The contribution of lot-to-lot variation to the measurement uncertainty of an LC-MS-based multi-mycotoxin assay

Abstract

Multi-mycotoxin determination by LC-MS is commonly based on external solvent-based or matrix-matched calibration and, if necessary, the correction for the method bias. In everyday practice, the method bias (expressed as apparent recovery RA), which may be caused by losses during the recovery process and/or signal/suppression enhancement, is evaluated by replicate analysis of a single spiked lot of a matrix. However, RA may vary for different lots of the same matrix, i.e., lot-to-lot variation, which can result in a higher relative expanded measurement uncertainty (Ur). We applied a straightforward procedure for the calculation of Ur from the within-laboratory reproducibility, which is also called intermediate precision, and the uncertainty of RA (ur,RA). To estimate the contribution of the lot-to-lot variation to Ur, the measurement results of one replicate of seven different lots of figs and maize and seven replicates of a single lot of these matrices, respectively, were used to calculate Ur. The lot-to-lot variation was contributing to ur,RA and thus to Ur for the majority of the 66 evaluated analytes in both figs and maize. The major contributions of the lot-to-lot variation to ur,RA were differences in analyte recovery in figs and relative matrix effects in maize. Ur was estimated from long-term participation in proficiency test schemes with 58%. Provided proper validation, a fit-for-purpose Ur of 50% was proposed for measurement results obtained by an LC-MS-based multi-mycotoxin assay, independent of the concentration of the analytes.Electronic supplementary materialThe online version of this article (10.1007/s00216-018-1096-5) contains supplementary material, which is available to authorized users.

Validation of an LC-MS/MS-based dilute-and-shoot approach for the quantification of > 500 mycotoxins and other secondary metabolites in food crops: challenges and solutions

“…Likewise, the two pistachio samples purchased from the supermarket exhibited higher extraction efficiencies for some analytes compared with the other five replicates deriving from field sampling. Although these lot-to-lot variations did not result in non-compliance as regards precision for most analyte/ matrix combinations, we think that their investigation should be an integral part of any validation process as these variations might significantly contribute to the overall measurement uncertainty of the respective LC-MS/MS method [17]. However, to the best of our knowledge, none of the existing guidelines explicitly demands the spiking of different individual samples for the determination of the lot-to-lot variation.…”

“…In order to challenge the repeatability of the extraction protocol and of matrix effects, a heterogeneous set of individual raw samples from field surveys and of processed samples from the supermarket was selected for each matrix (Table 1). Maize and fig samples were described in [17].…”

“…This neglects the investigation of relative matrix effects, although it has been observed that their extent might require matching the calibration on a variety level, as it was the case for white and red sorghum [16]. Even without going to such an extreme, we have recently shown that relative matrix effects might significantly contribute to the overall uncertainty of an LC-MS-based method and therefore need to be investigated during validation [17].…”

Adverse Effects, Transformation and Channeling of Aflatoxins Into Food Raw Materials in Livestock

“…Future research should aim at a deeper understanding of the high-complexity and multi-parameter processes influencing the aflatoxin contents of feeds and foods. Novel multilateral approaches are definitely needed to control mycotoxins and their disadvantageous agricultural, health care and economic impacts more effectively (Krska et al, 2016;Stadler et al, 2018).…”

Novel analytical methods to study the fate of mycotoxins during thermal food processing

“…Generally, in such cases, the increase of an analyte needs to be statistically significant. Since signal suppression/enhancement and recovery factors may even vary for different batches of the same matrix, the statistical significance of the result must be accompanied by a careful evaluation of the measurement uncertainty [20].…”