Ultrasound-assisted three-phase hollow fiber microextraction-based method for untargeted metabolomics

Sample treatment for untargeted metabolomics is still the bottleneck of the analytical procedure, especially for complex samples like food or biofluids. The main pitfalls are the suppression of minor compounds, low metabolite coverage and the impossibility to detect certain families, combined with the requirement of nonselective approaches for untargeted analysis. Herein, we developed a new analytical extraction method for untargeted metabolomics using a non-selective procedure based on three phase hollow fiber liquid phase microextraction aided by ultrasound (three-phase-UA-HF-LPME) followed by gas chromatography-mass spectrometry. As a proof of concept, the method has been validated for human milk (HM). The analytical method allows extracting organic nitrogenous compounds from HM, which was not possible using the conventional liquid-liquid extraction (LLE) and provided cleaner chromatographic profiles and mass spectra with lower background noise. Moreover, three-phase-UA-HF-LPME enhanced the extraction of a high number of metabolites from 1.65 to 3.27-fold compared to LLE, namely: fatty acids and derivatives (decanoic acid, lauric acid, palmitic acid, tetradecanoic acid, palmitoleic acid, stearic acid, myristic acid, linoleic acid, azelaic acid), lipids and lipidlike molecules (oleamide, monopalmitin), organic acids and derivatives (lactic acid, citric acid), amino acids and peptides (glycine, proline, glutamine). Otherwise, the extraction of carbohydrates and derivatives (urea, phosphoric acid, galactose) decreased the fold changes from 0.15 to 0.25. Results show that our proposed method is attractive owing to its limits of detection from 0.08 to 0.16 mg L-1, recoveries of 99-86 %, enrichment factors up to 123 and fold changes very satisfactory when compared with LLE. The validation by targeted metabolomics as well as the untargeted procedure showed that three-phase-UA-HF-LPME presents better sensitivity, enrichment factors, sample throughput and metabolite coverage than the conventional LLE method considered the goldstandard for untargeted metabolomics. Our results show insights into the possible application of HF for untargeted metabolomics with important analytical advantages in comparison with the LLE.