An all-solid-state potentiometric electronic tongue with 36 polymeric membranes has been used for the first time to detect gliadins, which are primarily responsible for gluten intolerance in people suffering from celiac disease. A linear discriminant model, based on the signals of 11 polymeric membranes, selected from the 36 above using a stepwise procedure, was used to semi-quantitatively classify samples of a “Gluten-free” foodstuff (baby milked flour), previously contaminated with known amounts of gliadins (<10, 20–50 or >50 mg/kg), as “Gluten-free”, “Low-Gluten content” or “Gluten-containing”. For this food matrix, the device had sensitivity towards gliadins of 1–2 mg/kg and overall sensitivity and specificity of 77% and 78%, respectively. Moreover, the device never identified an ethanolic extract containing gliadins as “Gluten-free”. Finally, the system also allowed distinguishing “Gluten-free” and “Gluten-containing” foodstuffs (15 foods, including breads, flours, baby milked flours, cookies and breakfast cereals) with an overall sensitivity and specificity greater than 83%, using the signals of only 4 selected polymeric membranes (selected using a stepwise procedure). Since only one “Gluten-containing” foodstuff was misclassified as “Gluten-free”, the device could be used as a preliminary tool for quality control of foods for celiac patients.
Financial support received from FCT, Lisbon, under the research project PPCDT/QUI/58076/2004, is gratefully acknowledged.
