An Enhanced Protocol to Expand Human Exposome and Machine Learning-Based Prediction for Methodology Application.

Area of Science:

• Environmental Chemistry
• Toxicology
• Analytical Chemistry

Background:

• Assessing the human exposome is challenging due to difficulties in detecting low-level endocrine-disrupting chemicals (EDCs) and their metabolites in biological samples.
• Current analytical methods limit the accuracy of EDC exposure assessment and understanding of cumulative health effects.

Purpose of the Study:

• To develop an enhanced protocol using multi-solid-phase extraction (multi-SPE) to expand the measurement of polar EDCs and metabolites in the human exposome.
• To train a machine learning (ML) model for predicting methodology based on molecular descriptors.

Main Methods:

• Implementation of a multi-SPE protocol for serum and urine sample analysis.
• Comparison of multi-SPE with hydrophilic-lipophilic balance (HLB) sorbent alone.
• Nontarget analysis of serum and urine from women of childbearing age within a larger cohort.
• Development of an ML model utilizing molecular descriptors for methodology prediction.

Main Results:

• The multi-SPE protocol significantly enhanced the measurement of EDCs in serum and urine compared to HLB alone.
• Nontarget analysis using multi-SPE identified a higher number of target EDCs and novel chemicals in serum and urine.
• The ML model predicted that multi-SPE could identify an additional 38% of the most bioactive chemicals.

Conclusions:

• The developed multi-SPE protocol substantially advances human exposome research by expanding the scope of measurable EDCs and metabolites.
• This enhanced analytical approach improves the identification of chemical exposure profiles and aids in understanding potential health impacts.