Wrangling Whole Mixtures Risk Assessment: Recent Advances in Determining Sufficient Similarity

Abstract

Human health risk assessments for complex mixtures can address real-world exposures and protect public health. While risk assessors typically prefer whole mixture approaches over component-based approaches, data from the precise exposure of interest are often unavailable and surrogate data from a sufficiently similar mixture(s) are required. This review describes recent advances in determining sufficient similarity of whole, complex mixtures spanning the comparison of chemical features, bioactivity profiles, and statistical evaluation to determine "thresholds of similarity". Case studies, including water disinfection byproducts, botanical ingredients, and wildfire emissions, are used to highlight tools and methods. Limitations to application of sufficient similarity in risk-based decision making are reviewed and recommendations presented for developing best practice guidelines.

Keywords: botanicals; complex mixtures; new approach methodologies; non-targeted chemical analysis; water disinfection byproducts; wildfire smoke.

Figure 1.. Risk assessment context for whole mixtures evaluation and sufficient similarity determination.

Data on the exact whole mixture of interest are rarely available, so methods to determine sufficient similarity of a data-rich surrogate mixture(s) are required.

Figure 2.. Sufficient similarity assessments for whole mixtures represent integrative approaches that bridge exposure, toxicology, and data science to inform human health risk assessments.

Illustrated here are proposed conceptual steps involved in such assessments. (A) First the primary question of whether or not a mixture is sufficiently similar to a known reference mixture is posed. Data surrounding the mixtures’ (B) exposure chemistry and (C) biological response profiles are (D) integrated and compared to inform the manner in which health risks are ultimately quantified. Figure generation in Biorender Software.

Figure 3.. Steps that are commonly employed in a sufficient similarity analysis of complex mixtures.

(A) A comparison across chemical signatures is commonly implemented, showcased here using global, non-targeted chemistry methods to yield unannotated fingerprints of molecular features that can be compared across samples. A reference sample is bolded throughout and represents the data-rich sample to which all other mixtures are compared. (B) A comparison across biological response signatures can also be implemented, here using select toxicity endpoints (Tox Endpoints 1–5) that are displayed using a heat map based on clustered profiles. (C) Combined similarity metrics can be displayed using line plots, with distances between samples (dots) scaled according to similarity distance to the reference sample. In this generic example, a similarity intersect is shown indicating which samples are similar (left) vs dissimilar (right) to the reference sample.