Characterizing the formation of process contaminants during coffee roasting by multivariate statistical analysis

Coffee is a relevant source of dietary exposure for neoformed furan, alkyl furans and acrylamide. In this study, different statistical methods (hierarchical cluster analysis, correlation analysis, partial least squares regression analysis) were used for characterizing the formation of these process contaminants in green coffee beans roasted under the same standardized conditions. The results displayed a strong correlation between sucrose levels and furans in relation to the other sugars analyzed, while acrylamide formation was strongly related to the free asparagine. The data suggest that a sufficiently large amino acid pool in green coffee favors Maillard-induced acrylamide formation from asparagine, while reactions amongst the carbonyl-containing sugar fragmentation products leading to furan formation are suppressed. If the pool of free amino acids is small, it is depleted faster during roasting, thus favoring the formation of furans by caramelization, basically a sugar degradation process in which reactive carbonyl substances are generated and react together.