Screening of deoxynivalenol contamination in durum wheat by MOS-based electronic nose and identification of the relevant pattern of volatile compounds

Abstract

Fungal volatile metabolites can be used as an indicator of mycotoxins occurrence in cereals. A rapid, easy-to-perform and non-invasive method using an electronic nose based on metal oxide sensors was developed to distinguish the quality of durum wheat samples in three classes based on the content of the mycotoxin deoxynivalenol (DON): class A ([DON] < 1000 μg/kg), class B (1000 ≤ [DON] ≤ 2500 μg/kg;) and class C ([DON] > 2500 μg/kg). Two different approaches were tested by performing 525 analyses of naturally contaminated durum wheat either on whole and ground kernels. Discriminant Function Analysis (DFA) provided calibration models that permitted to classify wheat samples with mean recognition percentages (R) for whole and ground kernels of 69.3% and 86.7%, respectively. Samples belonging to class A were better recognized than those belonging to classes B and C. The e-nose method applied to the analysis of ground wheat showed highest mean recognition percentage (R = 82.1%) in validation. By considering data with recognition percentage rate (Rr) higher than 80% in validation, performances of this method were further improved in terms of either mean recognition percentage (R = 90.1%) and percentage of misclassified samples (4%). An SPME-GC–MS method was also developed that allowed to characterize the pattern of volatile compounds of durum wheat samples. A positive (trichodiene, longifolene, 3-methyl butanal, tridecane, γ-caprolactone, 6,10,14-trimethyl-2-pentadecanone) or negative (hexadecane, 2,3,7-trimethyl-decane, 4,6-dimethyl-dodecane) association with DON content was observed. These results confirm that the developed e-nose method could be used as useful tool for high throughput screening of DON contamination in durum wheat.