Heterogeneity of mycotoxin contamination of crops RIKILT RD project 2015 Ine van der Fels-Klerx, Theo de Rijk, Martin Alewijn, Yannick Weesepoel
Mycotoxins Secondary metabolites from fungi that can infect crops Under conducive conditions Local weather; warm and humid weather Field management farmer Harvest and storage conditions Causing disease and deaths in humans and animals EC legislation setting maximum limits for presence of mycotoxins in commodities, feed and/or food
Mycotoxins Major classes of present concern Aspergillus mycotoxins, e.g. aflatoxins, ochratoxin A Penicillium mycotoxins, e.g. ochratoxin A and patulin Fusarium mycotoxins, e.g. trichothecenes, fumonisins and zearalenone Whitaker, Cannavan et al. 2010
Heterogenous distribution Spatial distribution of aflatoxin B1 in maize in the field, pre-harvest
Sampling & analysis Official & private control Sampling & analysis plan Number (sub)samples, size sub- and aggregate samples, sampling preparation Mycotoxin detection procedure, defined accept/reject limit (e.g. regulatory limit) 1- Sampling: taking a sample from a lot; 2- Sample preparation: milling the sample to reduce the size of its particles and take a subsample for analysis, and; 3- Analysis: extract the mycotoxin from the subsample and quantify it (Whitaker et al., 2011).
Variability mycotoxin sampling & analysis Reasons Inhomogeneity of distribution of mycotoxins in batch; sampling variance (largest source of variability) Detection methods: expensive and time-consuming, or less sensitive and accurate Limited costs for sampling (too few samples/ size) No 100% certainty of determined lot concentration Seller’s and buyer’s risk
Performance of sampling plan The performance of a sampling & analytical plan can be determined from an operating characteristic (OC) curve Good lots (lots with a concentration less than or equal to the legal limit) will test bad and be rejected by the sampling plan seller’s risk Bad lots (lots with a concentration greater than the legal limit) will test good and be accepted by the sampling program buyer’s risk Operating characteristic Curve
How to reduce variability ? Increasing laboratory sample size (or number of samples of a given size) Increasing the number of subsamples Increasing test portion size Grinding the sample into smaller particles Using a more precise analytical method (HPLC, LC-MS , ELISA, dipsticks, ..) Methods with red underline will be used to optimize sampling plan.
Aflatoxins in maize Aflatoxins: group of mycotoxins (B1, B2, G1, G2), A flavus and parasiticus. AFB1 most toxic mycotoxin, genotoxic and carcinogenic. In EC, AFB1 strictly regulated in food (EC, 2006b) and feed (EC, 2002) AFB1 often occurs in maize, e.g. 2013 Balkan incident Regulation (EU) No 691/2013 prescribes sampling, e.g. 100 subsamples
Problem statement Official procedure: Resource (time, money) demanding Could new methods such as simple video camera or ELISA be of help, e.g. for pre-screening? Quick, low costs, high throughput, local contamination More subsamples can be analysed at same costs. Will it improve OC curves?
Aim study Evaluate if the hyperspectral camera (near infrared, NIR 1000-2500 nm) can be used to estimate AFB1 contamination in maize batches
Reference set - 100 samples of maize kernels of approx. 800-1000 g Collected with at regular intervals from one shipment Analysed at RIKILT with LC-MSMS for AFB1 concentrations 3 samples with AFB1 >20 µg/kg (21.3, 25.7, 27.0 µg/kg)
Multispectrale imaging (MSI) - Videometer Co-operation with Bioscience (WUR)
MSI – schematic overview Camera & lens Captures reflectance data at each LED wavelength Integrating sphere Diffuses light onto sample 19 LED’s Narrowband illumination 375-970 nm Sequential strobes Sample Imaging time: 5-10 s
MSI – schematic overview
Principal component analysis 375 405 435 450 470 505 525 570 590 630 645 660 700 780 850 870 890 940 970 PC1 PC2 PC... Variables: wavelengths (nm) Reduces the amount of variables (wavelengths) Goal: find the directions (PC’s) that maximize the variance in the dataset. Variables: principal components (PC’s)
Results
Conclusion The use of the current simple videometer for detection of AB1 in maize is not effective enough Possibly because Lack of reference spectra of mycotoxins and–producing fungi in this matrix and with this wave lenght Too much grit in subsamples Experiences gained with data analysis and use of camera In future studies, we will investigate a more proven effective camera (300 wave lenghts), available at RIKILT in 2016
Thanks For Your Attention