Toxicologic Pathology

The sequencing of the human genome will have a major impact on the prevention, diagnosis, treatment, monitoring, and outcome of cancer. Progress will most likely occur in a stepwise fashion with the biggest initial impact in diagnosis and molecular targeting of new medicines. Advances in genomics and proteomics have already resulted in major findings that are facilitating earlier cancer diagnosis and disease stratification. New treatments that target specific pathways are well underway for many cancers and likewise, molecular analyses aimed at finding medicines with reduced toxicities are being conducted. Close interaction with the Food and Drug Administration will be very important to integrate these new technologies into product development and the approval of new medicines and diagnostics. Clearly, we are now approaching the era of personalized medicines in which one has the opportunity to maximize the efficacy and minimize the side effects of the cancer treatments. The ability to translate this new technology into improving cancer patient care and outcomes will require that scientists from academia, industry, and government work together closely to develop the framework and standards necessary to maximize the benefit of the human genome project.

The sequencing of the human and mouse genomes, and soon that of the rat, offers a foundation to evaluate biological phenomena, including toxicologic effects. Numerous tools are being developed to evaluate aspects of biology based on the DNA sequence. These tools can be utilized to evaluate absorption, distribution, metabolism and excretion, and effects of the toxicologically active product on the target organ. The genes involved can be broadly categorized as those affecting susceptibility to a toxicologic effect and those that are involved in the biologic response. For risk assessment to be performed in a rational manner, fundamental mechanisms of toxicologic processes must be ascertained. Based on successes already achieved, such as development of transgenic and knockout mouse strains, the application of aspects of the genomics revolution could be useful in developing a better understanding of mechanisms, and possibly in identifying specific markers of responses. In addition, genomics are likely to be useful in translating effects between species. However, genomics are being portrayed as the ultimate solution to all of toxicology. This is hardly the case. Basic chemistry, biochemistry, toxicokinetics, pharmacology, and pathology will continue to be needed in the overall weight of evidence approach to risk assessment. Genomics are likely to be of limited usefulness in predicting individual, in contrast to population susceptibility to various toxicological responses. Concordance of various diseases in identical twins, for example, different cancers, is rarely greater than 20% over the lifetime of these individuals. Similarly, genomics are likely to be of limited usefulness in screening for toxicologic end points. As with other tools of biology, those to be developed based on the genome are likely to provide greater usefulness in dissecting the mechanistic processes involved and defining the basis for susceptibility.

Identification of reliable markers to predict drug-related adverse events (DRAEs) is an important goal of the pharmaceutical industry and others within the healthcare community. We have used genetic polymorphisms, including the most frequent source of variation (single nucleotide polymorphisms, SNPs) in the human genome, in pharmacogenetic approaches designed to predict DRAEs. Three studies exemplify the principles of using polymorphisms to identify associations in progressively larger genomic regions: polymorphic repeats within the UDP-glucuronysltransferase I (UGT1A1) gene in patients experiencing hyperbilirubinemia after administration of tranilast, an experimental drug to prevent re-stenosis following coronary revascularization; high linkage disequilibrium within the Apolipoprotein E (ApoE) gene in patients with Alzheimer Disease (AD); and the polymorphic variant HLA-B57 in patients with hypersensitivity reaction after administration of abacavir, a nucleoside reverse transcriptase inhibitor for the treatment of HIV. Together, these studies demonstrate in a stepwise manner the feasibility of using pharmacogenetic approaches to predict DRAEs.

Toxicogenomics is an emerging multidisciplinary science that will profoundly impact the practice of toxicology. New generations of biologists, using evolving toxicogenomics tools, will generate massive data sets in need of interpretation. Mathematical tools are necessary to cluster and otherwise find meaningful structure in such data. The linking of this structure to gene functions and disease processes, and finally the generation of useful data interpretation remains a significant challenge. The training and background of pathologists make them ideally suited to contribute to the field of toxicogenomics, from experimental design to data interpretation. Toxicologic pathology, a discipline based on pattern recognition, requires familiarity with the dynamics of disease processes and interactions between organs, tissues, and cell populations. Optimal involvement of toxicologic pathologists in toxicogenomics requires that they communicate effectively with the many other scientists critical for the effective application of this complex discipline to societal problems. As noted by Petricoin III et al (Nature Genetics 32, 474-479, 2002), cooperation among regulators, sponsors and experts will be essential for realizing the potential of microarrays for public health. Following a brief introduction to the role of mathematics in toxicogenomics, “data interpretation” from the perspective of a pathologist is briefly discussed. Based on oscillatory behavior in the liver, the importance of an understanding of mathematics is addressed, and an approach to learning mathematics “later in life” is provided. An understanding of pathology by mathematicians involved in toxicogenomics is equally critical, as both mathematics and pathology are essential for transforming toxicogenomics data sets into useful knowledge.

Cancer affects approximately 1 in 3 individuals. An individual's susceptibility to cancer is partly determined by environmental exposures and by the combination of inherited cancer susceptibility and resistance genes. Initial mapping of these low penetrance cancer susceptibility genes has been done in the mouse because human low penetrance genes are extremely difficult to find using traditional methods due to heterogeneity and interacting factors. Also, the choice of candidate genes for human association studies can miss the unknown or unexpected. Mouse models also have limitations; it can be difficult to identify causal polymorphisms in the mouse because linkage disequilibrium often extends across several genes. To exploit the strengths of both systems, we outline a cross-species strategy to identify human variants associated with increased cancer risk. This approach uses linkage analysis and haplotyping, allelic imbalance in tumors, and gene expression studies in the mouse, combined with association studies and tumor imbalance studies in humans to identify causal cancer susceptibility variants. Allelic variants in both mouse and human can then be used to better understand the mechanisms behind cancer risk and as targets for intervention.

Modern pathologists have been challenged to “validate” mouse models of human cancer. Validation requires matching of morphological attributes of the model to human disease. Computers can assist in the validation process. However, adequate controlled, computer-readable vocabularies that can match terms do not currently exist in mouse pathology. Further, current standard diagnostic terminologies do not include the new concepts discussed here such as pathway pathology and mammary intraepithelial neoplasia. The terminologies must be revised and improved to meet the challenge. Human medicine has traditionally used “guilt-by-association” to validate interpretations of disease. Experimental pathology uses experimental verification exemplified by “test-by-transplantation.” Genetically Engineered Mice (GEM) develop unique tumor phenotypes bringing new structural-functional insights and reevaluation of concepts. Novel GEM-related tumors appear in all organ systems but mouse models of human breast cancer are prototypes. For example, mammary tumors induced by Mouse Mammary Tumor Virus (MMTV), chemical, radiation or other carcinogenic stimuli have limited phenotypes. These “spontaneous” or induced mammary tumors have never resembled human breast cancers. GEM tumors created with genes associated with human cancer are strikingly different. GEM tumors have unique histological phenotypes. Depending on the genes, the tumors may: 1) resemble MMTV-induced tumors, 2) display “signature” phenotypes, and 3) mimic human breast cancers. The phenotypes can be placed into structural and functional clusters with shared characteristics leading to the concepts of Pathway Pathology: tumor phenotype reflects the genotype.

Use of laboratory animals to identify carcinogenic potential of chemicals, mixtures, and other agents has a modern history of greater than 40 years from which much useful scientific and public health information can be derived. While laboratory animals differ from humans in some respects that may affect responses to hazardous exposures, use of such models is based on experimental evidence indicating that there are more genetic, genomic, physiological, biochemical, and metabolic similarities than differences among mammalian species. Issues of concordance of responses between rodent species and between rodents and humans as well as repeatability and site-specificity are important considerations in evaluating laboratory animal carcinogenicity results. Variables in experimental design such as animal strain, diet, route of exposure, and study duration as well as single-site versus multisite carcinogenic responses all influence interpretation and intelligent use of study data. Similarities and differences in site-specific laboratory animal and corresponding human cancers should also be considered in study evaluation. Recent attempts to explore genetically engineered mice and to humanize the mouse for more relevant identification of carcinogen hazard identification have yielded mixed results. In the end we are confronted by the realization that virtually all animal cancer models are useful but imperfect surrogates for humans. Assuming the percentage of chemicals currently in commerce that are estimated to be potent animal or human carcinogens is quite low, the task of identifying agents with significant carcinogenic potential is daunting and important. The biological conundrum of scientific debate regarding the relevance of carcinogenicity studies in laboratory animals is likely to continue. Nonetheless public health considerations must take precedence when deciding human safety issues.

Determining the carcinogenic potential of materials to which humans have significant exposure is an important, complex and imperfect exercise. Not only are the methods for such determinations protracted, expensive and utilize large numbers of animals, extrapolation of data from such studies to human risk is imprecise. Toxicologists have long recognized these shortcomings but the 2-year chronic rodent study has remained the gold standard. Recent developments in the field of molecular oncology and development of methods to insert or inactivate specific genes in animals have provided the tools with which to develop the next generation of carcinogenicity assays. With improved understanding of oncogene activation and tumor suppressor gene inactivation a number of animal models have been developed to dramatically reduce latency for chemically induced cancers. This has led to the development of shorter carcinogenicity assays. Also, because the spontaneous tumor frequencies in these animals are low during the in-life portion of the study, and studies are terminated well before the health complications of advanced aging are observed, it has been possible to reduce the group sizes and reduce animal usage. FDA's adoption of ICH S1B in 1997, (ICH, 1997) “Testing for the Carcinogenicity of Pharmaceuticals,” opened the door for the use of such transgenic models in regulatory toxicology. This presentation reviews the current state of the science and its application to regulatory issues.

Selection of mouse models of cancer is often based simply on availability of a mouse strain and a known compatible tumor. Frequently this results in use of tumor models long on history but short on homology and quality control. Other factors including genetics, sex, immunological status, method and site of tumor implantation, technical competence, biological activity of the tumor, protocol sequence and timing, and selection of endpoints interact to produce outcomes in tumor models. Common reliance on survival and tumor burden data in a single mouse model often skews expectations towards high remission and cure rates; a finding seldom duplicated in clinical trials. Inherent limitations of tumor models coupled with the advent of new therapeutic targets reinforce need for careful attention to design, conduct, and stringent selection of in vivo and ex vivo endpoints. Preclinical efficacy testing for anti-tumor therapies should progress through a series of models of increasing sophistication that includes incorporation of genetically engineered animals, and orthotopic and combination therapy models. Pharmacology and safety testing in tumor-bearing animals may also help to improve predictive value of these models for clinical efficacy. Trends in bioinformatics, genetic refinements, and specialized imaging techniques are helping to maintain mice as the most scientifically and economically powerful model of malignant neoplasms.

The objective of molecular profiling of cancer is to determine the differential expression of genes and proteins from human tissue in the progression from normal precursor tissue to preneoplastic tissue to cancer in order to discover diagnostic, prognostic, and therapeutic markers. With the development of high-throughput analytical techniques such as microarrays and 2-D PAGE as well as the development of tools for cell procurement from histological sections such as laser capture microdissection (LCM), it is now possible to perform molecular analyses on specific cell populations from tissue. Since recognition of specific cell populations is critical, there is a need to optimize fixation and embedding not only to improve preservation of biomolecules, but also to maintain excellent histology. We have shown that 70% ethanol fixation of prostate tissue improves the recovery of DNA, RNA, and proteins over routine formalin fixation and maintains histological quality comparable to formalin. There is also a need to develop new technologies in order to expand the range of tissue types that can be analyzed. The development and applications of Layered Expression Scanning (LES) for the molecular analysis of whole tissue sections are discussed.

Toxicology and the practice of pathology are rapidly evolving in the postgenomic era. Observable treatment related changes have been the hallmark of toxicology studies. Toxicogenomics is a powerful new tool that may show gene and protein changes earlier and at treatment levels below the limits of detection of traditional measures of toxicity. It may also aid in the understanding of toxic mechanisms. It is important to remember that it is only a tool and will provide meaningful results only when properly applied. As is often the case with new experimental tools, the initial utilization is driven more by the technology than application to problem solving. Toxicogenomics is interdisciplinary in nature including at a minimum, pathology, toxicology, and genomics. Most studies will require the input from the disciplines of toxicology, pathology, molecular biology, bioinformatics, biochemistry, and others depending on the types of questions being asked.

While classical histopathologic approaches are invaluable in classifying tumors and understanding aspects of cellular interactions, genomic approaches provide a means to molecularly dissect tumorigenesis. The relationship of gene expression to the development of neoplasia remains an area of intensive research. With the advent of large-scale genomic platforms, alterations in gene expression can be related to the morphological development of cancer. The feasibility of using large-scale genomic analysis platforms has dramatically changed the landscape of biological sciences, as cellular processes must be considered in the context of complex networks. Alterations in gene expression must now be understood in a systems approach in which the relationships between genes expression changes are studied by considering the interplay of multiple regulatory networks. Ultimately, such changes must be understood at the protein level. We have begun to apply this technology to determine changes in gene expression that differentiate various types of mammary cancers that arise in mouse models that have been initiated by different genetic alterations. Ultimately, a molecular catalogue of similarities and differences between rodent and human tumors can be created which will serve to validate or credential particular models for specific experimental purposes, such as preclinical testing. These approaches have led to new insights into molecular pathways involved in oncogenesis, new classifications of human breast cancer, and the identification of new genes that may be relevant to understanding and treating human cancer.

As many as 5% of human cancers appear to be of hereditable etiology. Of the more than 50 characterized familial cancer syndromes, most involve disease affecting multiple organs and many can be traced to one or more abnormalities in specific genes. Studying these syndromes in humans is a difficult task, especially when it comes to genes that may manifest themselves early in gestation. It has been made somewhat easier with the development of genetically engineered mice (GEM) that phenotypically mimic many of these inheritable human cancers. The past 15 years has seen the establishment of mouse lines heterozygous or homozygous null for genes known or suspected of being involved in human cancer syndromes, including APC, ATM, BLM, BRCA1, BRCA2, LKB1, MEN1, MLH, MSH, NF1, TP53, PTEN, RB1, TSC1, TSC2, VHL, and XPA. These lines not only provide models for clinical disease and pathology, but also provide avenues to investigate molecular pathology, gene-gene and protein-tissue interaction, and, ultimately, therapeutic intervention. Possibly of even greater importance, they provide a means of looking at placental and fetal tissues, where genetic abnormalities are often first detected and where they may be most easily corrected. We will review these mouse models, examine their usefulness in medical research, and furnish sources of animals and references.

The biotechnology revolution offers unprecedented opportunities for identification of mechanistically-based biomarkers that report and predict cancer and other pathologies. The combination of genomic technologies with a knowledge of gene sequence and sequence conservation has made available markers that facilitate the correlation of genetic variation with biological outcomes, and “-omic” technologies allow efficient biochemical characterization of functional pathways—providing new markers of the susceptibility of individuals to cancer development, and of tumor susceptibility to specific therapies. New therapeutic agents targeted to individuals with specific genetic or biochemical characteristics already exist. The powerful -omic technologies allow efficient monitoring of gene transcripts, proteins, and intermediary metabolites, making it possible to monitor a large number of key cellular pathways simultaneously. This has enabled the identification of key biomarkers and signaling molecules associated with cell growth, cell death, and cellular metabolism. These new markers are facilitating monitoring of functional disturbance, molecular and cellular damage, and damage-response. Improved imaging technologies have made it feasible to image some of these molecular events noninvasively. To meet the challenge of evaluating and developing consensus criteria for the application of these new technologies and biomarkers, consortium approaches are being increasingly undertaken to share resources and to build a common understanding among the research, industry, and regulatory communities. These developments promise more efficient pharmaceutical product development, safer and more efficacious drugs, and provide clinical practitioners with new and better biomarkers for cancer screening, patient monitoring, and choice of therapy.

Successful development and use of biomarkers will improve the productivity of oncology drug development. Recognition of the importance of biomarkers for speeding drug development is reflected in the precise definitions and concepts proposed by an NIH Working Group to standardize terminology and promote a more coherent and systematic approach to the development and use of biomarkers. Potential clinical biomarkers of drug efficacy are often identified through pre-clinical studies or basic research. Identification of potential biomarkers for use in oncology is moving rapidly forward through continuing advances in clinical imaging technologies, especially molecular and functional imaging. Other rapid advances are a product of the growing availability of new scientific reagents for established technologies and of high-throughput genomic and proteomic technologies that can generate hundreds of potential biomarkers for further evaluation. In certain cases, conventional clinical diagnostic techniques or assays can be adapted for use in pre-clinical models to evaluate their ability to serve as biomarkers for predicting clinical responses to new drug candidates. Evaluation (pre-clinical and clinical) of a potential biomarker is often the longest stage of biomarker development, and standards for evaluation or validation depend on the intended use and stage of clinical development. Biomarkers verified for use in preclinical studies can be used to help select appropriate animal models and lead compounds. Biomarkers verified for use in clinical trials can confirm a drug's pharmacological or biological mechanism of action, guide protocol design, aid patient and dose selection, and help to minimize safety risks. Oncology drug development can be optimized by using a tiered set of clinical biomarkers that predict compound efficacy and safety with increasing confidence at each rise in tier thereby aiding corporate decision-making about advancing compounds. In oncology, a special class of extensively evaluated biomarkers of efficacy (surrogate endpoints) that generally correlate with desired clinical outcomes can be used as a basis for corporate decisions as well as for gaining accelerated provisional regulatory approval of a drug.

An intended consequence of the significant investment to characterize the mammalian genome, and its alterations in neoplastic diseases, has been the discovery and commercialization of new approaches to anticancer therapy. As a result, the list of available targets, formerly directed against the processes of DNA replication and mitosis, or in hormonal regulation of growth of tissue, has been dramatically extended. Many of the newer targets represent normal or aberrant signaling pathways, and are present within the cancer cell as second messengers, on its surface (as receptors), or in the external milieu (as ligands). These targets are, therefore, important to the cancer cell phenotype, and affect proliferation, differentiation, and death options for the cell. Another group of targets encompass the cancer cell's relationship to the tissue environment (both stroma and nonneoplastic cells). These targets involve interactions with blood supply (angiogenesis), immune function (evasion), and matrix (invasion and metastasis). Thus, most targets are directly or indirectly critical to some aspect of cancer cell physiology, although a few additional targets are being approached as localization signals for the delivery of otherwise less specific chemotherapeutic or radiotherapeutic agents. The development of new therapeutic approaches has expanded the scope of research required to characterize the mechanism of action for anticancer agents in preclinical models and in clinical trials. In preclinical research, mechanism of action studies have supported the selection of therapeutic agents, appropriate models of efficacy, and experimental design, as well as rational characterization and prediction of nontumor (host) effects. In clinical research, mechanism of action studies have supported the identification of surrogate markers of efficacy (critical for determining adequacy of dose and latency of response), and the selection of patient subpopulations and tumor subtypes most likely to exhibit clinical responses. Finally, information on mechanism of action may suggest strategies for combination therapies and predict potential mechanisms of disease resistance to therapy.

Proteomics is more than just generating lists of proteins that increase or decrease in expression as a cause or consequence of pathology. The goal should be to characterize the information flow through the intercellular protein circuitry which communicates with the extracellular microenvironment and then ultimately to the serum/plasma macroenvironment. The nature of this information can be a cause, or a consequence, of disease and toxicity based processes as cascades of reinforcing information percolate through the system and become reflected in changing proteomic information content of the circulation. Serum Proteomic Pattern Diagnostics is a new type of proteomic platform in which patterns of proteomic signatures from high dimensional mass spectrometry data are used as a diagnostic classifier. While this approach has shown tremendous promise in early detection of cancers, detection of drug-induced toxicity may also be possible with this same technology. Analysis of serum from rat models of anthracycline and anthracenedione induced cardiotoxicity indicate the potential clinical utility of diagnostic proteomic patterns where low molecular weight peptides and protein fragments may have higher accuracy than traditional biomarkers of cardiotoxicity such as troponins. These fragments may one day be harvested by circulating nanoparticles designed to absorb, enrich and amplify the diagnostic biomarker repertoire generated even at the critical initial stages of toxicity.

Cell signaling is a term used to describe a complex interactive system of signals that act to regulate or mediate a cellular response. Therapies that target cell signaling pathways have the potential to effectively reverse molecular deregulation underlying disease. The inherent complexity of cell signaling presents a major challenge to designing such therapies however, because perturbation of pathways has the potential to produce dramatic adverse effects. Pathologists are in the primary position of detecting adverse responses in drug development and are essential members of teams whose goal is to determine the mechanisms underlying tissue responses. The pathologist therefore will be expected to integrate morphologic interpretation with data obtained from several laboratory-based methods and data derived from novel technologies. Approaches being used include several in silico tools that provide access to public databases and signal pathway visualization that can serve to focus on key mechanistic hypotheses. The main objective of this article is to discuss a basic mechanistic approach and methods that can be used to associate modulation of cell signaling pathways with pathologic endpoints. The approach suggested begins with diagnostic pathology and uses global gene expression analysis in conjunction with transcription factor profiling and confirmatory protein technologies, to elucidate pathways relevant to the biological mechanism. Another important objective is to highlight the use of in silico technologies to prioritize laboratory efforts and focus these efforts on key hypotheses.

With the vast technological and informational resources increasingly available from investments in “genomics,” toxicology and much of biological science, is faced with previously undreamed of opportunities and equally daunting challenges. The ability to generate the large quantities of data becoming routinely available could not be imagined a decade ago. The complexities of data analysis are increasingly the rate-limiting element in scientific advances. The expectations that these large scientific investments will reduce the incidence of human disease and improve health are very high. An emphasis on genetic variation and Toxicogenetics is expected to yield risk estimates for specific rather than average individuals and individuals with varied lifestyles and complex patterns of exposure. Examples from studies of polymorphic variation in DNA repair genes in the healthy population and cancer risk highlight the complexity and challenges of incorporating genetic variation into quantitative estimates of risk associated with environmentally relevant exposures. Similar issues exist in selecting the animal models most appropriate for predicting human risk from environmental exposures to toxic agents.