Sampling variation in the quantification of fumonisins in maize samples

Abstract

Fumonisins produced by F. verticillioides and F. proliferatum cause mycotoxicoses in horses, swine and rats and have been associated with oesophageal cancer in humans. Accurate measurement of mycotoxins is essential for determining the safety of grain and their products for consumption. Four sources of variation were studied, namely sub-sample size, variation within a single maize sub-sample, number of replicates and toxin detection techniques used by independent laboratories. Variation in detected fumonisin levels within a single maize sample was high using the 25 g sub-samples proposed in the Neogen Veratox protocols. A 250 g subsample significantly reduced variation in fumonisin levels of samples. An incremental increase in sample size also improved the number of positive samples recorded. Increasing the number of replicates using the recommended sub-sample size (25 g) did notreduce variation except when the sample had high fumonisin levels. Improved accuracy was recorded when a 250 g sub-sample was used in conjunction with increased replicates. Data from laboratory analyses indicated that ELISA reactions (Agricultural Research Council - Grain Crops Institute) correlated significantly with HPLC results of the Medical Research Council (MRC), but neither of these correlated with results from an independent laboratory. Concentrations determined using ELISA were consistently higher than those from the HPLC (MRC) technique. Quantification technique, sample size, replicate number and laboratory where analyses are conducted, appear to be important sources of variation for quantification of fumonisins.