Neurotoxicologyand Teratology, Vol. 14, pp. 199-203, 1992

0892-0362/92 $5.00 + .00 Copyright © 1992 Pergamon Press Ltd.

Printed in the U.S.A. All rights reserved.

Study Design Considerations in Developmental Neurotoxicology HUGH

A. TILSON

Neurotoxicology Division, Health Effects Research Laboratory, U.S. En vironmen tal Pro tection Agency, Research Triangle Park, N C 2 7711 R e c e i v e d M a y 1991; A c c e p t e d S e p t e m b e r 1991 TILSON, H. A. Study design considerations in developmental neurotoxicology. NEUROTOXICOL TERATOL 14(3) 199-203, 1992.- It is widely accepted that exposure to environmental factors during development can result in effects other than death, gross structural abnormality, or altered growth. One area of concern is the developing nervous system, which may be especially vulnerable to environmental perturbation. Testing chemicals for potential developmental neurotoxicity has received a high priority and testing guidelines have been published and recently revised by the U. S. Environmental Protection Agency. These guidelines are based on several principles of developmental neurotoxicity that have been developed during several years of research. In general, manifestation of neurotoxicity following developmental exposure can depend on the time at which exposure occurs and for the purposes of hazard detection, experiments should be designed to optimize the detection of neurotoxicity. In addition, maternal health and interaction with the offspring, as well as postnatal development are important design issues in developmental neurotoxicology. It is also widely accepted that several doses be used and multiple measures of neurotoxicity assessed in both genders at several points during the life span of the animal. Finally, the litter is usually regarded as the most appropriate statistical unit to control for genetic and maternal factors. Developmental neurotoxicology

Experimental design issues

Polybrominated biphenyls

Mice

METHOD

IT is widely believed that the developing nervous system is particularly vulnerable to perturbation by environmental agents (3,7,8,12). In addition, exposure to some agents having little or no effect in the adult can cause persistent alterations following exposure to the same agent during development. The potential risk to h u m a n health following developmental exposure to environmental agents has led to the promulgation of testing guidelines for premarket approval o f some chemicals. A draft version o f the Environmental Protection Agency (EPA) test guidelines for developmental neurotoxicity was recently discussed in a workshop sponsored by the E P A (2) and a final version of the E P A developmental neurotoxicity testing guidelines has recently been published (6). A l t h o u g h the effects o f developmental exposure on the nervous system have been studied for several years and protocols for developmental neurotoxicity are widely available, there are several experimental design issues that are sometimes overlooked, including the need for multiple endpoints and doses, evaluation at various times during maturation, possible influence of maternal toxicity, gender differences and several statistical considerations. To illustrate these points, the methods and results of a study concerning the effects o f perinatal exposure to a mixture o f polybrominated biphenyls is summarized. These data were reported in preliminary f o r m elsewhere (13).

Experimental Animals Twenty-four gravid C57B1/6 mice were housed individually in plastic-home cages in a r o o m having a constant lightdark cycle (light, 07:00 to 19:00 h), temperature (21 + 2°C), and relative humidity (50 + 10°70). F o o d ( N I H diet #31) and water were freely available throughout the experiment.

Dosing and Procedure The mice were dosed by gavage with 0, 3, or 10 m g / k g of FireMaster ( F F - I ) every other day during gestation and until weaning of the offspring at 21 days of age. F F - I , a commercial mixture of polybrominated biphenyls (PBB), was suspended in corn oil vehicle. Dams were weighed on gestational days 0, 7, 14, and on postnatal days 1, 7, 19, and 21. Offspring were counted at birth and culled to three pups per sex. The pups were weighed on days 4, 14, and 21. On postnatal day 21, the pups were weaned. One male and female pup per litter were selected for neurobehavioral testing, i.e., acoustic startle response, negative geotaxis, m o t o r activity, and body weights, at 30, 60, and 120 days o f age. One pup per sex was assigned for avoidance learning and neurochemical measurements at 30 days o f age, whereas the remaining animals were assessed for learning and

This article has been reviewed by the Health Effects Research Laboratory, U.S. Environmental Protection Agency, and approved for publication. Mention of trade names or commercial products does not constitute endorsement or recommendation for use. 199

200

TILSON

neurochemistry at 120 days o f age. Mice were housed individually after weaning.

TABLE 2 EFFECTS OF PERINATAL EXPOSURE TO PBB ON OFFSPRING PRIOR TO WEANING

Neurobehavioral Tests Average Value +_ SE"

Avoidance acquisition and retention. Mice were tested for the ability to acquire a one-way avoidance response in an apparatus consisting of a small Plexiglas chamber with grid floors. A speaker and white lamp were located at the rear of the chamber opposite a guillotine door leading to a larger, darkened chamber with a solid floor. Mice were given 4 acquisition trials that consisted of the presentation o f light, tone (110 dB, 5 kHz), and electric footshock (0.8 mA) applied to the grid floor. Presentation o f the stimuli coincided with the raising of the guillotine door. The trial ended when the mouse moved from the small cubicle and to the larger chamber. An arbitrary upper limit o f 40 s was established but never used. Acquisition was followed by 2 avoidance trials that consisted of the presentation o f the light and tone stimuli only for up to 20 s. If the mouse entered the larger chamber, the trail was terminated and an avoidance response recorded. If the mouse did not enter the larger chamber after 20 s, electric footshock was applied to the grids as in the acquisition trials. The latencies of responses made during the 4 acquisition and 2 avoidance trials were recorded using electronic timing circuitry. Five min separated each o f the 6 trials. T w o weeks later, the animals were given 2 retention trials that were methodologically identical to the avoidance trials. Acoustic startle reflex. Mice were placed on a Plexiglas platform mounted on a universal transducer. Output from the transducer was connected to a solid-state electronic instrument to measure the amplitude o f the peak startle response. Each animal was allowed to acclimate for a period o f at least 20 s and then was presented with a single 110 dB, 5 k H z tone having a duration of 100 msec. If no measurable response occurred on the first trial, up to 2 additional trials were given. If no response was elicited on these trials, an arbitrary score of 0.1 was assigned. Motor activity. Commercially available activity monitors placed in sound-attenuating outer chambers were used to measure m o t o r activity. Mice were placed singly into darkened test chambers and activity measured for 30 min. Negative geotaxis. Mice were placed with their head up-

TABLE 1 MATERNAL BODY WEIGHTS DURING GESTATION AND LACTATION Average Body Weight (G) ± SE °

PBB (MG/KG) Age

Control

3

10

18.4 _+ .3 20.7 + .4 28.5 +_ .4

18.4 _+ .4 20.8 +_ .8 29.0 + .4

19.1 _+ .3 21.0 _+ .3 28.5 _+ .4

25.4 28.4 30.7 31.0

26.0 28.7 29.6 32.0

26.8 28.4 30.2 30.9

Gestational day 0

7 14 Lactational day 1 7 19 21

PBB (MG/KG) Measurement

Control

3

10

Pups/litter Body weights (G) PN day 4 14 21

6.8 ± .2

6.9 _+ .4

6.8 + .3

2.3 _+ .1 8.6 + .1 11.4 + .4

2.4 ± .1

2.4 ± .1

8.8 + .3

8.8 + .4

12.1 + .4

11.8 _+ .7

aAverage number of pups and body weights of all pups in the litter were determined and the value used to obtain a group mean for each treatment.

ward onto the middle of a wire screen attached to a rectangular square frame positioned at a 60 ° angle. The frame was then rotated 180 ° and the time required for the mouse to orient to within 45 o o f facing up was measured with a stopwatch. An upper limit of 30 s was used and 1 trial was given.

Neurochemical Assays The biogenic amine content of the hippocampus, cortex, and striatum was assayed by high performance liquid chromatography using a method reported elsewhere (11). Filtrates from centrifuged brain homogenates were assayed for serotonin, dopamine, and their acid metabolites by reverse phase H P L C using an electrochemical detection system. C o m p o u n d s assayed were dopamine (DA), homovanillic acid (HVA), serotonin (5-HT), 5-hydroxyindoleacetic acid (5-HIAA) and norepinephrine (NE). Neurochemical assays were performed 48 h after the last retention trial.

Statistical Analyses A repeated measures analysis of variance ( A N O V A ) was used to determine the effect of treatment, gender, a n d / o r age on maternal body weight gain and neurobehavioral endpoints where appropriate (14). Posthoc analyses between groups were made using a Duncan's Range Test. N u m b e r of pups per litter and body weights of the pups prior to weaning were averaged for each litter and means for each group were calculated using the litter as the statistical unit. A repeated measures A N O V A was also used to evaluate the preweaning body weight data, whereas a one-way A N O V A was used to assess effects o f PBB exposure on litter size. Data from the learning and neurochemistry experiments were treated as separate studies and a two-way A N O V A was used to test for overall effects o f treatment and age. RESULTS

Maternal and Preweaning Toxicity + + ± +

.4 .6 .6 .4

aThere were 8 mice per group.

± .5 +_ .6 + .4 + .6

_+ ± ± +

.3 .7 .4 .4

Repeated measures A N O V A found that perinatal exposure to PBB had no significant effect on weight gain of the dams during gestation (Table 1). Furthermore, there were no effects o f treatment on body weight gain of the mothers up to postnatal day 21. A one-way A N O V A of the number of pups per

D E S I G N C O N S I D E R A T I O N S IN N E U R O T O X I C O L O G Y litter using the litter as the statistical unit found no significant treatment effect. In addition, repeated measures A N O V A of the preweaning body weights o f the pups based on the mean f r o m each litter found no significant treatment effect up to postnatal day 21 (Table 2).

201 TABLE 4 SUMMARY OF NEUROCHEMICAL EFFECTS OF PERINATAL EXPOSURE TO PBB Days of Age 30

Neurobehavioral Effects Table 3 summarizes the effects o f perinatal exposure to PBB on several neurobehavioral endpoints taken after weaning. Repeated measures A N O V A found that perinatal PBB exposure had no significant effect on body weights o f mice at 30, 60, or 120 days o f age. A similar statistical analysis found that perinatal exposure to PBB decreased acoustic startle responsiveness in high-dose males and females at 30 and 60 clays o f age. Negative geotaxis latencies were decreased in males and females in both dose groups at 30 and 60 days of age. There were no effects on acoustic startle or negative geotaxis at 120 days o f age. M o t o r activity was not affected at 30 or 60 days o f age but was decreased in high-dose females only at 120 days o f age. In the learning experiments, separate groups o f mice were tested at 30 and 120 days. Two-way A N O V A found no effects o f treatment on acquisition o f the one-way avoidance task (Table 3). However, perinatal PBB exposure increased the latencies of avoidance responses o f males and females in both dose groups at 30 and 120 days of age. There were no effects on retention 2 weeks later at either age o f testing.

Region Monoamine

Cortex 5HT 5HIAA DA HVA NE Hippocampus 5HT 5HIAA DA HVA NE Stfimum 5HT 5HIAA DA HVA NE

120

M

F

M

F

NE" NE NE NE

NE NE NE NE

NE NE NE NE

NE NE NE NE

NE NE NE NE NE

NE NE NE NE NE

NE NE NE NE NE

NE NE NE NE NE

NE NE NE NE NE

NE NE NE NE NE

NE NE ,b NE NE

NE NE NE NE

aNo effect. bStatistically significant decrease in content.

Neurochemical Effects The neurochemical data were analyzed for treatment effects on each m o n o a m i n e in each region o f the brain using a factorial A N O V A . The only significant finding was a decrease in dopamine content in the striatum o f high-dose mice o f both sexes at 120 days of age (Table 4). DISCUSSION The experiment summarized in this article serves to illustrate several issues important for the design o f experiments and interpretation o f data from developmental neurotoxicity studies (Table 5). The results o f the PBB studies support the

TABLE 3 SUMMARY OF EFFECTS OF PERINATAL EXPOSURE TO PBB Days of Age 30

60

120

Measurement

M

F

M

F

M

F

Body weights Acoustic startle Negative geotaxis Motor activity Learning Acquisition Avoidance Retention

NE" ,b ~ta NE

NE , ~d NE

NE l ~d NE

NE , ~O NE

NE NE NE NE

NE NE NE

NE 1"° NE

NE I"a NE

ND c ND ND

ND ND ND

NE To NE

NE Td NE

aNo effect. bT or ~, statistically significant increase or decrease. CNot done at that age. aSeen at both doses.

use o f several functional endpoints, including sensory, motor, and learning/memory. A recent workshop on the qualitative and quantitative comparability of human and animal developmental neurotoxicity found that the effects produced by developmental exposure to several human neurotoxicants, e.g., methylmercury, phenytoin, lead, ethanol, polychlorinated biphenyls, and ionizing radiation, were seen as alterations in cognitive, sensory, a n d / o r motor function (5). Similar to batteries used by several academic, governmental, and industrial laboratories (1), the battery of tests used in the present PBB study included assessments for these endpoints. It is important to use a battery of tests in hazard identification because it is not usually possible to predict a priori which neurobiological functions are affected by developmental exposure. Multifaceted test batteries optimize the probability o f detecting an effect if one is present. Another design issue illustrated by the present study is that developmental neurotoxicity should be assessed at different points in the lifespan o f the animal. Such measurements may

TABLE 5 SUMMARY OF EXPERIMENTAL DESIGN CONSIDERATIONS FOR DEVELOPMENTAL NEUROTOXICOLOGICAL STUDIES Evaluate multiple functional endpoints Measure at different points during lifespan Effects can depend on developmental period of exposure Determine contribution of material toxicity Gender differences Litter as statistical unit Range of doses

202 be necessary because effects can be transient or appear at a later age during development. In the PBB study, decreases in acoustic startle responsiveness and negative geotaxis latencies were present at 30 and 60 days of age but not at 120 days of age. Decreases in motor activity and brain dopamine concentrations were seen at 120 days of age but not at earlier ages. It is well known that the nervous system changes rapidly after birth and the manifestation of some effects may not be possible until the appropriate neural substrate has developed. Another design issue is that expression of neurotoxicity may depend on the time during which developmental exposure occurs. During the pre- and postnatal phase of development, the nervous system undergoes a relatively precise and timed sequence of cell division, differentiation, and migration. Chemicals could interfere with these processes, many of which occur at different times in various regions of the brain. Exposure to the same dose of a neurotoxicant at different times during development can result in different functional effects (9,10). In the present study, perinatal exposure to PBB resuited in several effects on neurobehavioral function later in life. In other experiments not reported here, prenatal exposure to PBBs had no significant effect on subsequent neurobiological development suggesting that PBBs have significant effects on processes that develop after birth in the rat. The PBB experiment also assessed developmental neurotoxicity in the context of maternal and postnatal toxicity endpoints. There is a widespread concern that agent-induced maternal toxicity can contribute to behavioral indicators of neurotoxicity in the offspring, confounding interpretation of the data (2). In the present study, developmental neurotoxicity was observed in the absence of any discernible effect on the health of the mother and weight gain of the offspring. Although it is possible that maternal toxicity could affect development of the offspring, it is not always the case. For example, food restriction to reduce weight gain of the dams (12%17%) does not produce the pattern of neurobehavioral toxicity observed with developmental exposure to PBB. Significant developmental neurotoxicity concurrent with maternal toxicity, however, is still considered to be a potentially adverse effect (6). Overlapping maternal and developmental neurotoxicity does not address the issue of possible differential sensitivities between the mother and offspring. Furthermore, adverse effects may be transient in the mother but persistent in the offspring (6). In the case where significant maternal toxicity may be a factor, other experimental designs should be considered including cross-fostering (12) or neonatal dosing. Another issue illustrated by the PBB experiment is that assessment of developmental neurotoxicity may require testing of both males and females. PBB-induced changes in motor activity, for example, were observed in female mice only. In cases of hazard identification where there is no a priori basis for predicting the sensitivity of one gender over the other, assessment of both sexes is recommended. Although the present example analyzed the two genders separately, the data could have been analyzed using gender as a main effect, allowing for a test of the gender x treatment interactions. This approach permits the investigator to identify chemicalinduced gender differences on a more appropriate statistical basis and is recommended. Developmental neurotoxicity studies require that maternal and genetic influences be controlled by using the litter as the statistical unit (4). In the PBB study, body weights of the offspring prior to weaning and the number of pups per litter were averaged and a value for the litter was obtained. Repeated measures or one-way ANOVA was then used to deter-

TILSON mine statistically significant treatment effects on these endpoints based on the litter as the statistical unit. The decision to use the litter as the statistical unit essentially determines the number of different categories of observations that can be made in any study. In the PBB study, one pair of litter mates was assigned for neurobehavioral tests at 30, 60, and 120 days of age, whereas the other two pairs were assigned to the learning and neurochemical studies. A repeated measures ANOVA was used to analyze the neurobehavioral data, whereas the learning and neurochemical data were analyzed using a two-way ANOVA. Addition of other endpoints, e.g., neuroanatomical measurements, would have required a separate pair of pups from each litter. Another statistical problem associated with developmental neurotoxicology studies is that multiple tests are performed at different ages. Such an approach may require some statistical correction because of the large number of statistical comparisons that are performed. The PBB study had over 100 tests for significance of the treatment effect or time x treatment interactions. If a nominal p level of 0.05 is used, then there could be at least 5 or more significant effects obtained based on chance alone (Type 1 errors). In the PBB study, there were 15 significant tests and it is possible that 1/3 of those were Type 1 errors. Some protection against Type 1 errors was attained by proceeding with post hoc tests only after a significant interaction with treatment was found in the overall ANOVA. Effects seen at multiple doses and in both genders should be given a greater weight than those seen at low, but not high doses, or seen only in one gender. Corrections for multiple tests (i.e., Bonferroni's) may also be considered. Finally, the study reviewed in this article utilized more than a single dose, which is highly recommended. For purposes of hazard identification, 4 treatment groups are usually employed, including 3 dose groups and 1 vehicle control group (6). Doses are usually selected on the basis of the toxicity to the mother. For example, the EPA testing guidelines for developmental neurotoxicology indicate that the highest dose shall be the dose which will not induce in utero or neonatal death or malformation sufficient to preclude a meaningful evaluation of neurotoxicity (6). It is recommended that lower doses be proportions of the highest doses. In summary, it is generally accepted that the developing nervous system may be vulnerable to the effects of environmental factors such as chemicals, biological agents, and physical factors. There are several design issues that are important for developmental neurotoxicity studies. Such studies should use multiple endpoints of neurobiological function taken at different points of the lifespan of the animals. It is also important to consider that the magnitude or presence of developmental neurotoxicity can depend on the time of exposure and the presence of maternal or postnatal toxicity in the offspring. It is generally appropriate to test both genders if there is no a priori basis for predicting that a chemical will have an effect on one of the genders. Finally, there are several statistical considerations relevant to developmental neurotoxicity studies. Of particular importance is the widely accepted practice of using the litter as the statistical unit in the analysis of the data. ACKNOWLEDGEMENTS

I thank Phyllis Keeter for her typing and editorial assistance. The critique of an earlier draft of this article by Dr. Robert Holson and Dr. Carole Kimmel is gratefully acknowledged. This article was presented at the Neurobehavioral Teratology Society Symposium entitled "Design and Analysis Issues in Developmental Neurotoxicology," in Boca Raton, FL on June 22, 1991.

DESIGN CONSIDERATIONS

IN N E U R O T O X I C O L O G Y

203

REFERENCES 1. Buelke-Sam, J.; Mactutus, C. F. Workshop on the qualitative and quantitative comparability of human and animal developmental neurotoxicity, Work Group II Report: Testing methods in developmental neurotoxicity for use in human risk assessment. Neurotoxicol. Teratol. 12:269-274; 1990. 2. Francis, E. Z.; Kimmel, C. A.; Rees, D. C. Workshop on the qualitative and quantitative comparability of human and animal developmental neurotoxicity: Summary and Implications. Neurotoxicol. Teratol. 12:285-292; 1990. 3. Hutchings, D. E. Behavioral teratology: Embryopathic and behavioral effects of drugs during pregnancy. In: Gottlied, G., ed., Studies in the development of behavior and the nervous system, Vol. 4, Early influences. New York: Academic Press; 1977:7-33. 4. Kimmel, C. A. Current status of behavioral teratology: Science and regulation. CRC Rev. Toxicol. 19:1-10; 1988. 5. Kimmel, C. A.; Rees, D. C.; Francis, E. Z. Qualitative and quantitative comparability of human and animal developmental neurotoxicity. Neurotoxicol. Teratol. 12:175-292; 1990. 6. National Technical Information Service. Neurotoxicity Testing Guidelines. US Department of Commerce, Springfield, VA 1991. 7. Ornoy, A.; Yanai, J. Central nervous system teratogenicity: experimental models for human problems. In: Persand, T., ed.,

8. 9. 10. 11.

12. 13. 14.

Neural and behavioral teratology, vol 4., Advances in the study of birth defects. Baltimore: University Park Press; 1980:1-21. Riley, E. P.; Vorhees, C. V. Handbook of Behavioral Teratology. New York: Plenum, 1986. Rodier, P. M. Critical periods for behavioral anomalies in mice. Environ. Health Persp. 18:70-83; 1976. Rodier, P. M.; Reynolds, S. S.; Roberts, W. N. Behavioral consequences of interference with CNS development in the early fetal period. Teratology 19:327-365; 1979. Rosecrans, J. A.; Hong, J. S.; Squibb, R. E.; Johnson, J. H.; Wilson, W. E.; Tilson, H. A. Effects of perinatal exposure to chlordecone (Kepone) on neuroendocrine and neurochemical responsiveness of rats to environmental challenges. Neurotoxicology 3:131-142; 1982. Spyker, J. A. Assessing the impact of low level chemicals on development: behavioral and latent effects. Fed. Proc. 34:18351844; 1975. Tilson, H. A.; Byrd, W. C.; Riley, M. T. Mice exposed perinatally to Firemaster (FF-1) show neurobehavioral alterations when tested during adulthood. Pharmacologist 21:187; 1979. Winer, B. J. Statistical principles in experimental design. New York: McGraw-Hill, 1971.

Study design considerations in developmental neurotoxicology.

It is widely accepted that exposure to environmental factors during development can result in effects other than death, gross structural abnormality, ...
471KB Sizes 0 Downloads 0 Views