (Francis et al. 2011), and muscle spindles (Krinke et al. 1978; Muller et al. 2008). A detailed consideration of such special procedures is beyond the scope of this article. Decisions regarding whether or not to deploy these methods should be guided by data showing the neurotoxic potential of a test article to the PNS, usually the presence of in-life neurological signs or prior knowledge that the test article or a related molecule produces morphological effects in the PNS. Another important reason for quantifying IENFD, motor end plates, and/or muscle spindles may be to confirm the absence of neuropathy to provide evidence that nerve signaling is intact. The choice regarding whether or not to use these special procedures should be left to the discretion of the institution.

Appropriate Use of Masked (“Blinded”) Analysis in Assessing PNS Neurotoxicity

The initial microscopic evaluation of PNS tissues from nonclinical toxicity studies for all four situations generally should be conducted in an informed (“unblinded” or “unmasked”) fashion. In other words, the study pathologist should receive in advance full knowledge of the dose level and group assignment for each animal as well as other data (macroscopic findings, organ weights, clinical observations, and outcomes of behavioral testing), all of which might help in interpreting the microscopic pathology data. This recommendation represents a consensus opinion among experienced toxicologic pathologists (Gosselin et al. 2011), including members of the current working group, and conforms to established STP best practices for histopathologic evaluation of tissues from toxicity studies (Crissman et al. 2004). The initial uncoded assessment may be limited to the control and high-dose groups or may include target tissues in all dose groups at the discretion of the institution. The rationale for this recommendation is that informed examination greatly enhances the quality of the pathology data set by (1) permitting development of more objective criteria for grading tissue changes, (2) increasing the likelihood of detecting subtle PNS findings, and (3) enhancing the speed with which the analysis may be undertaken (which reduces diagnostic drift). This logic is no different from that used when designing the assessment for any other organ or system.

Once a PNS finding has been identified, a masked (blinded or coded) post hoc assessment of specific changes in particular target tissues may be performed at the discretion of the study pathologist (or peer review pathologist; Morton et al. 2010). Such masked evaluations should be limited, performed only as needed to clarify the incidence of subtle findings, promote consistency of severity grade assignments, discern test article-associated exacerbation of background lesions, and/or establish a dose–effect relationship (including definition of a NOAEL). The choice of which dose groups and findings to include in a masked evaluation is not defined by existing regulatory guidelines but rather is chosen by the pathologist; for example, the blinded assessment may be limited to the control and low-dose animals and ignore any other dose groups if clear neuropathologic changes are evident in the mid-dose and high-dose animals. This same strategy—informed initial analysis followed if necessary by a supplemental masked evaluation—should also be the usual practice for neuropathology peer reviews oriented toward PNS lesions.

Assessment of PNS Recovery in Nonclinical Studies

Unlike neurotoxic lesions in the brain and spinal cord, PNS lesions often can undergo repair, but such restoration (especially of damaged myelin) may be incomplete. If the study has been designed to include recovery groups, restoration of PNS elements may be examined during the light microscopic examination. Microscopic evidence of PNS recovery differs with the primary target site (i.e., axons or myelin sheaths). Primary axonopathies exhibit extensive Schwann cell proliferation, resulting in solid columns of Schwann cells (termed “bands of Büngner”) to support axonal regrowth. These bands produce less myelin than normal until the axon once again extends through the affected region. Recovery from primary myelinopathies is associated with increased nuclear numbers (which is required because new Schwann cells typically are smaller and therefore cover less axonal area than did original Schwann cells) and gradual restoration of myelin sheaths around intact axons. Teased fiber preparations demonstrate that myelin regeneration leads to decreased Schwann cell size and thus greater numbers of nodes of Ranvier. This enhanced cellularity is the structural basis for the reduced nerve conduction velocity available to individuals who have recovered from PNS damage, as signal propagation down the axon slows measurably at each node separating the smaller Schwann cells.

While PNS lesions may regress over time, the working group has no specific data to support selection of a particular recovery period for use in nonclinical toxicity studies. Complete restoration will require an extended period of time. For example, chickens treated with an organophosphorus ester exhibit partial but not complete recovery of nerve structure at 7 weeks postexposure, but nearly complete recovery by 9 weeks postexposure (Jortner et al. 1989). In rats where a traumatic (crush) injury has been applied to the median nerve, the number of nerve fibers was restored by 24 weeks, but axon diameters, nerve fiber diameters, and myelin thicknesses remained diminished (Muratori et al. 2012). The working group posits that visible evidence of PNS restoration might be seen but is not guaranteed to be evident during standard recovery periods in typical nonclinical studies (i.e., one-third to one-half of the time spent on treatment). The rationale for this supposition is that the rate of axonal regeneration is limited to the pace of slow anterograde axonal transport (i.e., approximately 0.5–1 mm/day; Morfini et al. 2012), which equates to axon regrowth of approximately 1.5 to 3 cm during a 4-week recovery period. If attempting to gauge the level of recovery, researchers should know that the degree of PNS regeneration may be underestimated by light microscopic evaluation (Ronchi et al. 2014). With these parameters in mind, which suggest that demonstrating PNS restoration may be difficult to detect and interpret (especially in larger species with their longer axons), the working group recommends that assessment of PNS recovery should be considered on a case-by-case basis when setting the length of recovery periods for nonclinical toxicity studies.