Abstract
The continuing education course on Developmental Neurotoxicity Testing (DNT) was designed to communicate current practices for DNT neuropathology, describe promising innovations in quantitative analysis and noninvasive imaging, and facilitate a discussion among experienced neuropathologists and regulatory scientists regarding suitable DNT practices. Conventional DNT neuropathology endpoints are qualitative histopathology and morphometric endpoints of particularly vulnerable sites (e.g., cerebral, cerebellar, or hippocampal thickness). Novel imaging and stereology measurements hold promise for automated analysis of factors that cannot be effectively examined in routinely processed specimens (e.g., cell numbers, fiber tract integrity). The panel recommended that dedicated DNT neuropathology data sets be acquired on a minimum of 8 sections (for qualitative assessment) or 3 sections (for quantitative linear and stereological analyses) using a small battery of stains to examine neurons and myelin. Where guidelines permit discretion, immersion fixation is acceptable for younger animals (postnatal day 22 or earlier), and peripheral nerves may be embedded in paraffin. Frequent concerns regarding DNT data sets include false-negative outcomes due to processing difficulties (e.g., lack of concordance among sections from different animals) and insensitive analytical endpoints (e.g., qualitative evaluation) as well as false-positive results arising from overinterpretation or misreading by inexperienced pathologists.
The impact of developmental neurotoxicity on individuals, their families, and their communities is a growing concern around the world. Many factors enhance the vulnerability of the immature nervous system to xenobiotics. Complex processes for building and integrating new neural circuits are typically more sensitive to disruption than are the routine functions required of adult neural cells (Rodier 1994; Claudio et al. 2000). Protective systems (e.g., blood-brain barrier integrity, detoxifying enzymes) are less developed in newborns and juveniles. In addition, the developing brain has a neurochemical profile that increases its susceptibility to anesthetic-induced apoptotic neuronal degeneration (Jevtovic-Todorovic et al. 2003). Children are also more likely to encounter neurotoxicants (both by using taste to explore the world and by regular proximity to contaminated ground surfaces; National Research Council [NRC] 1993) and are less able to neutralize them (due to impaired elimination and detoxification pathways during prenatal and early postnatal periods of development; Schwenk et al. 2003). In the past few decades, developmental neurotoxicity “epidemics” in humans from environmental contaminants (e.g., lead [Winneke 1996], methyl mercury [Castoldi et al. 2008], pesticides [Jurewicz and Hanke 2008]) and lifestyle-related toxicity (e.g., drugs of abuse [Derauf et al. 2009] and fetal alcohol syndrome [Bearer 2001]) have increased societal awareness that a more comprehensive regulatory approach to developmental neurotoxicity testing (DNT) is desirable to protect public health.
The ongoing evolution of nonclinical DNT assessment necessitates periodic reevaluation of DNT neuropathology practices (Hass 2003; Kaufmann 2003; Bolon et al. 2006; Kaufmann and Gröters 2006; Bolon et al. forthcoming). A current challenge for DNT neuropathologists is to move beyond conventional qualitative descriptions of neuroanatomic changes (gross and/or microscopic) to incorporate more quantitative endpoints. The dilemma posed by the shift toward more quantifiable analysis stems from the conflict between the benefits (greater objectivity, less bias) and disadvantages (labor intensiveness, slowness) afforded by historically available quantitative techniques. Several recent technological innovations in quantitative neuroanatomy may engender modifications in the traditional DNT neuropathology battery in the future.
The continuing education course on DNT held at the 2010 joint scientific symposium of the Society of Toxicologic Pathology (STP) and the International Federation of Societies of Toxicologic Pathologists (IFSTP) was created with three objectives in mind. The first was to review current “best practices” for DNT neuropathology analysis. The second goal was to describe new advances in neural stereology and noninvasive neural imaging that promise to increase the efficiency and speed with which high-quality quantitative data may be acquired. The third aim was to consider the essential elements necessary to instill confidence that a regulatory review based on DNT neuropathology data is well founded. The first two ends were addressed in a series of four technical talks, while the last purpose was accomplished via a panel discussion.
Current and Potential Components of the DNT Neuropathology Analysis
The first talk by Dr. Wolfgang Kaufmann provided an “Overview of the Role of Neuropathology and Morphometrics in DNT Assessment.” At present, the neuropathology analysis for a typical DNT study emphasizes qualitative histopathologic scoring in conjunction with a small battery of straightforward quantitative measurements (usually total or region-specific brain weights as well as two-dimensional [2D] areal or linear morphometric measurements) for such sensitive brain regions as the cerebral cortex, cerebellum, and hippocampus (U.S. Environmental Protection Agency [EPA] 1998; Bolon et al. 2006; Kaufmann and Gröters 2006; Organisation for Economic Cooperation and Development [OECD] 2007). Both the qualitative and quantitative evaluations are commonly complicated by the difficulty in reliably obtaining comparable sections through the same brain structures from all subjects. This problem may be dealt with by using tools (e.g., brain molds) to produce tissue blocks with equivalent orientations but is most commonly solved by cutting multiple serial or step sections to ensure that structures are oriented equivalently for all animals. The conventional DNT approach reliably detects large-scale defects in brain size (e.g., altered weight) and structure (e.g., anatomic disorganization, cellular ectopia) but is less able to distinguish alterations in cell number and tissue volume; even experienced neuropathologists cannot discern subtle deficits in cell number (<25% to 30%). Two additional disadvantages are that conventional DNT neuropathology endpoints provide a limited survey of microscopic changes in 2D for a modest number of brain regions and are acquired only at necropsy, which may preclude recognition of early transient abnormalities that arise immediately after initial toxicant exposure. The three remaining technical talks presented updates on recent developments in DNT neuropathology that promise to help address such challenges.
The second lecture, delivered by Professor Hans Jørgen Gundersen, conveyed “New Developments in the Application of Stereology to DNT Assessment” for unbiased evaluation of object distribution, orientation, shape, and size. A decision to include three-dimensional (3D) stereological analysis in the DNT neuropathology evaluation is typically made retrospectively using a weight-of-evidence approach or prospectively using historical information regarding toxicity of the test compound or structurally related molecules. Although highly sensitive and precise, prior stereological methods and the equipment available to undertake them have been too cumbersome and time-consuming to implement on a routine basis during DNT. However, new advances in stereological theory, sampling protocols, automated software, and imaging devices have greatly improved the efficiency and speed with which such data may be acquired (Korbo et al. 1990; Madeira et al. 1995; Dorph-Petersen, Nyengaard, and Gundersen 2001). The preferred stereological approaches for DNT neuropathology are likely to be Cavalieri-based volume estimates and optical or physical disector-based neuron counts of selected brain regions (e.g., cerebral cortex, cerebellum, corpus callosum, corpus striatum [basal ganglia], diencephalon [thalamus and hypothalamus], hippocampus, and mesencephalon). Adequate precision is typically achieved by evaluating 25 to 50 fields in 6 to 8 serial, 3-µm-thick disectors (where a disector consists of two adjacent sections) to acquire about 100 object counts. This disector method is fairly rapid: approximately 4 hrs to serially section a rat brain in the sagittal plane, and 1 hr per site for each animal to perform systematic uniform random sampling at 400x. The discriminative strength of quantitative DNT neuropathology endpoints (brain weight → 2D linear morphometry → 3D stereology), and thus the probability for pinpointing the extent and location of developmental brain lesions, increases as one moves from gross to a more detailed and quantitative analysis (de Groot et al. 2005).
The third talk by Dr. Al Johnson described “Magnetic Resonance Microscopy (MRM) of the Developing Brain,” while the final talk by Dr. Kathleen Sulik discussed the “Application of MRM to Defining Toxicant-Induced Abnormalities in the Developing Brain.” The MRM approach and its variants, such as diffusion tensor imaging (DTI), are noninvasive means for the in vivo or ex vivo examination of structural changes in the immature brain (Lester et al. 2000; Huang et al. 2008; Lodygensky et al. 2009). Currently available MRM instruments acquire digital 2D “sections” through the entire brain at a sufficient resolution (typically 20 µm [Petiet et al. 2008] to 100 µm [Johnson et al. 2006], acquired in approximately 45 min per sample) to permit evaluation comparable to microscopic analysis at magnifications up to 100x (10x objective). Automated algorithms allow assembly of the 2D serial sections into 3D reconstructions that highlight the site and extent (linear or volumetric involvement; Parnell et al. 2009; Godin et al. 2010) of structural disruption in the brain. Processing conditions may be modified to optimize MRM data (Petiet, Hedlund, and Johnson 2007). Discrete neural domains may be highlighted in the 3D reconstructions to facilitate structural comparison (qualitative and quantitative; Johnson et al. 2006) between normal and toxicant-exposed conceptuses. Complete fiber tracts may even be followed as they course between brain regions, permitting qualitative and quantitative assessment of toxicant-induced disruption in structural development. This approach is a powerful tool for investigating the potential pathogenesis of developmental neurotoxicity.
Regulatory Experience with DNT Neuropathology Analysis
The discussion regarding “A Regulatory Perspective on New Developments in Morphometric Approaches to Developmental Neurotoxicity Assessment” included experienced DNT neuropathologists (Drs. Robert Garman, Georg Krinke, and Peter Little) and regulatory scientists (Susan Makris and Dr. Daniel Mellon) as panelists, with regulatory scientist Dr. Karl Jensen as moderator. The objective was to identify common themes that arise in the course of DNT neuropathology evaluations and, where necessary, to define potential approaches for addressing such issues.
One outcome recognized by the panel was that not all DNT neuropathology measurements are created equal. This conclusion is supported by a recent retrospective examination of data from 69 guideline DNT studies of acceptable quality that were submitted to the EPA Office of Pesticide Programs (EPA-OPP). At the lowest observed adverse effect level (LOAEL), linear morphometry of tissue sections was found to be a more sensitive index of structural damage to the immature nervous system than either brain weight measurements or qualitative histopathology scores, and can be as or more sensitive than behavioral data (Raffaele et al. 2010). In like manner, gross linear measurements of the untrimmed brain may detect significant treatment-related differences when linear morphometric analysis at the microscopic level was inconclusive due to the imprecise positioning of the “homologous” tissue sections (Senn et al. 2010).
A second conclusion discussed by the panel was that the application of “novel” stereological and imaging methods to DNT neuropathology analysis will require a considerable degree of additional validation before these approaches may be considered for routine implementation in a regulatory setting. Particular care will be needed to ensure that the predictivity, reliability, reproducibility, and sensitivity of nonclinical DNT models are applicable to human developmental neurotoxicity disorders. Extensive comparison between new methods and currently recognized standards (e.g., 3D stereology vs. 2D linear morphometry, high-resolution MRM vs. conventional histopathology of tissue sections) using a battery of known developmental neurotoxicants will be required to substantiate the utility of the new techniques. A further consideration that must be addressed for these innovative approaches is to ascertain whether the presumably greater statistical sensitivity at the low end of the dose-response curve is actually predictive of a biologically relevant adverse outcome.
Finally, the collective experiences of the panel members indicated that a spectrum of DNT neuropathology practices may be acceptable, depending on the specifications that are given in various DNT guidelines (EPA 1998; OECD 2007). The panel held that the neuropathology component of dedicated DNT studies should routinely incorporate certain practices:
A minimum set of 6 (and ideally 8) coronal sections to show major brain regions (especially such sensitive sites as the cerebral cortex, hippocampus, and cerebellum). However, some neuropathologists recommend evaluating 1 or more para-sagittal (longitudinal) sections near the midline (to better visualize the major fiber tracts connecting major brain centers) in conjunction with 4 to 6 coronal hemi-sections (which will permit evaluation using traditional brain landmarks). A minimal staining battery of hematoxylin and eosin (to assess general brain architecture) and Luxol fast blue (to evaluate myelin). Fluoro-Jade B or a silver degeneration method (to demonstrate dying neurons) are not typically employed for DNT neuropathology because early neuronal injury (e.g., apoptosis) is unlikely to be detected at the end of the exposure period. Immunohistochemical detection of glial fibrillary acidic protein (GFAP) may be used to reveal enhanced glial production at sites of damage, but this method is more discriminating in young adult animals (postnatal day [PND] 60-70) rather than juveniles (PND 11 or PND 21-22). The panel felt that, where permitted by guidelines, DNT neuropathologists should have discretion to select among various practices: The preferred time for early DNT assessment is PND 21-22 as neural tissues exhibit less structural variation and are less fragile relative to PND 11. Perfusion fixation is required for proper brain fixation at PND 60-70, but immersion fixation is an acceptable means of preservation at PND 11 and PND 21-22 due to the difficulty in reliably perfusing such small animals. The structure of peripheral nerves can be reliably evaluated in samples embedded in paraffin.
What kinds of background alterations are commonly reported in immature rodent (mouse and rat) brains, and how often do they occur? The most frequent findings are visible differences in brain size (evident at PND 11; Garman et al. 2001) and increased numbers of displaced neurons (heterotopia), which can occur in laminated regions at multiple sites. What kinds of positive control procedures are recommended (i.e., concurrent vs. historical), and what DNT aspect are they validating (the sensitivity of the test species, the sensitivity of the quantitative method, and/or the proficiency of the pathologist)? The panel felt that concurrent controls are necessary for confirming the existence of positive findings within a study but that historical data using a positive control agent serves as evidence that a particular institution and its personnel can reliably perform a DNT study. The panel recommends that a positive control agent be included from time to time in future studies to validate that new personnel and novel techniques can reliably discriminate toxicant-induced changes; however, a specific time frame for repeating the positive control group was not stated by the discussants. The intermittent addition of positive control groups also confirms the ongoing sensitivity of the test species.
Data sets to support the range of acceptable practices defined above using the experiences of the panel members have yet to be fully reported in the literature.
The panel felt that major concerns with DNT neuropathology data sets usually stem from inaccurate (false-negative or false-positive) conclusions rather than from analysis of insufficient tissues. False-negative outcomes arise from processing difficulties (e.g., lack of concordance among sections from different animals) and/or insensitive analytical endpoints (e.g., qualitative evaluation). False-positive results usually signify overinterpretation or misreading of background findings or artifacts by inexperienced pathologists. The panel felt that reports should detail exactly what brain structures were examined (e.g., cerebellum, deep cerebellar nuclei, pons) rather than using broad nonspecific terms (e.g., hindbrain).
The panel was impressed with the potential for noninvasive neural imaging to substantially improve the neuropathology evaluation of DNT studies in the future. The current state of the art does not permit high-resolution, longitudinal (i.e., time course) neural imaging studies of the brain in vivo because living tissues move during breathing and cardiac contractions. However, a welcome advantage once the technology matures will be the ability to obtain qualitative and quantitative DNT neuropathology data at multiple time points from one animal.
The panel consensus was that better standardization of DNT neuropathology analyses among sponsors and across regulatory agencies and geographic regions should be a primary goal in any future revision of DNT guidelines. The discussants felt that such harmonization in the near term would provide meaningful improvement in current efforts to assess the risk of developmental neurotoxicity. Regular inclusion of innovative stereological and imaging methods in the DNT process will likely be warranted in the future. However, given the newness of these novel technologies, the discussants could not suggest a timetable or any criteria for deciding when the time will have arrived to request routine inclusion of such data in product registration packages.