SOP xxx, Version 1.0

FOOD STANDARDS AGENCY

STANDARD OPERATING PROCEDURE (SOP) xxx

Version 1.0, August, 2006

STANDARD OPERATING PROCEDURE FOR DETERMINATION OF THE GEOGRAPHICAL ORIGIN OF UNCOOKED BEEF MEAT

Prepared by ______Date ______

Dr Simon Kelly, Institute of Food Research

Approved by ______Date ______

Page 1 of 12

SOP xxx, Version 1.0

CONTENTS

Page 1 of 12

SOP xxx, Version 1.0

1.HISTORY / BACKGROUND......

1.1Background......

2.PURPOSE......

3.SCOPE......

4.DEFINITIONS AND ABBREVIATIONS......

5.Principle of the method......

5.1Chemicals......

5.2Solutions, standards and reference materials......

5.3Equipment......

6.PROCEDURES......

6.1Sample preparation......

6.2Measurement of 13C‰ and 15N‰......

6.2.1Preparation of combustion and reduction reactors

6.2.2Preparation of Reference Materials and samples for 13C and 15N analysis

6.2.3Batch protocol

6.2.4Correction of measured 13C‰ and 15N‰ data

6.3Quality Assurance......

6.3.1Quality of ‘fat free dry mass’

6.3.2Absolute difference between duplicate 13C‰ and 15N‰ results

6.3.3Measurment of L-glutamic acid reference material (USGS40)

7.CALCULATIONS AND DATA ANALYSIS......

7.1Interpretation of FFDM 13C‰ data......

Page 1 of 12

SOP xxx, Version 1.0

1.HISTORY / BACKGROUND

1.1Background

As a result of concerns relating to bovine spongiform encephalopathy (BSE), Human variant Creutzfeldt-Jakob disease (CJD) and the impact on the internal market, the European Commission established far reaching legislation concerning the labelling of beef. The introduction of pan-European compulsory beef labelling rules, from the 1st September 2000 onwards [(EC) No 2772/1999] was designed to provide consumer’s with correct, complete and transparent information to enable them to make an informed choice on the type and origin of beef they purchased[1]. The natural variation, or fractionation, that occurs in the isotopic content of the bio-elements, hydrogen, carbon, nitrogen, and oxygen may be exploited to determine the geographical origin of beef and dietary patterns that can act as proxies for the identification of provenance. In addition multi-element screening may be used to identify macro-, micro- or trace-elements that indicate origin. In some instances comparison of one or two variables is sufficient to discriminate geographical origins; for example, carbon stable isotope ratios of beef fat-free dry mass indicate the quantity of maize in the diet and this leads to useful ‘screening’ discrimination of UK and Brazilian beef. Discriminant analysis of the stable isotope and multi-element data permits simultaneous comparison of several origins on a global and regional scale and quantitative assessment of correct classification and identification of the most useful elements for geographical discrimination.

2.PURPOSE

The SOP is designed to provide an initial rapid screening technique to identify beef that is derived from cattle that have been reared with a significant proportion of maize (or C4 component) in their diet. This forms the basis for rapidly identifying beef that is likely to have originated in Brazil. It has previously been acknowledged that there is a significant undisclosed trade in mislabelled Brazilian beef through UK ports.

3.SCOPE

The methods described in this SOP are for an initial measurement of the 13C‰ value of beef fat-free dry mass to identify beef derived from cattle fed predominantly maize in their diet. The supplementary determination of δ2H‰, 15N‰, trace element composition and 87Sr/86Sr ratio serve as further confirmation of origin.

4.DEFINITIONS AND ABBREVIATIONS

13C‰, 13C‰PDB"Delta carbon13 per mil''. The 13C/12C ratio expressed relative to the international standard Pee Dee Belemnite.

13C/12CThe ratio of the isotope of carbon with atomic mass 13 to the isotope of carbon with atomic mass 12

PDBPee Dee Belemnite Calcium carbonate used as an international standard for which the 13C/12C ratio is precisely known and is defined as 0 ‰ on the 13C scale.

15N‰,15N‰AIR"Delta nitrogen15 per mil''. The 15N/14N ratio expressed relative to the international standard AIR.

15N/14NThe ratio of the isotope of nitrogen with atomic mass 15 to the isotope of nitrogen with atomic mass 14.

AIRAIR atmospheric air used as an international standard for which the 15N/14N ratio is precisely known and is defined as 0 ‰ on the 15N scale.

EAElemental Analyser

ICMInter-Comparison Material

sdThe sample standard deviation of the mean, (n 1)

IRMSIsotope Ratio Mass Spectrometry

5.Principle of the method

The beef is cut into small pieces and is dried completely with the aid of a lyophiliser (freeze-drier). The dried pieces are homogenized with a suitable grinder. The resultant dry powder is extracted with petroleum ether for 6 hours in a soxhlet-apparatus. Afterwards the fat free dry mass (FFDM) can be stored in an appropriate container in a vacuum desiccator until measurement.

Sufficient organic FFDM is placed in a tin capsule to provide approximately 100 μg of carbon. The tin capsule is sealed and dropped into an elemental analyser reaction tube containing chromium oxide at 1020C. The sample is quantitatively converted into carbon dioxide, nitrogen oxides and water. The combustion products pass into a second reactor at 650C that quantitatively coverts the nitrogen oxides to dinitrogen gas. Water is then removed from the carrier stream by a chemical trap containing drying agent and the carbon dioxide and nitrogen are separated on a packed GC column. The GC effluent then flows into the stable isotope ratio mass spectrometer via an interface and the ratio of the isotopomers of nitrogen and carbon dioxide are determined against nitrogen and carbon dioxide reference materials or gasses of known 15N/14N and 13C/12C ratios versus accepted international standards.

MATERIALS AND EQUIPMENT

5.1Chemicals

5.1.1Petroleum ether, boiling range 40-60°C
Supplier: Riedel-de Haën, Cat. no. 32246-2.5L
Directions for Safe Handling: Do not breathe vapor. Avoid contact with eyes, skin, and clothing. Avoid prolonged or repeated exposure.
Conditions of Storage: Keep container closed. Keep away from heat, sparks, and open flames.

5.1.2Chromium oxide
Empirical Formula (Hill Notation): Cr2O3, Formula Weight: 151.99
Supplier: Elemental Microanalysis, Cat. No. B1179
Directions for Safe Handling: Avoid inhalation. Avoid contact with eyes, skin, and clothing. Avoid prolonged or repeated exposure.
Conditions of Storage: Keep tightly closed.

5.1.3Silvered copper oxide
Supplier: Elemental Microanalysis, Cat. No. B1070

5.1.4Copper wires, reduced
Empirical Formula (Hill Notation): Cu, Formula Weight: 63.55
Supplier: Elemental Microanalysis, Cat. No. B1213
Not hazardous according to Directive 67/548/EEC.

5.1.5Magnesium perchlorate (drying agent)
Linear Formula: Mg(ClO4)2, Formula Weight: 223.21
Supplier: Elemental Microanalysis, Cat. No. B1101
Directions for Safe Handling: Avoid breathing dust. Do not get in eyes, on skin, on clothing. Avoid prolonged or repeated exposure.
Conditions of Storage: Keep tightly closed. Store in a cool dry place. Unsuitable: Do not store near, nor allow contact with, clothing and other combustible material.

5.2Solutions, standards and reference materials

5.2.1L-glutamic acid is a non-essential amino acid, supplied by the International Atomic Energy Agency, with an accepted 13C‰ value of -26.2‰ versus Pee Dee Belemnite and 15N‰ value of -4.5‰ versus AIR.
Supplier: IAEA (Vienna), Cat. No. USGS40

5.2.2Collagen Inter-Comparison Material (collagen ICM working standard). This industrial porcine collagen material has been characterised by TRACE participant stable isotope laboratories and has a nominal 13C‰ versus Pee Dee Belemnite of -17.94‰ and a 15N‰ value of +6.2‰ versus AIR.

5.3Equipment

5.3.1 / Teflon chopping board
5.3.2 / (ceramic) knives
5.3.3 / wide neck bottles with filter cap (600ml) for freeze-drying
5.3.4 / freeze-dryer, e.g. Christ, Alpha 1-4 e.g. Christ Cat. No. 127415
5.3.5 / Mulinette food processor with insert
5.3.6 / pair of (ceramic) scissors
5.3.7 / Agate pestle and mortar
5.3.8 / extraction thimble, 33 x 118mm, e.g. Schleicher & Schuell Cat. No. 10350245
5.3.9 / quartz wool
5.3.10 / folded filters, 150 mm, e.g. Schleicher & Schuell Cat. No. 312545
5.3.11 / soxhlet- extraction apparatus (with 150ml extractor)
5.3.12 / Iso-mantle heater
5.3.13 / 250ml round-bottomed flask
5.3.14 / analytical mill, e. g. Fa. IKA
5.3.15 / Securitainers, 35ml
5.3.16 / rotating evaporator
5.3.17 / evaporator with heating module and nitrogen facility, e.g. Barkey
5.3.18 / tin capsules for solid samples (3,3 x 5 mm; or similar)
5.3.19 / Elemental analyser (EA)
5.3.20 / Isotope Ratio Mass Spectrometer (IRMS)

6.PROCEDURES

6.1Sample preparation

Cut 100g of beef meat into small pieces of approximately 1cm3 using a clean ceramic knife and a Teflon chopping board (5.3.1; 5.3.2). The chopping process is aided by using partially rather than fully thawed meat. These pieces are placed in a suitable receptacle (5.3.3) and connected to/placed in a freeze dryer (5.3.4).

The beef is freeze dried for 48 hours. Completeness of drying can be checked by removing samples from the freeze drier and checking the temperature of the sample container (by hand): a cold container indicates remnants of ice in the sample. Alternatively, in a gas-tight freeze-drier system the vacuum may be temporarily iso-lated from the drier and a sharp rise in pressure indicates a significant vapour pressure from the samples. The dry meat is homogenized afterwards with the help of a Mulinette (5.3.5) [IRMS only] or ceramic scissors (5.3.6) and an agate pestle and mortar (5.3.7).

Approximately 40g, of the beef dry mass is placed in an extraction thimble (5.3.8) and closed with quartz wool (5.3.9) in a folded filter (5.3.10).

NOTE 1: The quartz wool has to be separated from the sample with a piece of filter paper. Otherwise quartz wool fibres may contaminate the beef sample.

NOTE 2: The quartz wool should be placed tightly into the extraction thimble to prevent the petroleum ether and therefore the lipid becoming contaminated with the "dry matter" of the sample.

The fat is extracted for a minimum of 6 h with 200ml of boiling and condensed pe-troleum ether (5.1.1) in a soxhlet apparatus (5.3.11; 5.3.12) with round bottomed flask (5.3.13) or equivalent Soxtech instrument.

The round-bottomed flask with the extracted fat is stored and the extraction thimble containing the fat free dry mass (FFDM) stays over-night in the Soxhlet apparatus until the solvent has completely evaporated. Alternatively, to speed-up the drying process, remove the thimble from the Soxhlet apparatus, remove the quartz wool and filter paper, and leave the thimble containing the FFDM to stand for several hours in a fume cup-board to allow any remaining solvent to evaporate. Drying is also facilitated by breaking-up the compacted mass of FFDM before leaving it to air dry..

The solvent-free FFDM is homogenized in an analytical mill (5.3.14) [IRMS only] or homogenised using ceramic scissors and an agate pestle and mortar (5.3.6; 5.3.7). The samples are stored in air-tight container (4.15) until measurement.

6.2Measurement of 13C‰ and 15N‰

The information given in this section is for general guidance only. For example, It is accepted that the configuration of specific elemental anlaysers and gas isotope ratio mass spectrometers will vary according to manufacturer’s requirements.

Similarly the quantity of test material required to obtain reliable 13C‰ and 15N‰ results will vary according to the instrument manufacturer and the age of the instrument.

Consequently it is recognised that demonstration of and adherence to agreed Quality Assurance is essential.

6.2.1Preparation of combustion and reduction reactors

Combustion reactor / Reduction reactor / Moisture trap
Figure 1: Configuration of combustion, reduction and drying tubes for production of CO2 and N2 from bulk organic materials

6.2.2Preparation of Reference Materials and samples for 13C and 15N analysis

Weigh accurately about 1 mg of solid test or reference material ± 0.2 mg, or sufficient to give an equivalent of 100 μg of nitrogen for 15N analysis or 100 μg of carbon for 13C analysis into a pressed tin capsule (5.1.1). Fold and seal the capsule with a pair of flat nosed tweezers.

6.2.3Batch protocol

Blank capsules must be included in the batch of samples for analysis. A duplicate anlysis of the collagen Inter-Comparison Material (5.2.2) should be included every 20 analyses. This standard is included after every 10th sample pair (as shown below). Sufficient standards should be located in the batch to permit reliable drift correction over the duration of the analysis. Batches consist of multiples of this standard/samples/standard sequence. Each batch must also include at least one duplicate measurement of International Atomic Energy Association reference material L-glutamic acid (USGS 40) (5.2.1).

6.2.4Correction of measured 13C‰ and 15N‰ data

The measured 13C‰ and 15N‰ of the FFDM samples should be corrected according to the difference between the measured 13C‰ and 15N‰ values of the collagen ICM working standard and its accepted values of -17.94 ‰PDB and +6.4‰AIR respectively. In the example below the accepted 13C‰ of the collagen ICM is subtracted from the measured 13C‰ values across the analytical batch. These ‘correction factors’ are then plotted against corresponding positions (line numbers) at which the collagen ICMs were measured in the analytical batch. A correction for each of the 10 samples between consecutive collagen ICMs is then calculated by substituting the sample position (line number) into the equation of the regression line.

The mean of the two 13C‰ corrected values is then calculated and the absolute difference between the two results. 15N‰ data are corrected in the same way by making a linear interpolation between the values of the differences between the accepted value of +6.4‰ for the collagen ICM and the measurements.

6.3Quality Assurance

6.3.1Quality of ‘fat free dry mass’

In general the quality of the FFDM (protein fraction) can be checked by measuring the carbon: nitrogen ratio (C:N, which should be about 3.45 for meat protein and casein, 46 –47 % carbon and 14 % nitrogen). A good homogenisation ensures acceptable repeatability of the isotopic measurements as the FFDM still contains material of different isotopic and elemental composition (muscle tissue protein, soluble protein and collagen). This can be achieved routinely by taking the ratio of the combined area (‘area all [Vs]’) value for CO2 and N2 gas measured by the isotope ratio mass spectrometer. If these criteria are not met then the samples should be re-prepared from the original tissue.

6.3.2Absolute difference between duplicate 13C‰ and 15N‰ results

The absolute difference between duplicate measurements of FFDM and reference materials, should be less than or equal to 0.2‰ for 13C‰ and less than or equal to 0.3‰ for 15N‰. If these criteria are not met the sample should be re-analysed in duplicate.

6.3.3Measurment of L-glutamic acid reference material (USGS40)

The absolute difference between the measured values for the USGS40 reference material and the accepted 13C‰ value of -26.2‰PDB and 15N‰ value of -4.5‰AIR should be less than or equal to 0.2‰ and 0.3‰ respectively. If these criteria are not met the analytical batch should be repeated.

7.CALCULATIONS AND DATA ANALYSIS

7.1Interpretation of FFDM 13C‰ data

Using the mean and standard deviation values of FFDM 13C‰ obtained from authentic English and Irish beef, it is possible to generate normal distribution curves for the expected entire populations of each of these groups. The normal distribution is described by the equation,

Equation 1

where, y = frequency, x = 13C‰ value, σ = standard deviation,  = mean

In this case, the mean of each group (English/Irish beef) of the measured individuals () can be used as an estimate of the mean for each population) and the standard deviation of each dataset (σn-1) gives an estimate of σ for the population. The expected normal distributions of 13C‰ values for the English, Irish and Brazilian beef populations have been constructed in this way using the NORMDIST function in Excel which produces a normal distribution for a specified mean and standard deviation. These are shown in Figure 2 below. For a normal distribution, z-scores of 1.96 and 2.58 are the limits on either side of a population mean within which 95% and 99% of all observations will lie where,

Equation 2

where,x = value of an individual observation (in this case, the 13C‰ value of a beef FFDM sample),  = mean of the entire population,  = standard deviation of the entire population

The values of  and  are not known for the entire population but and σn-1 are known, respectively, as estimates from the datasets of measured beef FFDM 13C‰ values. To compensate for uncertainty in the estimates of  and  the values of 1.96 and 2.58 may be increased if the number of measurements is less than 50 (as in the case of the Irish samples, n = 26), i.e. set further out from the mean, and the symbol z is replaced by t. T-distributions are determined not only by the mean and standard deviation but also by sample size (i.e. the number of observations). Critical t-scores can be found in standard statistical tables (e.g. Fowler and Cohen[2]). The t-score for an observation ‘x’ can be calculated using Equation 4,

Equation 3

where, x = value of an individual observation (in this case, the beef FFDM 13C‰), = mean of the sample set, σn-1 = standard deviation of the sample set.

For an observation with a particular 13C‰ value (x), if the calculated value of t is larger than that tabulated for (n-1) degrees of freedom at p=0.05, it can be concluded that the observation is ‘unlikely’ to have been drawn from a population with the same mean and standard deviation as the authentic beef dataset (either the English or Irish beef dataset). If the calculated value of t is larger than that tabulated for (n-1) degrees of freedom at p=0.01 then it can be concluded that the observation is ‘highly unlikely’ to have been drawn from a population with the same mean as that from which the sample was drawn (Fowler and Cohen, 1992). The values in Table 1 are the critical t-values based on the number of authentic English and Irish beef samples in the gathered datasets – obtained by interpolating between the values given in standard statistical tables. Each t-value in the Table 1 below can be substituted into Equation 3, and the corresponding and σn-1 values input to find the values of x (value of 13C‰) within which 95% and 99% of all the observations are expected to lie. These have been plotted in Figure 2 below.

Table 1 : Critical t-values for the English and Irish FFDM 13C‰ datasets

n / Degrees of freedom (n-1) / Critical t-value at p=0.05 / Critical t-value at p=0.01
English / 78 / 77 / 1.96 / 2.58
Irish / 26 / 25 / 2.07 / 2.80

Figure 2 shows that ninety-five percent (p=0.05) of the modelled English beef population would be expected to lie between -21.05‰ and -28.82‰. Of the remaining 5% of the population, 2.5% would be expected to fall in each of the distribution tails. This means that 5 English beef samples out of 200 could be expected to have 13C‰ values greater than -21.05‰. A beef sample with a 13C‰ value of >-21.05‰ can be described as statistically “unlikely” to be drawn from a population with the same mean as the set of English beef samples analysed during the Geobeef project (Q01066). In the same way, it can be said that a beef sample with a FDM 13C‰ value of > -19.92 is statistically ‘highly unlikely’ to be drawn from a population with the same mean as the English beef dataset analysed during the Geobeef project. In this case, 0.5% of the entire modelled population would be expected to fall in each of the distribution tails (p=0.01), i.e. 1 English beef sample out of 200 would be expected to have a 13C‰ values of greater than -19.92‰.