Recording Pathology Data
Cheryl L. Scudamore
MRC Harwell, Harwell Science and Innovation Campus, Oxfordshire, UK
In pharmaceutical companies and associated contract research organisations (CROs), pathology data are usually generated by a toxicologic pathologist. Historically, toxicologic pathologists in the pharmaceutical industry have come from a number of scientific backgrounds, with different primary degrees in biological sciences, medicine, dentistry and veterinary medicine. Currently, the majority of toxicologic pathologists are veterinary graduates, usually with postgraduate training and qualifications in pathology (Bolon et al., 2010; van Tongeren et al., 2011). As well as recognising and recording pathological findings, pathologists should also be able to interpret these findings in the context of macroscopic observations made in-life and at necropsy, changes in organ weights and other available pathology data (e.g. clinical biochemistry and haematology changes). All of this information should be interpreted in a study pathologist’s report, taking into consideration what is already known about the compound, biopharmaceutical, gene or infectious agent under study (Crissman et al., 2004).
Pathologists often have multiple roles within an organisation, acting as study pathologists, project team members, researchers and managers. Whilst within the pharmaceutical industry the process of generating regulatory pathology (safety) data from necropsy, via slide production and through interpretation by a qualified experienced pathologist, is usually well regulated by Good Laboratory Practice (GLP) or an equivalent quality framework, this is not always the case in academia. It is therefore important when reviewing ‘discovery’ research data and published literature to be aware that the pathology may not have been reported by a qualified individual and that the terminology may vary from that used in toxicology studies (Cardiff et al., 2008).
Pathologists often refer to their results as ‘findings’, rather than ‘lesions’. A ‘finding’ is something that a pathologist considers worth recording and may or may not be a pathological lesion.
- 1) Pathologists may record ‘normal’ alterations and variations in tissue morphology, because these changes may be modified in treated animals. Examples of normal changes which might be recorded include:
- stage of oestrous cycle in females;
- presence/degree of extramedullary haematopoiesis in rodent spleens;
- presence of immature/pubescent reproductive organs, especially in studies involving large animal species (dogs and non-human primates).
- stage of oestrous cycle in females;
- 2) Pathologists may also note pathology that is incidental and which may or may not be specifically related to the study. For example, trauma related to fighting or accidental injuries from caging may occur in any study, whereas injuries related to gavage in rodents or cystitis following bladder catheterisation in dogs may be related to study procedures rather than an effect of treatment.
- 3) Tissues from all species will show spontaneous or background incidental findings that are recognised and expected variations for a given strain or species (McInnes, 2012). Pathologists will generally record these, because increases or decreases in the incidence of these lesions may be associated with a treatment-related effect. For example, inflammatory cell foci are common in many organs, but particularly the liver (Foster, 2005), and increased foci may be associated with minimal levels of hepatotoxicity, whereas decreased foci may be seen if a compound has anti-inflammatory properties.
- 4) Pathologists may, finally, note lesions that are uniquely induced by the experimental protocol itself. These may be related to the compound under test in a toxicity study, or they may be the result of a genetic modification in a study involving genetically altered animals.
Pathology is an observational science, and it is therefore, by definition, difficult to ensure exact standardisation in the recording of findings between different observers. Two major techniques are used to enhance the standardisation and reproducibility of pathology findings: semiquantitative analysis of lesion severity and harmonised terminology nomenclatures.
2.2.1 Semiquantitative Analysis
Traditionally, diagnostic pathology employs qualitative narrative reports to record what the pathologist sees down the microscope. A well-written qualitative description provides a lot of information about the morphological changes present in a tissue and can be very useful in creating a record of a novel induced lesion, which can subsequently be recognised by other pathologists. However, as with any other scientific observations, qualitative data are difficult to compare, subject to variation in style between individual analysts (pathologists) and hard to analyse statistically.
For many scientific parameters, quantitative (numerical) data are the expected output, but for pathology analysis, quantitation involves a significant input of labour and human intervention, in order to train computer systems and enable image-analysis tools to be used. In practice, semiquantitative analysis is almost universally used for high-throughput, non-neoplastic pathology studies, including toxicology studies, as it provides sufficiently reproducible data when used by experienced pathologists (Table 2.1). Comparison between semiquantitative and quantitative methods has shown that the overall conclusions are usually the same when analysing pathology data (Shackleford et al., 2002; Von Bartheld, 2002).
Table 2.1 Comparison of data types.
|Description of morphological changes seen in tissue||Extent or severity of lesions divided into a number of discrete scores or grades||Measurement of numbers of cells, lesions and areas affected|
|Useful for recording novel induced lesions||Useful for rapid data recording, where a yes/no or ranked answer is sufficient||Useful where precise data are required and in separating subtle differences|
|Time-consuming||Faster||Faster for individual parameters, once image-analysis system is trained|
Slower if used to measure whole range of possible lesions in a study
|Difficult to analyse||Can be analysed||Can be analysed statistically|
|Difficult to compare groups statistically||Allows statistical comparison using nonparametric tests|
|Relies on experience of operator||Relies on experience of operator||Less reliant on experience|
Neoplasms (i.e. tumours) are not normally graded semiquantitatively in terms of severity, but are usually recorded as being ‘present’ and categorised based on morphological criteria and biological behaviour as ‘benign’ or ‘malignant’ or as ‘fatal’ or ‘nonfatal’, to allow for Peto analysis (Peto et al., 1980).
Semiquantitative analysis involves allocating a grade or score to a lesion, based on its extent or severity. Ordinal grades or scores usually extend from 0 to between 3 and 5. Higher numbers of categories tend to be associated with less reproducibility, as it is hard for an individual to distinguish between and remember minor differences between grades. The system of scoring may be based on an approximately linear or nonlinear approach; some examples of descriptors for different grades are given in Table 2.2. A nonlinear approach is often preferred, as it allows background lesions to be acknowledged without overemphasising their significance (i.e. the assumption is that if a lesion is a background finding, it should not affect the tissue or organ function, and therefore should be present at a low grade at most; Mann et al., 2013).
Table 2.2 Examples of linear and nonlinear grading schemes that could be used for semiquantitative analysis of non-neoplastic lesions in laboratory animal tissues. Source: Scudamore (2014). Reprinted with permision from Wiley-Blackwell.
|0||NAD (WNL): No change recorded||0||NAD (WNL): No change recorded|
|1||Minimal: 0–20% of tissue affected by change||1||Minimal: The least change that is visible on light microscopy at ×20; small, focal or affecting <10% of tissue|
|2||Mild (slight): 21–40% of tissue affected by change||2||Mild (slight): Change is readily detected but not a major feature; may involve multifocal small lesions or affect <20% of tissue; may still be within background appearance for the species|
|3||Moderate: 41–60% of tissue affected by change||3||Moderate: Change is more extensive or involves more foci (e.g. seen in every ×20 field), beyond the usual background for the lesion in the species; may start to have relevance for organ function and may correlate with other changes (e.g. increased organ weight)|
|4||Moderately severe (moderately marked): 61–80% of tissue affected by change||4||Moderately severe (moderately marked): As for 3, but more of the tissue is affected (e.g. up to 75%); likely to have relevance for tissue/organ function|
|5||Severe (marked): 81–100% of tissue affected by change||5||Severe (marked): Virtually the whole tissue is affected by the change, which is likely to be functionally relevant/detrimental|
NAD, nothing abnormal detected; WNL, within normal limits.