Jump to main content.


Note: EPA no longer updates this information, but it may be useful as a reference or resource.


Summary of the Screening and Testing Work Group Meeting on Low Dose Considerations

February 2, 1998 in Washington, DC
Prepared by Penelope A. Fenner-Crisp, Chair, STWG

During the afternoon preceding the February 3-4 meeting of the EDSTAC, the Screening and Testing Workgroup (STWG) held a meeting to discuss low dose considerations in the screening and testing of potential endocrine disruptors. Attendees included members of the STWG, the Priority Setting and Communications and Outreach Work Groups, as well as approximately half of the members of the parent Committee.

Dr. George Lucier from NIEHS, an EDSTAC member, served as host for the session. He began the discussion by presenting a brief overview of issues pertinent to dose response, along with general issues related to the risk assessment for endocrine disruptors.

Dr. Chris DeRosa from the Agency for Toxic Substances and Disease Registry summarized the findings of a series of human epidemiology studies, most of which attempt to characterize the relationship between exposure to persistent and/or bioaccumulative toxic chemicals such as PCBs, DDT degradates, lead and mercury in a variety of populations which are consumers of fish from the Great Lakes. Some of these studies describe decrements in neurobehavioral parameters of children born to mothers considered to be high fish consumers. Other studies reported adverse reproductive effects in women. From these studies, it can be shown that some high consumers have exposure levels 3-4 times the general population. ATSDR's conclusion regarding these studies is that the weight of the evidence from the body of data suggests that there is cause for concern, especially among the "high consumers."

Dr. Dan Sheehan from FDA's National Center for Toxicological Research offered his hypothesis that there is no threshold for effects of exogenous endocrine disruptors since there exists background levels of endogenous hormones which are already biologically active. Thus, any additional exposure constitutes an exceedance of the threshold. To support his hypothesis, he described the findings of his study with turtle eggs in which very small amounts of additional estradiol shifted the sex ratio in a linear dose response manner. Additional data to support the no-threshold hypothesis included the study of DES in mice which showed decreased fertility and number of pups with an increase in dose and Earl Gray's work with vinclozolin and TCDD in male rats. He noted that the observation of non-monotonic and no-threshold dose response curves for endocrine effects may drive the risk assessment process. He offered suggestions as to how dose selection for testing could be determined. The first was to use predictive models along with estrogen receptor and serum binding data. A second suggestion was to add more endpoints for evaluation in the study.

Dr. Fred vom Saal, University of Missouri, presented his position that "things happen at physiological levels" that are of concern and, usually, this is not taken into consideration when conducting toxicity testing in the traditional manner. (The consequences of this could be that current risk assessment methodologies would underestimate risk and lead to the development of allowable exposure levels that were not protective of human and environmental health). He summarized his data on the effects on prostate weight and sperm production in male offspring of female mice treated with estrogen, DES. and bisphenol A (the inverted U-shaped dose response). He recommended that, when choosing doses for testing, doses known or expected to represent human exposure should be included.

Dr. Renate Kimbrough from the Institute for the Evaluation of Health Risks offered her view of the evidence put forth in the epidemiology studies described above by Dr. DeRosa. She disagreed with ATSDR's viewpoint, stating that she believed that they did not show anything of clinical significance (e.g., non-concordance of effects across studies, inadequate determination of exposure, and difficulties in sorting out confounding factors). She stated that most of the neurobehavioral tests that were administered have no predictive value, that the thyroid function tests conducted in these infants were normal and, therefore, there is no plausible mechanism for the tested chemicals to have neurobehavioral effects. She concluded that, in the aggregate, the studies raised questions, but did not support conclusion of real concern.

Dr. John Ashby, from the Central Toxicology Laboratory-Zeneca, UK, described the difficulties he is having in trying to duplicate the work of vom Saal, Richard Sharpe, and others who have described low dose effects in various types of studies. He is asking the question of, rather than the chemical exposure being responsible for the effects observed, could they be due to confounding factors such as handling stress, lighting pattern, group vs. individual housing and hand feeding? He raised the issue of hormesis, the occasionally seen phenomenon that presents as the U-shaped lower end of a non-monotonic dose response curve. Noting that hormesis is observed for a variety of effects other than those which are endocrine-related, he acknowledged that the phenomenon is poorly understood and deserves further investigation. He also argued that the traditional manner of toxicity testing would, in most cases, be adequate for the evaluation of the endocrine-mediated effects of interest to EDSTAC and that additional work should be done to identify any exceptions to that view.

Dr. Jim Lamb from Jellinek, Schwartz and Connelly presented an overview of general principles that should be considered when designing, conducting, and evaluating toxicity studies. He also provided a listing of industry-sponsored studies underway in the U.S. which should be available within the next year or so, and which should shed further light on the issue of low dose effects.

Dr. John Bailer, Miami University of Ohio, presented areas for consideration when designing toxicity studies, from the statistical point of view.

In order to get the General Discussion started, Dr. Lucier posed nine questions regarding dose selection for the attendees to consider. While no formal consensus was expected, the general sentiment of the group was developed on each. The questions are noted below, along with a brief summary of the group's opinions.

1) (Should a test have a ) wide dose range?

Sometimes. A wide dose range should be employed when effects are expected to occur at the low(er) doses. How one concludes that effects could/would occur was a matter of vigorous discussion, and while no specific directions could be given at this time, it was agreed that additional research should be done promptly to characterize those circumstances under which it would be appropriate/necessary to incorporate low doses in the study design.

2) Should Tier 1 Screening data inform dose selection for testing?

Yes. Most of the in vivo screens will be conducted using only one dose. However, each screen will also employ a positive control, so some measure of comparative potency can be estimated. Also, it is possible that range-finding studies may be conducted in order to select the single dose. In addition, the receptor binding screens afford an opportunity for the development of dose response data, and comparison against a positive control.

3) Should the high(est) dose be abandoned?

No. This question spoke to the issue of whether or not the traditional high dose in tests should be abandoned in favor of low(er) doses. As currently required, tests must include a dose which causes some toxicity. It was agreed that this dose should remain a part of tests, in order to continue to have an "upper bound" on the dose response and give assurance that we can understand the full hazard potential of the chemical under evaluation.

4) What should the low dose be?

It should be determined case-by-case. Its choice would be predicated upon prior information from the data base existing before screening and testing for endocrine disruption potential begins, and from the data that are derived in the course of screening and testing. There was sentiment for including dose(s) that are environmentally relevant, that is those to which humans and/or wildlife are or may be expected to be exposed, on a case-by-case basis.

5) What should the high dose be?

As discussed above in Question 3, the traditional high dose should remain a part of the study design. Thus, the current decision logic used to determine the maximum tolerated/tolerable dose should continue to be used.

6) How many doses (should be used in the test)?

At least three treatment groups plus a control. This study design (3 + 1) represents the current requirement. If it is determined that "low" doses are needed to adequately evaluate the CSM, then additional doses should be included. Modification of the current requirements for group size could be considered. Flexibility is the keyword: encourage tailoring the study at the "low" doses, perhaps using fewer animals, fewer endpoints; be statistically creative.

7) How should the doses be spaced?

This should be determined case-by-case. Choices would be determined by information on pharmacokinetics, the endpoints to be evaluated, and other prior information.

8) Does dose selection influence the selection of the endpoints for evaluation?

Yes. For example, if the CSM being evaluated is known/anticipated to produce estrogenic effects, it may be decided that evaluation of such effects at low doses should occur. In this case, then, endpoints known to be responsive to estrogenic (or anti-) mediation should be evaluated in the study.

9) What are the implications for EDSTAC? (That is, for its meeting February 3 and 4?)

A. The nature of the discussion argues against skipping T1S, if equivalent data do not already exist. This is an important point, in light of the proposal by both the STWG and PSWG that T1S could be bypassed under certain circumstances.

B. The timeline for implementation of the Endocrine Disruptor Screening and Testing Program (EDSTP) incorporates time to determine the specific designs of the tests with regard to the low dose question. For example, early activities such as the High Throughput Pre-Screening (HTPS) demonstration project and subsequent performance of the HTPS program must occur before priority setting can be completed. Also, each screen must be subjected to validation and standardization before being put "online". In the meantime, a substantial body of data will be developed via the research and testing now underway within and outside of government.


Local Navigation


Jump to main content.