Evaluation of polarity switching for untargeted lipidomics using liquid chromatography coupled to high resolution mass spectrometry

Untargeted lipidomics using liquid chromatography high-resolution mass spectrometry (LC-HRMS) was performed using polarity switching, and in positive and negative polarity separately on the same set of serum samples, and the performances of the methods were evaluated. Polarity switching causes an increase in the cycle time of the HRMS measurements (1.18 s/cycle vs 0.27 s/ cycle), resulting in fewer data points across chromatographic peaks. The coefficient of variation (CV) was on average lower for the added isotopically labelled standards in pooled samples (QC) and patient samples using separate polarities (QC = 5.6%, samples = 12.5%) compared to polarity switching (QC = 8.5%, samples = 13.4%), but the difference was not statistically significant. For the endogenous features measured in the QCs polarity switching resulted in on average significantly higher CVs (3.80 (p = 4.25e-30) and 3.3 percentage points (p = 6.84e-40), for positive and negative modes, respectively) however still acceptable for an untargeted method (mean CVs of 17.9% and 12.2% in positive and negative modes respectively). A slightly larger number of endogenous features were detected using the separate polarities, but the large majority of features (> 95%) were detected with both methodologies. The overlap of features detected in both positive and negative polarities was low (4.1%) demonstrating the importance of using both polarities for untargeted lipidomics. When investigating the effects of a treatment on multiple sclerosis patients it was found that both methodologies gave highly similar biological results, further confirming the applicability of polarity switching.

Automated summary

Untargeted lipidomics analysis using liquid chromatography high-resolution mass spectrometry was performed with polarity switching and separate polarities, showing that using separate polarities can reduce the coefficient of variation and detect slightly more endogenous features, with low overlap between positive and negative polarities, highlighting the importance of using both polarities for untargeted lipidomics analysis. The results also indicated that both methodologies provided highly similar biological results, confirming the applicability of polarity switching.