Science Inventory

Insider Secrets for Design and Analysis of Defined-Mixture Experiments

Citation:

Simmons, J. Insider Secrets for Design and Analysis of Defined-Mixture Experiments. Society of Toxicology Annual Meeting, NA, NA - Virtual Meeting, March 12 - 26, 2021.

Impact/Purpose:

The purpose of this presentation is to provide a continuing education course at the Society of Toxicology Annual meeting (held virtually in 2021) with the goal of improving the quality, and thus the usefulness to risk assessors and decision makers, of defined-mixture toxicological experiments.

Description:

Design, conduct, analysis and interpretation of mixtures experiments are daunting challenges. Frequently, defined-mixture experiments investigate whether the response of a mixture is predictable from the dose-responses curves of the component chemicals. Experimental toxicologists have found that guideline study designs, while extremely valuable for intended purposes, are often not useful for investigation of consistency or lack of consistency with various definitions and forms of additivity (e.g. dose/concentration addition, response addition). Not taught typically in toxicology courses, individuals seeking knowledge on experimental design for mixtures generally sort through sometimes bewildering literature, where sources seemingly, or actually, contradict one another. There is a long history of poorly designed and analyzed studies; the ability to use available literature to understand the potential for nonadditive interactions is hampered by these design and analysis issues. This sunrise presentation will shed light on the poorly illuminated topic of mixture experimental design. Attendees will leave the course informed on fundamental factors and important elements to consider when constructing defined-mixture experiments. Benefits of incorporating multi-disciplinary expertise (the essential trio) will be discussed. The advantages of working with a qualified data analyst in advance of executing the experiment will be contrasted to the inefficiency of statistical consultation only after data are in hand. Areas of focus will be the low-dose/low-effect region, particularly important when concerned with environmental agents; designs useful when higher dose regions are of interest, such as combinations of pharmaceutical agents; and, ensuring utility of results for risk assessment, risk management and regulatory decision making. Both frequently used and less common but important designs with associated analysis strategies will be covered as will those that allow insight into biologically interpretable dose-response models. Key factors requiring consideration during construction of the design will be emphasized, including power, overall experimental size, dose level spacing and placement of experiment units within dose groups. The design impact(s) of testing for greater than additive versus less than additive outcomes will be covered. The concepts and strategies covered apply to traditional in vivo, traditional in vitro (e.g. Salmonella mutational assays) and new approach methodology (NAM) experiments. Attendees will be provided a curated, annotated bibliography for future reference. Example mixtures covered in the course and/or the annotated bibliography include mixtures of chemicals known or thought to act either by a common mechanism/mode of action/adverse outcome pathway or by dissimilar mechanisms/modes/pathways. While design and statistical considerations will be illustrated with mixtures relevant to occupational, pharmaceutical and environmental exposures, the concepts are broadly and generally applicable. At the conclusion of the course, attendees will be better equipped to answer the perennially vexing question: What is the optimal defined-mixture experiment for my goals? Attendees will acquire a foundation of knowledge equipping them to participate more fully in selection or construction of experiments suitable to the goal(s) of the study, yielding data that meet the criteria for appropriate statistical analyses. In addition to toxicologists interested in defined-mixture experiments, this course will be of value to those who evaluate or use the results of such experiments. Due to the multi-disciplinary collaboration required for fit-for-purpose high-quality defined-mixture experimentation, the presentation will be given jointly (in true mixtures fashion).

Record Details:

Record Type:DOCUMENT( PRESENTATION/ SLIDE)
Product Published Date:03/19/2021
Record Last Revised:08/03/2021
OMB Category:Other
Record ID: 352456