Document G6Dvp70aDXw5jxEg6xKYbgwL7

OCT 1 1980 .... EXPERIMENTAL AND CLINICAL 9 Neurotoxicology edited by PETER S. SPENCER, P h . D . Associate Professor of Neuroscience Director, Institute of Neurotoxicology Albert Einstein College of Medicine, Bronx, N.Y and HERBERT H. SCHAUMB Professor and Vice-chairman of Neurology 1nst i t ute of Neurotoxicology Albert Einstein College of Medicine, Bronx, N.Y. RG, M.D. WILLIAMS 8. WlLKlNS Baltimore/London Copyright 0,1980 The Williams & Wilkins Company 428 E. Preston Street Baltimore, Md. 21202, U S A . + All rights reserved. This book is protected by copyright. No part of this book may be reproduced in any form or by any means, including photocopying, or utilized by any information storage and retrieval system without written permission from the copyright owner. Made in the United States of America Library of Congress Cataloging in Publication Data Main entry under title: Experimental and clinical neurotoxicology. Includes index. 1. Nervous system-Diseases. 2. Neurotoxic agents. 1. Spencer, Peter S. 11. Schaumburg, Herbert H. [DNLM: 1.Environmental pollutants-Toxicity. 2. Nervous system-Drug effects. 3. Nervous system diseases-Chemically induced. WL100.3 E961 RC347.E99 616.8'04'7 79-23302 ISBN 0-683-07854-2 Composed and printed at the Waverly Press, Inc. Mt. Royal and Guilford Aves. Baltimore, Md. 21202, U S A . SUBSECTION 2: Experimental Methodology for the Detection and Screening of Chemicals with Neurotoxic Properties - Chapter 48 Experimental Design for Animal Toxicity Studies GERARD F. EGAN, STEVEN C. LEWIS, and ROBERT A. SCALA INTRODUCTION This chapter will consider some of the conceptual and practical considerations in the design of wholeanimal toxicity studies. The conceptual issues will be concerned with the definition and evaluation of the problem, selection of approaches, consideration of variables, and the like. The practical elements of whole-animal experimental design, nominally worthy of a separate text, will be summarized and highlighted with respect to types of studies and design elements such as animal model, mode of exposure, selection of controls, and the sequence of steps in data evaluation. Biological studies seem more likely to fail as they become removed from clear purposes and stepwise evoiution of understanding. At each point in the development of the experiment, the nature and function of a control must be recognized. As one asks "Why are you doing this experiment?", it is also necessary to be able to determine whether the intended effect or observation has been made. This is ordinarily measured by comparison with the appropriate concomitant control. PURPOSES OF LABORATORY ANIMAL STUDIES GENERAL CONSIDERATIONS Although it seems to be unduly obvious to make such a statement, it is necessary to have a clear idea about why the specific study is to be done. The investigator needs more than an overall concept of the total research program or the kinds of information desired. He must have a clear, explicit, and brief statement of why a specific experiment is to be performed. When the data are in hand and the interpretation is uncertain or ambiguous, one explanation may be an uncertain purpose. Normally, concise statements of purpose are defeated by succumbing to the very common temptation of adding just one more component to the study. In these instances, what starts out as a clear, straightforward experiment, becomes encumbered with side issues, add-ons, supplemental or secondary purposes, and the original intent is lost, obscured, or thwarted. Whole-animal laboratory studies in toxicology and related disciplines may be divided into three major types. These are simulation, characterization of adverse effect, and evaluation of safety. SIMULATION The earliest experiments in toxicology are thought to have been simple simulations, either predictive or diagnostic. In its most rudimentary mode, a simulation experiment involves dosing an animal with the same amount of material as the human would be expected to receive (or did receive). The route is likewise the same, and there follows a period of watchful waiting. From the earliest use of royal "tasters," where the experimen- ' tal subjects were, unfortunately, other humans, through contemporary forensic toxicology assays, the simulation experiment has answered the ques- tion of "What would happen i f . .. " More sophis- 1. - ' Section D EXPERIMENTAL DESICN CONS1DERATIO NS 709 , ticated simulation studies stem from unanticipated, j but not necessarily lethal, effects in man from en' vironmental or other agents, and seek to reproduce the effects first grossly and then on successively more elemental levels. CHARACTERIZATION OF TOXICITY Modem-era toxicology has strongly emphasized characterization of toxicity as the initial stage in the evaluation of a new material. Though fiequently confused with safety evaluation, the characterization of toxicity has different objectives and, consequently, differences in design. The major emphasis is on the identification of target organs and systems through the use of multiple doses and routes of exposure. By these means, the investigator attempts to understand the quantitative relation between dose by any given route and the magnitude of response in the whole animal, and in any organ or subsystem which can be measured by invasive or noninvasive means, including postmortem examination. Not to be ignored are differences in effects which may result from the use of different routes of exposure and the possibility of reversing the effect by concluding the exposure. During these studies, the investigator is also concerned with identifying those functional or other measures which are early indicators or measures of adverse effect. From this array of findings,it is possible to design an adequate series of safety evaluation studies. EVALUATION OF SAFETY Once the investigator has identified the specific toxic effects of a material, additional studies to determine safe conditions of use or exposure are in order. The design principles for such experiments are the use of a route which is comparable to the anticipated means of exposure. Food additive chemicals should be fed and not injected, for example. The doses should reflect use levels and reasonable exaggerations of use levels. The duration of study should be related to the anticipated pattern for man. Food additives, to return to the earlier example, should be fed for the lifetime of the test animal. Products of incidental-or infrequent use may be tested for shorter times, although chemical structural or other considerations could dictate long-term postexposure observations. The safety evaluation tests should employ those predictive clinical or other tests identified in the earlier studies. These should be regarded as the most efficient criteria of effect. Perhaps the subject of greatest controversy and the most progress in recent times with respect to experimental design has been the choice of animal species. The traditional choices of animals for testing were based on cost, dvailability, size, suscepti- bility to the effect under study, and extent of the historical data base. In the majority of safety evaluation studies in the United States, the result of this selection process was the use of beagle dogs and white rats. For the last several years, however, the driving issues have been susceptibility to the desired effect (i.e. cancer-resistant uersus cancersusceptible strains) and the pharmacokinetic similarity to man. Pharmacokinetic Considerations Increasingly, studies on characterization of toxicity have included assessments of the effect of the experimental animal on the material under study, in addition to the more obvious study of the effects of the material on the animal. Measurements of uptake, distribution, storage,bioconversion, and excretion, provide clues to the understanding of the mechanism of adverse effects, the identity of the active chemical agent, and possible species differences in effect levels on target organs. Most particularly, any pharmacokinetic study which can rule in or out a particular animal species with regard to relevance to man becomes a singularly powerful tool in the selection of test speciesfor whole-animal experiments designed to evaluate the safety to man of a particular agent or material. UNDERSTANDING VERSUS MEASUREMENT The ultimate purpose of a whole-animal study should be an understanding of the effects produced by the agent administered, and not merely the compilation of data from certain tests, or the satisfactory completion of a protocol fixed according to some governmental or other regulation. While the latter may give the comfort of having marked all the boxes on a checklist, the experiments will have failed if they did not provide first a character- ization of the effects and second, an understanding of underlying mechanisms. The checklist approach assures that hitherto unidentified or untested effects will be missed. Striving for understanding while not assuring that such effects will be found, certainly provides for the prepared and sensitive experimental setting in which they are more likely to be recognized. These statements are as axiomatic as the need for a firm purpose for each experiment, yet the literature gives generous examples of inattention to these principles. SCIENTIFIC JUSTIFICATION FOR WHOLE-ANIMAL STUDIES AS A USEFUL MODEL FOR HUMAN DISEASE PROCESSES Disease states can be produced in experimental animals by a variety of means. There are genetic 710 EXPERIMENTAL AND CLINICAL NEU ROTOXICOLOGY.p Section D diseases whose prevalence and degree can be controlled to some extent by the breeding process: the availability, for example, of a colony of animals with a genetic trait toward hypertension provides a model in which to study the physiologic or pathologic consequences of the disease, and may be useful in screening of potential therapeutic agents. Disease states may be produced through surgical intervention and the ligation or extirpation of critical organs or tissues. The literature on adrenal cortical functions, and the consequences of the loss of these functions, is based, in part, on such animal studies. Replacement therapy experiments have followed and, particularly with endocrine disorders, the use of the animal model has led to successful therapies in man. The third mode of inducing disease states in animals is from exposure to chemical or physical agents. One of the most widely known is the production of diabetes mellitus subsequent to the destruction of the p-cells of the pancreas by the chemical agent alloxan. This animal model yields most of the clinical features of the disease in man, although the degree of similarity between experimental alloxan neuropathy and human diabetic neuropathy is in question. Given the fundamental similarity of mammalian functions across many species, and the essential commonality of many cellular mechanisms and reactions, it is ordinarily to be expected that there would be a commonality of lesions whether genetically, surgically, or chemicaUy induced. This obvious simplification becomes the opening premise in any specific instance and must be carefully tested by comparative study of the disease of "spontaneous" or unknown etiology in man and the induced disease in animals. Unfortunately, such whole-animal modeling is more likely to be confirmatory of the human disease rather than predictive. HAZARD VERSUS SAFETY EVALUATION (DOSE-RESPONSE ASSAYS) ACUTE TOXICITY INVESTIGATIONS The foregoing discussions have been concerned with the purpose of conducting toxicity experiments and the rationale supporting the use of animal models to predict the potential deleterious effects of chemicals on Man or his environment. In general, toxicity investigations are designed to uncover and to characterize a range of harmful effects produced in intact biological systems as a consequence of chemical exposure. These studies also may be structured to provide some evidence of the dose-response relationship that may exist between a chemical and an observed biological change. The assem- blage of all such toxicity data then can be employed to assess the degree of risk associated with exposure to, or use of, a selected material. Thus, the goals of toxicity experiments may be divided into two main ; categories: .- 1 A. Those which attempt to determine the range of specific toxic effects produced by a chemi- cal, e.g.skin or eye irritation, behavioral mod- ification, detrimental changes in reproductive function, cancer, etc. B. Those designed to quantify the relationship that may exist between exposure a t different levels and the degree of biological response in ~ the test organism. The former are useful in . characterizing a number of potential effec while the latter are employed in safety eval- uation procedures. Among the most simple and perhaps most m ligned of all toxicity tests is the so-called LDso or LCm experiment. The expression is a calculated value representing a mathematical estimation of i the dose (D) or concentration (C) a probability range or a confidence limit. These limits, arbitrarily selected by the investigator, describe the probability of obtaining similar results the test is repeated. The LDm test is conducted by administering the test material to groups of animals over a r mal, intraperitoneal, s nous. The dermal test provides information on skin-penetrating properties of the test substanc properly conducted, dermal tests may also yield valuable information regarding the irritating corrosive potential of the chemical. The LCWtest is a series of acute inhalation wherein the duration of the exposure is held constant, e.g. 15 min, 1 h, 4 h, 6 h, etc., whil concentration of test substance in the breathing ` atmosphere is varied for each exposure. A variation of this procedure involves adjusting the time while holding the concentration constant. The so-caIl LTw test then measures the time required to p duce mortalities in 50%of the animals exposed a given concentration of material. Most LCdLTm tests rely on a nominal calc tion of the chamber concentration rather than actual analytical determination of the expos level. In those instances where the test species highly reactive, it is imperative to carry out ro analyses of chamber atmospheres, to control e sure levels, and to correlate mortalities care with exposure concentrations. The LDw or LCsovalue, and the graphical rep resentation of such parameters, allows the investi Section D EX PER1MENTAL D ESIC N C O NSI DERATIONS 711 gator to learn something about the toxicity and the potency of a chemical, particularly with respect to other chemicals. Figure 48.1 illustrates the doseresponse relationships of three chemicals (A,B, and C), each having different slopes or different x intercepts. Plots A and B indicate the two materials have the same LDm values. The dose required to produce 50%mortalities with both chemicals is the same. However, the difference in their slopes suggests that for a given dose increment, Compound B will produce more mortalities than Compound A. Furthermore, a comparison of the LDss of these materials indicates that, compared to Compound B, smaller doses of Compound A will produce mortalities less than 50%. Hence, A is said to be more potent. Plots B and C illustrate the dose-response relationships for two chemicals whose relative toxicities remain comparable over the range of doses tested. In this instance, the LDw of C is less than that of B and, for any given dose, its (C) toxicity would be regarded as being lower than that of B. The LD or LCWvalues are useful, then, in comparing the toxicities and potencies of chemicals. These lethality tests also can provide some information regarding functional signs of intoxication as well as the time of onset and the duration of such signs. By careful attention to the character and progress of these signs, coupled with a thorough gross pathological examination of surviving animals (14-day postexposure), as well as those dying shortly after dosing, it may be possible to determine the mechanism of toxic action and the principal target organ. These gross examinations would routinely involve the heart, lungs, liver, kidney, spleen, gastrointestinal tract and, when deemed appropriate, the brain, gonads, adrenals, etc. DOSE OR CONCENTRATION 'LOC SCnLE Figure 48.1 Hypothetical dose-response curve for three chemicals ( A ,B , and C) administered to a uniform population of biological specimens. c The misuse or maligning of the LD/LCm value referred to earlier stems from a lack of appreciation of those variables which can have a significant influence on the outcome of the experiment. Although a more thorough.discussion of experimental design variables will be presented in the succeeding section of this chapter, illustrative examples will be offered here of how such variables can impact on the outcome of acute, subacute, and chronic toxicity investigations. In the case of the LD/LCm test, age, weight, caging conditions, dietary programs, species, route of administration, etc., can all exert a significant effect on the calculated LD/LCm value. Unless it is possible to know the nature and extent of these variables as they influence the outcome of an LD/LCm test, then, strictly speaking, any attempt a t comparing such values would be highly suspect and without scientific base. The estimation of the LD/LCm value constitutes a frrst step in the process of assessing the toxicity of a material. The test has the advantage of being relatively simple, and it lends itself readily to statistical analysis. The calculated lethality index can serve as a yardstick by which to gauge the toxicity of that material. It also allows for the comparison of the relative toxicities of a number of chemicals. This comparative ranking has led to the use of such common descriptors as highly toxic, moderately toxic, nontoxic, etc. The LD/LC, test is, by definition, a parameter of acute toxicity. The latter term is used to describe those adverse effects occurring during an exposure and/or for a period of time up to 14 d postexposure. More often than not, the exposure or dosing regimen is a singular event whose duration may be measured in a matter of seconds, e.g. intravenous, intragastric, intramuscular, subcutaneous, topical administration, etc., or hours, e.g. inhalation exposure. Acute (short duration) exposures may result in toxic effects which, over a period of time, will be self-resolving. Such brief exposures, however, can also produce an effect which is still in evidence months to years later. For example, animals or humans exposed to elevated concentrations of hydrocarbon vapors for a period of minutes to hours will experience a range of central nervous system (CNS) effects. These may include headache, nausea, vomiting, dizziness, drowsiness, sleep, coma, and perhaps even death. If exposure is halted during the early stages of central nervous system involvement, the effects will gradually subside and the animal or individual will recover completely. Should the exposure continue to the point where brain oxygen levels have been seriously compromised, due to an interruption in the available supply of oxygen-rich blood to the brain or to an interference in the metabolism of oxygen by the 712 EXPERIMENTAL AND CLINICAL NEUROTOXICQLOGY Section D brain cells, then nerve cell damage could begin and the resulting toxic effect might be permanent. The design of acute toxicity studies must take into consideration several significant factors, among which the following are probably the most critical: A. Selection of species B. Route of administration C. Concentration of test material in vehicle D. Duration of exposure and observation period. The selection of species in acute toxicity investigations is more often than not based on practical considerations. A wide range of genetically homogeneous rats, mice, guinea pigs, hamsters, and even rabbits, is available to choose from. Depending on the nature and goals of the experiment, a given species is generally employed because of the ease in handling, space considerations, cost, the number of animals which are to be treated, etc. Rats and mice are typically selected for the various peroral and parenteral LDmstudies. The calculated lethal- ity index can then be compared to values for other chemicals which are commonly derived from rat or mouse studies. Because of the ease with which they can be manipulated, and the fact that they have sufficient body surface area to permit a variety of dermal toxicity investigations, rabbits are often selected to act as human surrogates in a number of dermal irritation, penetration, and sensitization studies. Species variation in acute toxicity experiments has been well documented (4,25);particularly with respect to the two most commonly used species for LDW estimation, namely the rat and the mouse. Morrison and co-workers (20) reported that the oral LDm in male albino mice for 5-(N-piperidine)10,11-dihydr0-5H-(a,dcycloheptenweas 1160 mg/kg, while in male albino rats the value was found to be 6.6 mg/kg. The classic example of norbormide species sensitivity is also offered for consideration (rat oral LDw-5-15 mg/kg uersus 2250 mg/kg for mice) (20).Such differences tend to be the exception rather than the rule, as noted by Wed and Wright (36) and Weil and associates (34). The question of intraspecies variation is also one that must be considered. Thus, the LDm for thiourea in the Hopkins rat strain was 4 mg/kg, while a value of 1340-1830 mg/kg was noted in the Norwegian strain (20). One could speculate that the differences in toxicity recorded among species or within a given species may be a function of relative liver microsomal enzyme activity. Such differences in enzyme activity may ultimately be genetically determined. Other biological factors which can influence the toxicity of a material, e.g. age, sex, etc., should also be weighed carefully when deciding on the elements of an acute toxicity protocol. The reader is referred to articles by Conney and co- workers (6), Yearly and Benish (38), and Lu and i associates (18),for discussion of the impact of sex . and age on the outcome of acute LDm determina- tions. A brief review of their findings suggest the following: 4 IA. Young animals do not have a fullcomplement of mixed-function oxidase (MFO) enzymes in 1 the liver. These enzymes are thought to be [ ! responsible for the metabolism and ultimate f detoxification of a number of xenobiotes and !' a wide variety of endogenous metabolites, e.g. 1 T steroids, etc. B. Older animals experience a gradual dimuni- tion of MFO enzymes. As a consequence, their ability to detoxlfy exogenous materials f is far less efficient than the also tend to be more obese counterparts. Their fat tissues .c the deposition and retention material for an extended period in some instances, the period of C. Young adult (e.g.sexually mature) animals a given species tend toward greater me t. 0 I; r. and enzymatic homogeneity. This is pr a function of the stabilization ity, sex hormone influence, an sensitivity (9, 15, 28). D. In general, males and females tend to re similarly in acute toxicity tes those instances where the test materials affects, or is affected by, metabolism of endogenous sex hormone L. LI T ai w cc in. tl: is conceivable the resulting sex related. The route of administration, z.e. oral, derm intravenous, inhalation, etc., is o portance in determining the ac chemical. The manner in which a test material isl administered in acute studies s h o d most likely route($ of exposure for humans, p ularly as in those cases where the resul to be employed in safety evaluation exerc ever, when the purpose of the test ns inl Tl- wk m: chi 1ns -.- identify the acute toxic properties of a more often than not, the route of administratio peroral, dermal, or one of a number of routes. When selecting a route of exposure, necessary to consider the impact that certain o systems can have on the metabolism of that material. Thus, the toxicity of compounds a istered via the oral route can be modified number of factors. Certainly uptake and tr :eT. ha1 ora the con aP.F1-- via the enterohepatic system allows the li exert a major influence (detoxification or t tion/activation). The changing pH profil res: . .itf .. 'e! gastrointestinal tract can influence the relative :Cir sorption rates of ionizable test species. Stoma rher $ 7D m r: .-L. ent -in be ate 2nd 9.g As ley ger for .ble :ng, on. -. -L1\ NS :act ", in the the . it -I.&> .r : imif a 3i is tne rticare ou- 1: I Jn 15 eral it is rgan test ninJ? a port r fit ;LA 1 abach- 1 Section D EX PER1MENTAL DESlCN CONS1DERATION S 713 emptying time and gastrointestinal food content' the acute test protocol. For example, the LD, test also will play a role in the ultimate expression of a does not provide an indication of toxicity per se. chemical's toxicity when administered via the oral Rather, it is an experiment designed to determine route. It is evident that an intravenous administra- the association between dose or extent of exposure tion of a material allows for the most efficient and and a predetermined end point, e.g. death. No a rapid distribution of a test chemical throughout the information is developed regarding the persistence body. Intramuscular and intraperitoneal adminis- or the gradual development of any lesions, be they trations are less efficient, while the subcutaneous physiological or morphological in nature. The ques- , route permits a very slow release of material to the tion of nonlethal doses bringing about the devel- systemic circulation. It is also important to weigh opment of an irreversible lesion at a time subse- the relative vascularity of the area to be treated. quent to the routine postexposure observation pe- The inhalation route represents a different set of riod is also not addressed by the routine acute LDm technical problems. To conduct a proper, well con- test. Thus, the carcinogenicity of 3-methylcholan- itrolled, acute inhalation trial, it is important and threne, 1,2:5,6-dibenzanthracene,and 3,4-benzpyr- useful to know the test material's vapor pressure, ene, which can be evidenced after a single dose boiling range, and the lipid/water partition coeffi- treatment and a 1-2-y observation period, is not at cient, particle sues for aerosols and dusts, etc. With all reflected by the LD, value or by observations 1I such information, it should be possible to determine made during the 7-14-d postexposure period (14). the theoretical dose administered to the test species Other examples include the clinical effects of cou- g over a period of time. This determination presup- marin-based anticoagulants whose action is only poses a knowledge of the respiratory dynamics of manifest after the depletion of endogenous supplies that test animal, e.g. respiratory rate, tidal volume, of prothrombin. The same may be offered regarding f1 minute volume, etc. the neurotoxic effects of tri-o-cresyl phosphate, Respiratory exposures are often complicated by lead, methyl mercury, and phenacemide (13, 14). the tendency of test animals, rodents in particular, That lesions induced in survivors of acute toxicity to ingest a considerable quantity of test material. tests may become manifest after more than a year This occurs as a consequence of grooming activities of observation is illustrated by the work of Schoen- f" along with the ingestion of food and water which, while maintained in the inhalation chamber, be- g Ji come contaminated by the test material. In these instances, any observed toxicity would have to re- 'f flect the combined contribution of ingested and inhaled test material. 5 The size of the dose (volume),the concentration ! of the test material in its vehicle (diluent),and the t nature of the vehicle may all have a significant tal and Magee (27) and Wheeler and colleagues (37). SUBACUTE AND SUBCHRONIC EXPERIMENTS The subacute/subchronic experiment is a logical and necessary extension of the acute test, particularly regarding any effort to assess the safety of a chemical in the work environment, as a human food influence on the outcome of the acute experiment. component or as a household product. In many Thus, the administration of an oral dose in a volume instances, the acute toxicity study serves as a means which approaches or exceeds that of the test ani- of establishing exposure levels, i.e. concentrations '; mal's stomach will only guarantee the artifactual in dietary regimens or in inhalation trials for the character of the observed toxic response. In this subacute/subchronic test. instance, the biological response is but a manifes- Within the broad category of subacute/sub- tation of the unphysiological nature of the dosage chronic experiments, there are essentially two volume employed. Regarding the impact of dilu- classes of studies: tion, Griffith (11)has demonstrated that the con- A. Routine tests designed to determine the ad- centration of test material in a diluent (vehicle) can verse effects of regularly repeated exposures have a major impact on the outcome of an acute over a period ranging from several days to 6- oral experiment. Increasing the concentration of 9 months. In these studies, the intent of the 5 the test compound by a factor of 2.5 results in a experiment is to produce no observed effect(s1 concomitant decrease in toxicity by a factor of a t one level and adverse effects in at least one .approximately 2. z Although a well planned acute LD/LCx, toxicity higher exposure level. The parameters monitored may include biochemical changes, i.e. a test can provide useful information regarding a enz-me assays, hematology, clinical chemis- chemical's relative toxicity and the influence of 2 species, age, sex, route, and mode of test material tries, urine analyses, organ and body weights, food intake, and physiological/functional, be- administration on the toxicity of that chemical. havioral, and histopathological determina- there are some major shortcomings associated with tions. 714 EXPERIMENTAL AND CLINICAL NEUROTOXICOLOCY Section D B. Specialized studies designed to determine the levels be adjusted weekly or biweekly to re- metabolism and pharmacokinetics of a ma- flect any variations in total body weight. terial, its mutagenic/teratogenic or neuro- pathogenic potential and, in some instances, the impact of repeated exposure on the im- CHRONIC OR LIFE-TIME TEST mune mechanisms of the body. Chronic or lifetime toxicity tests are performe The selection of species for these extended stud- to identify and characterize toxic effects whose ies is generally pragmatic. Ideally, it would be pref- pathogenesis requires several months before be- erable to choose a species whose metabolism of the coming manifest as a frank lesion, i.e. tumor, nerve test material is comparable to that in man. Rats, fiber degeneration, etc. Such tests also may be mice, and rabbits usually have been employed in carried out to quantify the nature and extent of such tests, because of their size and the extensive those toxic responses requiring prolonged and/or background on husbandry and nutritional needs repeated exposure before the disease state become that has been developed on them during the course evident. of several years of experimentation. The dog is The carcinogenesis bioassay is a special kind of often used, particularly when the use of a nonrodent long-term toxicity investigation designed to detect 8 I i i species, which is phylogenetically closer to man a specific kind of lesion, namely a tumor in a.given than rodents, may yield more meaningful and sig- organ system, e.g. hepatoma, lung adenoma, skin nificant information. papilloma, etc. In contrast to this specialized test, The route of exposure ideally should reflect the the chronic toxicity investigation is geared toward anticipated route of contact in man. For some ma- detecting a wide range of pathological changes terials, where the inhalation route represents the cluding tumors, blood dyscrasias, decrements nature of the real-world exposure, it may be possi- reproductive capacity, modifications in central an ble to use a feeding study in lieu of the more costly peripheral nervous system homeostasis, i.e. a , - 1 inhalation trial. This is usually done only if it has behavioral changes, etc., and fibrotic changes E. been demonstrated that the metabolism and phar- pulmonary tissue, modifications in basement mem macokinetic profiles of the test material are the brane architecture, induction of autoimmune dis same regardless of the route of administration, i.e. eases, etc. With regard to the selection of test inhalation versus oral. species for chronic toxicity investigations, including Testing involving rats should be initiated just the carcinogenesis bioassay, it is usually impossible after weaning so observations can be made during to determine with any degree of confidence the' the period of most rapid growth. If no data are fidelity with which a given animal model mimics available regarding the metabolism of the test ma- the metabolism and pharmacokinetics of a test terial in man, it is often deemed appropriate to chemical in man. Nor is it possible to know whether employ a species which has been shown to be most the dose of the chemical at the active receptor site sensitive in preliminary acute toxicity tests. would be similar in the chosen test animal to that When the test material is to be administered as occurring in man. Consequently, long-term toxicity part of the dietary regimen, it is essential to keep investigations are often designed using an animd the following in mind: model which may be insensitive or otherwise ina& A. Some test materials are altered or modified propriate for predicting human health hazards. by interaction with dietary components or as This certainly should be borne in mind when , a consequence of exposure to air or moisture. investigation yields negative results for, in this in: .,I ,. > ` In those instances where it is imperative to stance, the test results should be considered as a determine the toxicity of the chemicalper se, "false negative." On the other hand, for the very it is then essential that chemical analyses be same reason, a positive finding in the animal model carried out to determine the stability of that may represent a "false positive" result. To minimize chemical in the diet. Should these tests reveal the prospects of such a dilemma, it is recommended ! 1that alterations in test material can and do that two unrelated species (rodent and nonrodent) 7' occur with time, it may be necessary to re- be employed in chronic toxicity experiments, pa: place the food supplies on a more frequent ticularly when it is not possible or practical. basis or to administer the chemical to the test determine the similarity in metabolic and pharma: animals by means of gavage or in an encap- cokinetic profiles of the chemical in the test species sulated form. and man. B. When test chemicals are added to the diet as The duration of treatment in either carcinog a fraction of the total diet to achieve a pre- esis bioassays or in chronic toxicity investigat determined dose level, e.g. mg/kg body may include a 1-year or half-life time exposure . weight per day, it is essential that the dietary period followed by a 1-year to lifetime postexpos- $ i , Section D EXPERIMENTAL DESIGN CONSIDERATIONS 715 observation period. Other long-term tests are designed to carry exposures until no survivors remain I while in stiU others, sacrifices are initiated at some I arbitrary point near the natural end of the test I animal's lives, i.e. 24-30 months for rats, or when the survivors number 10%of the original test group, t etc. I I PROGRESSIVE AND RETROSPECTIVE STUDIES Toxicology investigations are often carried out , on materials for which there are no toxicity data. Such is typically the case for the vast majority of :i materials currently in commerce. Experimental investigations of this kind are termedprogressive. In s contrast to this class of study, an investigation may be triggered by the detection of an adverse effect during the manufacture or processing of a chemical, ' or as a consequence of its use in commerce. When a test is initiated under such circumstances, it is termed retrospective. In some prospective animal studies, covert func- f tional changes may go undetected. Often, such de- f fects are only uncovered after human experience 'i with the compound. Thus, Hawkins and Laurie (12) i discovered that aminoglycoside antibiotics can 'eE cause hearing impairment, an effect which went undetected in progressive animal investigations. g Furthermore, it is reasonable to assume the metabE olism (detoxification) of an exogenous material could be compromised if the liver is diseased, as in f the case of alcoholics. The change in metabolism in this instance, i.e. a prolongation of the biological half-life of the material, would probably enhance ' the toxicity of that substance and/or prolong the manifestations of its toxicity (2).Any such potential, yet highly probable, effect would not be predicted in a routine toxicity test. Retrospective studies also have their own limitations. The association between prolonged exposure to elevated levels of benzene and the production of leukemia has never been confirmed in ret- , rospective animal studies. The "beer drinkers myocarditis" observed in heavy beer drinkers in Canada and the United States in the mid-1960s) was diffkult to confirm in animal tests until it was noted that protein-deficient diets, high levels of cobalt salts,and beer, were required in the test diets (10, 24). In short, the progressive and retrospective animal toxicity studies must be pursued carefully to ensure an accurate interpretation of the test results. An awareness of the potential covariables, e.g. diet, preexisting disease, metabolic predisposition, etc., must all be given careful consideration before those interpretations can be deemed valid and useful. EXPERIMENTAL DESIGN CONSIDERATIONS At the outset of any experiment in toxicology, two questions need to be posed: (a) What is the purpose of the experiment or what information is being sought? (b) What is the nature and potential impact of those variables which, of necessity, become an integral part of any experimental design? The preceding section attempted to provide information needed to answer the fust question. This section will concern itself with a detailed review of the major design variables which have a significant impact on the outcome of a toxicity study. THE ANIMAL MODEL As has already been noted, the choice of the animal model for a toxicity investigation is often based on practical considerations, i.e. the rat and mouse are frequently employed because of their size, ease of handling, short life spans and relatively low costs. These animal models have other characteristics which are important in the design of an experiment: (a) Their lack of a regurgitation reflex makes them suitable for those experiments requiring the administration of a test material intragastrically. (b) Certainly, it is advantageous to know the tumor incidence and the nature and extent of other serious lesions in a given animal model before electing to use it in a chronic investigation. Many rat and mouse strains have been inbred for over 50 years. In so doing, the gene pools in these lines have become quite stable and homogeneous. This, in turn, has provided for a variety of rodent models with highly predictable incidences of tumors and other serious disease states. The notion of predisposition to selected pathologic processes deserves some consideration here. The pathogenesis of many disease entities requires several months before evidence of any detrimental change becomes manifest. Knowledge of the incidence, age of onset, and time-to-median-lesion development, is important in the selection of a species for a chronic toxicity test, particularly in the case of carcinogenesis bioassays. If an animal model has a high incidence of cancer in a specified organ, e.g. hepatocarcinoma, then it would be very difficult to obtain any meaningful data regarding the hepatocarcinogenic potential of a test material in that animal model. High background levels can be important, however, when the antitumor potential of a chemical is to be assessed. Also, a species with a predisposition to atherosclerosis would be quite useful for the study of drugs designed to modify cholesterol qynthesis. There exist today numerous animal species which have been developed because of their predisposition toward selected disease -s 716 EXPERIMENTAL A N D CLINICAL NEUROTOXICOLOGY Section D states. One has only to decide the purposes of an experiment before making the model selection. However, it has been argued that since Man is a random-bred animal, it would be appropriate to utilize an animal model with a heterogeneous gene pool. Although this argument is sound, it has its practical limitations. Thus, because studies generally involve the comparison of a limited (40-100) number of exposed animals with a control population of comparable size, it is important that there is not a large variation in the incidence of a particular disease in the control population. Otherwise, life expectancies may be compromised and the real significance of treatment-related disease processes might be masked by background pathological changt?sT Interspecies differences in metabolism are ais0 important variables to consider in the design of experiments. As indicated in the preceding section, differences in toxicity often can be associated with variations in metabolism. Most of these differences probably are linked to the genetic makeup of a species (8). For example, dalmations may be the only member of the canine species which metabolizes purines to uric acid and/or allantoin, a process similar to that occurring in man. Microsomal-mediated hydroxylations are often species-specific, e.g. 7-hydroxylation of coumarin in man uersus 3-hydroxylation in rats and rabbits (6, 8). Along these same lines, the potential impact of exogenous materials on t h e microsomal enzyme system should be given careful consideration. The formation of a toxic metabolite or the detoxication of a chemical by this system can be affected significantly by pre-exposure to exogenous chemicals, e.g. phenobarbital, polychlorinated biphenyls, etc., which act to induce or otherwise enhance microsomal enyzme activity. In a similar fashion, it has been demonstrated that volatile hydrocarbons emitted from cedar wood chips used for rat and mouse bedding can induce the microsomal enzymes. Thus, in addition to the need to control species selection because of the importance of relative microsomal enzyme activity, it is equally important to control the environment of the animal cage as this may also affect that activity. The selection of an appropriate animal model for a toxicity study also should reflect a consideration of relative species sensitivity. For example, the chicken is generally regarded as appropriate for the assessment of organophosphorus neurotoxicity in Man since both species respond with neuropathy after a similar dose and time delay ( 5 ) .The chicken is also very sensitive to other chemicals such as methyl n-butyl ketone (19), but it is the rat which is generally regarded as a suitable mammalian species for neurotoxicity testing (32). In addition to the problems inherent in the selection of an appropriate species and strain of animal, the importance of body weight, age, and sex should be assessed, particularly as these variables are among the easiest to control, and because they may often exert a detrimental impact on the outcome of the study. Thus, Balazs (1, 2) reported that the subcutaneous LDm of isoproterenol hydrochloride is about 800 mg/kg in 200-g male Sprague-Dawley rats, but only 0.35 mg/kg in 600-g males of the same strain. Sex and strain, however, had no influence on the outcome of this LD50 experiment. Weight differences in animals of approximately the same age would suggest differences in total body fat mass. As such, the heavier animals have greater storage capacities for lipid-soluble materials which could prolong the period of intoxication associated with a chemical. Mention already has been made of the significant differences in metabolic (microsomal) activity between adult animals and those that are very young or very old. Age-related differences in toxicity also can be due to factors such as an incompletely developed blood-brain barrier. For example, the translocation of quaternized nitrogen-containing compounds across the blood-brain barrier is inhibited in the adult but not in the immature individual. In contrast to this age-related sensitivity is the finding that conjugation reactions are more fully developed in most young animals than they are in infants (1,8, 20). This variable could p e w i t the regulatory approval of material as a drug for pediatric use when it might be highly toxic in the human infant. Finally, it should be appreciated that animals have age-associated degeneration of the Liver and kidneys as well as degeneration and regression of other organs and tissues. It would be reasonable to employ such older animals in tests designed to assess the bioactivity of drugs developed for geriatric use. Wherever possible, chronic and subchronic toxicity investigations should be carried out using male and female animals, unless there are data to suggest sex-related sensitivities do not exist, or unless the goals of the program dictate the use of males uersus females. EXPOSURE PARAMETERS Among the important categories of variables having sigdicant impact on the outcome of an experiment are those concerning the character of the exposure, i.e. route of administration, the length or duration of exposure and its frequency, the dose o concentration, etc. Route of Admintstration , z? In acute toxicity studies, the route of administration is generally selected on the basis of conveai- n D I Section D EX PER1MENTAL DESI CN C O N SID E R A T I O N S 717 .,.I ence unless the stated objective of the study is to mucous linings of the respiratory system, particu- determine toxicity vis-a-vis a specified route of larly a t low values, ie. less than 30% when mild exposure, e.g. dermal, oral, etc. However, when it is irritation may result. dye necessary only to assess the general acute toxicity Chamber temperature is important in maintainlay of a material, the latter may be administered intra- ing chemicalsin a predetermined physical state,i.e. - of peritoneally, subcutaneously, intramuscularly, in- vapor versus liquid (aerosol). Furthermore, some :he travenously, or intragastrically. The intravenous polymeric materials which can exist in the vapor ide route bypasses absorption problems associated with state a t room temperature, e.g. dicyclopentadiene, ley the other listed routes. The use of the intraperito- can generate monomers (cyclopentadiene, in this me neal route results in a rapid absorption of the test instance) at slightly elevated temperatures. Cham- on compound due to the high degree of vascularization ber temperature is also of consequence to the health in the peritoneal cavity and to the relatively large of the test animals. surface area of absorption. However, the use of this Chamber air pressure provides an indication of i4. f route allows for the passage of the test material test chamber stability relative to the exterior envi- `a- through the liver via the hepatic portal route. In so ronment. As such, any potential leaks Wiu be into ro- doing, the test material may be metabolized (detox- rather than out from the chamber. Measurement I! a :he f5 S e d or activated) rapidly. This will also hold true for chemicals administered via the oral route. The of chamber pressure also provides an indication of airflow stability in the chamber. .a11 oral administration of a material also may present ire problems beyond those associated with the impact Duration and Extent of Exposure 1i? 1 of the enterohepatic circuit. It is conceivable that The duration of exposure may represent a varithe rate of absorption from the gastrointestinal able requiring control, particularly in acute toxicity ; lumen to the surrounding vascular bed could be investigations. For example, the rate of infusion of :... influenced by the quantity and quality of the food a test material could affect the intravenous toxicity ~ 15 content in the tract. Furthermore, it is possible that of a chemical. A relatively slow rate of administra- .ire ; the chemical composition of the test chemical could tion via this route could be equal to the rate of TIt?. * be affected by that food and/or by the intestinal detoxification of a compound. A rapid rate of infu- )re f flora. For these reasons, it may be prudent to con- sion, however, might also prevent the effective in- :ey sider employing more than one parenteral route of teraction of a test chemical with the reactive site(s) .lit i i exposure in subchronic and chronic toxicity studies. in the body. 01' * In acute tests, this problem can be avoided by Acute dermal toxicity investigationsoften require : fasting the animals for the 12-h period immediately the application of a test material to the skin of f* preceding treatment. rabbits for a 24-h period under an occlusive ban- Chemicals administered via the subcutaneous or dage. Such a protocol is deliberately biased to max- n e intramuscular routes will be absorbed more slowly imize the likelihood that test materials will be ab- 'De than those introduced intraperitoneally. If the test sorbed through the skin. However, in the case of 5 t S material has any vasoactivity, i.e. vasconstriction solvent evaluations, the defatting properties of the el- or dilatation, then the rate of absorption via the test material can produce a breakdown in the pro- subcutaneous or intramuscular route can be mark- tective epidermal and dermal tissues, particularly )X- edly affected. under occlusive conditions. Thiscould enhance der- 31e The most complex route of administration is that mal absorption resulting in an apparent increase in ,:- presented in inhalation exposures. In addition to systemic toxicity. If the same volume of solvent was problems involved with the generation of contami- applied over a larger surface area under occlusive nant (test chemical) concentrations at a predeter- conditions, then the dermal toxicity probably would mined level and at a constant rate, inhalation tests be decreased. require a fairly extensive analytical control of the Finally, mention should be made of the utility of QS inhalation chamber test environment. Among the chronic (l-2-year) toxicity studies in determining 9V- parameters which should be monitored, particu- "no effect" doses. With the exception of cancer, er- larly in subchronic and chronic studies are: relative most toxic effects recorded in chronic trials can be -he humidity, temperature, chamber air pressure, con- detected in studies as short as 90 d. Weil and or centration of test material and, in those studies McCollister (35)have provided an extensive review ; involving aerosols or other airborne particles, an on this subject wherein they concluded that a dose analysis of particle size distribution and configura- producing no toxic effects when thoroughly studied tion (shape). Often, test chemicals are very hygro- for 90 d will produce no effects, cancer excepted, scopic and may, as a consequence of this interaction when administered over the lifetime of the test `2- with water molecules, undergo spontaneous hy: animal. These authors also concluded that the most - .- -.il- drolysis. Chamber relative humidity can affect the sensitive criteria of systemic toxicity are body i 718 EXPERIMENTAL A N D CLINICAL NEUROTOXICOLOCY * Section 0 weight, ratios of liver and kidney weight to body weight, and kidney pathology. The dose or concentration of a test substance must be well quantified. In the case of peroral or parenteral administration of test chemicals, it must be demonstrated that vehicles or diluents do not interact with, or in any way modify, the physical or chemical nature of the test material. The impor- to include standards which permit the investigator to know with a high degree of precisio meaning of any observed changes in the expo treated animals. Adequate controls in an e ment are particularly important with respect animals, the test material, and the environment which the animals are treated and/or maintain throughout the course of the experiment. tance of dilution also must be weighed. Thus, the oral toxicity of chemicals can increase when administered in dilute solutions (1).This is thought to be due to enhanced intestinal absorption of the chemical. However, in some cases, increases in concentration bring about subsequent increases in toxicity primarily due to local irritation. The total volume of test material and vehicle also should be controlled in acute studies. In general, the injected volume should be kept small in relation to the s u e of the animal mode. Oral doses should not exceed 2-3% of the body weight while intravenous doses should be controlled to an upper limit of 0.5 mlfor rodents and 2 ml for larger animal models. (8). Should it be necessary to inject volumes greater than those recommended, then multiple dosing at several injection sites should be considered. Often, the relative toxicity of a test material is quite low. In such cases, it may be necessary to increase the concentration of the chemical in chronic toxicity test diets. In so doing, however, the relative percentage of nutrientain the diet may be compromised significantly. To control for this problem, paired feeding of control groups should be considered. Here, the paired controls are allowed only the amount of food consumed by the test group. Another approach is to include an inert material in the diet of the controls so that test and control groups experience the same dietary restrictions vis-a-vis daily intake of nutrients. Some final comments are offered regarding the diluent or test vehicle. This material should have a very low order of toxicity and it should not interfere with, or in any way modify, the physical and/or chemical properties of the test material. The diluent should not enhance, inhibit, or otherwise interfere in the uptake, distribution, metabolism, or excretion of the test material. Itsosmolarity, as well as the combined osmolarity of the test material and vehicle, should not affect local tissues. In short, the selection of a vehicle for use with those materials requiring dilution, solubilization, or suspension in a fluid media, should reflect a potential for significant modification in the actual toxicity of the test, regardless of the route and/or duration of exposure. In general, there are three major types of a control groups: negative, sham-treated, tive. Negative controls are selected a t ra the same population of animals as is use any test or treatment groups. The age (we range) and sex should be similar to the group($. Such controls are maintained in holding rooms and are examined, and sam body weight, hematological and clinical che determinations, etc., at the same time are the treated groups. The negative c then sacrificed as per protocol requirements, ally a t a time coincident with the sacrifice o treated animals. A complete gross and histopa ogy work-up comparable to that employed on the$ aexposed group(s) is then carried out on these con-'" trols. ,.+j& Sham-treated controls are selected in the sames manner as described above for negative These animals are then treated in exactly tchoentrols.*3 manner as are the test or exposed animals, exce the sham controls never encounter the material test. For example, subacute or subchronic te requiring intubation of the test material in a dilue would have a parallel control group which receiv the diluent via intubation. The volume of the ve- hicle so administered to these controls would be equal to the largest amount administered to the treatment group. Sham-treated controls for an in- halation test, regardless of its duration, would be maintained in an inhalation chamber and exposed to the same temperature, pressure, relative humi ity and air flow characteristics as maintained the treatment chamber. This group also would exposed to a vehicle should the latter be incorp rated in the test exposure regimens. An exam was offered in an earlier section of this review regarding the need to use paired feeding in those feeding studies where the test material, non-nutri- tional in nature, represents a sigmficant PO the diet. Such paired feeding studies also w considered to be sham-treatment controls. T h pose of the sham-treatment is to determine method of administering the test material co Utes in any way to the toxic effects observed in th EXPERIMENTAL CONTROL'S .treatment group. Therefore, every effort should be The design of any toxicity experiment should made to incorporate this control group in studies, reflect a concern for standardization, that is, a need particularly when the nature of the dosing tech- Section D investigato ;ion the trl le exposed 0. n an expenh respect to vironment in 1- maintained ent. Animals pes of anima! -.d, and pos random fro: used to form age (weight to the test ,ed in animal 1 sampled for ~ aclhemistry ie interval as 1 controls are .ements, us' icrifke of ti i histopatholdoyed on the on these con- i in the same .tive controls. .ctly the same iimals, except le material c )chronic te: .alin a diluen .hich received me of the ve-01s would be stered to the rols for an inion, would be r and exposed dative humidmaintained 1also would be ar be incorpo- .. An example ,f this review ?ding in those ial, non-nutriant portion of also would be rrols. The pur .tennine if tki .terial contrii bserved in the fort should he )up in studies. 3 dosing tech- Section D EXPERIMENTAL DESIGN CONSIDERATIONS 719 nique could be expected to influence the outcome i of the study. 1 Positive controls are included in studies which Z focus on a specific biological event, e.g. sensitization, cancer, mutation, teratogenesis, neuropathy, etc. In such investigations, it is often considered appropriate, and indeed prudent, to maintain a control group of animals which is treated in the same time frame as are the test group(s) with a material known to produce the toxic effect under study. A positive control group allows the investigator to titrate the level of sensitivity of the treated groups and to compare this relative sensitivity to that recorded in other similar tests. It is then possible to know with greater confidence that a negative test result is more a reflection of the inherent low toxicity of that test material than a relative lack of animal responsiveness. The use of positive controls also permits the comparison of oncogenic potency between positive controls and treatment groups. Finally, the use of positive controls provides an indirect check on the reliability of a test laboratory. The choice of the positive control for an investigation should reflect an awareness of the chemical structure and possible similarities in metabolism, distribution, and excretion, between that a control material and the test compound(s).This is particularly so with carcinogenicity bioassays and mutagenicity screens. However, teratology tests tend to rely on a handful of dissimilar compounds for the use as positive controls, 2.e. aspirin, 6-aminonicotinamide, etc. For neurotoxicology investigations, it would be appropriate to select a material whose chemical composition or tissue response most closely approximates that of the test substance. The selection of positive controls should reflect variable sensitivities that exist among animal species vis-a-vis a particular positive control. For example, aspirin is not an effective teratogen in rabbits. Some nitrosamines cause a specific form of tumor in some animal models but not in others. Organophosphorus compounds are relatively weak neurotoxins in rats and guinea pigs but they are - quite potent in chickens. It is well to point out that some investigators do not use positive controls because of the potential for contaminating the test The Test Chemical Among the variables that need careful consideration in toxicity investigations are those concerning the actual chemical to be tested. Sample purity and stability certainly should be ascertained before the onset of any investigation. Any such effort, however, presupposes the availability of an analytical + technique which has sufficient detectability. selec- i tivity, and reproducibility. As man). studies will involve the dosing of animals with a compound in t the parts-per-million or -billion range, the method of analysis will have to be capable of detecting extremely low quantities of the test compound while, at the same time, being able to discriminate that compound from the vast array of other chemicals potentially coexisting in the same environment, i.e. diet, chamber atmosphere, etc. The analytical technique should also be free from interference from any other chemicals existing in that environment. Finally, it should be possible to reproduce the quantitation of a material repeatedly and with time. Once such an and-yticd capability is in hand, then it is necessary to analyze the material to be tested to determine the nature and extent of any contaminants likely to be present with the test compound. Some studies are designed to assess the toxic properties of a pure chemical without interference from, or contribution by, contaminants present with the test compound. Other tests involve the evaluation of a chemical or product as used in commerce. In this instance, it is appropriate to examine the toxicity of the pure chemical and any associated by-products from its manufacture. The importance of an adequate analytical technique is nowhere more important than when applied to the control of an inhalation toxicology experiment. In the course of preparing for, and actively carrying out such a study, the investigator must be able to monitor the chamber atmosphere for the contaminant or level of test material. It is necessary to be able to make rapid adjustments in the delivery of test material to the chamber to compensate for excursions that occasionally occur in chamber contaminant concentrations. Where feasible, a continuous analysis of contaminant levels should be provided. This can be accomplished with the use of infrared and ultraviolet spectrophoto- metric procedures. Gas chromatographs can be em- ployed, but the analytical results lag by several minutes behind the actual sampling process. Regarding the latter, it is important to evaluate any procedure used to collect and/or transmit the contaminant from the chamber to the analytical unit. It is possible that the sampling train may introduce some interference in the actual level of contami- nant. This can be accomplished by altering the contaminant's physical/chemical properties, thus interfering with the actual measurement of the test material. In other instances, the test material can interact with, or plate-out on, the surfaces of the sampling train. In short, the latter can represent a major factor in any effort to monitor chamber contaminant levels effectively. Experiments involving the generation of aerosols or particulates present an added dimension to the problems of sampling and analyzing chamber atmospheres. These studies require the measurement 720 EXPERIMENTAL A N D CLINICAL NEUROTOXICOLOCY Section 0 of particle (aerosol)sizes, and their distributions, in order to obtain some information regarding the respirable load presented to the test animal. Experiments of this type usually require some preliminary measurements of chamber contaminant levels to assure the adequacy of the contaminant generation apparatus and to demonstrate that the sampling and analytical systems do not interfere with the calculation of the true contaminant level. These preliminary tests are conducted in chambers without the complement of test animals, yet, it is conceivable that the presence of animals can have a significant effect on measured contaminant levels. Animal fur, cage surface areas, animal-generated humidity (respiratory and waste material) all can play a role in decreasing the expected level of test material in the chamber. Environment The last major category of experimental design variables is that involving environmental conditions. Included in this grouping are temperature, pressure, himidity, light cycles, caging conditions, and diet. Doull (8) and Morrison and associates (20) provide a thorough discussion of the impact of these environmental parameters on the outcome of toxicity investigation. Temperature changes have variable effects on the rate at which exogenous materials are metabolized. For example, the mouse intraperitoneal LDm for amphetamines has been recorded to be about 6.0 mg/kg at 35"C, 155 mg/kg a t 21C and 84 mg/kg at 10C. On the other hand, increases in relative humidity decrease the LDm values for a number of chemicals. Pressure is more likely to present a problem in inhalation toxicity studies than it will in other areas of toxicology. The relative chamber airflow rates represent the source of this problem. Lighting conditions can play a major role in affecting the internal homeostatic patterns or biological rhythms of animals. This is particularly so with rodents. The latter are nocturnal and yet tend to be maintained in rooms which are lit during daylight or working hours. Although these animals can and do adjust rapidly to permanent changes in light cycles, they tend also to be sensitive to sporadic shifts in the light cycle. This sensitivity is reflected by changes in LDm values, mixed function oxidase activity, circulating blood elements, etc. The question of humidity already has been raised during the discussion of the control of inhalation chamber. contaminant levels. This environmental factor also may play a sigmficant role in the establishment of infections in the respiratory apparatus though, to be sure, the humidity would have to be decreased well below 40%. In contrast to this, some investigators have reported that increases in rela- tive humidity will result in a contaminant decrease in L D m values. Regarding caging conditions, Doull(8) notes that : sensory input to the animal brain is capable of modifying hormonal and peripheral nervous system ~ output. That input can be affected by the number of animals per cage, the nature of the cage, i.e. solid floor uersus wire mesh, the type of litter material, relative noise levels, activity patterns within the animal holding room, etc. As studies extend beyond the 2-week (subacute) type to the 2-year-plus (chronic) program, these factors take on greater significance regarding their potential impact on the outcome of the study. Finally, diet control represents a key variable in the overall consideration of an experiment. A chemical analysis of the diet should be conducted to determine the level of metal contaminants, potential microsomal enzyme inducers, i.e. polycyclic aromatic hydrocarbons, carcinogen, i.e. aflatoxin, etc., protein and fat contents, etc. Depending on the goals of the study, these dietary variables could offer a significant contribution to any observed toxicity. Mention already has been made of the',? potential for interaction between dietary compo- T:- nents and test materials. Such interaction could easily result in an inactivation of the t In short, by careful consideration and attentio to the potential effects associated with these design variables, the investigator can expect to ha greater sense of confidence that those biolo effectsobservedin conjunction with a toxicitys are indeed due to exposure to the test compo and not to an uncontrolled variable. DATA EXTRACTION GENERAL STUDIE Any well-conducted toxicology study, whethe directed primarily toward evaluatio icity or not, should include observati of a number of non-neurologic parameters most commonly monitored of these indicato whole body weight, organ weight, ble functional signs, hematologic and c istry analytical data, and the observati at the time of gross necropsy and subsequent topathologic examination. The following is a supe ficial discussion of each of these general data are For single-dose probe studies, b erally need be taken only at the of the experiment. For subchro proximately 2 week duration, body weights be taken on a daily basis; for subchronic stu the order of 3-month duration, w should be recorded. Beyond the 6th month of for a chronic study (i.e.,2 years/lifetime in rodents -. .- =a Section D EXPER1M ENTAL DESIC N CONSIDERATlONS 721 biweekly or monthly body weight determinations should be adequate. Although comparison of control and treatment groups by Student's t test (29, 30). .. evaluation of mean group body weights at selected time intervals on test. .. is most commonly encountered in the literature, a more sensitive test can be applied. A linear approximation of the body weight plotted versus the logarithm of the time on test should be constructed for each animal. The slope of each of these lines may be determined graphically. A mean slope and the standard error of that mean for each treatment group then can be caiculated. These mean slopes may be treated as normally distributed independent data, and the mean slopes of the weight gain functions for each treatment group may be compared by Student's t test. A significant reduction in the slope of the body weight gain function for the "test group" should be considered indicative of nonspecific systemic toxicity. Both absolute organ weight data (simple wet weight of the carefully dissected organs) and relative organ weight data (absolute organ weights normalized for total body weight) are commonly used by toxicologic investigators. Considerable controversy persists over the relative merits of each method of expression. With adequate attention paid to data variance, they are probably equivalently ; sensitive (when sex is taken into account and over` all body weights and ages are roughly equivalent). : Nonparametric statistical tests are ideally applied with either absolute or relative organ weight data (videinfra). In the particular case of the use of brain weight data as a measure of neurotoxicity, several comments are in order. First, the CNS has a limited capacity for reactive hyperplasia (vide infra),in contrast to the kidney, heart, etc.; therefore, increases in CNS weight are quite unlikely except in the case of neoplasia or massive edema, both of which would be detectable by gross or histopathologic examination. Second, the weight of the brain is generally taken to be one of the most nearly constant of all the organ weights; second only to total body weight in its frequency of use, the brain weight is a basis for normalization of other organ weights. Third, the CNS apparently replaces many of the neuronal structures, lost through whatever process, with connective tissue (glialcells);the likelihood of detecting subtle changes by determination of gross weight is therefore very low. Fourth, organ weight is by its very nature a relatively widely distributed variable and so is not likely to be as sensitive an indicator of toxicity as are either functional or morphologic evaluation. Measurement of hematologic parameters generally is taken to be a sensitive indicator of both nonspecific cytotoxic effects on rapidly dividing cells and of disturbances of mature cell morphology and physiology (i.e. typical erythrocyte biconcave conformation, osmotic fraghty). It should be remembered that general anemias develop and progress only relatively slowly, this being the particular case with agents that are only weakly active. Reactive hyperplasia of the bone marrow (considerable reserve exists) may delay the appearance of, or completely ablate, the toxic response if only peripheral blood analyses are performed; examination of bone marrow is indicated in any toxicology study wherein hematotoxicity is suspected. Total and differential leukocyte counts may indicate only the intercurrent state of the subject's overall health. Interpretation of these data should be the exclusive province of a hematologist trained and experienced with the species under study. These hematologic data very rarely can be considered pathognomonic, but rather they provide an opportunity to identify qualitatively the scope of a toxic response. In no cases should highly sophisticated statistical hypothesis testsbe applied to these data; simple Student's t test or analysis of variance (29, 30) may be performed when n L 30 and the following are kept in mind: Changes in these parameters are highly nonspecific. There is considerable variation within what is considered the normal range. There is considerable quantitative uncertainty inherent in the methods of determination Most grossly observable signs do not fulfill the criteria either of objectivity or of independence, the latter of which is required for statistical analysis. Compromise is struck in the case of grading (for severity) of histopathologic lesions and staging (for severity) of grossly observable signs. Because the intervals separating the discrete classifications are not uniform, only simple nonparametric statistical tests should be applied (26).The principal weakness of such scoring systems is classically demonstrated in the fact that there exists no uniformity of opinion among clinicians and/or pathologists as to the precise dividing lines between the classes/grades/ stages. The best course is to restrict evaluation of these data to the judgment of the trained and experienced clinical professional and abandon statistical testing of data on grossly observable functional signs. Analysis of the following clinical chemical parameters may be considered a standard diagnostic profile for descriptive toxicology: Blood glucose concentration Blood urea nitrogen concentration Serum glutamate-pyruva te transaminase (SGPT)activity 722 EXPERIMENTAL A N D CLINICAL NEUROTOXICOLOCY Section D Serum glutamate-oxaloacetate transaminase have documented the preferred histologic prepa- (SGOT) activity ration protocol (see also Chapter 50). Creatine phosphokinase activity Physioiogical abnormalities in the absence of Lactate dehydrogenase activity morphological change have been documented. Such Alkaline phosphatase activity is the case when animals are exposed to acute1 Serum (or plasma) bilirubin concentration narcotizing concentrations of volatile low mole Serum cholesterol concentration weight hydrocarbon solvents. These are appar Serum triglyceride concentration entirely pharmacologic/biochemicalevents an Serum (or plasma) inorganic phosphorus con- presumably completely reversible. These fun centration decrements in the absence of histopathol Serum calcium concentration customarily associated with acute overexpo Serum potassium concentration although the signs may recur frequently wit Serum sodium concentration peated overexposure (and thus may m h c Serum chloride concentration chronic intoxication). Finally, very little Serum total protein concentration dence can be brought to bear on the co Senun uric acid concentration sensitivities and specificities of morpholoec u Urine specific gravity functional parameters. Urine protein concentration Behavioral toxicology (a subunit of neurotoxic Microscopic examination of urine for cells/ ogy) can be divided principally into two classes o casts. studies. The first includes investigations of an Each of the above tests is intended to survey the ological character in which the amount and q functional status of a particular organ system, but tative type of spontaneous motor activity is mo vanishingly few are so specific. For instance, both tored. The advantages of these studies are that + transaminase activity assays (SGOT and SGPT) require no special equipment (given the dispos are taken to reflect the hepatotoxicity of an agent. of the investigator to catalog behaviors However, both may be elevated by result of trau- the variables are relatively easy to m matic injury, myocardial infarction, or by acute or behaviors are spontaneous, and the tests have chronic diseases of heart, muscle, and liver tissues. validity (23). However, these tests are relati Other tests are exquisitely sensitive to the recent nonspecific; the data are heavily influenced b prior state of the subject; glucose and triglyceride experimental environment (equipment,biorhyt assays are normally performed bn 18-h-fasted sub- of the animals, and the presence of distracti jects. Some variables exhibit fairly narrow "nor- and the age, health, nutritional status, and pn mal" ranges (2.e.st5% for serum calcium concentra- status of the subject. Furthermore, the face validi tion), whereas others vary widely (2.e. k60% for of the test is in question; a variety of agents (i serum triglyceride concentrations) (3). It is the chlordecone, scopolamine) may produce incr evaluation of the complex, but coherent, sum of all in activity in one situation and decreases in act of these data which forms the basis for interpreta- in another (22).A particular behavioral change tion; individual atypical values should not be over- rarely is diagnostic of a specific neurotoxic eve emphasized. A discussion of the significance of the The prudent investigator monitors more than results of each of these tests in both healthy and behavioral indicator and under more than one unhealthy individuals can be found in Docurnenta perimental control situation. More than perha Geigy ( 7 ) . any other of the data areas considered in t As with any definitive toxicology experiment, a cussion, the prudent investigator employs o detailed gross dissection with all notable findings "two-tailed tests" of the nonparametric type recorded at the time of necropsy is absolutely re- evaluation of data from such behavioral toxic01 quired. Here, as with other details of protocol, a studies. balance must be struck in the judgment of the The second principal class of behavioral to investigator between a comprehensive, but super- ogy research is the study of the disintegration ficial, necropsy and an exhaustive, even semimi- operant conditioned response. "A great stre . .L croscopic, dissection of specific organs and tissues. operant conditioners lies in their ability to I? SPECIALIZED STUDIES facture behavior to specifications, manipulatm way an animal's responses pay-off" (16). In th Although adequate for most histopathologic eval- studies, a subject is "taught" by conventional uations in general toxicology studies, conventional ods of behavioral shaping and operant condit histologic techniques for use of the light microscope to present a particular behavior in a parti are generally inadequate for determination of subtle environmental situation. Following exposure t neuropathologic changes. Spencer and Bischoff (31) putative toxic agent, the subject is presented ~1 Section D EXPERIMENTAL DESIGN CONSIDERATIONS 723 the specific environmental condition, and performance of the anticipated conditioned response is documented and analyzed. Thompson and Moerschbaecher (33) have shown that the slope of the learning curve for an operant condition may be a more sensitive indicator of behavioral toxicity than is disintegration of the pre-established condition after exposure to the putative neurotoxin. This observation is equivalent to the more sensitive evaluation of the time-dependent body-weight-gain function as discussed earlier. Maximum nerve conduction velocity is useful for detecting the effects of certain neurotoxic agents. This test is most sensitive for identification of agents which preferentially damage the largest and most heavily myelinated motor fibers. The sensitivity of the test decays in inverse proportion to this specificity. That is, if large diameter nerve fibers are not affected by a toxic agent, regardless of size or degree of myelination, the maximum nerve conduction velocity as monitored by recording muscle contraction, would be expected to be affected very little, if a t all. Early neurotoxic damage produced by most axonal neurotoxins (hexacarbons excepted) would go undetected by determination of maximum nerve conduction velocity, whereas such a test would be chosen to detect neurotoxic damage produced by agents which primarily damage the myelin sheath. In cases where neurotoxicity of the axonal type is suspected, determination of the amplitude of the sensory nerve action potential is indicated (17). STATISTICAL TESTS In studies of the effects of chemical agents on physiologic function, it is possible (and desirable) to use the "within subjects" design. With employment of the so-called Latin-square full-crossover design (20),each subject can serve as its own control in the study of partially or completely reversible neurologic decrements. In studies of irreversible functional decrement or in studies of morphologic damage (requiring the sacrifice of the test animal following treatment), the "between-subjects" design is required. In either case, a sufficient sample should be drawn to control cost effectively the probability of committing a type I1 statistical error (a failure to identify an experimental difference as real). The error variance (random noise) may be quite different for each of the above designs, even when the same agent is tested a pilot study (t-ypically of about 10 animals in each of the control and treatment groups in the between-subjects design, or 20 animals total in the within-subjects design) will aid greatly in estimating the population variances, which can, in turn, be used to estimate the optimum number of subjects to be studied in the final experiment (30). The diligent investigator should always consider a pilot study when dealing with a new variable (either independent or dependent). No study should ever be undertaken which does not include a rigorous concurrent control group (untreated or sham-operated) against which the subjects treated with the test agent are compared. As a practical midline in the selection of appropriate statistical tests, one can apply well L. E. Moses' "Principle of the Blunt Ax" (21). "If the ax chopped down the tree, it was sharp enough." Freely paraphrased, this can be taken to say that statistical overkill is almost invariably counterproductive. Since many, if not most, neurotoxicologic end points are non-normally distributed, since aberrations of hypothesis testing are introduced by such non-normally distributed data, and since the nonparametric statistical tests do not assume normally distributed data, one should select from the cadre of nonparametric statistical tests for hypothesis testing in neurotoxicology. In general, the nonparametric tests are less powerful than are their parametric counterparts; said simply, one is less likely to discover a subtle difference by use of nonparametric hypothesis test than by use of an equivalent parametric method. The sacrifice, however, is mandated by the non-normally distributed data. When "n" 530, the simplest test is best. The Fischer Exact Probability Test (from a "so-called" 2 x 2 contingency table) is fully explained in Nonparametric Statistics (26).Since, in most cases, the problem to be examined is one of events which differ in kind rather than in magnitude (that is, "were the subjects normal or not normal?" rather than "were exposed subjects less able to perform a task than were the unexposed?"), the Fischer Exact Probability Test is the most appropriate to apply. When "n"exceeds 30, the chi-square test should be used; the single excepting qualification is given in Nonparametric Statistics. (20): Note: When the variable under scrutiny is the presence or absence of a phenomenon which by its very presence (neurohistopathology) is incontrovertible proof of effect, then the use of any statistical analysis is unnecessary (and undesirable). For statistical evaluation of data for variables along a scale of relative intensities ( i e . ,a > b > c, and a, b, c # 0), the Mann-Whitney U-test is an excellent choice. The test takes full advantage of all the information from ordinal scaled data and is, therefore, exceptionally powerful among nonparametric tests. Among the data for which the U-test might be applied are the quantitative morphologic 1 1. i, 724 EXPERIMENTAL A N D CLINICAL ;UEUROTOXIGOLOCY I Section 0 parameters suggested by Norton (221, nerve con- Each of these particulars can be generalized to a duction velocities, or sensory action potential am- set (i.e. of agents or of environmental circum- plitudes. stances) or to a range (i.e. of exposure levels). There is no magic formula for determination of Safety, in the common usage of the word, is an an ideal sample size for experimental test. For a absolute term describing a state in which either the given level of power in any statistical test, there is agent is absolutely ineffective (nontoxic) or the an inverse relationship between the intrinsic vari- target is absolutely insensitive (immune). If Para- ation in the measured variable and the sample sizes celsus were presumed to be correct when he wrote of control and treatment groups which are required "Allsubstances are poisons; there is none which is to discriminate a difference of any given magnitude. not a poison," then safety is achievable only In the absence of pilot trial data (vide supra), the through "zero exposure." Restated, some nonzero purely apocryphal rule of thumb, n = -30, may be risk, however immeasurably small, is associated applied; the warning to be sounded is that such with a nonzero exposure to a toxic agent. arbitrary experimental design may seriously com- The "toxicity test" is conducted-usually includ- promise the power of any hypothesis test. ing high and low dose exposure levels, for the purpose of qualitatively identifying the entire set of SIGNIFICANCE OF ANIMAL TESTING deleterious effects which an agent is capable of causing in a given set of exposure circumstances. In The proper application of animal testing data to its most complex form, the toxicity test may also - the human situation begins with distinctions among attempt to demonstrate a "no discernable effect the various terms which classify the data base. level" which is presumed to represent a combina- There is almost universal disagreement on the exact tion of the effective biological threshold and the meanings of the words toxicity, hazard, risk, and minimum detectable limit (i.e. sensitivity) of the , . - safety; but, for the present purpose, these defini- test. tions are offered. Of the several elements of "risk," two are the Toxicity is the extrinsic property of a particular principal foci of animal testing, namely, toxicity material to cause damage (morphologic or func- and potency. The outcome of toxicity test deter- tional) to living organisms. Prefures to the word mines whether or not the agent is capable of pro- more narrowly define the target-i.e. neurotoxicity, ducing a toxic response at any dose under any nephrotoxicity, hepatotoxicity. Tqxicity is a nomi- circumstances. Such tests are commonly referred nal scale; a particular toxic capacity either exists or to as hazard evaluation bioassays by virtue of the it does not. However, the magnitude of toxic re- fact that they are directed toward a particular kind sponse is assumed to be straightforwardly depen- of toxic response (hazard). dent upon the magnitude of exposure. The so-called "safety test" (more appropriately Hazard is the seriousness of the particular toxic named "relative safety test") is more complex and response. For instance, a material which is capable may employ several dose/exposure concentrations of causing only transient CNS depression might be and schedules. The principal purpose of the test is stipulated to be less hazardous than a material to focus resources on the suppression of the mini- which is capable of producing irreversible paralytic mum detectable limit (sensitivity) of the test and neurotoxicity. the consequently improved certainty of identifica- Risk is the complex product of toxicity, exposure, tion of the effective threshold. The safety test de- sensitivity of the target, potency of the agent, and termines potency. the coincidence of other environment circum- The absolute value of the empirically determined stances which may alter the agent or target. Risk is threshold and the confidence interval on that value the practical likelihood (finite probability) that a may then be permuted with what is known or particular agent will cause a particular undesirable believed of the relative seriousness of the effect, of effect of a particular magnitude in a particular the sensitivities of the target organisms, and of the organism at a particular exposure level under a environmental circumstances of encounter to esti- particular set of environmental circumstances. mate relative risk. REFERENCES 2: f &i & Iu ~ E=: 1. BALAZS T iMeasurement of Acute Toxlcology. In: Methods in Toxicology, edited b y G. E. Paget. Blackwell Scientific Publications, Oxford, 1970. 2. BALAZS T: Assessment of the Value of Systemic PI 1 2 .Toxicity Studies in Experimental Animals. In: A d - z!uances in Modern T O X L C O V~oOl. ~I,~N,ew Concepts in Safety Evaluation, edited by M. A. Mehlman, R. (; E. Shapiro, and H. Blumenthal. Hemisphere Publish- b t Section D EXPERIMENTAL DESIC N CONS1DERATIONS 725 ing Corporation, Washington, D. C., 1976. 3. BIO-SCIENCE LABORATORIES: The Bio-Science Handbook, Specialized Dingrwstic Laboratory Tests. Bio-Science Laboratories, Van Nuys, Calif, 1975. 4. BRODIE BB: Distribution and Fate of Drugs; Ther- apeutic Implications. In:Absorption andDistribution ofDrugs, edited by T. B. Binns. Williams & Wilkins, Baltimore, 1964. 5. CAVANAGH JB: Peripheral neuropathy caused by ~~ chemical agents. CRC Critical Reviews in Toxicob& 2.365. 1973. 6. C O N h Y AH, SCHNEIDERMAN K, JACOBSON M, KUNTZMAN R Drug-induced changes in steroid metabolism. Annals of the New York Academy of Sciences 123, 98,1965. 7. Documenta Geigy, Scientific Tables, ed. 7, edited by 1 I I K. Diem and C. Lentner. J. R. Geigy S. A., Bade, 1970. 8. DOULL J: Factors Influencing Toxicology. In: Toxi- cology: The Basic Science ofPoisons, edited by L. J. ! Casarett and J. Doull. Macmillan Publishing Company, New York, 1975. 9. FOUTS JR, HART LG: Hepatic drug metabolism i i ! during the perinatal period. Annals of the New York Academy of Sciences 123,245, 1965. i 10. GRICE H, MUNRO IC, WIBERG GS, HEGGTVEIT HA: The pathology of experimentally-induced cobalt i: cardiomyopathies. A comparison with beer drinkers cardiomyopathy. Clinical Toxicology 2,273, 1969. ! . 11. GRIFFITH J F Inter-laboratory variations in the de- termination of acute oral LDm. Tonicology and Ap- plied Pharmacology 6, 726,1964. 12. HAWKINS JE, LAURIE MH: The ototoxicity of streptomycin. Annals of Otology, Rhinology, and Larynogohgy 61,789,1952. 13. HAYES WJ Jr.: The 90-dose LDm and chronicity factor as measures of toxicity. Toxicology a n d Ap- ' plied Pharmacology 11,327,1967. 14. HAYES WJ Jr.: Tests for Detecting and Measuring Long-Term Toxicity. In: Essays in Toxicology, Vol. 3, edited by W. J. Hayes, Jr. Academic Press, New York, 1972. 15. JONDORF WR, MAICKEL RP, BRODIE BB: Insta- bility of newborn mice and guinea pigs to metabolize drugs. Biochemical Pharmacology 1,352,1959. 16. LATIES VG: How operant conditioning can contrib- ute to behavioral toxicology. Environmental Health Perspectives 26, 29, 2978. 17. LeQUESNE PM: Clinical expression of neurotoxic injury and diagnostic use of electromyography. En- vironmental Health Perspectives 26, 89, 1978. 18. LU FJ, JESSUP DC, LAVALLEE A Toxicity of pesticides in young versus adult rats. Food and Cos. metic Toxicology 3, 591, 1965. 19. MENDELL JR, SAIDA K, GANANSIA MF, JACK- SON DB, WEISS H, GARDIER RW, CHRISMAN C, ALLEN N, COURI D, O'NEILL J , MARKS B, ' HETLAND L Toxic polyneuropathy produced by methyl n-butyl ketone. Science 185, 787, 1974. 20. MORRISON JK, QUINTON RM, REINERT H: The Purpose and Value of LDm Determinations. In: Mod- ern Trends in Toxicology,Vol. I, edited by E. Boyland and R. Goulding. Appleton-Century-Crofts, London, 1968. 21. MOSES LE: Principle of the Blunt Ax.In: Biostatis- tics, p. 62, edited by A. Goldstein. Macmillan Publishing Company, New York, 1964. 22. NORTON S: Is behavior or morphology a more sensitive indicator of central nervous system toxicity? Environmental Health Perspectives 26,21, 1978. 23. REITER L Use of activity measures in behavioral toxicology. Environmental Health Perspectives 26,9, 1978. 24. RONA G, CHAPPEL CI: Pathogenesis and Pathology of Cobalt Cordiomyopathy. In: Recent Advances in Cardiac Structure and Metabolism, edited by E. Bajusz and G. Rona. University Park Press, Baltimore, 1973. 25. RUMKE C L Some Limitations in Animal Tests. In: Evaluation of Drug Activities Pharmacokinetics, Vol. I, edited by D. R. Laurence and A. L. Bacharach. Academic Press, London, 1964. 26. RUNYON RP: Non-Parametric Statistics, A Contemporaq Approach. Addison-Wesley Publishing Company, Reading, Mass., 1977. 27. SCHOENTALR, MAGEE PN: Chronic liver changes in rats after a single dose of lasiocarpine, a pyrrolizidine (senecio) alkaloid. Journal of Pathology and Bacteriology 74, 305, 1957. 28. SETNIKAR I, MAGISTRETTI MJ: The toxicity of central nervous system stimulants in rats of different ages. Proceedings of the European Society for Drug Toxicity 4, 132, 1964. 29. SNEDECOR GW, COCHRAN WG: StatisticalMethods, Ed 6. IowaState University Press, Ames, Iowa, 1967. 30. SOKAL RR, ROHLF FJ: Biometry. W. H. Freeman and Company, San Francisco, 1969. 31. SPENCER PS, BISCHOFF MC: Contemporary Mor- phological Techniques for Evaluating Peripheral Nerves. In: Management of Per&heralNerve Problems, edited by 0.K. Omer and M. Spinner. Saunders, Philadelphia, 1980. 32. SPENCER PS, SCHAUMBURG HH: Distal axonopathy: one common type of neurotoxic lesion. Enuironmental Health Perspectives 26, 97, 1978. 33. THOMPSON DM, MOERSCHBAECHER JM: Op- erant methodology in the study of learning. Environmental Health Perspectives 26, 77, 1978. 34. WEIL CS. CARPENTER CP, WEST JS, SMYTH HR J R Reproducibility of single oral dose toxicity testing. American Industrial Hygiene Association Journal 27,483, 1966. 35. WEIL CS, McCOLLISTER DD: Relationship between short and long term feeding studies in designing an effective toxicity test. Journal of Agricultural Food Chemistry 11,486,1963. 36. WEIL CS, WRIGHT GJ: Intra- and interlaboratory comparative evaluation of single oral test. Toxicology and Applied Pharmacology 11,378,1967. 37. WHEELER AG, DANSKY D, HAWKINS HC, PAYNE HG, WEIKEL JH: A toxicologic and hematologic evaluation of cyclophosphamide (CYTOXAN) in experimental animals. Toxicology and Applied Pharmacology 4, 324, 1962. 38. YEARLY RA, BENISH RA: A comparison of the acute toxicities of drugs in newborn and adult rats. Toxicology and Applied Pharmacology 7,504, 1965.