Impact of extraction methods on the determination of protein, fat, and carbohydrate content in long-life plant-based milk alternatives

Milk-alternatives manufacturers strive to produce milks that meet regulatory requirements and consumer expectations. Fat, protein and sugar content are assessed using extraction methods whose efficiency is not clear. In this study, we evaluated one physical (ultracentrifugation) and four chemical (Carrez reagent, acetic acid/ methanol, pH-based extraction, and methanol/chloroform) extraction methods on four milks (soy, almond, oat, and bovine) by analysing the content of major components. We made a reference soymilk to estimate recoveries. Each extraction method typically produced two distinct phases: clear/cloudy and pellet. Despite literature assumptions, none of the methods achieved complete extraction of proteins, sugars, or fats. Sometimes, as with fats, this was due to partial solubility in 'the other' phase, perhaps associated with another component. Sometimes, this was due to analyte degradation. The methanol/water phase of methanol/chloroform/water gave highest protein levels (similar to 93 % recovery). The clear phases of ultracentrifugation and Carrez extraction gave the best sugar recoveries (similar to 80 %). The pellets from Carrez and acetic acid/methanol gave the highest fat recoveries which were about twice the reference methanol/chloroform. We used bovine serum albumin (BSA) at the standard for measuring the protein content with a modified Bradford assay. Using the reference soymilk, we found BSA underestimates protein content by similar to 10 %.