The Vital Role of Pathology in Improving Reproducibility and Translational Relevance of Aging Studies in Rodents

Abstract

Pathology is a discipline of medicine that adds great benefit to aging studies of mice by integrating in vivo, biochemical, and molecular data. It is not possible to diagnose systemic illness, co-morbidities, and proximate causes of death in aging studies without the morphologic context provided by histopathology. To date, many rodent aging studies do not utilize endpoints supported by systematic histopathology, which leaves studies incomplete, contradictory, and difficult to interpret. Similar to traditional toxicity studies, if the effect of a drug, dietary treatment, or altered gene expression on aging is to be studied, systematic pathology analysis must be included to determine the causes of age-related illness, moribundity, and death. In this Commentary we discuss the factors which should be considered in the design of aging studies in mice, with the inclusion of robust pathology practices modified after those developed by toxicologic and discovery research pathologists. Investigators in the field of aging must consider the use of histopathology in their rodent aging studies in this era of integrative and preclinical geriatric science (geroscience).

In contrast to long-lived humans, laboratory mice have a relatively short lifespan of less than three years for most inbred strains.^45,53,80 As laboratory mice are the most frequently used animal model system in biomedical research, many reagents, assays, and technologies are available for detailed analyses of these animals as they age. Understanding aspects of aging at the molecular, cellular, and tissue levels that impact the pathogenesis of chronic disease is the goal of the Trans-National Institute of Health Geroscience Interest Group (GSIG). The GSIG “strives to understand how aging enables chronic disease” and recognizes the critical role for histopathology in translational aging studies (http://www.nia.nih.gov/sites/default/files/geroscience_summit_2013_outcomes-recommendations_v2.pdf).³⁴ In this Commentary, we discuss the factors which should be considered in the study design and provide a framework for designing aging studies in rodents to include robust and consistent systematic pathology analysis, practices similar to those developed by toxicologic and discovery research pathologists. As aging research is increasingly focused on interventions with the goal of developing anti-aging therapeutics, well-designed pathology investigations are a critical and mandatory step in the preclinical process.^18,41,42

Basic experimental design and considerations

There are two basic designs of aging studies; cross-sectional (serial necropsy) and end-of-life (lifespan) studies.⁵¹ Cross-sectional analyses select specific ages for serial necropsy to study progression of aging changes in gene expression, biochemistry, histopathology, imaging, and functional in life (in vivo) testing.^50,71,80 common end point for most commonly-used strains is 20–24 months of age, or 24 months for carcinogenicity studies.^26,77

Where defining the natural lifespan is the goal, end points are based on morbidity and mortality. Mice age until they are found dead or, ideally, are euthanized when moribund. Kaplan-Meier plots²⁵ can be used to compare length of life between controls and test animals or between inbred strains.⁸⁰ Institutional Animal Care and Use Committees may not allow death as an endpoint and may require specific clinical criteria for euthanasia. These criteria and their thresholds can vary dramatically among countries, institutions, and individuals conducting clinical assessments of the mice even within the recommendations established by the NIH (http://oacu.od.nih.gov/ARAC/documents/ASP_Endpoints.pdf) or Federation of European Laboratory Animal Science Associations (http://eslav-eclam.org/legislation-and-guidelines/felasa-guidelines-and-recommendations/). This variability may contribute to the lack of reproducibility between studies.

Another source of variability is in defining the mice that are selected for pathologic analysis. If mice are found in a moribund state and euthanized, a high quality analysis can be done. Despite the rapid onset of post-mortem autolysis in mice, valuable pathology data may still be obtained from animals found dead and studies should be designed to examine all carcasses to assess for overt disease and tissue preservation. Given the financial and scientific value of aging mice as they approach their natural death, it should be a major goal of aging studies to perform systematic histopathological analysis to determine causes of illness and death, and to detect subclinical diseases which enables assessment of overall disease burden. Research staff should have basic training in performing a necropsy and have appropriate fixative available to store carcasses prior to histopathological assessment.^37,57,66,73

Conducting aging studies: the urgent need for pathology support in the analysis of aging rodents

Numerous papers describe the nuances of designing and conducting research on aging in rodents.^{11,19,53–55} Aspects to consider include animal strain, endpoint, and estimated number of animals required.^11,51 The inevitable loss of animals over time must be taken into account when planning studies in older mice.⁴⁸ Statistical analyses in aging studies can also be complicated by early censure of mice for humane reasons.⁷⁴ Studies should balance statistical power against financial and welfare concerns, as using insufficient numbers of mice is as inappropriate as using too many.¹ Several references provide details on experimental design and principles for conducting aging studies in rodents.^{11,19,35,37,51,67,71} Here, we focus on the aspects of study design that impact pathologic analysis.⁶⁷

Most critical and often unachieved, histopathological assessment requires appropriate funding and a commitment to maintaining available funds for pathologic analysis at the end of the study. Aging studies are, by their nature, long term and expensive. The failure to plan for histopathology analysis out of a false sense of cost containment can significantly impact the final interpretation of the remaining molecular and functional data, much of which is much more expensive and less comprehensive than a routine morphologic assessment. To illustrate, aging a cage of up to 5 mice at a per diem rate of $1 USD per day will cost $730 USD over 2 years. For large phenotyping projects, it may cost $50,000 USD to generate and in vivo phenotype a mouse line.⁶⁵ At an estimated $250–350 USD per mouse for necropsy, histology, and expert interpretation by a certified pathologist, pathologic assessment provides a wealth of integrative data which enables understanding of diseases by linking morphology (lesions) to pathogenesis.^2,63,65 The GSIG calls for “increased pathology measures in multiple animal models with age so as to assess the presence of co-morbidities, and the role of aging”.^12,34 While this is encouraging, the reality of the current financial status of funding agencies makes it unlikely that budgetary concerns will be alleviated in the near future, so studies must be designed with realistic budget, feasible timelines, and flexibility to collect as much reproducible data from the animals as possible. Minimally, tissues should be collected properly for archiving in the form of fixed wet tissues or paraffin blocks even if there is no immediate histopathological analysis funded.^37,66,73 The proper archiving of tissues is a form of insurance against having to repeat laborious and expensive studies just to collect tissues.

While the robust data set that histopathology provides to studies is well recognized as a cornerstone of research by the pharmaceutical industry, research outside of drug development does not consistently adopt this postulate. This is for financial and logistical reasons as well as a perceived lack of access to expertise or failure to realize that histopathology contributes meaningfully to an understanding of treatment intervention effects.^2,41,64,65 The powerful multi-disciplinary datasets generated in preclinical testing are obtained through in-life and terminal metrics, consensus terminology and diagnostic criteria, lesion scoring, and data management.^32,33,64,65 Protocols and experimental designs should be standardized and well described.^{8,15,52,58,67,35} Data must be appropriately collected, assessed, and interpreted by experienced professionals with formal pathology training who understand the limitations of the study design, model, and types of data sets.^7,18,76 The guidelines established for pre-clinical Best Practices that were set out by the Society of Toxicologic Pathology (http://www.toxpath.org/positions.asp) provide a framework for investigators studying aging to understand the integrative nature of pathology. Protocols can be modified to suit the needs of aging research studies.

Historically, the majority of published rodent aging studies evaluated survival and gene expression.^12,44 Many of these studies provided little to no pathologic data given the focus on life span rather than cause of death. A review of longevity studies involving genetically modified mouse models found that only 7 of 24 studies had some pathology data reported.⁴³ However, 3 of these were limited to selected tissues,⁷⁹ diseases,²⁸ or gross necropsy.¹⁴ In most studies, the causes of death and spectrum of lesions associated with illness and death were not examined. Recently, as candidate genes or interventions have been identified, follow-up studies to define healthspan and characterize possible mechanisms of lifespan extension have been initiated.^36,50,74 Healthspan is of increasing interest, and disease or absence of expected disease must be documented to define health. As noted by Flurkey and Harrison,¹⁹ without histopathology one cannot study senescence, as illness due to undetected lesions can be misattributed to another cause. Also, it is critical to document aging-associated lesions that may be modulated by interventions.³⁰

The contributory causes of death and the total burden of disease are often much more informative about the biology of aging and chronic disease than the immediate cause of death. As such, recent publications (for example those examining the role of rapamycin in aging) have incorporated some pathology data, yet tissues and description of the methods are generally limited.^{27,49,56,78,81} To compare datasets and to achieve reproducibility of studies across studies, pathology-based methods and endpoints must be standardized, systematically applied, and well described. Aging studies should strive to document all lesions present within an individual as it is the total burden of disease (or absence thereof) that defines health and enables investigations into senescence in non-diseased models.^11,19,41,59 Mechanisms of delayed aging or pathogenesis of altered disease onsets or total burden may be more fully characterized if a complete data set is generated and published. An aging population (human or rodent) typically has more than one independent disease process (co-morbidities) which may contribute to cause of death, although death may also be due to a single disease entity. While a catastrophic illness may be identified in some individuals, assigning a single cause of death often oversimplifies the disease burden and fails to capture co-morbidities that impact health span and lifespan.⁷⁴ Assigning cause of death is fraught with challenges and is greatly impacted by the formal training and personal experience of the individual assigning the cause (see Snyder et al. this issue).^38,39 All lesions present within an individual should be documented and scored when appropriate. The Interventions Testing Program’s (ITP) rapamycin studies document no change in disease patterns and presumptive causes of death²⁷ or decreases in non-lethal mild tissue changes⁷⁸; however, limited tissues were examined, lesions were not scored, and neither study reported co-morbidities.⁷⁸ Additional studies from the ITP⁸¹ and Neff et al.⁵⁶ have included more pathology data but did not calculate disease burden or the role of neoplasms in morbidity/cause of death. Further, inflammatory lesions were not discussed in detail despite the known role of inflammation (“inflammaging”) of elderly humans,^20,21 reported age-associated chronic inflammatory lesions in old mice,^4,46,69 the immune modulating effects of rapamycin,^4,46,69 and predicted anti-inflammatory mechanism of action of some of rapamycin’s anti-aging or morbidity-compressing effects.⁶⁸ Questions regarding the true effect of rapamycin on healthspan remained unanswered.^17,31

Pathology Data: Necropsy and Histopathology

The critical first step in capturing data on changes in tissues, organs, and organ systems is the systematic gross examination of tissues and collection of tissue samples at the time of necropsy, with systematic and unbiased reporting of lesions.⁶⁷ When morphologic data are coupled with in-life functional assays, laboratory, medical, and molecular findings, accurate phenotypes or pathophysiologic mechanisms are characterized and models are validated.^2,59,65

Necropsy

When planning terminal tissue collections, most principal investigators overlook the time and expense of sample collection, processing, and interpretation. Inventory, tracking of samples, and distribution of materials to various investigators for specialized assays all complicate the process. For example, organs of mice are small and often there is a demand to sample tissues for multiple assays, such as gene or protein expression as well as pathology. Tissue collection must be triaged according to the study’s critical questions. Technical approaches to necropsy are well referenced in numerous publications and on-line resources.^9,37,66,75 For routine collection of mouse tissues for histology, a systematic and effective necropsy can be performed in less than 5–10 minutes by an experienced prosector with minimal equipment and supplies.^66,73 The most critical point regarding necropsy is that tissues are irretrievable once they are disposed of, improperly collected or inappropriately fixed. The proper observation of gross lesions (especially focal lesions that would not be captured in a random "blind" sampling of standard organs for later histopathology) is also critical to correct final diagnosis.

Histopathology

The interpretation and definitive diagnosis of gross lesions generally requires histopathology. For example, multiple white masses within the liver may be a result of neoplasia or inflammation.⁵⁹ Also, important lesions seen microscopically may not be visible on gross examination. The multi-disciplinary design of aging studies requires that experts are involved at all levels. This is especially true with histology: it is clear that attempts at “do-it-yourself pathology”^8,13,29 have resulted in over- and mis-interpretation²⁹ in many areas of research. Histopathology by a trained, board-certified, and experienced veterinary pathologist or medical pathologist with mouse expertise, remains the gold standard to relate molecular and biochemical findings to morphologic observations.⁶⁰

Nomenclature and Lesion Scoring Systems

A rational lesion scoring system using consensus definitions of lesions, nomenclature (https://www.toxpath.org/inhand.asp)^10,23,32 and severity criteria should be utilized to address particular study needs. Scoring systems can be used to subjectively generate semi-quantitative data or by computerized image analysis to objectively compare data across groups or studies.^64,65 Critically, published reports should define and illustrate the pathology scoring system to ensure reproducibility between studies. Scoring can be affected by issues such as tissue sampling and diagnostic drift, subjects familiar to most pathologists but often unappreciated by researchers with little formal pathology training.^24,64

Data management

Evaluation and reporting of histopathologic data requires databases to collect and analyze the data in a systematic manner. There are a number of systems available. One that is available at no cost and allows for integration into other databases is Mouse Disease Information System (https://www.jax.org/research-and-faculty/tools/mouse-disease-information-system).^70,72 This allows for rapid collection of data in which anatomic location, disease process, and definitive diagnoses can be entered by typing the first letters of the term which will then be autocompleted and consistently coded from the mouse anatomy (MA) and mouse pathology (MPATH) ontologies.⁶² Data can be moved into relational databases such as MySQL for sophisticated analyses such as genome-wide association studies.^5,6 Industry-standard (as used for toxicity studies), purpose-designed pathology data collection systems are available, and may be accessed in some instances on an individual subscription basis. Commercial programs can also be used for recoding and retrieving pathology data (e.g. Microsoft Access, Word, and Excel). Statistical analysis of pathology data should be conducted under the supervision of a biostatistician familiar with pathology datasets. Statistical analysis of pathology and other phenotyping data is covered elsewhere.^40,61

Integration of pathology data with other phenotype data

Necropsy and histopathology data should be integrated with other in-life phenotype or imaging data to determine causes of death and differences between experimental groups. With physiological phenotyping, findings may not fully contribute to determining the cause of morbidity or death without histopathological confirmation, and indeed important phenotypes may be entirely missed.^2,3,16 For example, a recent study by Adissu et al² on a random selection of mutants from the International Mouse Phenotyping Consortium (IMPC) primary phenotyping pipeline found that histopathology added correlating morphological data in 63% of cases for which the primary physiological screen detected a phenotype, and 14% presented significant histopathology findings that were not predicted by the standard primary physiological screen. On the other hand, if pathology findings do not discover the cause of illness, other phenotype data may aid in the determination of the cause of morbidity and mortality. For example, high glucose level in the urine is a historic diagnostic criteria for diabetes mellitus for which histopathologic lesions can be less specific. Thus, morphologic assessment and physiologic data each provide critical and non-redundant information that together defines the contributory causes of death.

Conclusion

Pathology must be an integral and required part of aging rodent studies as it provides contextual data, validates the model, improves translatability, and reduces wasted effort. Histopathologic analysis conducted systematically by experienced pathologists can reveal phenotypes and insights to disease and intervention mechanisms that may be missed or misinterpreted with other in vivo phenotyping assays.^2,3 The lack of standardized, well planned, and well-executed pathology assays in long-term and resource-intensive aging studies is problematic and should be addressed by developing and adopting consistent pathology protocols, terminology, and scoring systems.^41,42 Systematically employed, consistently applied pathology end points will increase accuracy, reproducibility, and translatability of preclinical interventional aging studies.^13,22