Bayesian approach to peak deconvolution and library search for high resolution gas chromatography - Mass spectrometry.

A novel probabilistic Bayesian strategy is proposed to resolve highly coeluting peaks in high-resolution GC-MS (Orbitrap) data. Opposed to a deterministic approach, we propose to solve the problem probabilistically, using a complete pipeline. First, the retention time(s) for a (probabilistic) number of compounds for each mass channel are estimated. The statistical dependency between m/z channels was implied by including penalties in the model objective function. Second, Bayesian Information Criterion (BIC) is used as Occam's razor for the probabilistic assessment of the number of components. Third, a probabilistic set of resolved spectra, and their associated retention times are estimated. Finally, a probabilistic library search is proposed, computing the spectral match with a high resolution library. More specifically, a correlative measure was used that included the uncertainties in the least square fitting, as well as the probability for different proposals for the number of compounds in the mixture. The method was tested on simulated high resolution data, as well as on a set of pesticides injected in a GC-Orbitrap with high coelution. The proposed pipeline was able to detect accurately the retention times and the spectra of the peaks. For our case, with extremely high coelution situation, 5 out of the 7 existing compounds under the selected region of interest, were correctly assessed. Finally, the comparison with the classical methods of deconvolution (i.e., MCR and AMDIS) indicates a better performance of the proposed algorithm in terms of the number of correctly resolved compounds.

Hypoglycin A Concentrations in Maple Tree Species in the Netherlands and the Occurrence of Atypical Myopathy in Horses.

BACKGROUND: Atypical myopathy (AM) in horses is caused by the plant toxin hypoglycin A, which in Europe typically is found in the sycamore maple tree (Acer pseudoplatanus). Owners are concerned about whether their horses are in danger if they graze near maple trees. HYPOTHESIS/OBJECTIVES: To measure hypoglycin A in the most common maple tree species in the Netherlands, and to determine whether concentration of toxin is a predictor of AM in horses. METHODS: A total of 278 samples of maple tree leaves, sprouts, and seeds were classified by species. Mean concentrations of hypoglycin A were compared for the type of sample, the season and the occurrence of AM in the pasture (non-AM versus AM). Statistical analysis was performed using generalized a linear model (SPPS22). RESULTS: Almost all Acer pseudoplatanus samples contained hypoglycin A, with concentrations differing significantly among sources (P < .001). Concentrations were significantly higher in seeds from the AM group than in seeds from the non-AM group (856 ± 677 and 456 ± 358 mg/kg, respectively; P = .039). In sprouts and leaves this was not the case. Acer platanoides and Acer campestre samples did not contain detectable concentrations of hypoglycin A. CONCLUSIONS AND CLINICAL IMPORTANCE: Acer platanoides and campestre seem to be safe around paddocks and pastures, whereas almost all Acer pseudoplatanus samples contained hypoglycin A. In all AM cases, Acer pseudoplatanus was found. Despite significantly higher concentration of hypoglycin A in seeds of pastures where AM has occurred, individual prediction of AM cannot be made by measuring these concentrations because of the high standard deviation.

Guidelines for the validation of qualitative multi-residue methods used to detect pesticides in food.

There is a current trend for many laboratories to develop and use qualitative gas chromatography-mass spectrometry (GC-MS) and liquid chromatography-mass spectrometry (LC-MS) based multi-residue methods (MRMs) in order to greatly increase the number of pesticides that they can target. Before these qualitative MRMs can be used for the monitoring of pesticide residues in food, their fitness-for-purpose needs to be established by initial method validation. This paper sets out to assess the performances of two such qualitative MRMs against a set of parameters and criteria that might be suitable for their effective validation. As expected, the ease of detection was often dependent on the particular pesticide/commodity combinations that were targeted, especially at the lowest concentrations tested (0.01 mg/kg). The two examples also clearly demonstrated that the percentage of pesticides detected was dependent on many factors, but particularly on the capabilities of the automated software/library packages and the parameters and threshold settings selected for operation. Another very important consideration was the condition of chromatographic system and detector at the time of analysis. If the system was relatively clean, then the detection rate was much higher than if it had become contaminated over time from previous injections of sample extracts. The parameters and criteria suggested for method validation of qualitative MRMs are aimed at achieving a 95% confidence level of pesticide detection. However, the presence of any pesticide that is 'detected' will need subsequent analysis for quantification and, depending on the qualitative method used, further evidence of identity.

Occurrence of Fusarium Head Blight species and Fusarium mycotoxins in winter wheat in the Netherlands in 2009.

Most recent information on the occurrence of Fusarium Head Blight species and related mycotoxins in wheat grown in the Netherlands dates from 2001. This aim of this study was to investigate the incidence and levels of Fusarium Head Blight species and Fusarium mycotoxins, as well as their possible relationships, in winter wheat cultivated in the Netherlands in 2009. Samples were collected from individual fields of 88 commercial wheat growers. Samples were collected at harvest from 86 fields, and 2 weeks before the expected harvest date from 21 fields. In all, 128 samples, the levels of each of seven Fusarium Head Blight species and of 12 related mycotoxins were quantified. The results showed that F. graminearum was the most frequently observed species at harvest, followed by F. avenaceum and M. nivale. In the pre-harvest samples, only F. graminearum and M. nivale were relevant. The highest incidence and concentrations of mycotoxins were found for deoxynivalenol, followed by zearalenone and beauvericin, both pre-harvest and at harvest. Other toxins frequently found--for the first time in the Netherlands--included T-2 toxin, HT-2 toxin, and moniliformin. The levels of deoxynivalenol were positively related to F. graminearum levels, as well as to zearalenone levels. Other relationships could not be established. The current approach taken in collecting wheat samples and quantifying the presence of Fusarium Head Blight species and related mycotoxins is an efficient method to obtain insight into the occurrence of these species and toxins in wheat grown under natural environmental conditions. It is recommended that this survey be repeated for several years to establish inter-annual variability in both species composition and mycotoxin occurrence.

Screening of plant toxins in food, feed and botanicals using full-scan high-resolution (Orbitrap) mass spectrometry.

A generic method based on LC with full-scan high-resolution (Orbitrap) mass spectrometry (MS) was systematically investigated for the simultaneous detection of a wide range of plant toxins in a variety of food and feed matrices. For a selection of 150 substances, representing various chemical classes, the limit of detection was established using fixed LC-MS conditions. Ion suppression effects and selectivity were evaluated using generic extracts from representative and relevant matrices (food supplement, honey, silage, compound feed). The majority of the substances could be measured as positive ions after electrospray ionisation (ESI(+)). Using a mass resolving power of 100,000 a reliable high mass accuracy was obtained despite the high abundance of co-extractants in the sample extracts. This enabled the use of ±5 ppm mass extraction windows, which in turn resulted in a high degree of selectivity. On the other hand, except for honey, strong ion suppression effects were frequently observed which adversely affected the detection limits. Nevertheless, for the majority of the substances the detection limits were in the range 0.01-0.05 mg kg(-1). Since non-selective sample preparation and non-targeted data acquisition were performed, the presence of plant toxins initially not targeted for during data review can be subsequently investigated, which is a very useful option because for many known toxins no analytical reference standards are yet available. The applicability of the method was demonstrated by analysis of a variety of real-life samples purchased on the market or from cases of intoxication. These included honey, herbal tea, food supplements, poppy seeds, traditional Chinese medicines, compound feed, silage and herb-based feed additives. Plant toxins that were detected included various pyrrolizidine alkaloids, grayanotoxins, opium alkaloids, strychnine, ricinine (a marker for ricin), aconitine, aristolochic acid and cardiac glycosides (e.g. digitoxin, digoxin).