RESUMEN
Soybean containing products are widely consumed, thus reliable methods for detection of soy in foods are needed in order to make appropriate risk assessment studies to adequately protect soy allergic patients. Six methods were compared using eight food products with a declared content of soy: a direct sandwich ELISA based on polyclonal rabbit antibody (ab) to raw soy flakes, a commercial and an in-house competitive ELISA both based on ab to denatured, 'renatured' soy protein, an enzyme-allergosorbent test (EAST) inhibition based on two sera from soy allergic patients, histamine release (HR) using basophils passively sensitized with patient serum and a PCR method detecting soy DNA. Eight food products were selected as model foods to test the performance of the methods. There was an overall good agreement between the methods in terms of ranks of soy content but not the quantity. The sandwich ELISA aimed at native soy proteins had the lowest detection limit of 0.05 ppm, but only identified soy in 5/8 products, and generally in lower amounts compared to other methods. The competitive ELISA had a higher detection limit of 21 ppm, but seemed more successful in detecting processed soy. Only HR, EAST inhibition and PCR detected soy in all eight products. In spite of a general good correlation in terms of ranks of soy content, more than a single method may be necessary to confirm the presence of soy in foods.