Overview (To Print: pdf 3 p)

Introduction
Methods
Site Organization

Key Questions

INTRODUCTION

The threshold dose-response model has long been recognized as the most dominant dose-response model in the biological sciences, including pharmacology and toxicology (Clark 1926, 1933, 1937).   The threshold model dominates the leading pharmacological and toxicological textbooks (Hayes, 2000; Klaassen 2003), study designs that drive hazard assessment procedures for pharmaceutical and chemical agents and risk assessment processes used by regulatory and public health agencies world-wide.   Despite its general acceptance in biological disciplines, the centrality of the threshold model has been recently challenged by the hormesis model with a number of papers providing evidence of its generalizibility by biological model, endpoint measured and chemical class tested (Calabrese, et al., 1999, Calabrese and Baldwin 2001, 2001a; Calabrese and Baldwin 2003, 2003a, Calabrese 2004 Calabrese and Blain, 2005) and high frequency in the toxicological literature (Calabrese and Baldwin,   2001).   Using a priori entry and evaluative criteria, Calabrese and Baldwin (2003) reported that the distribution of responses to doses below the NOAEL for approximately 800 dose response relationships was more consistent with the hormetic rather than the threshold model.  

We describe new research to examine evidence for a hormetic model using a large set of dose response studies conducted as part of the National Cancer Institute's (NCI) anti cancer drug screening program.   This research examines response below a no observed adverse effect level (NOAEL) using the NCI Yeast Chemical Screening data base.
(top)

METHODS

We utilized data from the NCI yeast anticancer drug screen.   The yeast data is one arm of a larger program conducted by NCI to screen chemicals that includes screening in 70 additional human cancer cell lines.   Initial analyses aimed at evaluating evidence for hormesis were conducted in some of these human cell lines.   The results of these early analyses helped determine the methods applied to the yeast strains.   As a result, we describe some of these analyses results as part of this research.   However, most of this research is focused on the results from the dose response studies conducted in yeast strains.

A detailed description of the data base is provided at the NCI website and Holbeck (2004).   Briefly, yeast data from Stage 2 were selected for evaluation.   This stage contains the most promising compounds based on preliminary testing; the agents are tested at five concentrations (1.2, 3.7, 11, 33, and 100 micromoles) with 13 yeast strains.   The yeast strains are a panel of Saccharomyces   cerevisiae strains altered in DNA damage repair or cell cycle control genes.   The NCI website contains a description of the specific type of genetic alteration in each of the yeast strains employed.   The data used in the drug screen studies are derived from the fraction of growth of the yeast strain exposed to the compound relative to the growth of the same yeast strain treated with solvent control.  

Using a 96 well plate system, there are eight control wells employed against which all chemicals tested on the specific yeast strain are compared. The dose response for a chemical is replicated on the same day on a different 96 well plate with the yeast derived from a separate colony.   The same chemical which is   tested is intentionally derived from a different shipment/source.  

The solvent used was DMSO ( <    0.5%).   The yeast were grown in liquid Complete C Media overnight and then assessed for growth when the control yeast displayed an OD within the range of 0.4-0.6.

The results of the findings of these two five concentration dose response relationship studies were then averaged.   This process was then repeated on a different day and designated as the replicate evaluation for that chemical and yeast strain.   The use of original colonies for each evaluation, chemicals from different sources and different day for replication were intended to maximize possible variation in the model test system.

Dose response relationships for 2189 different chemicals are were obtained from testing on the 13 yeast strains yielding 28,457 total dose response relationships of five doses.   The objective of this research is to determine the extent to which such dose response relationships provided evidence of hormetic effects.   To investigate such possible effects, each of these doses responses was subjected to a priori entry criteria.   The a priori entry criteria required the concentration-response relationship to have a lowest observed adverse effect level (LOAEL), a no observed adverse response level (NOAEL), and at least one concentration below the NOAEL.   All responses below the NOAEL are reported as a percentage of the control value.  

Our goal is to identify doses below the NOAEL for a particular dose response study, and to evaluate the evidence of an elevated response relative to control at such doses.   Although data from many dose response studies are available, data from each study is limited.   Algorithms were used to categorize response at different dose response studies as the LOAEL and NOAEL.   The algorithms were constructed so that they could be related to previous research, and so that they could be practically implemented with the available data.   We examined the properties of the algorithms via simulation studies to identify and correct for possible biases.   This process not only led to refinements in the algorithms, but also to an appreciation of the limitations of the methods.

For dose response studies where response was available at doses below the NOAEL, we examined the distribution of response.   We also critically reviewed such examination to explore potential biases that could result from such a process.   The examination is based on hypothesizing dose response models, and then evaluating the performance of the algorithm via simulation studies assuming that such models hold.   The simulations revealed biases that can occur as a result of the identification of LOAEL and NOAELs.   Subsequent work is aimed at quantifying such biases, and using estimates to adjust analysis.
(top)

SITE ORGANIZATION

This site is organized into sections that describe

• research and development methods,
• data sources and examples of data and codebooks, and
• reports, figures, and tables that contain results.

The reference section contains a bibliography of articles that are of interest to this area.

Most of this research is focused on dose-response studies conducted either in human cell lines, or yeast strains using a wide variety of chemicals (see NCI web site). There are some differences between the type of data available for the human cell lines and the yeast strains. As a result, the data, methods, and results sections are organized by human and yeast sections.
(top)

© 2005 University of Massachusetts Amherst . Site Policies .
This site is maintained by the SPH&HS/Dept. of Public Health/ Biostatistics and Epidemiology Program .