Factorial design applied to LC-ESI-QTOF mass spectrometer parameters for untargeted metabolomics

Abstract

Investigations of untargeted metabolomics are based on high-quality data acquisition usually from multiplatform systems that include high-resolution mass spectrometry equipment. The comprehensive set of results is used as data entry for bioinformatics and machine learning sciences to access reliable metabolic and biochemical information for clinical, forensic, environmental, and endless applications. In this context, design of experiments is a powerful tool for optimizing data acquisition procedures, using a multivariate approach, which enables the maximization of a high-quality amount of information with reduced number of tests. In this study, we applied a 3³ Box–Behnken factorial design with central point triplicate for optimizing the ionization of an HPLC-ESI-QTOF method used for screening urine samples. Nozzle voltage (V), fragmentor voltage (V) and nebulizer pressure (psig) were the factors selected for variation. The response surface methodology was applied in the molecular features extracted at each level, resulting in a statistical model that helps evaluating the synergic interaction between these factors. Together with the qualitative analysis of the resulting total ion chromatograms, we came across a reproducible (6.14% RSD) and highly efficient method for untargeted metabolomics of human urine samples. The proposed method can be useful for applications in several urine-based metabolomics-driven studies, as the factorial design can be applied in the development of any analytical protocol considering different LC-MS setups.