5
BIOLOGIC MARKERS IN EXTRAPOLATION

Extrapolation is common in many scientific disciplines—so common that it can easily go unrecognized. Extrapolation is concerned with the translation or transfer of relationships (e.g., clinical measures and mathematical variables) observed in one setting to another setting. When we extrapolate from experimental systems to humans, we assume that some cause-effect relationships are the same in humans as in the experimental systems. The more we understand about the variables in an experimental situation, the more we will understand about the validity of such extrapolation to human situations. In the context of biologic markers and urinary toxicity, our goal is to gain a better understanding of the overall relationship between exposure and disease by examining those markers. The relationships among markers that can be observed and tested in experimental systems might be extrapolatable from those systems to situations of concern with respect to human health risks, such as occupational or environmental exposures to urinary toxicants.

The main purpose of extrapolation in any context is to make it possible to predict. In the clinical setting, the observation of particular symptoms in a patient leads clinicians to conclude that the patient has a particular disease or will soon manifest other symptoms. In that case, the clinicians are extrapolating from their experience with some patients to a new patient. Extrapolation is required to support prediction and the design of a suitable treatment.

Extrapolation from epidemiologic studies is commonly used to predict risks to other cohorts. Every epidemiologic study is restricted to some population. To extrapolate from a study population to other potentially affected populations, one must consider the differences between the study population and the potential target populations. Variables related to and affecting the development of the health effect under investigation must be considered; differences between the study and target populations with respect to those variables will influence how the extrapolation is completed by influencing or modifying



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 153
Biologic Markers in Urinary Toxicology 5 BIOLOGIC MARKERS IN EXTRAPOLATION Extrapolation is common in many scientific disciplines—so common that it can easily go unrecognized. Extrapolation is concerned with the translation or transfer of relationships (e.g., clinical measures and mathematical variables) observed in one setting to another setting. When we extrapolate from experimental systems to humans, we assume that some cause-effect relationships are the same in humans as in the experimental systems. The more we understand about the variables in an experimental situation, the more we will understand about the validity of such extrapolation to human situations. In the context of biologic markers and urinary toxicity, our goal is to gain a better understanding of the overall relationship between exposure and disease by examining those markers. The relationships among markers that can be observed and tested in experimental systems might be extrapolatable from those systems to situations of concern with respect to human health risks, such as occupational or environmental exposures to urinary toxicants. The main purpose of extrapolation in any context is to make it possible to predict. In the clinical setting, the observation of particular symptoms in a patient leads clinicians to conclude that the patient has a particular disease or will soon manifest other symptoms. In that case, the clinicians are extrapolating from their experience with some patients to a new patient. Extrapolation is required to support prediction and the design of a suitable treatment. Extrapolation from epidemiologic studies is commonly used to predict risks to other cohorts. Every epidemiologic study is restricted to some population. To extrapolate from a study population to other potentially affected populations, one must consider the differences between the study population and the potential target populations. Variables related to and affecting the development of the health effect under investigation must be considered; differences between the study and target populations with respect to those variables will influence how the extrapolation is completed by influencing or modifying

OCR for page 153
Biologic Markers in Urinary Toxicology the relationships between variables and the health effect. Consider the situation of a new chemical that is proposed for use or that will be a byproduct of some new operation. Direct evidence that the chemical causes adverse effects in humans is lacking. Some important considerations include the potential for adverse health effects in humans exposed to the chemical and, if adverse effects do result from exposure, the magnitude of the effects after exposures of different severities. It is clear that predictions are required. However, the basis for the predictions cannot be the previous human experience; there is no previous experience. It might be possible to extrapolate, but the model from which the extrapolation is made will of necessity be a nonhuman model. In the scenario just described, the need for extrapolations from experimental systems to humans is apparent. Many other scenarios, both clinical and "population-based," will require the prediction of human responses from data obtained in nonhuman test systems, especially in light of the thousands of chemicals that are produced, used, and released into the environment as byproducts of our way of life. Indeed, it is the desire to be predictive that drives the need to develop and apply good experimental systems. Such systems have at least four advantages: they allow predictions of human health effects and the magnitude of those effects before human exposure occurs or before adverse effects are manifested in exposed populations; they can be altered to clarify aspects of the process leading from exposure to adverse health effects when similar experimentation in humans would be unethical; they can be designed to eliminate many factors that confound the determination of cause-effect relationships in epidemiologic studies; and they can suggest directions for epidemiologic investigation by providing the hypotheses that epidemiologic studies might be able to test. Previous chapters have focused on markers of susceptibility, exposure, and effect (particularly early effect) and their value in clinical situations; the sooner a disease state or precursor of a disease state can be identified, the greater the chance of successful therapy or treatment. This chapter focuses on the prediction of effects, not in an individual patient but rather in a (hypothetical) population of humans potentially exposed to a supposed toxicant. In this context, one is concerned about maintaining the health of the population by predicting whether an activity or an exposure is likely to produce harmful consequences in that population—often without previous observations of humans exposed at the magnitudes of interest. The objective is to learn how to tie chemical exposure under various scenarios to the dose or amount of the chemical that reaches the body, to the amount that is absorbed and distributed to target tissues, and ultimately to the effect. How does one make such predictions? The discipline of risk assessment addresses that question. Human-health risk assessment is a complex,

OCR for page 153
Biologic Markers in Urinary Toxicology multifaceted process that relies on data and scientific principles from many disciplines to determine whether a chemical is toxic and the likelihood of manifestation of detrimental effects under specific conditions. Risk assessment draws on a variety of methods and models to examine and evaluate information about the toxicity of a chemical. A useful and well-described system for looking at toxicity information is the risk-assessment paradigm originally depicted by the National Research Council in the 1983 publication Risk Assessment in the Federal Government: Managing the Process. That system for organizing and analyzing risk information follows an understandable series of steps portraying qualitative and quantitative aspects and typically includes some or all of the following: hazard identification, dose-response assessment, exposure assessment, and risk characterization (NRC, 1983). The result is a characterization of the potential adverse health effects of human exposures to a chemical. The characterization of potential adverse human health effects requires extrapolation. That typical of many risk assessments includes extrapolation from animal test species to humans, from large exposure to small exposure, and from one route of exposure to another. Risk assessments are not infallible, and the relative accuracy of a risk assessment depends not only on the scope and quality of the scientific data but also on the reliability of the methods and the validity of the models used. The degree of confidence in a risk assessment depends on how well data and model quality are validated and on the extent to which uncertainty is quantified. As described in Chapter 2, various aromatic amines are recognized human bladder carcinogens (NRC, 1981). Consider, for example, an assessment of human bladder-cancer risk associated with dermal exposure to one of those aromatic amines—4,4'-methylene-bis(2-chloroaniline), or MOCA. For such an assessment, we might be required to extrapolate relationships that were observed in dogs exposed to MOCA in their diet (Stula et al., 1977) at doses far greater than the human exposures of interest. The conclusion that the occurrence of bladder cancers in dogs exposed to MOCA implies a bladder-cancer risk for humans assumes that the qualitative relationship between MOCA exposure and bladder cancer can be extrapolated from dogs to humans. Such cross-species extrapolation is typical of the hazard-identification component of risk assessment, that is, the determination of the existence of a cause-effect relationship between chemical exposure and adverse health effect. Mathematical extrapolation is particularly relevant to risk assessment of a most useful kind, i.e., quantitative risk assessment. Quantitative risk assessment is a means of providing a measure of the risk of some harm as a result of a specific exposure to some substance or activity (Almeder and Humber, 1987), and mathematical extrapolation can be conceived of as the transfer of the quantitative relationships estimated in one scenario to another scenario, whether

OCR for page 153
Biologic Markers in Urinary Toxicology those scenarios differ in species (animal-to-human extrapolation), in magnitude of exposure (dose extrapolation), or in route of exposure (route-to-route extrapolation). Again, it is assumed, lacking data to the contrary, that a relationship observed in one scenario is valid in the other. Risk can be stated as the magnitude of exposure that is estimated to be without substantial likelihood of harmful effect, or stated as the probability of occurrence of harmful effect. The remainder of this chapter concerns answers to two questions: What is the basis for concluding that extrapolation is reasonable? Can it be said that one approach to extrapolation is better than another approach? It is in relation to the second question that the relevance and utility of biologic markers in risk assessment become apparent. BASIS OF EXTRAPOLATION Animal Studies A fundamental principle of toxicology is that results of animal studies can be applied to humans. The scientific basis for assuming that animals are good surrogates for humans and therefore a suitable basis for extrapolation to humans is overwhelming. If one considers that the genetic makeup of a mouse or a rat is more than 95% and of a monkey is more than 99% identical with that of a human, it is reasonable to assume that these animals in particular and mammals, in general, will react to infectious agents and chemical stressors much as humans will. Among mammals, most of the host defense mechanisms (barrier and immune) and metabolic (anabolic and catabolic) systems are similar. In particular, the urinary systems of most mammals are very similar. Although specific, often subtle, differences between humans and other animals with respect to renal function have been demonstrated, the vast majority of human renal responses to xenobiotics mimic what has been observed in other species. Biologic markers of renal transport, concentrating, and metabolic functions of the kidney are reproduced in many species, including humans, although quantitative differences have been demonstrated. Therefore, it is reasonable to use animal models for extrapolation to humans unless specific information on specific chemicals dictates otherwise. Many epidemiologic investigations have, in fact, been suggested as a result of animal studies, and the epidemiologic findings have tended to support the results of the animal studies. For example, several of the current epidemiologic studies of heavy-metal toxicity, including small exposures, were initiated in response to urinary toxicity observed in a large number of animal studies and were undertaken specifically because of the likelihood that the human response would mimic that seen in animal test species. Mechanistic studies of chemically induced nephrotoxicity in animals influenced or stimulated epidemiologic studies of a variety of substances, including many of the halogenated hydro-carbons, such as chloroform, hexa-

OCR for page 153
Biologic Markers in Urinary Toxicology chlorobutadiene, and bromobenzene. Much of the epidemiologic investigation and mechanistic understanding of anesthetic, analgesic, and antibiotic nephropathy has been driven by observations of nephrotoxicity in animals. Classic examples of the direct application of animal studies to epidemiology of known nephrotoxicants are the use of antineoplastic and immunosuppressant drugs (e.g., cisplatin and cyclosporin, respectively). Studies of carcinogenic responses in laboratory animals and in humans have revealed substantial correlations (Crump et al., 1989). The results suggest that there are reasonable approaches to extrapolating cancer responses observed in test species—approaches that appear to predict fairly well the responses in humans. Identification of chemical hazards should include assimilation and evaluation of all relevant information. Appraisal of physical and chemical properties and structure-activity relationships can sometimes provide important indications of potential toxic characteristics. Markers of urinary function and chemical toxicity in experimental animals can be studied at various levels of tissue structure and organization (Table 5-1). This is in contrast with human studies, in which only noninvasive studies of renal function are possible. Markers identified through in vitro studies of systems that use animal and human cells or tissues in culture can often give insight into potential toxicity. However, because of the intricacy of the body, only whole-animal studies or observations in humans provide information on the operation of multiple cells, tissues, and organs under the influence of complicated feedback mechanisms. Animals are necessary in the study of chemical-induced toxicity and for the development and validation of markers because studies that involve modulation of cellular responses and tissue sampling cannot be performed in humans. Traditionally, experimental animal Studies have been of most value for identifying markers that can be used to predict target-organ effects, understand dose-response relationships, and study mechanisms of action. The studies include measurements of function, blood chemistry, urinalysis (including cytology), histopathology, and electron microscopy. Metabolism and transport peculiar to the kidney are often routine parts of such studies. Animal models developed as surrogates for humans in the study of renal and urinary function should conform to some general principles, which are applicable to any organ system, although they are discussed here in the context of renal and urinary toxicology. They include the following: The animal model should be reproducible within and among laboratories. It should not be so complex that only a few laboratories could do the study. The model should be peculiar to the part of the urinary tract under consideration. This characteristic can be realized only with sophisticated procedures that permit study of discrete nephron segments.

OCR for page 153
Biologic Markers in Urinary Toxicology The model should be sensitive enough to differentiate normal from abnormal changes in structure or function. The model should be able to measure alterations in renal structure or function caused by exogenous agents. Whole-animal studies are usually the first step in evaluating the potential toxicity of a given agent, and these studies involve assessment of urine and plasma for markers indicative of organ function and toxicity. Noninvasive or nondestructive studies can be followed by application of histopathologic techniques to determine markers of target-organ or tissue-site injury. In most cases, rodents suffice, but there might be instances when only a primate can properly represent the human situation. For example, the route of administration might be important if direct extrapolation to humans is likely. With nephrotoxicants as with other toxicants, acute, subchronic, and chronic exposures are used to determine potential toxicity. The determination of markers of urinary toxicity is generally easier with acute protocols than with subchronic or chronic exposures. Studies often involve single exposures of both sexes of at least two species, usually rodents. Depending on the results of the rodent studies and the questions being asked, one might decide to study the agent in higher mammals, such as dogs or monkeys. The use of subchronic or chronic exposure regimens is usually driven by the nature of the potential human exposure, the agent being studied, and the possibility of chronic toxicity, including carcinogenicity. Renal Parenchymal Injury The difficulties in diagnosing renal injury and predicting its health consequences are considerable. That is primarily because the kidney can undergo substantial chemically induced injury without any clinical manifestation; subtle injury can be negligible, given the considerable functional reserve of the kidney. For example, the single cross-sectional measurement of glomerular filtration rate (GFR) might show only severe acute or chronic renal damage, as discussed in Chapter 2. Most studies indicate that quantitative urinary-enzyme secretion patterns cannot reveal either the type or the severity of renal injury, and often they do not correlate with structural or functional changes, as discussed in greater detail in Chapter 4. The need, therefore, is for standard diagnostic criteria that are sensitive enough to serve as markers of renal damage in the presence of renal functional reserve. Much of the nephrotoxicity that follows the administration of inert, relatively nontoxic chemicals is related to the formation of reactive electrophiles during their metabolism (Ford and Hook, 1984). It is thought that the electrophilic products can react covalently with various nucleophilic sites on renal macromolecules and, by some mechanism yet to be defined, lead to renal damage. Measurement of the re-

OCR for page 153
Biologic Markers in Urinary Toxicology active electrophiles or the covalently bonded compound with sensitive techniques might yield markers of renal damage (Harris et al., 1987; Omichinski et al., 1987; Reddy et al., 1984; Tyson and Mirsalis, 1985). However, those procedures require renal tissue; although they might yield useful markers of renal damage in experimental studies, they are not suitable for use in humans exposed to potentially electrophilic products of nontoxic chemicals. Glomerular Filtration Rate Evaluation of the blood-urea nitrogen concentration (BUN) is a common procedure for the indirect assessment of the GFR in experimental animals. As discussed in Chapter 4, it is unsuitable for quantitative purposes but might have utility in establishing the course of chronic renal failure in an experimental setting if renal damage is severe enough. Measurement of the serum creatinine concentration and urinary creatinine excretion, with calculation of the creatinine clearance, is generally preferred as an indicator of GFR. However, in some animal models, variable amounts of creatinine can be excreted via tubular secretion, and that reduces its utility as a marker of GFR. More subtle changes in GFR can be assessed by evaluating the clearance of various exogenous substances, such as inulin, EDTA, and iodothalamate. Sensitive analytic procedures are available for measurement of those markers and GFR. Again, however, the extent of reduction of GFR in the face of a nephrotoxic insult might be hidden by the inherent renal reserve, and even measurements of GFR often are not sensitive enough to detect modest renal damage. GFR can be assessed in either conscious or anesthetized animals. In both cases, the same markers can be used and their clearance determined with standard renal physiologic techniques. Both creatinine and inulin are used commonly to determine GFR. The use of anesthetized animals permits a more accurate determination of urinary flow than the use of conscious animals housed in metabolism cages. The use of anesthetized animals also permits the collection of precisely timed blood samples for determination of the marker under study. However, if conscious animals are used, GFR can be determined with reasonable accuracy with subcutaneous injection of the marker in a concentrated gelatin solution and collection of a single blood sample at the end of a 60-mill urine-collection period. Some researchers have suggested that the anesthetic agents by themselves can alter renal function. Tubular Function Tubular dysfunction in experimental animals can be assessed through relatively simple and inexpensive tests, such as those for the measurement of glucosuria, enzymuria, and osmolality. Some are sensitive enough to detect relatively small effects on the kidney after acute

OCR for page 153
Biologic Markers in Urinary Toxicology administration of a nephrotoxicant, but caution must be exercised in predicting specific effects on transport processes or cell viability on the basis of the data obtained from these types of in vivo tests (Berndt, 1981). Some researchers (Daugaard et al., 1988; Dieperink et al., 1983) have used the renal clearance of lithium as a more subtle technique for evaluating renal damage that occurs during chronic studies; this noninasive method is applicable to humans (Thompson et al., 1984), as well as to animals. The loss of renal tubular function can also be assessed with test conditions that impose stresses on renal function, e.g., maximal urinary dilution or concentration or urinary acidification or alkalinization. Similarly, tests of maximal tubular reabsorption of glucose or maximal tubular secretion of p-aminohippurate (PAH) can be valuable in assessing tubular damage. They can also be applicable to humans and yield relatively sensitive markers of renal damage. However, these approaches require carefully controlled experimental studies and are not suitable for casual observations in the workplace. Proteinuria Proteinuria is the appearance of proteins in the urine after increase in the permeability of the glomerular membranes, reduction in tubular reabsorption of filtered proteins, shedding of specific constituents into urine as a consequence of cellular turnover or selective renal tubular damage, or a combination of the above. Glomerular or tubular damage can occur in the absence of a substantial reduction in GFR, so it has long been thought that the evaluation of proteinuria can be useful in detecting renal dysfunction at either the glomerular or tubular level. Although this topic is discussed in considerable detail in Chapter 4, a few comments concerning proteinuria. as a marker of renal dysfunction are incorporated here for the sake of completeness. Although one can measure total protein in urine, it is a relatively insensitive assessment of renal damage. Total-protein measurement also offers no insights into whether one is assessing damage to glomerular membranes or to tubular membranes. A more rational approach is the use of electrophoretic separation of single proteins to provide a comprehensive approach to chemically induced renal dysfunction. Proteins of relatively high molecular weight (over 45,000 daltons) usually are retained in the vascular compartment by the various glomerular membranes. Those membranes also serve as charge discriminators and tend to retain negatively charged proteins in the plasma compartment as well. Proteins of low molecular weight pass the glomerular barrier with various degrees of efficiency and are later (more or less efficiently) taken up by proximal tubular cells. Indeed, the reabsorption of proteins that pass the glomerular membranes is very efficient; even slight decreases in tubular fractional reabsorption due to tissue damage increases the excretion of relatively low-molecular-weight proteins.

OCR for page 153
Biologic Markers in Urinary Toxicology Electrophoretic patterns of urinary proteins can reveal glomerular damage, tubular damage, or both. Careful assessment of electrophoretic patterns can reveal selective glomerular damage (loss of glomerular polyanion) or unselective damage (glomerular hyperfiltration). Studies in experimental animals must be undertaken cautiously because in some species large variations in urinary protein excretion can lead to incorrect conclusions. For example, sex-, age-, and diet-related changes that occur in mate rats are not related to glomerular damage (Neuhaus et al., 1981). Young male rats can show "tubular" proteinuria, whereas aging rats show "glomerular" proteinuria; the proteinuria in the first instance is essentially physiologic, and that in the second is attributable to spontaneous nephropathy, which can be controlled in part by reducing dietary protein. Tamm-Horsfall Protein Excretion of Tamm-Horsfall protein (discussed in detail in Chapter 4) is increased after damage to the distal part of the nephron and is decreased when the renal mass is reduced. Enzymuria Several investigators have used enzymuria as a marker of nephrotoxicity, but Dubach et al. (1989) have indicated that none of the enzymes studied experimentally satisfies all the criteria of a nephrotoxic response. Because renal enzymes are not distributed uniformly along the nephron, it might be possible to localize renal damage within the nephron on the basis of the pattern of enzymuria. The site selectivity of single enzymes is questionable. Other factors that complicate the use of enzymuria as a marker of renal dysfunction have been suggested. For example, early renal changes induced by chemicals might be less selective, in which case the predictive value of enzyme markers would be compromised. Many procedures for analyzing urinary enzymes are poor and, rather than pinpointing specific nephron sites, might give rise to nonspecific patterns that are difficult to interpret. Most urinary enzymes are stable only over a narrow pH range, and their activity in urine could be affected by inhibitors, some of which can alter urinary pH (Price, 1982). Thus, enzymuria studies in experimental animals must be carried out under very carefully controlled experimental conditions. Contamination of urine with food or microorganisms must be minimized, and urine must be collected in vessels that are then stored in ice (Berlyne, 1984). (See Chapter 4) Monoclonal Antibodies Other potential markers of renal damage are immunoreactive tissue constituents that are released into urine because of increase in cellular turnover or cell death. Those constituents can be detected immunochemically, and monoclonal antibodies have been produced

OCR for page 153
Biologic Markers in Urinary Toxicology against both rat (Tokoff-Rubin, 1986) and human (Mutti, 1989; Mutti et al., 1985) brush-border antigens. The selectivity of these markers for identifying specific nephron segments is still debated, although experimental results suggest that they can be useful. The earlier studies suggested that the BB-50 brush-border antigen was also localized in peritubular capillaries; Mutti et al. (1988) identified a monoclonal antibody that reacted with an antigen that was peculiar to the brush border—the so-called brush-border antigen. Monoclonal antibodies to the S3 segment of the nephron, where alkaline phosphatase is, also have been produced (Verpooten et al., 1989). They might prove useful as markers of the effects of chemicals, such as mercury, that act selectively on the straight part of the proximal tubule. Bladder Toxicity Xenobiotics On the basis of results of long-term carcinogenicity studies of 358 xenobiotics (Barrett and Huff, 1991), the bladder is among the 10 most prevalent sites of cancer development in rodents. According to histopathologic findings after chronic exposure, 16 xenobiotics (4%) caused bladder tumors in at least one sex of either rats or mice. Histologic evaluation of rat bladders after various doses and durations of exposure to 4-butyl-(4-hydroxybutylnitrosamine) (BBN) and N-(4-[5-nitrofuryl]-2-thiazolyl) formamide (FANFT) demonstrated a series of changes in structure that are good models of the changes noted in human bladder cancer. A considerable amount of information based on those models and the results obtained with 2-acetylaminofluorene (2-AAF) is available on the potential of xenobiotics to influence the development of bladder cancer (Ito et al., 1989; Soloway and Hardeman, 1990; Staffa and Mehlman, 1980). It should be noted, however, that there is not complete concordance across species, even for the genotoxic bladder carcinogens. That fact, by itself, makes extrapolation to humans difficult. The text that follows should be viewed in this light. For example, BBN produces tumors at a lower rate in mice than in rats, and at a lower rate in hamsters than in mice, and guinea pigs do not develop bladder tumors after exposure to BBN (Hirose et al., 1976). Furthermore, in the long-term carcinogenicity studies, 10 chemicals produced bladder tumors in rats, but only six were associated with tumors in mice (Barrett and Huff, 1991). It is also noteworthy that the first xenobiotic (or group of xenobiotics) shown to induce bladder cancer in humans, the aromatic amines (see Rubber et al., 1985, for discussion), readily induced bladder cancer in dogs but did not induce bladder tumors in rats until massive doses were given chronically by oral gavage (Hicks et al., 1982). In fact, it has been stated that the failure of the rat bladder to respond to the aromatic amines was one factor that led to the use of the maximum tol-

OCR for page 153
Biologic Markers in Urinary Toxicology erated dose (MTD) in rodent bioassays (Wiseburger, 1992). The capacity of chemicals and of various physical injuries to induce hyperplasia in rat bladders has been measured by histopathologic means (Ito et al., 1989) and on the basis of increased DNA replication and increased fresh and dry organ weight (Anderson, 1991; Cohen and Ellwein, 1991; Ito et al., 1989). When those techniques have been applied after the same treatments, they yielded similar results and appeared to be equally valid ways to measure cell division in the bladder. The use of the techniques in humans is questionable because they depend on using the isolated organ. It has been reported that one can ascertain the potential of a chemical to induce bladder damage by determining its ability to enhance concanavalin A's agglutination of bladder cells from rats treated with the chemical (Kakizoe et al., 1981). That technique is reportedly capable of distinguishing between complete carcinogens and tumor promoters (R.L. Anderson, Procter and Gamble, unpublished material, 1987). Complete carcinogens increase cell agglutination directly. Tumor promoters do not produce the response when given alone, but they can sustain the response in animals that were first exposed to a known initiator carcinogen. Extracellular Calcium Studies with nitrilotriacetate (NTA), a nongenotoxic compound that causes bladder tumors in rats but not in mice, have demonstrated that high doses of this metal-chelating chemical cause an increase in urinary calcium and a coincident decrease in bladder-tissue calcium in rats (Anderson and Alden, 1989). That state is accompanied by the presence of crystalline calcium-sodium NTA in collected urine. Uncomplexed NTA in urine extracts calcium from the urothelial extracellular pool more rapidly than it can be replenished from the circulation. The removal of extracellular calcium reduces cell-cell contact in the urothelium and results in increased cell loss and increased urothelial replication. If this process is continued chronically, it can result in urothelial tumors (at a low rate). Only one other rat-bladder carcinogen, terephthalic acid, which forms calcium terephthalate crystals in urine, is known to cause bladder tumors by the mechanism demonstrated for NTA (Chin et al., 1981). Attempts to demonstrate a broader base for this mechanism with other treatments known to induce bladder tumors have not been successful (Anderson, 1991). Zinc There are few data to support the notion that tissue concentration of zinc is related to the development of bladder cancer. What data are available are the result of a single study with short-term exposure (four weeks) to BBN that demonstrated that the treatment caused increased bladder-tissue weight and tissue zinc without a change in any other mineral (Anderson et al., 1986a). It should

OCR for page 153
Biologic Markers in Urinary Toxicology al., 1992), characterization of chromosomal deletions induced by bladder carcinogens in SV40-immortalized human uroepithelial cells (Meisner et al., 1988; Reznikoff et al., 1988; Wu et al., 1991), and characterization of second-messenger mechanisms and the carcinogenic process in SV40-immortalized human uroepithelial cells (Jacob et al., 1991). Chopra and colleagues (Chowdhury et al., 1989) have developed cell culture methods for prostatic epithelial cells from mouse. Cells were derived from the ventral prostate of normal adult mice. Primary cultures and serial propagation were performed in serum-free, hormonally defined media to optimize differentiation. The cells grew and exhibited tissue-specific markers, including prostatic acid phosphatase activity and prostate-specific antigen. The issue of markers of exposure and susceptibility, which is a major focus of this report, has not been addressed directly in the in vitro systems. Tissue samples, such as those used as starting material for human uroepithelial-cell cultures (Jacob et al., 1991; Meisner et al., 1988; Messing et al., 1988; Pratt et al., 1992; Reznikoff et al., 1983, 1986, 1988; Wu et al., 1991), can be readily obtained and analyzed for transformation. Cytogenetic analysis should reveal the presence of chromosomal changes that indicate exposure to a mutagenic or carcinogenic agent. For prostatic epithelial cells, examination of tissue-specific markers as described by Chowdhury et al. (1989) might be a useful means of detecting exposure of the cells to toxic agents. IMPROVED RISK-ASSESSMENT EXTRAPOLATION Biologic markers are the key to improving risk-assessment extrapolation. Both qualitatively and quantitatively, biologic markers are crucial in determining the best available basis for and approach to cross-species, low-dose, and route-to-route extrapolation. Relevance of Test Systems The major qualitative question facing a risk assessment concerns the relevance of an experimental system or animal model to the human situation. Simply stated, one can address this concern by examining the sequence of markers that occurs in humans—the sequence that is associated with exposure to a toxicant and that indicates progression to a disease state. If that sequence or a portion of it is observed in the test system, that system can be regarded as relevant for human risk estimation. Or examination of the markers, particularly markers of susceptibility, associated with disease progression in an animal model might indicate substantial similarities or dissimilarities to humans and thus help to determine relevance. In the case of renal tumors mediated by alpha2u-globulin (see Chapter 6), for example, the rat model might not be appropriate for assessing the risk of human renal tumors associated with some chemicals. Other examples of similarities and dissimilarities are discussed below. It is imperative that the relevance of proposed test sys-

OCR for page 153
Biologic Markers in Urinary Toxicology tems be established to enhance the scientific basis of risk assessment. Metabolism and Kinetics The concept of metabolism as a means by which chemicals are biotransformed to reactive and toxic species is central to an understanding of chemically induced nephrotoxicity. Mammalian kidneys have numerous enzymatic activities that can activate a diverse array of chemicals. An important concept is that many activation and detoxification pathways are present simultaneously and therefore compete for the same substrates. The balance between competing pathways is often the primary determinant of the ultimate biologic response. Knowledge of the biochemical regulation of the pathways and of the interorgan and intrarenal distribution of the various enzymes involved is necessary for correlation of in vitro data with the in vivo situation and of data from experimental animals with human risk assessment and development of markers of human exposure. The importance of bioactivation in determining toxicity is not a new idea, but it is critical to understand the prevalence of enzymatic activation reactions in chemical-induced toxicity (Miller and Miller, 1985). Many toxic or carcinogenic chemicals do not produce their effects directly but must be metabolized by cellular enzymes to generate reactive, electrophilic intermediates. The metabolites are responsible for the interactions with cellular components that lead to toxicity. Much of the target-organ specificity of many toxic chemicals is due to the tissue-specific distribution of activation pathways. Other important factors in determining risk are the tissue-specific patterns in types and concentrations of protective molecules, such as glutathione (GSH) and alpha-tocopherol; in activity of detoxification enzymes, such as catalase and GSH peroxidase; and in types of and activities of transport systems that provide access to intracellular sites. When the relative roles of all those factors have been assessed in an experimental species, one must consider how differences in one or more of them will alter susceptibility to injury. This point is central to realistic assessments of human risk because differences between humans and laboratory animals and between several laboratory animal species in activation and detoxification enzymes in specific tissues have been documented. As a consequence of qualitative and quantitative species differences in drug-metabolizing enzymes and transport activities, patterns observed and conclusions reached in one species might not be applicable to another laboratory animal species or, more important, to humans. Study of nephrotoxicants is complicated further by the pharmacokinetics and interorgan pathways involved in their disposition. Several chemicals are initially metabolized by the liver or other tissues and then, by enterohepatic and renal-hepatic pathways, reach the kidneys where they are processed further and cause nephrotoxicity. As de-

OCR for page 153
Biologic Markers in Urinary Toxicology scribed in Chapter 2, the kidneys are a frequent target of chemical toxicants because of high blood flow that reaches them; the large numbers, high activities, and overlapping substrate specificities of membrane transport systems; the high basal metabolic requirements; and the presence in renal epithelial cells of several activation enzymes, some of which are peculiar to kidneys (Anders, 1980, 1989; Commandeur and Verneulen, 1990; Jones et al., 1980; Lash et al., 1988; Rush and Hook, 1986; Rush et al., 1984). The mammalian kidney is very active in numerous pathways of drug metabolism, and renal metabolism plays a quantitatively important role in overall metabolism of a large number and variety of chemicals (Table 3-1). The kidneys are sometimes overlooked as important sites of drug metabolism for several reasons: the liver has a quantitatively large role in metabolizing a huge number of xenobiotics, the kidneys make up only 1–2% of total body weight and so are thought not to contribute importantly to metabolism, and there is pronounced heterogeneity in the distribution of enzymes in the cell populations that constitute the nephron, so detection of enzyme activities in a tissue homogenate or in a mixed population of renal cells can be difficult if the enzymes are localized to a discrete cell population. For example, activities of some isozymic forms of cytochrome P-450 in proximal tubular cells from the rat kidney are comparable with those in the rat liver (Commandeur and Verneulen, 1990; Jones et al., 1980). Some phase II enzymes, particularly those in the GSH conjugation pathway, are found at very high activities in the renal proximal tubule but are nearly absent in other regions of the nephron. Species and strain differences need to be evaluated for each chemical or class of chemicals. It needs to be determined whether the biochemical or physiologic responses obtained in test species will be the same as those in humans. Differences in isozymic forms or the presence or absence of metabolic pathways can have effects that vary from chemical to chemical. In some cases, the differences can produce entirely different pharmacokinetics and hence different biologic effects; in other cases, the species-dependent differences might produce only quantitative differences but not yield different overall biologic effects. Differences that are noted between responses in humans and in animal species do not necessarily invalidate an animal model, nor do they imply that an appropriate animal model is unavailable for a specific toxicant. Rather, the differences highlight that we do not have much mechanistic information about many urinary toxicants and that we have not yet developed an appropriate animal model. Species differences and resulting difficulties in extrapolating to humans also highlight the need for additional research. The mercapturic acid pathway (see Chapter 3) and associated activation and detoxification pathways illustrate many of the above principles. Although the GSH S-transferases, which catalyze the initial reaction in this pathway, are

OCR for page 153
Biologic Markers in Urinary Toxicology common, the succeeding enzymatic reactions show considerable variability. The variability occurs within a given species, with respect to specificity and activity, and occurs between species for a given tissue. The tissue distribution of γ-glutamyltransferase, which is the first enzyme in the pathways of GSH S-conjugate metabolism, is the primary determinant of the pattern of interorgan metabolism (Lash et al., 1988). This initial, hydrolytic reaction determines which metabolite reaches the target organ. Whether the GSH S-conjugate is metabolized to the cysteine S-conjugate or to the N-acetylcysteine S-conjugate determines how actively the protoxicant is transported into renal cells for further metabolism to a reactive intermediate. Because most biochemical studies on the metabolism and transport of GSH, GSH S-conjugates, and related metabolites have used rats, the view that has evolved from numerous investigations is based on the extremely high γ-glutamyltransferase activity of rat renal proximal-tubular brush-border membranes and the nearly complete absence of the enzyme in rat hepatic canalicular membranes (Lash et al., 1988). Liver and biliary epithelium of humans and other mammals, such as rabbits and guinea pigs, contain substantial γ-glutamyltransferase activity (Ballatori et al., 1988; Hinchman and Ballatori, 1990). In species that have higher activities of hepatic γ-glutamyltransferase, the liver will make a larger contribution and the kidneys a correspondingly smaller contribution to interorgan metabolism of GSH and GSH S-conjugates. For reactive and therefore toxic electrophiles that are metabolized by GSH conjugation, it is the final step in the pathway that converts the metabolite to a highly polar N-acetylcysteine conjugate that is readily excreted in urine. λ-Glutamyltransferase also shows marked species differences in velocity and substrate specificity and is absent in guinea pigs. Data on metabolite distribution and interorgan metabolism of GSH S-conjugates cannot be directly extrapolated from laboratory animals to humans without consideration of those potential differences. Therefore, although pathways studied in rats or other laboratory animals will occur in all mammals, the relative importance of each pathway will differ considerably from species to species. Mechanism of Action Not as much is known about the mechanisms of toxicity as is known about metabolism and kinetics. Furthermore, the relevance of some findings from animal studies to prediction of human risk is controversial because it is not clear in some cases that the mechanism or mode of action by which toxicity is produced in the animals emulates what occurs in humans. Nevertheless, in recent times, more sophisticated scientific data have provided useful information for postulating mechanisms of toxicity in test animals for many chemicals that are potentially toxic to humans. Understanding mechanisms of toxic-

OCR for page 153
Biologic Markers in Urinary Toxicology ity in animals can lead to new methods for estimation of human risk, although exactly how to interpret scientific advances is not always clear and so how to incorporate the information into risk assessment has not found general agreement. It is evident from the mechanistic data, however, that different risk-assessment methods might be used to model human risk associated with exposure to different chemicals. Progress in the elucidation of underlying mechanisms of toxicity indicates possible high-dose phenomena), whereas comparative biology suggests differences among species. In some cases, markers in animals might not be relevant to humans; in other cases, only indirect extrapolation from animal studies may be inappropriate. For example, there is considerable discussion among scientists regarding chemicals that induce cancer in rodent bioassays but do not exhibit classical genotoxicity and regarding the role of cell proliferation in the development of cancer, particularly for these ''nongenotoxic'' chemicals (Ames and Gold, 1990; Cohen and Ellwein, 1991). Whether events related to cell proliferation induced by high doses of a specific chemical are limiting in carcinogenesis and whether these events are likely to occur in humans is important in predicting the likelihood of carcinogenesis in humans exposed to low doses of the chemical. Most important, if cell proliferation, particularly in response to toxicity, is integrally involved in the induction of tumors, then only the dosages and mechanisms that cause toxicity will produce a proliferative response and result in tumor formation. Quantitative Issues The qualitative issues discussed above are related primarily to the hazard-identification component of risk assessment. Secondarily, consideration of those issues focuses attention on the experimental results suitable for quantitative extrapolation. Traditionally, quantitative risk assessment that has involved dose-response modeling (almost all of which has until recently been cancer risk assessment) has relied on extrapolation of the relationship between administered dose and observable disease outcome (e.g., cancer incidence). Lately, with the consideration of physiologically based pharmacokinetic models and biologically based dose-response models, that has begun to change. The role of biologic markers is fundamental to the progress that has been made and will be crucial to further progress. Consider the simplified flowchart of classes of biologic markers shown in Figure 5-1. The figure shows the sequence of markers paralleling the processes leading from exposure to clinical disease and, more important, indicates the relationships between the markers. The pictorial representation can be translated into semiquantitative form as follows: DI = f1(E, s1), DE = f2(DI, s2), BE = f3(DE, s3), A = f4(BE, s4), and C = f5(A, s5), where E, DI, DE, BE, A, and C are exposure magnitude, internal dose, biologi-

OCR for page 153
Biologic Markers in Urinary Toxicology Figure 5-1  Simplified flow chart of classes of biologic markers (indicated by boxes). Solid lines indicate progression, if it occurs, to the next class of marker. Dashed lines indicate that individual susceptibility influences the rates of progression, as do other variables.  Biologic markers represent a continuum of changes, and the classification of change  might not always be distinct. Source: Adapted from Committee on Biological Markers  of the National Research Council, 1987. ically effective dose, early effective dose, early biologic effect, altered structure or function, and clinical disease, respectively. The functions relating the markers, fi (i = 1, ..., 5) are represented simply but might be complex systems of equations (e.g., a pharmacokinetic model). In this context, the markers of susceptibility, si (i = 1, ..., 5), which might include biologic and nonbiologic components, can be considered indicators of all the factors that can modify the relationships among the markers of exposure and effect. In most cases, it is more appropriate to consider the probability of clinical disease (and perhaps of altered structure or function, of early biologic effect, or of the dose markers) to be functions of the other variables—e.g., P(C) = f5(A, s5), where P(C) is the probability of clinical disease—but the important features of this discussion are not altered in either case. An example is provided by the development of bladder tumors as a result of exposure to MOCA. In that case, the markers can be defined as follows: DI: absorbed MOCA, DE: concentrations of MOCA in urine, BE: DNA adducts in bladder epithelial cells, A: increase in red blood cells in urine, and C: bladder cancer. The traditional estimation of the risk of bladder cancer in humans would have relied on extrapolating from animals a relationship of the form P(C) = g(DI), where g(DI) is some function of administered dose expressed in terms like mg/kg/day or milligrams per kilogram per day or milligrams per square meter per

OCR for page 153
Biologic Markers in Urinary Toxicology day. It is assumed here that the conversion from exposure magnitude (e.g., atmospheric concentration to which the animals were exposed) to internal dose, DI, is appropriate, although in practice the estimation of an absorbed dose has typically been rather crude. The function g(DI) might be the multistage model routinely used in cancer risk assessment (Anderson and the Carcinogen Assessment Group of the U.S. Environmental Protection Agency, 1983; Crump, 1984) fitted to the observed cancer results in the animal test species. If we refer to the relationships between the markers (the functions f1 to f5,) we see that the overall relationship P(C) = g(DI) ought to be representing the marker-to-marker relationships from an internal-dose marker to a marker of clinical disease. In fact, it will be noted that P(C) = g(DI) = f5(f4(f3(f2(DI, s2), s3), s4), s5). This convoluted form indicates why traditional risk-assessment extrapolations are problematic: the relatively simple functional relationship extrapolated from animals to humans should be representing a variety of biologically important processes. Moreover, given the variety of processes and differences among species with respect to markers of susceptibility, it is unlikely that the overall relationship is the same in humans as in animals. The utility of an approach that considers biologic markers of effective dose, early biologic effect, and altered structure or function is related to our ability to model the relationships among those markers. The fundamental assumption of extrapolation—i.e., that relationships observed and modeled in one setting hold in another setting—can be applied at the more specific level of those marker-to-marker relationships. It might be possible to rely on extrapolation from experimental systems to humans for only a subset of those relationships. In the case of MOCA, suppose that the observation of occupationally exposed people has provided sufficient information to derive a direct human model linking exposure and urinary MOCA concentrations. A physiologically based pharmacokinetic (PBPK) model derived from human data would serve that purpose; human PBPK models have been developed for other workplace contaminants, such as trichloroethylene (Allen and Fisher, 1993). The human PBPK model would provide the relationships represented by f1 and f2 in the above scheme. Suppose also the availability of human data that relate the extent of hematuria with the probability of developing bladder cancer. Those data would not necessarily need to be specific to MOCA. This is another advantage of biologic markers in risk assessment: the relationship between two markers, especially markers of effect, need not be chemical-specific, so a wider base of information can be used to conduct an assessment. If the links between exposure and

OCR for page 153
Biologic Markers in Urinary Toxicology urinary concentration and between hematuria and bladder-cancer risk are established from human data, then the only relationships that need to be extrapolated from experimental systems are those between urinary concentration and adduct formation and between adduct formation and the degree of hematuria. Animal-based test systems—in vivo, ex vivo, or in vitro—can be used to elucidate those relationships. They can then be extrapolated to humans on the assumption that the relationships observed in the test systems are the same as in humans. That is the fundamental assumption of risk assessment, but in this context it is a much more particular and specific level than is typical in risk assessment. Because that is the case—i.e., because the extrapolations involve fewer variables and less "gap" between cause and effect—there is a greater chance of appropriately accounting for the modifying factors of susceptibility (e.g., species differences in DNA repair) and therefore a greater likelihood that the assumption is justified. Even when human data are less direct than in the example just cited, the application of the extrapolation assumption to specific links in the chain leading from exposure to disease will improve the quality of risk-assessment estimates. The links in that chain can often be studied with experimental techniques or systems. Results of the studies can then be incorporated at the appropriate point, allowing even greater consideration of species differences or other differences that can be represented as differences in susceptibility. As mentioned above, the ability to separate the chemical-specific from the chemical-neutral relationships facilitates the incorporation of a larger base of data. The way to improve risk assessment is to give it a firmer scientific foundation. The use of biologic markers to break the disease process into manageable research pieces is a way to do that. In risk-assessment practice, there will be a need to compromise between the ideal (all the markers one could want and a full understanding and representation of the sequence of events) and the real (less than complete information and less than perfect representation). The application of the basic assumption of extrapolation might not be at the level that one thinks is best; it might have to be at a level as close as possible to the best, given the data at hand. That introduces additional uncertainty, which can be reduced as additional research is conducted to provide more fundamental data, to understand more clearly how cause-effect relationships can be represented, or to elucidate the role of modifying factors. For assessing risk, any model should account for individual variability in response and health or environmental status. Differences in those factors will alter susceptibility to potential chemical-induced injury. SUMMARY Extrapolation from animal models is a common and necessary component of risk assessment for humans. To im-

OCR for page 153
Biologic Markers in Urinary Toxicology prove the validity of such extrapolations, a better understanding of the relationship between these markers and disease is needed. In most cases, the scientific basis for assuming that animals are good surrogates for humans, and therefore a suitable basis for extrapolation to humans, is overwhelming. It is reasonable to use animal models for extrapolation to humans unless specific information on specific chemicals indicates otherwise. Identification of chemical hazards should include assimilation and evaluation of all relevant information, including appraisal of physical and chemical properties and structure-activity relationships, which can often provide important indications of potential toxicity. Difficulties in diagnosing renal injury and predicting its health consequences are considerable, primarily because the kidneys can undergo substantial chemically induced injury without any clinical manifestation, and subtle injury can be negligible because of the considerable functional reserve of the kidneys. Standard diagnostic criteria are needed that are sensitive enough to serve as markers of renal damage in the presence of renal functional reserve. Only whole-animal studies or observations in humans can provide information on the operation of multiple cells, tissues, and organs under the influence of complicated feedback mechanisms. Animals are necessary in the study of chemically induced toxicity, because studies that involve modulation of cellular responses and tissue sampling cannot be performed in humans. The first stage of any investigation of the nephrotoxicity of a xenobiotic should be in vivo studies. In the absence of any knowledge about potential toxicity and target-organ specificity, the first step should be to determine whether toxicity occurs and the tissue distribution of the toxic response. More detailed studies, both in vivo and in a variety of in vitro models, can then be pursued to elucidate modes of chemical action, specific mechanisms of toxicity, and potential protective or preventive strategies. Animal models developed as surrogates for humans in the study of renal and urinary function should conform to some general principles: the animal model should be reproducible within and among laboratories, the model should be specific to the part of the urinary tract under consideration, the model should be sensitive enough to differentiate normal from abnormal changes or functions, and the model should be able to measure alterations in renal function caused by exogenous agents. A variety of experimental model systems are available for study of renal metabolism, renal function, and nephrotoxicity. They range from whole-animal studies to those in the isolated perfused kidney, kidney slices, isolated nephron segments, isolated tubule fragments, and isolated renal cells. Each model has advantages and limitations that must be taken into account when developing conclusions and extrapolating animal data to human risk assessment. In vitro models of nonrenal urinary tract epithelia have also been developed and applied primarily toward examination of carci-

OCR for page 153
Biologic Markers in Urinary Toxicology nogenesis. Development of markers of exposure and susceptibility has not been addressed directly with such nonrenal models and should be pursued for better extrapolation of data for risk assessment. The importance of enzymatic activation of toxic chemicals is central to an understanding of chemically induced renal injury. Species and strain differences in amounts and tissue distribution of various enzymes can be critical in determining the ultimate toxic response. Consequently, patterns observed and conclusions reached in one species might not apply to another species. We recommend that species and strain differences in disposition and metabolism be evaluated for each chemical or class of chemicals. For assessing risk, any experimental model should account for individual variability in response and in health and environmental status. Differences in those factors will alter susceptibility to potentially toxic chemicals.

OCR for page 153
Biologic Markers in Urinary Toxicology This page in the original is blank.