Article pubs.acs.org/JAFC
Comparison of Data from a Single-Analyte and a Multianalyte Method for Determination of Urinary Total Deoxynivalenol in Human Samples Paul C. Turner,† Michele Solfrizzo,*,§ Allison Gost,† Lucia Gambacorta,§ Monica Olsen,# Stina Wallin,# and Natalia Kotova# †
Maryland Institute for Applied Environmental Health, School of Public Health, University of Maryland, College Park, Maryland, United States § Institute of Sciences of Food Production (ISPA) of the National Research Council (CNR), Bari, Italy # The National Food Agency, Box 622, SE 751 26 Uppsala, Sweden ABSTRACT: Deoxynivalenol (DON) exposure is estimated by the combined measures of urinary DON and DONglucuronides. In this study, data from single-mycotoxin (SM) and a multimycotoxin (MM) methods were compared for 256 Swedish adult urine samples. Both methods included β-glucuronidase predigestion, immunoaffinity enrichment, and LC-MS/MS. However, the specific reagents, apparatus, and conditions were not identical in part because the MM method measures additional mycotoxins. DON was detected in 88 and 63% of samples using the SM and MM methods, respectively, with the following mean and median concentrations: SM, mean = 5.0 ng/mL, SD = 7.4, range of positives = 0.5−60.2 ng/mL, median = 2.5 ng/mL, IQR = 1.0−5.5 ng/mL; MM, mean = 4.4 ng/mL, SD = 12.9, range of positives = 0.5−135.2 ng/mL, median = 0.8 ng/mL, IQR = 0.3− 3.5. Linear regression showed a significant, albeit modest, correlation between the two measures (p = 0.0001, r = 0.591). The differences observed may reflect subtle handling differences in DON extraction and quantitation between the methods. KEYWORDS: deoxynivalenol, multimycotoxin method, single-mycotoxin method, exposure
■
INTRODUCTION Deoxynivalenol (DON), 1 (Figure 1), also known as vomitoxin, is a B-trichothecene type mycotoxin produced by various Fusarium species.1 DON is one of the more frequently detected mycotoxins in cereal crops, including wheat, maize, and barley, and tends to occur more frequently in temperate regions of the world. 2 In animals DON and other trichothecenes adversely affect the immune system via binding to ribosomes. Additionally, they suppress appetite, reduce food utilization, and interfere with the concentration of serum insulin-like growth factor.3−6 There are numerous incidences of human toxicosis linked to Fusarium mycotoxin contamination of cereal crops, and DON contamination of these cereals is associated with some of the larger incidences, involving tens of thousands of individuals from countries including India, China, Japan, and Korea. Typical symptoms include abdominal pain and fullness, nausea, diarrhea, vomiting, fatigue, and fever,7−10 although causal relationships are not always clearly defined. In part, this reflects the acute nature of the events, limited possibility to confirm levels in contaminated food consumed at the time of poisoning, and inherent difficulties in accurate exposure assessment for mycotoxins based on food sampling. In 2003, Meky et al.11 suggested that urinary measurement of DON and a DON-glucuronide (now termed total DON (tDON)) may improve exposure assessment. Subsequently, urinary t-DON was validated as a reliable tool to assess recent DON exposure,12−14 based on analytical precision using immunoaffinity enrichment and LC-MS and demonstration of both dose−response relationships between exposure and the measure, combined with stability surveys through collection © 2016 American Chemical Society
and long-term cryo-storage of urine. Novel approaches to analytical detection are now reported, for example, GC-MS and direct analysis of DON and DON-conjugates in urine by LCMS,15−17 and some approaches report quantitative data for several mycotoxins, and metabolites thereof, in a single analysis.18−22 It is important to understand and be able to compare biomarker levels and frequencies from different laboratories, especially when subtle methodological differences exist; yet to date such data are mostly absent for mycotoxin biomarkers. The aim of this study was to compare the urinary tDON concentrations in a single set of samples, initially reported using the single-mycotoxin (SM) approach,23 with data independently obtained using the multimycotoxin (MM) method used to measure six distinct mycotoxins, including total DON.24 One smaller study of a similar nature was conducted using samples from South Africa,25 although here we have 5 times the number of samples. The parent toxin, 1 (Figure 1), and the two major DON-glucuronide conjugates, 2 and 3 (Figure 1), can be observed in the urine of individuals exposed to DON,11−16 although determinants of their relative contribution to t-DON in urine have not been reported. Special Issue: Public Health Perspectives of Mycotoxins in Food Received: Revised: Accepted: Published: 7115
October 25, 2016 November 14, 2016 November 16, 2016 November 16, 2016 DOI: 10.1021/acs.jafc.6b04755 J. Agric. Food Chem. 2017, 65, 7115−7120
Article
Journal of Agricultural and Food Chemistry
Figure 1. Structures of deoxynivalenol, 1, deoxynivalenol-15-glucuronide, 2, and deoxynivalenol-3-glucuronide, 3.
Table 1. Summary of Results of Urinary t-DON Concentrations Obtained by the Two Methods for Total and Common Samples data > LOQa
all data
a
mean, ng/mL (SD)
median, ng/mL (IQR)
N, ng/mL (%) (range)
mean, ng/mL (SD)
median, ng/mL (IQR)
SM total samples, n = 299 MM total samples, n = 278
5.3 (7.4) 4.3 (12.5)
2.8 (1.2−5.9) 0.7 (0.3−3.4)
266 (89%) (0.5−60.2) 172 (62%) (0.5−135.2)
5.6 (7.6) 6.8 (15.4)
2.9 (1.7−6.0) 2.4 (1.0−6.2)
SM common samples, n = 256 MM common samples, n = 256
5.0 (7.4) 4.4 (12.9)
2.5 (1.0−5.5) 0.8 (0.3−3.5)
226 (88%) (0.5−60.2) 160 (63%) (0.5−135.2)
5.6 (7.6) 6.9 (15.9)
2.9 (1.7−6.0) 2.5 (1.0−6.2)
Data > LOD for MM.
■
samples. DON was analyzed using a Waters 2795 HPLC separation module and MS detection with a Micromass Quattro Micro (Waters, Milford, MA, USA). The column used was a 150 mm × 4.6 mm, 5.0 μm, Luna C18 column (Phenomonex, Macclesfield, UK); all conditions, values, and concentrations were as documented by Turner et al.12 The range of the standard curve was equivalent to 0.5−62.5 ng DON/mL urine, additionally containing the 13C15 DON as an internal standard; standard curve R2 was always >0.990. The limit of detection (LOD) was 0.1 ng/mL, and the limit of quantitation (LOQ) for this analysis was 0.5 ng DON/mL urine. The mean value for the in-house QC was 9.9 ng/mL with a CV of 2.5% (n = 32). Data presented were automatically fully adjusted for recovery and that is presented. To allow method comparison to the MM, we additionally report “recoveries” without such adjustment; thus, the mean recovery through the entire extraction process was calculated as 65% (range = 61−73%). Multimycotoxin (MM) Method Performed at ISPA. This laboratory analyzed 278 urine samples using the MM method described by Solfrizzo et al.19 Briefly, frozen urine samples were thawed and centrifuged at 3000g for 5 min at 4 °C to remove particulate matter. Enzymatic deconjugation of DON-glucuronides was performed by adding 300 μL of β-glucuronidase/sulfatase solution containing 11,700 units of β-glucuronidase and 64 units of sulfatase type H-2 from Helix pomatia (Sigma-Aldrich, Milan, Italy) to 6 mL of urine, which was then incubated at 37 °C overnight. The digested urine samples were then diluted with 6 mL of ultrapure water and purified on Myco6in1 IAC (Vicam, Watertown, MA, USA) and Oasis HLB columns (Waters, Milford, MA, USA) connected in cascade. After sample elution, the two columns were separated and treated separately as follows. The Myco6in1 IAC was washed with 4 mL of water and then vacuum-dried for 15 s. DON was eluted with 3 mL of MeOH and 2 mL of water in a vial. The Oasis HLB column was washed with 1 mL of MeOH/water (20:80) and then vacuum-dried for 15 s. DON was eluted from the column by gravity with 1 mL of MeOH/water (40:60) and collected in the vial containing the eluates from the Myco6in1 IAC. The combined final eluates were dried under an air stream at 55 °C, reconstituted in 200 μL of MeOH/water (20:80) containing 0.5% acetic acid, filtered with a regenerated cellulose filter (0.20 μm), and analyzed by UPLC and a triplequadrupole API 5000 system MS/MS (Applied Biosystems, Foster City, CA, USA) by injecting 10 μL, equivalent to 300 μL of urine. The column used was a 150 mm × 2.1 mm i.d., 1.7 μm, Acquity UPLC
MATERIALS AND METHODS 23
The study population was previously described. In brief, the majority of the participants took part in Riksmaten, a Swedish national survey investigating dietary habits among adults (18−80 years), conducted from May 2010 to July 2011.26 In addition, 32 participants from a pilot study for Riksmaten, from September and October 2009, were included. The designs of the studies were very similar, and the sampling procedures were identical. The urine samples were stored at −20 °C before shipment. Urine of DON-exposed individuals contains a mix of free DON and one or more DON-glucuronides.27,28 Both analytical methods in this comparison used an enzymic digest step to convert any DON-glucuronide to DON prior to quantitation of tDON. Data collection for each approach and the analytical differences are described in brief below, although data from each study have already been separately published.23,24 The two published studies had different objectives but included subsets of samples from the main study involving 29923 and 27824 samples each, with 256 samples common to both surveys. The analysis by the SM method in the United Kingdom was conducted approximately one year before those in Italy using their MM method. Separate aliquots were sent to each laboratory, so whereas no freeze/thaw issues are relevant, the samples for MM had an additional 12 months of storage at −20 °C. Two independent surveys reported negligible loss of urinary total DON either following 2 months at −20 °C21 or after both a 1 year and 3 year storage at −40 °C.29 Single-Mycotoxin (SM) Method Performed at University of Leeds. This laboratory analyzed 299 urine samples using the SM method described by Wallin et al.23 Briefly, frozen samples were thawed and centrifuged at 2000g for 15 min at 4 °C. The supernatant was collected, and 1 mL (adjusted to pH 6.8) was treated with 250 μL of 75 mM phosphate buffer (pH 6.8) containing 7000 units of βglucuronidase type IX-A from Escherichia coli (Sigma, Poole, UK) and 20 ng of 13C15 DON internal standard (Sigma). Samples were digested overnight at 37 °C, then centrifuged, and the supernatants were diluted with 3 mL of phosphate-buffered saline (pH 7.4). Diluted supernatants were passed through wide-bore DON immunoaffinity columns under gravity, according to the manufacturer’s instructions (Vicam, Milford, MA, USA). DON was eluted with 4 mL of MeOH and dried in vacuo. The residue was reconstituted in 250 μL of 10% (v/v) ethanol and analyzed by LC-MS by injecting 5 μL, equivalent to 20 μL of urine. Two quality control (QC) samples (urine spiked with 10 ng/mL DON) and two PBS blanks were run with each batch of 20 7116
DOI: 10.1021/acs.jafc.6b04755 J. Agric. Food Chem. 2017, 65, 7115−7120
Article
Journal of Agricultural and Food Chemistry BEH phenyl column (Waters, Milan, Italy). Quantitation was performed by using a five-point matrix-matched calibration curve prepared by using five purified urine extracts obtained from a pool of urine samples. The range of the calibration curve was 0.5−120 ng/mL for DON, and results were reported as nanograms per milliliter after correction for method recovery (77%). The LOD was 0.45 ng/mL, and the LOQ was 1.5 ng/mL.
samples, the concentrations of DON measured with the SM method were all higher than those obtained with the MM method, having mean values of 5.3 and 1.7 ng/mL, respectively. For 60 other samples the DON concentrations of the SM were all lower than the MM method, mean values being 5.5 and 12.6 ng/mL, respectively. For seven samples the same DON concentrations were obtained by using both methods. The range of DON concentrations found in the common samples permitted comparison of the two methods over an analytical range of 2 orders of magnitude. As shown in Figure 3, a significant albeit modest linear relationship (p < 0.0001) was observed between the two methods (R2 = 0.35).
■
RESULTS AND DISCUSSION All data below the LOQ were assigned a value of half the LOQ and if below the LOD, a value of half the LOD. The data obtained with the MM method were corrected for the method recovery, whereas the data obtained with the SM method were generated using an internal standard, and thus reported data are individually adjusted. Recoveries and repeatability of results (RSDr) for both methods were within the criteria for methods established by EU Regulation No. 401/200630 (RSDr ≤ 20%, recovery range = 60−110%). In addition to DON, the MM method analyzed fumonisins B1 and B2, zearalenone, α- and βzearalenols, ochratoxin A, and nivalenol, published previously.24 Table 1 reports the summary of results obtained in the two laboratories. For all samples analyzed t-DON was detected in 89% of 299 samples analyzed by using the SM method (overall mean = 5.3 ng/mL; range = 0.5−60.2 ng/mL; median = 2.8 ng/mL, interquartile range (IQR) = 1.2−5.9 ng/mL), whereas it was detected in 62% of 278 samples analyzed by using the MM method (mean = 4.3 ng/mL, range = 0.5−135.2 ng/mL, median = 0.7 ng/mL; IQR = 0.3−3.4 ng/mL). Within the 256 common samples comparable mean values but not median values were obtained for the two methods, with the median value significantly lower using the MM method compared to the SM method (Table 1). The t-DON concentration was below the LOD for 30 samples (12%) and for 97 samples (38%) by the SM and MM methods, respectively, whereas 146 (57%) of samples were above the LOD and 25 (10%) were below the LOD by both methods. For the 97 samples below the LOD using the MM method, 78 (80%) were above the LOD by the SM method; 60 (77%) of which were
