Practices to Optimize Generation, Interpretation, and Reporting of Pathology Data from Toxicity Studies

2.1

Ensure a Deep Understanding of Regulatory Guidance and Best Practices

Governmental regulatory agencies such as the FDA in the United States and the EMA in Europe as well as international consortia like the OECD have issued GLP standards specifying that all procedures used in the toxicologic assessment of new drugs intended for human use must be clearly defined and accountability accurately documented. These guidelines should be followed when toxicity assessment is conducted in order to satisfy the regulatory requirements for the registration and marketing of new drugs.

Although each country may have unique regulatory requirements for the assessment of the safety of new chemical or biological entities, international efforts for harmonization have resulted in the publication of standardized approaches by the ICH ( Table 28.2 ; https://www.ich.org/page/safety-guidelines ). This Council is composed of representatives from regulatory authorities and bio/pharmaceutical firms.

Toxicologists and toxicologic pathologists working in the bio/pharmaceutical industry are expected to know and understand the appropriate regulatory guidelines. Understanding these guidelines is the basis for the appropriate design of studies intended to assess the safety of new chemical and/or biological entities. For instance, for the pharmaceutical industry, the ICH guideline M3(R2) describes the nonclinical safety studies that are recommended to support human clinical trials as well as marketing authorization of bio/pharmaceuticals. It is the basis for planning safety assessment packages and the specific toxicity studies that compose those packages. Specific guidelines pertaining to various toxicity studies (e.g., genotoxicity, carcinogenicity, acute toxicity, repeated-dose toxicity, reproductive toxicity, etc.) are also available and are discussed elsewhere in this textbook (see The Role of Pathology in Product Discovery and Safety Assessment , Vol 2, Chap 2; The Role of Pathology in Evaluation of Reproductive, Developmental, and Juvenile Toxicity , Vol 1, Chap 7 ; and Carcinogenicity Assessment , Vol 2, Chap 5).

In some cases, a deep understanding of regulatory guidance and pathology best practices needs to be augmented with additional consultations to ensure that the pathology aspects of a nonclinical toxicity study will be sufficient to enable dosing in humans, predict potential risks to exposed humans, or permit registration. In these cases, proactive consultations with third-party subject matter experts and/or regulatory agencies will be helpful, if not essential. Examples of these cases include carcinogenicity studies, programs with expected complex toxicity profiles, programs with low margins of safety, or test articles with novel mechanisms of action or unusual therapeutic modalities. The communication may include the submission of background project information and the proposed protocol or study design elements for key studies together with the key questions that the development team is submitting for advice. Such proactive consultations increase the chances that the study design is scientifically robust, will provide meaningful data to enable dosing or predict potential risk in humans, and will be acceptable to regulators. Not consulting with regulators or subject matter experts may result in nonclinical toxicity studies that are not acceptable to regulatory agencies. Communication with regulatory agencies will help the development team to incorporate the perspective and experience of regulators into study designs, especially since regulators may have knowledge of toxicity risks unique to a class of test articles. Understanding the regulatory perspective on a specific program is helpful for the most efficient allocation of time and resources as well as for the appropriate design of studies for that program.

2.2

Clearly Define Realistic Study Goals, Design the Study to Accomplish Those Goals, and Minimize Study Complexity

The study design starts with identifying the goals and objectives that the study team wants or needs to achieve with the experiment. These goals must be realistic and achievable, and the study team must understand the problem and what question(s) the study can and cannot address. It is often most efficient to start with the end in mind and then work backwards when designing the study. The identification of a hypothesis, or hypotheses, is a critical part of defining study goals and is easier to achieve if the team understands the details of the research program and the current status of the development program. For instance, knowing the pharmacologic target, the intended therapeutic indication, the proposed population to be treated, the literature reporting the biology of the target, and the potential implications of manipulating the target may guide the team as to what nonstandard endpoints must be included in the study. Understanding the current point in the development program helps determine if a short-duration or a dose-ranging study should only include survival and tolerability endpoints rather than including full batteries of in-life, clinical pathology, and anatomic pathology endpoints. Sometimes pathology endpoints, such as microscopic evaluation of tissues, are not essential to the goals of a study and avoiding inclusion of these endpoints will allow the use of nonnaïve animals to avoid the unnecessary use and termination of naïve animals. Having a clear understanding of the main goals of a toxicity study will help the toxicologist and/or toxicologic pathologist fine-tune the risk assessment of new chemical entities and avoid the conduct of unnecessary studies or the addition of unnecessary, complex, or unachievable goals or endpoints to toxicity studies.

The likelihood of ending up with an unsuccessful, confounded, invalid, or inaccurate study increases when an experiment has too many variables, requires an excessively complex design to test the hypotheses, if the goals are not sufficiently clear or specific, and/or has goals that are unlikely to be achievable. An example of the latter would be not planning for a sufficiently long reversibility period to determine whether or not complete recovery takes place for findings that require longer periods to reverse (e.g., testicular toxicity). Although being mindful of the 3Rs (Replacement, Reduction, and Refinement) in animal research is important, one should avoid the temptation to “get as much as possible” from any particular study by including too many endpoints. Currently, there is a strong push in industry for drug development teams to accelerate their research and development programs, prompting the common request for “innovative” ways to design nonclinical toxicity studies that maximize the amount of information obtained from a single study. Although innovation in such omnibus studies is welcomed and encouraged, study teams must always ensure that any novel endpoints or designs do not undermine the integrity of their studies. For example, the development team, which includes scientists outside of the toxicology or pathology disciplines, may request that the team “take advantage” of a toxicity study to answer questions regarding the pharmacology of the test article, which may greatly complicate the study design. Interest in “taking advantage” of a toxicity study is common within multidisciplinary teams that have a limited understanding of toxicology risk assessment or the limitations of nonclinical toxicity studies. If the toxicologist and toxicologic pathologist are not careful, agreeing to these requests to address side issues may introduce nonessential goals to the toxicity study, triggering the need for the collection of samples or data that are not commonly collected in standard toxicity studies. For instance, development teams may be interested in collecting tissue samples that require special surgical procedures to be performed during the in-life phase of the study in order to have special assessments of the pharmacology of the test article. Those procedures could introduce new variables into the study such as postprocedure complications, medications, and/or anesthesia that might jeopardize the interpretation of the core pathology data. Before introducing those new variables, the toxicologist and/or toxicologic pathologist must evaluate the impact on study procedures, the tissue samples, and the quality, accuracy and validity of the study. Features of an excessively complex study design may also include an unnecessarily large number of experimental groups, more than one test article, endpoints for which tests are not standardized or well understood, excessive sampling time points, and/or the collection of samples that require special procedures that have not been practiced in “production mode.” If the inclusion of any of these “extra” items is elected in a study, it is important to keep in mind that the main goal of a toxicity study is the toxicologic characterization of the test article. Therefore, study designs should emphasize acquisition of data to serve that purpose without introducing errors or external factors that may jeopardize the validity of the study. There may be times where designing separate studies rather than a single all-encompassing study will be better for the evaluation and interpretation of clinical pathology and anatomic pathology data.

2.3

Choose the Appropriate Animal Species, Genetic Background, and/or Disease Model for Toxicity Testing

The choice of which animal species to use in a nonclinical toxicity study is dictated by a number of factors including regulatory guidance, the test species' physiological and metabolic similarities to humans, the pharmacological activity (or lack thereof) of the test article in the species, availability of historical control data for the species, typical lifespan of the species (relevant mainly for chronic toxicity or carcinogenicity studies), adaptability of the species to experimental conditions, and feasibility of using the species under the intended experimental conditions. In addition to selecting the appropriate animal species for investigating the test article, the development team must select the appropriate genetic background or strain, geographic origin of animals, and/or determine if an animal model of disease will be used in the toxicity studies. The use of the appropriate animal species, genetic background, and (when necessary) disease model are key elements of a well-designed toxicity study that minimizes variability in toxicologic responses and facilitates the generation of high-quality pathology data. Irrelevant, confounding, or potentially misleading pathology data may be generated if inappropriate animal species, background, or disease models are used during the development program. For each animal species, strain, or model selected, the toxicologic pathologist must become deeply familiar with the types of spontaneous findings in the animals under investigation ( ). This knowledge will be increased with experience and awareness of the many scientific publications dedicated to these topics. Armed with experience and knowledge, the toxicologic pathologist will be able to better assess if microscopic and/or clinical pathology findings are spontaneous or test article related.

The growing use of biological entities has evolved to the point of increasing the need for nonhuman primates (NHPs) as test species. This choice is dictated because such human-derived test articles may not be pharmacologically active in nonprimates, thus limiting the use of rodents as part of the safety assessment package. Alternatively, other approaches might be needed, such as surrogate molecules (e.g., a rodent-derived biomolecule with a similar pharmacologic action in rodents as the human-derived test article), nonstandard animal species such as marmosets, or genetically modified animals (e.g., which express the human molecule that is targeted by the human-derived test article) to test the potential for pharmacologically driven toxicity ( , and Genetically Engineered Animal Models in Toxicologic Research , Vol 1, Chap 23 ). If a biologic is not cross-reactive with rodents, the development team must also consider if assessing carcinogenicity in alternative models will be a necessary course of action ( ). The toxicologic pathologist must be aware of the limitations, complications, and influence of these alternative approaches on the generation of pathology data. For example, the use of nonstandard animal species may complicate the interpretation of test article effects because of the lack of historical experience at the test facility or lack of a robust historical control database of spontaneous findings.

The use of nonstandard animal species may also be needed for test articles that are not biotherapeutics. Rodents have many metabolic pathways, physiological processes, and pathological responses similar to those in humans. However, there are some responses in rodents that have been demonstrated to not be relevant to humans due to differences in the PK, TK, PD, metabolism, genetic differences, and other factors (see Animal Models in Toxicologic Research: Rodents , Vol 1, Chap 17 ). Furthermore, these responses may also differ among rodent species (and strains) and may result in selection of nonstandard rodent species as the appropriate rodent species for testing, such as the Syrian Golden hamster for carcinogenicity ( ; , ). The toxicologic pathologist must be part of the discussion when the use of nonstandard animal species is being considered and be prepared to generate pathology data without the availability of robust historical control data.

In addition to the appropriate selection of a rodent species, the selection of a proper stock or strain is an essential piece of planning for a toxicology program and ensuring that relevant pathology data are generated. Being aware of the characteristics of the rodent outbred stocks and inbred strains or genetic lines used for toxicological assessment is important, and the development team must define, based on the nature of the test article, whether chronic toxicity and/or carcinogenicity studies will be required. If the toxicology package will include chronic toxicity and carcinogenicity studies, it is highly recommended to use the same rat stock, strain, or genetic line throughout the entire toxicology program due to the differences occurring among different strains or genetic lines ( ). Examples of the impact of rodent genetic background on the generation pathology data were highlighted in a recent publication ( ) where pancreatic microscopic changes occurring only in rats after administration of a Bruton's Tyrosine Kinase inhibitor were strain dependent. Other strain-dependent spontaneous findings in rodents, such as aortitis in BALB/c mice ( ), can influence the assessment of test article–related effects. Similarly, genetic background can influence spontaneous tumor incidence in rodents of different stocks and strains ( ). Rodent strains with high spontaneous tumor incidences in certain tissues may make the assessment of test article–related carcinogenicity in those tissues more difficult for the toxicologic pathologist.

The source of the animals is an important consideration in study design. When utilizing NHPs in toxicity studies, the geographic source (e.g., China vs. Mauritius vs. other regions of Southeast Asia) may be an important factor since it is known that there is slight genetic and phenotypic variability among populations of monkeys ( ; ; ). The use of different geographic sources across a development program may impact the toxicologic pathologist by introducing variation in the incidence of spontaneous microscopic changes or clinical pathology parameters and thus potentially confound the interpretation of test article–related changes ( ; ). It is recommended that when possible study teams utilize animals from a consistent geographic origin across the toxicity studies supporting a drug program, and that the toxicologic pathologist be well versed in the common spontaneous findings that occur in the animals from that geographic origin (see Animal Models in Toxicologic Research: Nonhuman Primate , Vol 1, Chap 21 ). Familiarity with the common spontaneous findings will enable the pathologist to more accurately separate test article–related findings from incidental findings. For rodents, we also recommend that the study team use a consistent source and breeding facility to minimize sources of variability across studies. However, over time, continued inbreeding of rodents at a specific location likely may result in generation of a substrain (e.g., C57BL/6NCrl vs. C57BL/6NTac substrains maintained by Charles River Laboratories and Taconic, respectively) ( ).

Animal models of disease are not routinely used to assess the toxicity of test articles. However, the thoughtful use of these disease models can sometimes help investigate pathogeneses and inform the evaluation of test article–related findings and risk to humans ( ; ; ). For example, some targets that are present only during a disease process, such as the expression of beta-amyloid (Aβ) in Alzheimer's disease patients, have prompted the use of engineered animal models of this disease, such as the PDAPP mouse model (which overexpresses mutant human amyloid precursor protein), as one of the species used to assess specific potential risks that are present only when Aβ is deposited in brain ( ). Similarly, spontaneous or engineered rodent and nonrodent models of many inherited genetic diseases are employed in combined efficacy/toxicity studies to examine biologic or gene therapy test articles. When a decision is made to utilize an animal model of disease, the toxicologic pathologist should be familiar with the limitations of the animal model including model-related microscopic or clinical pathology changes; inconsistency or variability in the phenotype or severity of the pathologies; different types of spontaneous findings; and limited or no historical control information (see Genetically Engineered Animal Models in Toxicologic Research , Vol 1, Chap 23 ). The toxicologic pathologist can gain experience in the typical pathologic findings in the animal model via literature reviews, microscopic evaluation of prior (e.g., efficacy or pharmacology) studies, and the use of pilot toxicity studies ( ). Although not animal models of disease, engineered mice with enhanced vulnerability to test article–mediated carcinogenicity, such as p53 hemizygous and Tg.RasH2 mice, are used in accelerated (i.e. six month) bioassays to assess carcinogenicity of pharmaceuticals ( , Carcinogenicity Assessment , Vol 2, Chap 5). In determining the most appropriate carcinogenicity testing approach, study teams should assess if the test article's pharmacologic actions interfere with a molecular pathway (e.g., Ras signaling for the Tg.RasH2 mouse) in a way that could alter the sensitivity of the bioassay such that a two-year carcinogenicity study in wild-type mice would be a better test system. Understanding potential limitations of confounding factors related to the animal models used in a safety assessment program is important for ensuring that the toxicologic pathologist generates high-quality pathology data.

During the study planning phase, the age and health of the test animals are important items to consider since these can have a negative impact on the generation of pathology data. In our experience, these issues are of greater importance in studies using nonrodents. The toxicologist and toxicologic pathologist should know the most relevant age or sexual maturity state of the animals for safety assessment in any study. Sometimes this factor is not considered early enough in the planning process, and limited availability of sexually mature animals may delay the start of the study. The utilization of a sexually immature animal when a sexually mature animal should have been utilized may lead to incorrect conclusions about the reproductive safety of a test article. The prestudy health of the animals may be dependent on the source of the animals, transportation-related stress, and other factors. Toxicologic pathologists and/or veterinary staff should assist in screening animals prior to inclusion in a study to ensure that those with suboptimal health are not included in the study. Suboptimal health of the animals may negatively impact the generation of pathology data by introducing the confounding effects of preexisting illnesses and lesions, or by altering the response of less robust animals to test articles. Some preexisting key health issues in NHPs and dogs that the authors have encountered over the years include diarrhea, inability to maintain or gain weight, and inability to maintain serum glucose concentrations within the accepted reference intervals prior to inclusion in a study.

2.4

Use Optimal Communication Practices

Proactive and collaborative communication by the entire study team is an important practice that facilitates the generation of high-quality pathology data. Suboptimal communication at the time of protocol preparation can occur due to the practices, institutional cultures, and business models of the diverse, and often complex, network of organizations involved in toxicity studies/programs ( ). Some bio/pharmaceutical companies conduct all or most of the toxicity studies at different test facilities operated by CROs. Thus, sponsor representatives, including sponsor pathologists, have a key role in implementing good communication practices. Good communication must start with the members of the study team, and certainly must include proactive communication between the study director, toxicologist, the study pathologist(s), peer review pathologist(s), and when applicable, sponsor representatives such as the sponsor pathologist. In outsourced studies, the sponsor and CRO must establish an effective working relationship and maintain open and proactive communications. The interactions should start before placing any studies ( ). During the preparation of the protocol sponsors should clearly communicate their goals and expectations, including communication practices and preferred procedures/processes for when in-life findings arise such as moribundity/mortality, and the CROs should communicate to sponsors when proposed study designs and sponsor expectations are not realistic. At this stage of the study, sharing critical background information such as the nature of the test article (e.g., small vs. large molecule), intended pharmacology, pharmacologic target, potential target tissues, and data obtained from previous studies will help the study team obtain a complete understanding of the test article and the potential toxicity that might be observed. Awareness of the potential toxicity can have an impact on protocol-specified study procedures.

2.5

Estimate and Communicate the Number of Potential Target Tissues

The toxicologic pathologist plays an integral role in the identification of potential target tissues in any toxicity study, which are tissues that are expected to demonstrate test article–related changes, regardless of if these changes are direct (i.e., primary) or indirect (i.e., secondary) effects due to the test article. Indirect changes include soft tissue mineralization due to altered calcium/phosphorus ratio resulting from renal disease and atrophic changes in fat secondary to test article–induced decrements in body weight and food consumption. Providing the expected number of potential target tissues (if known) is one of the optimal communication practices that benefits the generation of pathology data and facilitates the generation of timely pathology reports. The biggest benefit of this communication during the protocol preparation phase is the ability to proactively schedule enough time for the study pathologist(s) and the peer review pathologist(s) to complete their tasks without excessive and unnecessary timeline pressure. Not communicating target tissues thoroughly and proactively could limit the ability of the timeline to absorb unexpected target tissues and may cause delays in finalization of the study report. Sometimes, delays in finalization of the study report can have a negative impact on the progression of the test article in development, such as delays to planned regulatory filings. Timely communication of the expected target tissues may also influence the scheduling of sample collections (e.g., blood), preparation of the laboratory for nonstandard tissue collection or nonstandard procedures, or result in other modifications to protocol-specified procedures. Although not much can be done to decrease the risk of truly unexpected target tissues, it is possible to identify the likely potential target tissues at the time of the preparation of the study protocol by having a good understanding of results of previous toxicity studies with the test article or related compounds and by researching the potential toxicity risks associated with the chemical structure, excipients, chemical modifications or conjugates, and risks inherent in modulating the therapeutic target. For example, toxicologic pathologists should be aware of the potential findings related to administration of biomolecules conjugated to polyethylene glycol ( ), some excipients such as cyclodextrins ( ; ), and therapeutic platforms such as antisense oligonucleotides ( ); in some cases, findings associated with these materials can result in clinical pathology or microscopic findings in many tissues that are unrelated to the pharmacologic action of the test article.

2.6

Develop and Practice Novel or Nonstandard Pathology Procedures and Other Methods Prior to the Start of the Study and Use Accepted Practices for Incorporation of Nonstandard Endpoints

The use of standardized tissue lists and/or procedures across toxicity studies greatly helps maintain consistency in toxicity data sets in product development programs. Standardization also helps laboratories minimize errors, deviations, and delays. However, one of the most common dilemmas that a toxicologist or toxicologic pathologist may face during the study design phase is whether or not to include the assessment of nonstandard tissue samples, assays, or procedures in toxicity studies, particularly if these are not standard for a particular test facility or sponsor. Examples of nonstandard procedures may include perfusion fixation, immunohistochemistry (IHC) or in situ hybridization (ISH), transmission electron microscopy (TEM), in vivo assessments of immune function, and nonstandard biomarkers. As alluded to earlier, it is essential for the study team and/or for the toxicologic pathologist to ponder the true need of including those endpoints in the study and to confirm if the inclusion of those endpoints is essential for the risk assessment of the test article. If nonstandard samples, assays, or procedures are indeed needed in the study, we strongly recommend that nonstandard procedures be practiced to proficiency prior to implementation in definitive GLP toxicity studies. Such newly developed procedures may become the subjects of new standard operating procedures (SOPs), or these may be recorded as study-specific procedures.

Perfusion fixation prior to tissue collection requires planning and practice as the procedure adds significant complexity and time to necropsy workflows. Improper perfusion can easily ruin tissues due to excessive perfusion pressure or incomplete perfusion. Furthermore, perfusion fixation changes the normal postmortem appearance of tissues, which can greatly distort the appearance of highly vascular organs (e.g., lung, spleen) or interfere with the identification and collection of very small structures such as sympathetic or parasympathetic ganglia. Although assessment of ganglionic neurons or other small tissues may be important for a thorough toxicology evaluation in a particular study, it is important to define and practice how the samples will be collected, preserved, processed, and evaluated ( ).

In situ molecular pathology assays (e.g., IHC and ISH) and TEM are great investigative tools that toxicologic pathologists can utilize, if warranted. However, successful implementation of these procedures in toxicity studies requires planning, practice, and/or proactive method development to ensure the generation of high-quality pathology data. For example, associated with the development of new biotherapeutics, the incidence of immunogenicity-related events in nonclinical toxicity studies has increased. Characterizing these immunogenicity-related reactions may include confirmation of the occurrence of immune complex disease (ICD) with the use of ancillary tests such as IHC or TEM. Therefore, it is important to allocate additional time for sample processing and evaluation as well as to add special directions and trained personnel for sample collection, preservation, processing, and evaluation. Starting method development as soon as possible for the special procedures will prevent undesirable delays in toxicity studies.

Inclusion of in vivo assays of the immune response in routine nonclinical toxicity studies can help demonstrate and further characterize test article–related effects and avoid the need for stand-alone studies (see Immune System , Vol 4, Chap 6). However, inclusion of these assays may require practice and piloting. One methodology to assess cellular immune function is by studying the influence of a test article on the immunogenic response to the administration of tuberculin in NHPs. This assessment introduces a nonstandard procedure consisting of vaccination with bacille Calmette–Guérin (BCG) vaccine before and/or during the dosing phase. The assessment of the response includes the microscopic evaluation of the cellular reaction in skin samples taken during the in-life phase. Although implementation of these ancillary methods is feasible, their application in a toxicity study may introduce additional stress to the animals and potential consequences that may affect the overall toxicity assessment in the study. Specifically, the administration of BCG in NHPs may cause the formation of microscopic granulomas in multiple organs ( ; ). This phenomenon, commonly known as “BCGitis,” may complicate the interpretation of pathology results.

Nonstandard or novel biomarkers, such as Kidney Injury Molecule-1 (KIM-1) or Neutrophil Gelatinase–Associated Lipocalin (NGAL), can be valuable tools for the characterization of toxicity in nonclinical studies and/or can help guide the implementation of biomarkers in the clinic (see Biomarkers: Discovery, Qualification and Application , Vol 1, Chap 14 ). If the toxicologist and pathologist decide that the evaluation of those endpoints is warranted for the toxicological assessment of the test article, they must understand how and when in the study the samples must be collected, preserved, and processed in order to obtain as much accurate information as possible. Not communicating those requirements clearly in the protocol might affect the quality, accuracy, and validity of the results. A specific example on this would be the need for quantifying reproductive hormones (e.g., luteinizing hormone, follicle stimulating hormone, testosterone, and estradiol). The blood concentration of these hormones varies over different days during the estrous cycle in females or even throughout a single day in males (testosterone) and females (estradiol) ( ; ). Understanding these characteristics of the biology of the biomarkers helps in determining the number and timing of when samples need to be taken in order to have an accurate determination of the blood concentration of each hormone. Another example is the quantification of cardiac troponin I (cTnI), which is widely accepted as a biomarker for cardiac injury in several animal species, including rats, dogs, and NHPs. However, in order to have confidence in the validity of cTnI data, the study team must understand the analytical method used and craft a protocol that describes in detail appropriate sample handling procedures and the appropriate time points for sample collection. The half-life of cTnI is short, and early timepoints (e.g., 4 h or less after dosing) should be included to maximize the chances of identifying test article–related increases. The toxicologic pathologist also should be aware that simple stimuli such as restraining animals can induce stress and affect the concentrations of serum cTnI ( ).

2.7

Ensure that Study Procedures Are Adequately Described in the Study Protocol or Standard Operating Procedures

A clearly written study protocol that is easy to understand and follow greatly enables generation of pathology data and the success of nonclinical toxicity studies. The toxicologic pathologist must ensure that any procedures related to pathology endpoints are adequately, but not excessively, described in the protocol prior to initiation of the study. Protocols provide the minimum necessary amount of information to guide the collection of pathology endpoints, while detailed descriptions of standard pathology processes are generally defined in permanent institutional SOPs or a study-specific procedure document (see Pathology and GLPs, Quality Control and Quality Assurance , Vol 1, Chap 27 ). Such documents commonly include written descriptions and, if warranted, schematic diagrams, flowcharts, photographs, or tables to facilitate understanding by the personnel responsible for performing the task. Practices that should be documented in this fashion include the processes for tissue or blood collection, sample preservation, handling, processing, and analysis. Well-written protocols and SOPs facilitate the reproducibility and accuracy of the pathology data in a study and across studies for a given development program. Study teams must be aware that procedures differ among research facilities; therefore, detailed descriptions of the methodologies to be used in the study are important for the generation of reproducible and reliable data over time.

Tissue processing is an important step in toxicity studies. It is critical to clearly define in SOPs how processing will be performed and that these processes avoid the introduction of variables that complicate the microscopic evaluation by the study pathologist. Furthermore, not following well-written tissue processing SOPs may introduce artifacts, such as tissue vacuolation, that complicate the histopathological interpretation. If the study team has decided to include special procedures such as IHC or TEM, the relevant special directions for tissue collection, preservation, and processing must be included in the study protocol and/or SOPs. Poorly detailed protocols and/or SOPs reduce the chances of obtaining high-quality samples that are suitable for interpretation.

Most toxicity studies collect data that are subjected to statistical analysis. It is recommended to define in the study protocol which endpoints will be statistically analyzed and what kind of statistical tests will be applied. In some studies, a statistician should be consulted early in the study planning process to discuss what standard statistical analysis should be performed and what modifications may be needed based on the emerging data (see Experimental Design and Statistical Analysis for Toxicologic Pathologists , Vol 1, Chap 16 ). For instance, in carcinogenicity studies, early mortality in control or experimental groups may prompt the application of different statistical methods. Discussing the application of those methods either during the planning of the study or during the conduct of the study with the study director, toxicologist, statistician, study pathologist, the peer review pathologist, and, if applicable, sponsor representatives is an essential practice.

2.8

Plan for Pathology Peer Review

Reappraisal of a portion of the study pathologist's initial histopathologic observations is a useful quality control (QC) procedure (see Pathology Peer Review , Vol 1, Chap 26 ). We recommend that study teams incorporate pathology peer review ( ) in their GLP-regulated toxicity studies at the time of protocol preparation rather than waiting to see draft pathology results before deciding to add pathology peer review. Pathology peer review as set forth in a study protocol typically is performed by a single peer review pathologist. Peer review or less formal data review processes can also be utilized in non-GLP toxicity studies to increase confidence in the pathology data.

2.9

Do Not Blind the Study Pathologist

The practice of performing a histopathological assessment without knowledge of treatment and/or related demographic and clinical data, also known as an evaluation under “blinded” or “masked” conditions, can be used as a tool to reduce bias in the interpretation of findings. In product development programs, this practice is most commonly utilized in experimental studies used for standardization of biomarker identification or for assessment of efficacy of novel therapeutic entities ( ). However, this practice is not recommended in GLP-compliant nonclinical toxicity studies, and its use is discouraged by the STP ( ; ). Accordingly, we recommend that study protocols for GLP nonclinical toxicity studies do not include “blinding” of the study pathologist. Since the histopathology assessment in GLP studies is performed with the goal of identifying all effects related to the administration of the test article, it is important for the pathologist to know which animals are controls since they provide context regarding the range of normal incidental findings (or “within normal limits”) to be expected in the test animals under the conditions of the study that is being evaluated. The response to the administration of a test article, stress, infections, or other pathologies is variable, and may induce subtle changes that may not be possible to distinguish from test article–related effects if the pathologist does not understand what “normal” is in the animals of the study. Blinding the pathologist to treatment without the ability to define this normal range of background lesions may prevent the accurate identification of subtle test article–related changes such as atrophy, hypertrophy, increased or decreased cellularity, apoptosis, necrosis, and reproductive tissue estrous cycle abnormalities. An informed (“nonblinded” or “open”) evaluation in which the treatment group and clinical observations for each animal are known allows the pathologist to have an increased sensitivity in the detection of test article–related effects, helps in developing appropriate criteria for grading the severity of microscopic findings, and permits the more efficient evaluation and documentation of changes occurring in the study. This last item is of essential importance because a diagnosis is not simply the identification of one change but a holistic understanding of its variability from animal to animal and its impact in each affected group and the interpretation of its significance under the conditions of the study. Therefore, before making a histopathologic diagnosis, the pathologist must know all information from each individual in the study, including the treatment group allocation, time of termination during the study, clinical observations, gross findings, body and organ weights, and clinical pathology findings.

Although blinded histopathologic analysis is discouraged for the primary evaluation, the STP acknowledges that it can be appropriate during the pathologist's reassessment of the slides in order to fine-tune diagnoses and severity grades, minimize diagnostic drift, and/or confirm a subtle test article–related effect and threshold value, such as a NOAEL ( ). This masked reevaluation is performed on a case-by-case basis at the discretion of the study pathologist and usually is performed informally (i.e., by hiding the slide labels) rather than formally (i.e., by having the original slide labels covered by new coded labels).

2.10