Control Food

RUPP: JOURNAL OF AOAC INTERNATIONAL VOL. **, NO. *, 2002 1355

FOOD CHEMICAL CONTAMINANTS

Determination of Deoxynivalenol in Whole Wheat Flour and

Wheat Bran

HEIDI S. RUPP

U.S. Food and Drug Administration, Pacific Regional Laboratory Northwest, 22201 23rd Dr, SE, Bothell, WA 98021

A liquid chromatographic (LC) method was devel- a maximum DON level of 2 ppm in cereals. There is also a con-

oped for determining deoxynivalenol (DON) in cern that moldy cereals may be diverted from human food pro-

whole wheat flour and wheat bran. A 15 g test sam- cessing into pet food processing.

ple was extracted with acetonitrile water To ensure the safety of the wheat and corn supply, the level

(84 + 16, v/v) and applied to a Romer MycoSep of DON is determined by liquid chromatography (LC) with

cleanup column. The eluate was dried and then re- UV detection (1). At present there are no official methods for

constituted in a 0.1M phosphate buffer, pH 7.0, and the LC determination of DON in wheat, bran, and corn, nor is

applied to a Vicam DONtest-LC cleanup column. there an official confirmation (alternative analysis). The cur-

The methanol eluate was chromatographed with a rent AOAC method that is used for regulatory analysis (1), al-

methanol water (17 + 83, v/v) mobile phase on a though fast and simple, does not adequately clean up all given

C18 column with UV detection at 220 nm. Five repli- samples, which may give an irregular baseline, preventing

cates at each of 5 fortification levels (0.25, 0.50, 1.0, confident identification and quantitation of the analyte. The

2.0, and 4.0 ppm), plus 5 controls, were determined tolerance level of DON is not much more than the limit of de-

for both whole wheat flour and wheat bran. For tection (LOD) of the current detection method, which permits

flour, the average recoveries were 72.2 91.5% with only low ppm determination. Although analysis of pure stan-

relative standard deviations (RSDs) of 4.9 18.4%. dard gives a curve with a correlation near unity, the introduc-

The intra-assay flour recovery was 82.4% with tion of matrix modifiers in the chromatographic system causes

9.8% RSD. A 5 replicate sample of naturally in- the retention time of DON to shift dramatically and inconsis-

curred wheat had an average of 1.1 ppm DON with tently, adding to the uncertainty of identification.

6.7% RSD. For bran, average recoveries of fortified

DON does not have a strong absorbance (LOD is low ppm)

samples were 69.5 99.7% with RSDs of 1.7 18.8%.

or a very unique UV spectrum. Identification relies solely on

The intra-assay bran recovery was 81.5% with

retention time of peaks monitored at 220 nm a wavelength

8.9% RSD. The limit of detection (about 3 noise)

that is very nonspecific. Matrix peak(s) at or near the retention

for the method is 0.05 ppm; the correlation coeffi-

time of DON are very common and can often be mistaken for

cient (linearity) was >0.9995. The DON peak was

DON. Therefore, every sample must be spiked, which in-

clearly identified and easily integrated in the

creases sample preparation and analysis times. Furthermore,

chromatograms.

the instrumental run time for DON samples is 40 min because

coextractants must be flushed off the column after each injec-

tion so that they do not carry over to the next injection.

ycotoxins are analytes of interest under the U.S. Food

M Numerous methods in the literature address the analysis of

and Drug Administration s (FDA) Food Safety Initia-

DON, including several that explore charcoal column (2, 3)

tive. Deoxynivalenol (DON) is a common contami-

and solid-phase extraction cleanup (4, 5), and a few that ex-

nant of wheat and corn and is very stable to processing. It is also

plore immunoaffinity cleanup, such as the Vicam

called vomitoxin because exposure causes feed refusal, weight

DONtest (1, 6 8). Derivatization for fluorometric analysis has

loss, and vomiting. DON belongs to the B-trichothecene class

been explored (9), as well as derivatization for gas chromato-

of Fusarium mycotoxins (Figure 1). As such, DON is known to

graphic (GC) analysis (5, 10 12). A few authors have devel-

adversely affect the gastrointestinal and immunosuppressive

oped methods for mass spectral analysis (12, 13). Because

systems. Wheat and corn products (cereals) are a staple food for

DON is a common contaminant of a number of important

humans and animals. DON is a common contaminant in cereals

foodstuffs, the literature also boasts several articles dealing

around the world, including North America. In a 1993 survey,

with method reviews (14 18) and general concerns about

40% of U.S. wheat samples contained DON at levels

DON (19 22).

>2 ppm (1). The World Health Organization has recommended

Although there has been a fair amount of method develop-

ment research involving DON, the above methods standing

alone do not adequately meet our needs. The positive features

Received April 1, 2002. Accepted by AP June 12, 2002.

Corresponding author s e-mail: *****@***.***.***. of the various applicable methods have been combined in the

1356 RUPP: JOURNAL OF AOAC INTERNATIONAL VOL. 85, NO. 6, 2002

(g) Water bath/nitrogen evaporator. Meyer N-Evap,

12 stations, set at 60 C (Organomation Associates, Inc.,

Berlin, MA).

(h) Vortex mixer. Vortex Genie (VWR Scientific, San

Francisco, CA).

(i) Glassware. Stoppered 125 mL Erlenmyer flasks;

13 150 mm glass culture tubes; LC sample vials with

low-volume inserts; funnels.

(j) Pipettors. Adjustable 10 100 and 100 1000 L

Figure 1. Deoxynivalenol (vomitoxin).

(Calibra, Wheaton Science Products, Millville, NJ).

(k) Instrumentation. Agilent 1100 series quaternary LC

with diode-array UV detector, monitor at 220 nm (Agilent

method presented here. Our method makes use of both Romer

Technologies, Palo Alto, CA). Flow 0.7 mL/min, oven at

and Vicam cleanup columns to provide clear chromatograms

40 C, 50 L injection size.

of difficult samples without the need for derivatization for de-

(l) Analytical column. Waters Nova Pak C18,

tection with common UV instrumentation. However, we ulti-

300 3.9 mm 5 m, No. WAT011695 (Waters Corp.,

mately chose to use 2 cleanup columns in tandem to ensure the

Milford, MA).

cleanest possible extract for subsequent confirmational analy-

sis by liquid chromatography/mass spectrometry (LC/MS;

Standards

method in preparation). Our method was validated on both

whole wheat flour and wheat bran (data presented here). Pre- DON standard was purchased from Sigma-Aldrich (Mil-

liminary results indicate that our method may be useful for waukee, WI) in either 1 or 5 mg bottles (with septa). A 1 2 mL

corn (data not presented). volume of 50% methanol solution was injected through the

septum of the standard bottle, the bottle was mixed on a Vor-

Experimental

tex mixer, and the solution was transferred by using a 1 mL

gas-tight syringe to a volumetric flask (10 or 50 mL, respec-

Apparatus and Reagents tive to the mg used). This addition of 50% methanol to the

standard bottle, rinse, and transfer to volumetric flask was re-

(a) Solvents. LC grade acetonitrile (ACN), methanol

peated 3 more times. The contents were diluted to volume

(MeOH); anhydrous ethanol (EtOH); deionized water.

with additional 50% methanol, which gave a stock standard of

(b) Buffer. Phosphate, 0.1M, pH 7.0. Prepared from so- 100 g/mL. (Standard may be refrigerated up to 4 months.)

dium dihydrogen phosphate, NaOH or HCl as needed, and The solution was transferred to a screw-capped vial for stor-

deionized water. age and easy access for pipettors. Working standards, at 0.3,

(c) Methanolic solutions. Mobile phase: water metha- 1.5, 3.0, 12.0, and 18.0 /mL, were made by diluting 30, 150,

300, 1200, and 1800 L (using pipettors) of stock standard, re-

nol (83 + 17; mixed on instrument). Stock diluent: 50% meth-

anol. Sample phase: 20% methanol. spectively, to 10 mL with 20% methanol. These standards

(d) Cleanup columns. MycoSep No. 225 Columns, give equivalent sample concentrations of 0.1, 0.5, 1.0, 4.0, and

No. C2225 (Romer Labs, Inc., Union, MO). DONtest-HPLC, 6.0 ppm, respectively (see calculation below). The working

No. G1005 (Vicam, Watertown, MA). Supplemental 20 mL standard solutions were transferred to screw-topped culture

polypropylene reservoirs and column adapters to fit. tubes and stored in the refrigerator (up to 2 months). An

(e) Filter paper. Fluted, 24 cm diameter, No. 588 aliquot of each solution may be taken as needed for analysis.

(Schleicher & Schuell, Keene, NH). The calculations to determine the working standard con-

centrations are as follows:

(f) Blender. Multispeed Waring or Osterizer style with

pint-sized Mason containers. Extraction process:

Table 1. Deoxynivalenol in whole wheat flour

Spike levela 0.25 ppm 0.50 ppm 1.0 ppm 2.0 ppm 4.0 ppm Intra-assay average

NAb

Avg. found, ppm 0.20 0.43 0.70 1.5 3.4

Avg. recovery, % 91.5 85.8 72.2 76.7 85.6 82.4

18.4c

RSD, % 8.6 7.6 9.5 4.9 9.8

a

Five replicate extracts at each of the 5 levels were analyzed.

b

NA = not applicable.

c

If recalculated for n = 4 without high outlier, then recovery would be 79.5% with RSD of 10.1%. However, Q Test gives

RUPP: JOURNAL OF AOAC INTERNATIONAL VOL. 85, NO. 6, 2002 1357

Table 2. Deoxynivalenol in wheat bran

Spike levela 0.25 ppm 0.50 ppm 1.0 ppm 2.0 ppm 4.0 ppm Intra-assay average

NAb

Avg. found, ppm 0.20 0.41 0.80 1.5 2.8

Avg. recovery, % 99.7 82.6 81.9 73.6 69.5 81.5

18.8c

RSD, % 5.2 11.1 7.5 1.7 8.9

a

Five replicate extracts at each of the 5 levels were analyzed.

b

NA = not applicable.

c

If recalculated for n = 4 without low outlier, then recovery would be 89.2% with RSD of 3.8%. Q Test gives >96% confidence level.

600 L stock standard, respectively. Five replicates of each of

15 g flour/100 mL solution =

0.15 g/mL 5 mL aliquot = the 5 spike levels of both whole wheat flour and wheat bran

were analyzed. Five control or blank replicates of both com-

0.75 g flour/0.25 mL final volume =

modities were also analyzed. This provided a population of

3.0 g flour/mL solution

n = 30 for each commodity. Additionally, 5 replicates of natu-

Working standard equivalency: rally incurred whole wheat flour were analyzed.

DON working standard solution of 3 g DON / mL solution Each day, the 5 working standards were injected prior to the

= samples. A standard curve was constructed daily, and peak areas

3.0 g flour / mL solution

of the test samples were compared with the curve to determine

1 g DON/g flour = 1 ppm

the concentration in ppm. Reported recoveries are absolute.

Sample Spiking and Method Design

Procedure

Whole wheat flour for analysis was from 3 combined lots

of previously analyzed flour that was found to contain no Weigh 15 g ground sample into blender jar. Add 100 mL

DON by the traditional method (1). Wheat bran was of 2 com- ACN water (84 + 16, v/v). Blend for 3 min. Filter through

bined lots found to be clean. Incurred whole wheat flour (hard fluted paper, collect in stoppered flask. According to manu-

red wheat from Red River Valley, 1999) was provided by facturer s directions, pass portion(s) through Romer MycoSep

General Mills (Minneapolis, MN). column (combine portions if necessary) to collect >5 mL.

Individual 15 g samples were spiked to contain 0.25, 0.50, Transfer 5.0 mL aliquot to a tube. Dry under N2 over a near

60 C water bath (add EtOH to help dry if needed) to near dry-

1.0, 2.0, and 4.0 ppm DON with 37.5, 75.0, 150, 300, and

Figure 2. DON peak at about 10.6 min. Top line (x) shows whole wheat flour extract by AOAC method (1). Other

lines, by our method, are (a) sample (2.2 ppm), (b) spike (1 ppm), (c) matrix control (known trace), and (d) reagent

blank. Peak at about 12.5 min by our method is believed to be Vicam column artifact. Sample and matrix control are

from a different source than that used for method validation.

1358 RUPP: JOURNAL OF AOAC INTERNATIONAL VOL. 85, NO. 6, 2002

ness. Solubilize residue in 2 1 mL 0.1M phosphate buffer, Vicam columns. The part of the method that required the most

pH 7.0. Prepare Vicam DONtest-HPLC immunoaffinity col- care was the application of solutions to the Vicam columns

umn by washing with 2 mL 0.1M phosphate buffer, pH 7.0, through the top of the adduct reservoir without having an air

and attaching small reservoir to top of column. Transfer bubble prevent the solution from flowing. It was helpful to in-

solubilized residue to reservoir and pass through Vicam col- sert a glass Pasteur pipet through the solution in the adduct

umn at a gentle rate of not more than 2 mL/min. Wash column reservoir body and neck, down into the Vicam reservoir, and

with 2 mL distilled water. Remove extra reservoir. Gently as- to use it to squeeze out the excess air.

pirate off excess water for 2 3 s. Elute column with 4 1 col- Overall, variability (around 10% RSD on average) and re-

umn bed volumes (ca 4 mL total) methanol at gentle rate of covery (

Contact this candidate