The Center for Alternatives to Animal Testing is an academic center affiliated with the Division of Toxicological Sciences in the Department of Environmental Health Sciences of the Johns Hopkins University Bloomberg School of Public Health.

 

Johns Hopkins School of Public Health

Proceedings for TestSmart -- Endocrine Disruptors

Breakout Group I - Thoughtful Science

Chair

L. Earl Gray, Jr.
United States Environmental Protection Agency

Panel Members

Robert Bigsby
Indiana University

Errol Zeiger
Consultant

Focus Question. What methods exit from increasing the quantity and quality of data obtained from whole animal testing (e.g. Can additional endpoints be included in whole animal test?)

Introduction, Earl Gray:

The concerns of animal protectionists about the EPAs EDSP program were presented in Plenary by Martin Stevens, Vice president for Animal Research Issues, The Humane Society of the United States. For background, some comments from his abstract are presented here. There is concern that the EDSP "program will be one of the largest animal testing ventures in American history. An estimated 600,000 to 1.2 million animals will be killed for every 1,000 chemicals tested." While much of the focus is in promoting alternative (non-animal) methods for inclusion in EDSP to replace the use of animals there also is an interest in refining the approach to screening and testing to reduce animal use.

It is the opinion of the chair that these numbers are too high and that a more "thoughtful" approach (presented in the poster) can reduce the numbers of animals from that proposed above by five- ten fold. This estimate does not take into account further reductions that might be achieved by implementing some of the ideas discussed subsequently in the Breakout Session.

The overheads presented included (pdf)

  1. A description of the EDSTAC Tier 1 Screening Battery (T1S)
  2. The Objectives of T1S and Tier 2 Testing 9T2T) in EDSP
  3. A definition of Validation provided under the Food Quality Protection Act of 1996
  4. Alternative assays recommended by EDSTAC for possible inclusion in T1S
  5. A description of how different endocrine activities are detected in T1S
  6. Comments on standardization and validation of the Uterotropic and Hershberger Assays in 1962
  7. A figure demonstrating how simple "thoughtful" changes in screening and testing could reduce the numbers of animals used in the EDSP T1S and T2T (from Gray et al., in press).

I have organized the discussion comments into those that refer to 1) Priority setting, 2) Tier 1 Screening, 3) Tier 1 Screening- Repeat/Characterization Phase or 4) Testing. Presentation of these comments here does not infer that the group reached consensus on an issue, as this was not an objective of the Session, but rather, the issue was raised some which should be considered being implemented by the USEPAs as it moves toward with the EDSP effort.

Prioritization was not a major focus of discussion, but several comments did relate to this process.

  1. EPA should obtain better exposure data than it currently has and use this information to assure that only chemicals that present a potential exposure would be evaluated in the Screening and Testing Batteries.
  2. Use all available animal and human data to determine need for Tier 1 Screen. Animal data might come from OECD 407 or other studies, while human data may be available from accidental or worker exposure studies.

In vivo Screening Battery

  1. Use the results of T1S in vitro battery to tailor the execution of T1S in vivo studies. For example, a strong ER agonist should be evaluated in the uterotropic and pubertal female assays, but the Hershberger assay may not be needed if these assays were positive. On the other hand, a strong AR antagonist positive both in vitro and in the Hershberger assay may not need to be tested in the pubertal female or uterotropic assays.
  2. Several suggestions were made about adding additional endpoints to existing screening assays to improve their diagnostic ability. Such endpoints included, additional organ weights, preserve tissue for histopathology, save serum for hormone analyses, examine hormone production ex vivo, biochemical or molecular analyses for alterations of endocrine sensitive gene or protein functions. For example:
    1. Examine thyroid hormone levels in animals from the uterotropic and Hershberger assays. They may provide supportive data for the effects in the pubertal female assay or eliminate the need to rely on the pubertal female assay to assess thyroid function.
    2. Save vaginal tissue for histological examination in the uterotropic assay. This would help detect chemicals like Tamoxifen with mixed agonist/antagonist activities without conducting an additional experiment with more animals.
    3. Examine adrenal histology and function in the Hershberger assay for alterations in steroidogenesis. The early steps in steroid hormone synthesis are similar in the gonads and adrenals. Chemicals that altered CYP450 enzyme activities could be detected from the adrenals of males from the Hershberger assay which might eliminate.
    4. Examine serum LH, prolactin, TSH and FSH levels from rats from the Hershberger and uterotropic assays. This may provide supportive data for the effects in the pubertal female assay on hypothalamic-pituitary (HP) function and could eliminate the need to rely on the pubertal female assay to assess alterations of HP function or enable one to use fewer animals in the pubertal assay.
  3. Use the most sensitive route of exposure for screening but use a more relevant route of exposure in the same short-term assay before "triggering" testing.
  4. Report SDs, CVs and SEs for all endpoints such that performance criteria can be developed for interlaboratory studies. It was noted that training and practice will be required to attain reasonable levels of variability within a lab and this is critical to detecting effects at low dosage levels, especially with smaller sample sizes. Genomic and proteonomic techniques should be evaluated for their utility in the screening of EDCs but their utility as this time is unknown. Issues of reproducibility, sensitivity, specificity have not been adequately addressed for these new methods and it has yet to be determined if these inclusion of these techniques would ultimately lead to a significant reduction in animal use or an enchantment of the diagnostic capacity of the Tier 1 battery.
  5. The group discussed that the endpoint with the smallest CV or the endpoint that showed the largest change in response to a hormone were not necessarily the most "sensitive" or useful. For example, body weight has a CV of about 5% and shows the largest absolute increase with testosterone treatment but higher dosage levels of androgens are required to increase body weight than weights of the sex accessory tissues.
  6. Use the information gained from the standardization and validation phases of assay development to optimize the experimental design, statistical analyses and reduce sample sizes as warranted.
  7. Minimize the numbers of dosage groups evaluated in each assay. Dose-response is not an issue in screening. One must use enough dosage levels to be certain that effects are seen at or below the M.D. (a 10% reduction in necropsy weight) for positives or that no endocrine-related effects are seen at the M.D. for negatives.
  8. The relevance of an endpoint in an assay to an adverse effect in the rat and humans should be established. The implication of this is that T2T will not be triggered by changes in T1S assays that are meaningless. This is particularly germane in the case of biochemical and molecular (microarray or etc) markers. They must be clearly linked to an adverse effect before they can be used as a trigger for testing (i,e, the endpoint must be "validated").

Testing should not be triggered by changes in T1S of unknown biological significance. This should be resolved in short-term mechanistic studies, not in the testing phase. Tier 1.5, Repeat or Elucidate Phase Between T1S and T2T.

Recently, NHEERL, ORD, scientists proposed adding a phase between TS and T2T (Gray et al., in press). The purpose of this step is primarily twofold. First, equivocal responses from T1S assays should be repeated. It is not necessary to repeat the entire T1S Battery. Many chemicals evaluated in T1S will present false positive responses due to the large numbers of endpoints evaluated using a statistical value of p < .05 as a positive response. For a single endpoint the false positive rate would be 5%, but this would rise to about 10% with two independent endpoints. As the numbers of independent variables in T1S is in the dozens, the false positive rate could be very high. In fact, if single statistically significant effect in T1S was enough to trigger T2T then most of the chemicals being tested in multigenerational assays would be false positive. This problem can be virtually eliminated in the proposed T1.5RE phase. An assay showing a weak, unconfirmed responses would be reevaluated in T1.5RE phase, which would eliminate false positives. This short-term study also could be expanded to provide a more thorough evolution of the potential effect seen in T1S.

The second major objective of T1.5RE is to allow for the execution of additional short-term studies to provide additional information for T2T. For example, if a chemical was estrogenic in vitro and in the uterotropic assay (with sc administration) but not in the pubertal female assay it would not be useful execute a dietary testing study with rats if the chemical had no effect via the oral route. T2T might still be warranted for other vertebrates, like fish, however, in which the sc route may be more relevant than the oral route of exposure. If a chemical was an AR antagonist and positive in the Hershberger assay, one might want to execute the pubertal male study. In addition to examining the effect of differences in toxicity between different routes of exposure this phase also could alter the duration of the study to see if effects became more evident with extended dosing (having a greater AUC). For example, some of the effects of xenoestrogens are more obvious after several weeks of dosing than in a 3 day protocol.

The idea also was presented that T2T multigenerational studies using rats should not be executed if it was determined in T1.5RE that the effects in T1S were not relevant to humans.

Dose-Range Finding Studies

The EPA's EDSP has yet to describe how doses will be selected for the T1S in vivo assays. It was proposed that the M.D. be set at a 10% reduction in final body weight or that a limit dose be used. A chemical that did not reproducibly alter an endocrine-sensitive endpoint in T1S would be considered negative for EAT activities.

Dose range finding studies should use a minimum of animals to determine the M.D. for body weight in the in vivo assays. These studies should use of minimum of animals (possibly as few as two per dose group) for longer term studies or an "up-down" approach for shorter 3-4 day studies. Group sizes as large as ten per group should be avoided, being unnecessary to detect body weight differences. Experimental designs should be considered that minimize variance and increase the ability to detect small changes in body weight with small sample sizes. In addition to body weight, investigators must assure that doses selected for the screening assay does not induce any overt signs of toxicity including neurotoxicity.

Testing - T2T

The following ideas were discussed in the Session that are related to the Testing phase of T2T. Chemicals clearly positive in T1S or repeated in T1.5RE should be tested in T2T unless it is apparent that the route of exposure to be used in the test is totally ineffective (i.e. oral for the rat) or it is apparent that the mechanism of action is not relevant to humans. While some expressed the opinion that the mechanism of action must be relevant to humans if the chemical was going to be tested in a mammalian animal model, others felt that this would delay testing too long as often the mechanism of action is difficult or takes time to identify and testing should not be delayed for active EDCs. It was held by some that in some cases animals differ so greatly in their endocrine physiology that the effects in rats are meaningless, while others disagreed in part indicating that there is a large data base showing that chemicals that act via the estrogen or androgen receptor or inhibit steroidogenesis affect humans and laboratory animals in a predictable manner. This discussion highlights the need to identify mechanisms of toxicity and establish their relevance to species of concern. With respect to both the uterotropic and Hershberger assays, these were standardized and validated for screening chemicals for EA activities in rats in 1962 and the relevance to other species has been established.

The benefits and limitations of tailoring T2T based upon prior information were discussed by the Session. The general topics were 1) expand the numbers of endpoints and 2) increase the numbers of offspring that are thoroughly evaluated while reducing the numbers of litters.

Other Issues

The role of dietary components on the results of Tiers I and II have been raised repeatedly, but remain a hypothesis. It is well known that phytoestrogens, critical fatty acids and other components vary greatly from diet to diet, depending upon the original formulation, season and source of materials. On occasion, animal diets have been accidentally contaminated with EDCs like DES, zearalenone or unidentified substances that clearly altered the reproductive status of control animals. The degree to which phytoestrogens in diets affects the results of these assays is controversial and needs to be established in studies carefully designed to address specific hypotheses. Merely varying the diet from one formulation to another is inadequate because in such an approach several dietary factors vary at once and the contribution of any one of them is confounded with the other.

Identification of new endpoints for inclusion in existing assays for detection of endocrine activity. A lack of endpoints specific to thyroid hormone alterations at low dosage levels was specifically noted.