Matrix effects influence biochemical signatures and metabolite quantification in dried blood spots

Dried blood spots (DBS) represent a convenient clinical sample material, offering low infection risk, easy transport, and long-term metabolite stability. However, applying samples such as whole blood, serum or plasma onto filter paper introduces an additional matrix, potentially affecting metabolite extraction. Here, we compare metabolite recovery from liquid samples and their filter paper analogue using both targeted (acylcarnitines, amino acids) and untargeted metabolomics. Significant matrix effects were observed for some compounds, especially for dicarboxylic acylcarnitines (C3DC–C6DC) and specific amino acids (cystine, cystathionine). We did not identify specific metabolite characteristics that may predicted altered recovery. In a cohort of 229 authentic DBS samples — including patients diagnosed with inherited metabolic disorders, obesity or under a ketogenic diet — untargeted profiling combined with random-forest machine learning led to an effective stratification. Notably, C4DC, despite strong matrix effects, was ranked in the top ten variables of this random-forest model. With adequate validation, DBS can be safely used for diagnostic purposes despite possible matrix effects, but care must be taken in the comparison of values obtained when different sample materials are used.