Food Chemistry 169 (2015) 305–313

Contents lists available at ScienceDirect

Food Chemistry

journal homepage: www.elsevier .com/locate / foodchem

Analytical Methods

A multiplex real-time PCR method for the quantification of beef and porkfractions in minced meat

http://dx.doi.org/10.1016/j.foodchem.2014.07.1390308-8146/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Tel.: +49 9131 6808 5158; fax: +49 9131 6808 5458.E-mail address: azuka.iwobi@lgl.bayern.de (A. Iwobi).

A. Iwobi ⇑, D. Sebah, I. Kraemer, C. Losher, G. Fischer, U. Busch, I. HuberBavarian Health and Food Safety Authority, Veterinaerstrasse 2, 85764, Germany

a r t i c l e i n f o

Article history:Received 24 December 2013Received in revised form 1 July 2014Accepted 30 July 2014Available online 8 August 2014

Keywords:Real-time PCRSpecies identification and quantificationBeefPorkValidation

a b s t r a c t

One popular staple food in many lands is minced meat, traditionally prepared from beef and/or pork frac-tions. While beef is the more expensive of the two meat fractions, the possibility exists for manufacturersto fraudulently declare higher proportions of it. Additionally, the need exists to protect consumers who,out of medical or ethical reasons, reject specific meat fractions.

In this work, we report on a quantitative triplex real-time PCR approach for the quantification of meatin minced meat products. With the method, beef and pork fractions are quantified employing primer andprobe sequences that specifically recognise cow and pig components, against the backdrop of myostatin,a universal sequence commonly found in mammals and poultry species. The limit of detection of theqPCR method was 20 genome equivalents, while the measurement of uncertainty was determined at1.83%. The method was validated on several commercially available minced meat products and per-formed well in terms of handling, reproducibility and robustness.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

An integral part of the duties of a food control agency is the rou-tine surveillance of food products, including meat-based edibles toensure that their actual composition correlate with declared com-ponents. Meat products comprise a significant proportion of theprotein intake of millions worldwide, with global consumption ofmeat rising steadily. While demand for beef was at its peak inthe early 60s’, accounting for 40% of the global meat consumption,its dominance has declined steadily, with consumption falling to23% in 2007. Pork accounts for the most commonly consumedmeat fractions today, partly because of its relative cheapness,abundance, and lower production costs (The Economist Online,2012).

Authentic declaration of meat products may be particularlyimportant to several members of the community, for example indi-viduals who as a result of religious persuasions or health reasons,reject certain types of animal fractions (Ali, Hashim, Sabar Dhahi,Mustafa, & Bin Che Man, 2012). Additionally substitution of moreexpensive meat with cheaper derivatives might violate consumertrust and confidence. The foregoing emphasizes the importanceof the implementation of reliable analytical methods by therelevant regulatory bodies for the determination of the exact

composition of meat products. Recent scandals like the horse meatscandal that spread across Europe in early 2013, show the impor-tance of analytical tools not only for detection of the meatconstellation in a particular product, but also for quantitativedetermination of the individual components. This is important indistinguishing inadvertent contamination from deliberate adulte-ration of meat products, with accompanying legal consequences.

PCR-based methods, from singleplex reactions to multiplex sys-tems (mostly real-time PCR assays) have increasingly become rel-evant in the analysis of food products including meat samples(Girish, Haunshi, Vaithiyanathan, Rajitha, & Ramakrishna, 2013;Mane, Mendiratta, & Tiwari, 2012; Köppel, Eugster, Ruf, &Rentsch, 2012 and Köppel, Daniels, Felderer, & Brünen-Nieweler,2013). Multiplex PCR reactions offer the distinct advantages oflower costs and expenditures, coupled with a time-saving feature.Such methods however, typically quantify the DNA of the animalspecies present in the meat product (López-Andreo, Aldeguer,Guillén, Gabaldón, & Puyet, 2012; Eugster, Ruf, Rentsch, &Köppel, 2009; Drummond et al., 2013). While such results are use-ful, a direct correlation between DNA content and actual meat per-centages is more desirable and this may not always be possibleconsidering the complexity of tissues utilised for meat prepara-tions, with accompanying variations in the extractable DNA. Forreliable quantification of actual meat contents, reference materialssuitable for each meat product under examination would berequired. Production of such appropriate meat standards is

Table 2The table depicts precision (relative repeatability standard deviation, RSDr), accuracyand trueness results obtained from analysis of minced meat containing definedproportions of beef and pork. Results were compiled from at least 5 different runswith an average of 21 measurement points or test results.

Actual porkproportion (%)

Measured porkproportion (%)

Precision(RSDr)%

Accuracy(%)

Trueness(%)

50 51.78 2.79 2.37 2.4430 33.35 4.88 6.71 7.1070 65.72 6.12 7.07 4.6120 21.37 5.61 7.59 4.5480 75.55 2.48 4.36 4.16

5 5.71 11.56 13.82 11.3395 93.21 0.28 1.36 1.36

306 A. Iwobi et al. / Food Chemistry 169 (2015) 305–313

however time-consuming and laborious. Additionally, due to thecomplexity in the manufacture of several meat products, withaccompanying variations in manufacturers’ recipes and productionstyle, generating reference materials appropriate for each commer-cial meat product might not be feasible.

In this work, a multiplex real-time PCR assay for the quantita-tive determination of beef and pork fractions in minced meat isdescribed. The triplex assay utilizes previously published animalspecies - specific primers and probes, relative to the proportionof the reference gene myostatin present in most mammals and birdspecies (Laube, Zagon, Spiegelberg, et al., 2007; Köppel, Ruf,Zimmerli, & Breitenmoser, 2008). The meat contents of the samplesare accordingly computed as percentage compositions. Resultsfrom comparison of the triplex method with two other previouslydescribed assays are presented and discussed.

2. Materials and methods

2.1. Production of reference minced meat samples

For validation of the presented triplex real-time qPCR method,approximately 6 kg of analytically pure beef and pork minced meatwere prepared in a professional environment at the BavarianHealth and Food Safety Authority (LGL). 300 g of minced meat frac-tions derived from varying proportions of beef and pork were pro-duced to cover a dynamic range of 5–95% beef/pork (50beef/50pork,70beef/30pork, 80beef/20pork, 45beef/55pork, 5beef/95pork) and vice versain a first series, and a second series of beef and pork mixtures tocover the trace regions of 0.1–2% of beef and pork respectively(98beef/2pork, 99beef/1pork, 99.5beef/0.5pork, and 99.9beef/0.1pork andvice versa). Homogenisation was carried out in a dedicated ther-momixer (Thermomix TM21, Vorwerk, Germany) at mode 2 forup to 5 min. Mixtures were typically stored at �20 �C untilrequired.

2.2. Minced meat and other meat products

Additional to the reference minced meat samples describedabove, the performance and robustness of the presented quantita-tive triplex real-time PCR was tested on 50 commercially availableminced meat samples randomly selected by the official food mon-itoring and surveillance authority. More than thirty meat productswith varying composition and matrices were additionally includedto assess transferability of the method to other meat matrices (seeTables 3 and 5).

2.3. DNA extraction

Four grams each of the examined meat samples was subjectedto DNA extraction procedures, employing a modified CTAB proto-col previously described (ISO 21571:2005, modified). Additionally

Table 1Primer and probe sequences used for the quantitative triplex real-time PCR assay.

Name Target gene Sequence

Bos-PDE-f Cyclic-GMP-phosphodiesterase ACTCCTABos-PDE-r TGTTTTTABos-PDE-probe (ROX) AACATCA

Sus1-F_pork Beta-actin CGAGAGGSus1-R_pork TGCAAGGSus1_TMP (HEX) TCTGACG

My-f Myostatin TTGTGCAMy-r ATACCAGMy-probe (6-FAM) CCCATGA

a commercially available silicon-column based DNA extractionkit (Surefood Animal X Kit, Congen Biotechnology, Germany) wasused to extract DNA in parallel from a subset of meat products ofother composition. The two extraction methods were comparedto determine the suitability and efficiency of the commercial kitagainst the time-intensive CTAB Extraction protocol. FollowingDNA extraction, the purity and concentration of the DNA sampleswere confirmed either by conventional photometry, employingNanodrop technology (Nanodrop 1000, Peqlab, Germany) or byPicogreen measurement. DNA samples were typically diluted1:200, resulting in a final template concentration of at least10 ng pro PCR reaction.

2.4. Primers and probes

The primers and probes described in this work have been previ-ously reported and are listed in Table 1. Beef and pork fractionswere quantified over dedicated primer and probe sequencesagainst the backdrop of a universal sequence commonly found inmammals, namely the housekeeping gene myostatin. For each ofthe three targets the specific TaqMan probe was labelled with a dif-ferent fluorescence dye (see Table 1). The primer and probe sys-tems applied in this work all target single copy, chromosomallyencoded gene sequences. The 6-FAM, HEX and ROX – labelledprobes were quenched with a Blackberry quencher (BBQ, TIB Mol-biol, Berlin, Germany) on their 30-end. Preliminary titration exper-iments were initially carried out to determine the optimal primerand probe concentrations for the multiplex reaction, without neg-ative impact on the sensitivity of the assay.

2.5. Specificity

Specificity of the applied primer and probe constellation is animportant prerequisite for any real-time PCR system. Althoughthe primers and probes applied in this work had been previouslyreported by other workers, an exhaustive specificity test was car-ried out against the backdrop of several animal and plant species

50–30 References

CCCATCATGCAGAT Laube, Zagon, Spiegelberg, et al. (2007)AATATTTCAGCTAAGAAAAA

GGATTTTTGCTGCATTTGC

CTGCCGTAAAGG Köppel et al. (2008)AACACGGCTAAGTGTGACTCCCCGACCTGG

AATCCTGAGACTCAT Laube, Zagon, Spiegelberg, et al. (2007)TGCCTGGGTTCATAAGACGGTACAAGGTATACTG

Table 3Quantification of 50 commercial minced meat products randomly selected for validation of the triplex qPCR (P/B stands for pork before beef meat declaration, depicting a greaterproportion of pork in the meat mixture, while B/P depicts beef for pork, indicating a greater proportion of beef in the mince). The calculated measurement of uncertainty (MU) andthe expanded MU are also indicated.

Quantification (%) Declaration (%) MU (%) Expanded MU (%)

Pork Beef Pork Beef

mm12_01 3.50 96.50 0 100 2.12 3.95mm12_02 36.09 63.91 B/P 4.62 6.45mm12_03 39.79 60.21 B/P 3.81 5.64mm12_04 60.75 39.25 40 60 7.64 9.47mm12_05 34.89 65.11 40 60 3.95 5.78mm12_06 67.60 32.40 45 55 2.15 3.98mm12_07 46.90 53.10 50 50 4.76 6.59mm12_08 47.48 52.52 50 50 8.22 10.05mm12_09 68.36 31.64 50 50 4.57 6.40mm12_010 67.40 32.60 50 50 3.16 4.99mm12_011 71.64 28.36 50 50 2.77 4.60mm12_012 40.89 59.11 50 50 3.30 5.13mm12_013 57.20 42.80 50 50 3.39 5.38mm12_014 57.24 42.76 55 45 2.32 4.15mm12_015 64.85 35.15 55 45 3.93 5.76mm12_016 66.47 33.53 55 45 3.26 5.09mm12_017 58.45 41.55 55 45 3.50 5.33mm12_018 62.51 37.49 55 45 4.26 6.09mm12_019 65.68 34.32 55 45 3.51 5.34mm12_020 71.32 28.68 55 45 5.22 7.05mm12_021 64.01 35.99 55 45 3.46 5.29mm12_022 46.36 53.64 55 45 3.97 5.80mm12_023 65.79 34.12 55 45 3.47 5.30mm12_024 65.18 34.82 55 45 5.40 7.23mm12_025 59.43 40.57 60 40 5.10 6.93mm12_026 54.41 45.59 60 40 4.46 6.29mm12_027 75.83 24.17 60 40 3.41 5.24mm12_028 70.15 29.85 65 35 8.51 10.34mm12_029 72.17 27.83 65 35 3.16 4.99mm12_030 72.86 27.14 65 35 5.88 7.71mm12_031 67.65 32.35 65 35 3.63 5.46mm12_032 49.61 50.39 65 35 4.01 5.84mm12_033 70.27 29.73 65 35 3.23 5.06mm12_034 67.09 32.91 P/B 5.27 7.10mm12_035 65.86 34.14 P/B 5.86 7.69mm12_036 56.19 43.81 P/B 3.09 4.92mm12_037 59.13 40.87 P/B 7.17 9.00mm12_038 53.59 46.41 P/B 7.07 8.90mm12_039 45.17 54.83 P/B 6.18 8.01mm12_040 39.79 60.21 P/B 2.38 4.21mm12_041 50.45 49.55 P/B 3.17 5.00mm12_042 62.29 37.71 P/B 3.99 5.82mm12_043 62.95 37.05 P/B 2.61 4.44mm12_044 50.95 49.05 P/B 3.34 5.17mm12_045 60.24 39.76 P/B 4.18 6.01mm12_046 59.70 40.30 P/B 3.33 5.16mm12_047 64.45 35.55 P/B 5.12 6.95mm12_048 59.59 40.41 P/B 3.47 5.30mm12_049 26.46 73.54 P/B 3.37 5.20mm12_050 100.00 0.00 100 0 1.83 3.66

A. Iwobi et al. / Food Chemistry 169 (2015) 305–313 307

(see Supplementary information) especially relevant in the foodindustry, employing 1 ng/ll of template DNA with the presentedtriplex qPCR.

2.6. Preliminary animal species screening

In order to exclude the presence of other commonly consumedmeat products, a preliminary PCR screening was carried out. Thiswas important because the reference gene myostatin is presentin all mammals, and most bird and poultry species. For this screen-ing PCR, the method of choice was the Chipron LCD Animal ArrayKit (Chipron, Berlin Germany), which simultaneously detects thepresence of up to 24 animal species in meat products (see Supple-ment). An alternative approach was the application of the commer-cial AllMeat multiplex qPCR kit (Microsynth, Switzerland, Köppelet al., 2008), which detects the presence of beef, pork, chicken

and turkey. Both the Chipron biochip kit and the AllMeat multiplexreal-time PCR method were employed as screening approaches,strictly according to manufacturers’ instructions. Following exclu-sion of the presence of other meat species apart from beef andpork, the triplex qPCR method was carried out.

2.7. Quantitative triplex real-time PCR

The optimised real-time PCR assay described in this work wascarried out and validated on an Mx3005P real-time PCR cycler(Agilent Technologies, USA). Typically, a 25 ll reaction volumewas employed containing the following components: 2� Quanti-tect Multiplex PCR NoROX reagent (Qiagen, Hilden, Germany),5 ll template DNA and primer and probe at appropriately opti-mised concentrations. For the beef and pork specific systems, pri-mer concentrations were optimised at 300 nM for beef and

Table 4Comparison of the triplex qPCR with two other real-time PCR approaches approved for use by official food monitoring agencies. The first is based on the Laube Protocol (Laube,Zagon, Spiegelberg, et al., 2007) while the AllMeat is a commercially available PCR Kit for quantification of beef, pork and poultry fractions in food. The latter was used inconnection with a DNA dilution series for quantification. Both methods were compared in terms of handling for routine analysis of meat products against the triplex qPCRdescribed in this work. B/P denotes more beef than pork in meat product, while P/B implies more pork than beef fractions in mince.

Sample Declaration (%) Laube, Zagon, Spiegelberg, et al. (2007) (%) AllMeat (%) Triplex (%)

Beef Pork Beef Pork Beef Pork Beef Pork

mm12_01 100 0 100.00 0.00 97.00 3.00 96.50 3.50mm12_02 B/P 65.70 34.30 65.30 34.70 63.90 36.10mm12_04 60 40 53.50 46.50 9.00 91.00 39.30 60.80mm12_05 60 40 73.60 26.40 55.20 44.80 65.10 34.90mm12_08 60 40 61.10 38.90 48.40 51.60 52.50 47.50mm12_011 45 55 24.30 75.70 21.90 78.10 28.40 71.60mm12_012 50 50 45.50 54.50 29.30 70.70 59.10 40.90mm12_016 45 55 23.50 76.50 21.80 78.20 33.50 66.50mm12_018 45 55 40.70 59.30 26.20 73.80 37.50 62.50mm12_019 45 55 38.90 61.10 21.90 78.10 34.30 65.70mm12_020 45 55 19.60 80.40 18.70 81.30 28.70 71.30mm12_022 45 55 54.20 45.80 44.30 55.70 53.60 46.40mm12_025 40 60 50.30 49.70 28.20 71.80 40.60 59.40mm12_026 40 60 37.50 62.50 19.70 80.30 45.60 54.40mm12_027 40 60 23.90 76.10 14.50 85.50 24.20 75.80mm12_028 35 65 23.10 76.90 23.00 77.00 29.90 70.20mm12_030 35 65 34.80 65.20 17.10 82.90 27.10 72.90mm12_031 35 65 32.30 67.7 22.10 77.90 32.40 67.70mm12_033 35 65 24.90 75.10 20.20 79.80 29.70 70.30mm12_038 P/B 18.20 81.80 20.10 79.90 46.40 53.60mm12_050 0 100 0.10 99.90 0.00 100.00 0.00 100.00

Table 5Application of the triplex qPCR method for the quantitative determination of various meat products.

Product description Declaration (%) Sample ID Beef (%) Pork (%) MU (%) Expanded MU (%)

Salami Beef H12-1 100.00 0.00 1.83 3.66Frankfurter sausages Beef H12-2 99.98 0.02 1.84 3.67Hamburger Beef H12-3 100.00 0.00 1.83 3.66Cevapcici Beef H12-4 100.00 0.00 1.83 3.66Veal sausage Beef H12-5 100.00 0.00 1.83 2.66Noodle sauce Bolognese 27% Beef H12-6 27.00 0.00 1.83 3.66Salami Pork/beef H12-7 17.02 82.98 2.93 4.76Salami with chili Pork/beef H12-8 20.63 79.37 2.94 4.77Salami with herbs Pork/beef H12-9 22.86 77.14 3.46 5.29Bavarian veal sausage 50% (Pork/beef) , bacon, rind H12-10 2.58 47.42 2.58 4.41Cappellaci 26% Pork, 27% Beef H12-11 33.03 19.94 1.97 3.80Veal liver sausage 35% Pork, 25% Pig liver, bacon, 15% calf H12-12 4.40 70.60 2.66 4.49Beef meat loaf 43% Pork, 30% pork liver, 15% veal H12-13 13.75 74.25 3.68 5.51Regensburger small sausage 45% Pork, 7% Beef, bacon H12-14 3.67 48.33 2.18 4.01Beer sausage 50% Pork, 35% beef H12-15 39.27 45.73 3.02 4.85Sausage 50% Pork, 35% beef, bacon H12-16 38.08 46.92 3.49 5.32Bifteki 51% Pork, 39% beef H12-17 29.56 60.44 2.87 4.70Salami with cheese 56,2% Pork, 32,8% beef H12-18 57.49 42.51 1.94 3.77Sausage 58% Pork, 24% beef H12-19 19.33 62.68 4.05 5.88Mortadella with pistachio 60% Pork, 5% beef, bacon H12-20 0.06 64.94 1.92 3.75Cheese kransky 69% Pork, 22% beef H12-21 27.55 63.45 5.89 7.72Lyoner sausages 70% Pork, 5% beef, bacon H12-22 0.02 74.98 1.85 3.68Mortadella with paprika 70% Pork, 2% Rind, bacon H12-23 0.04 71.96 1.88 3.71Lyoner sausage 70% Pork, bacon, 5% beef H12-24 1.89 73.13 2.67 4.50Lyoner sausage 75% Pork, bacon H12-25 0.00 100.00 1.83 3.66Ham sausages 80% Pork, 5% beef, bacon H12-26 0.00 85.00 1.83 3.66Walnut salami 86% Pork, 7% beef H12-27 1.72 91.28 2.17 4.00Pfälzer 86% Pork/4% beef H12-28 89.40 0.60 2.12 3.95Corned beef 87% Beef, pork rind, -gelatine H12-29 60.11 39.89 4.34 6.17Meat loaf 89% (Pork/beef) H12-30 3.94 85.06 2.60 4.43Ham Pork H12-31 0.00 100.00 1.83 3.66Rolled ham fillet Pork H12-32 0.00 100.00 1.83 3.66

308 A. Iwobi et al. / Food Chemistry 169 (2015) 305–313

200 nM for pork, with 200 nM for the beef specific probe and80 nM probe concentrations for the pork detection system. Forthe myostatin specific system, forward and reverse primers wereemployed at 300 nM and an optimised probe concentration of200 nM was employed. The PCR thermo profile consisted of an ini-tial denaturation and activation of the polymerase at 95 �C for15 min, followed by 45 cycles with 30 s denaturation at 95 �C,and 60 s annealing at 60 �C.

The MxPro software (Agilent Technologies, USA) was employedfor data analysis on the Mx3005P cycler. The three fluorescencechannels (FAM, HEX and ROX) were analysed separately.

2.7.1. Generation of standard curves for quantification of the DNA-content of the animal species

For the generation of standard curves, the DNA concentrationfrom pure beef and pork mince (or pure genomic DNA from both

A. Iwobi et al. / Food Chemistry 169 (2015) 305–313 309

meat species) was measured with the Nanodrop or with the Pico-green method and the genomic DNA was diluted with nuclease-free water to yield a DNA dilution series. Two separate calibrationcurves were created for beef and pork respectively with the follow-ing defined genome copy equivalents: 156250, 31250, 6250, 1250,250, and 50.

2.7.2. Quantification strategyThe method described here exploits the principle of relative

quantification: the generated copy numbers for beef or pork frac-tions are extrapolated against the calculated copy numbers forthe endogenous reference gene myostatin, to give the proportionof beef or pork in percentages as exemplified below:

x% ¼ 100%� bos-PDE-cpmy-cp

ð1Þ

y% ¼ 100%� sus-cpmy-cp

ð2Þ

where x% and y% denote the proportion of beef and pork in percent-ages, respectively, and bos-PDE-cp and sus-cp depict the generatedcopy numbers of beef and pork fractions in the samples as calcu-lated from the respective standard curves, against the generatedcopy numbers of the endogenous universal gene myostatin (my-cp).

2.8. Validation of the quantitative triplex real-time PCR

The triplex real-time qPCR assay presented in this work wascritically assessed against recommended validation guidelines pro-posed in national and international documents such as the MIQEGuidelines governing the publication of quantitative real-timePCR experiments and the ENGL Criteria governing the assessmentof precision and LOD of an analytical method (Bustin et al., 2009;ENGL, 2008). The efficiency, robustness, reproducibility as well aslimit of detection and quantification (LOD, LOQ) of the assay wereextensively tested in an in-house validation process.

2.8.1. Efficiency and precision of the triplex real-time qPCR assayPCR reaction efficiency was extrapolated from the slope of the

line of best fit drawn to the standard curve. The standard curveplots the log of starting template vs. PCR cycle number, which isgenerated by the MxPro Software (Agilent Technologies, USA).Acceptance criteria were PCR efficiencies between 90 and 110%,typically corresponding to a slope of regression between �3.1and �3.6, and Rsq value of P0.98.

2.8.2. Limit of detection (LOD6)Four grams each of beef and pork meat mixtures derived from

the following beef/pork meat constellations: 100beef/0pork and 0beef/100pork respectively were subjected to DNA isolation procedures.The extracted DNA was accordingly diluted to yield 20,000, 5000,1250, 250, 50, 20, 10, 5, 2, 1 und 0.1 genome copy equivalentsper PCR reaction of cow and pig respectively. Two runs werecarried out under repeatability conditions for reliable generationof the LOD6 (AFNOR Standard, 2008). In an extended experiment,the LOD95% was determined which is defined as the LOD at whichthe analytical assay detects the presence of the analyte at least95% of the time (thus ensuring 6 5% false negative results)(AFNOR Standard, 2008; EURL Report, 2009).

2.8.3. Limit of quantification (LOQ)In the context of this work, the LOQ is defined as the lowest

amount of the analyte in a sample that can be reliably quantifiedwithin an acceptable level of precision and accuracy (ENGLCriteria, 2008). Generally, the relative standard deviation under

repeatability conditions should be within the 25%–30% range. Toassess the LOQ of the triplex qPCR, DNA extracts from the referenceminced meat prepared for the analysis of trace samples were sub-jected to the triplex real-time PCR reaction under repeatabilityconditions.

2.8.4. Precision–relative repeatability standard deviation (RSDr)To evaluate the precision of the method, DNA from reference

minced meat containing the following meat combinations weresubjected to analysis: 50pork and 50beef, 70pork and 30beef, 30pork

and 70beef, 20pork and 80beef, 80pork and 20beef, 5pork and 95beef,95pork and 5beef. As acceptance criterion, the relative repeatabilitystandard deviation was at least 625% over the whole dynamicrange of the assay. Estimates of repeatability were obtained on suf-ficient number of test results, typically 18 or greater (ENGL Criteria,2008, ISO 5725-3).

2.8.5. Robustness and reproducibility of the quantitative triplex PCRassay

The robustness of the triplex qPCR described in this work wasdetermined by assessment of the transferability of the methodon different real-time cyclers. The procedure for determination ofthe LOD as described above, and originally carried out on theMx3005P (Agilent Technologies) was repeated on the Rotorgene-Q-Cycler (Qiagen, Germany). Additionally, three reference meatsamples (95pork5beef, 95beef5pork and 70pork30beef) were analysed inparallel on the Mx3005P (Agilent Technologies), the Rotorgene-Q-Cycler, and on the CFX384 real-time PCR Cycler from BioRad(Germany).

Through application of different DNA extraction procedures, thevariability of the DNA extraction procedures and its impact on theextraction efficiency and amplicability in the PCR reaction was alsoassessed. In this regard, the modified CTAB method was comparedwith other DNA extraction methods such as the commerciallybased silica-column DNA extraction kits: Surefood Prep Animal(Congen Biotechnology, Germany) in the extraction of DNA frommeat of other matrices analysed in this work.

2.8.6. Measurement of uncertainty (MU)The measurement of uncertainty was determined by analysis of

the in-house generated reference minced meat products with vary-ing beef/pork composition (see Table 1 above). The referenceminced meats (with meat mixtures in the 5–95% range) were sub-jected to the triplex qPCR method here described in three indepen-dent runs, with each sample analysed six times. The standarddeviations for all measurements points were determined and themeasurement of uncertainty accordingly computed. For the com-mercial minced meat products quantitatively analysed in thisstudy, the measurement of uncertainty was determined separatelyfor each sample result and added to the doubled MU determinedfor the reference mince samples to generate the expanded mea-surement of uncertainty.

2.9. Comparison of the proficiency of the real-time qPCR with otherreal-time based quantification strategies

In order to assess the efficiency of the triplex qPCR method, itsproficiency was compared with two other real-time qPCR systemsthat had been previously validated, published and partly adoptedas analytical methods by various Official Food Monitoring and Sur-veillance Authorities. The first method was the AllMeat Kit(Microsynth, Switzerland, Köppel et al., 2008), a tetraplex real-timePCR method employing a DNA dilution series for quantification.The second was the relative quantification approach (two single-plex PCRs) published by Laube, Zagon, Spiegelberg, et al. (2007).

310 A. Iwobi et al. / Food Chemistry 169 (2015) 305–313

Both methods were carried out strictly according to the prescribedmode of practice.

3. Results and discussion

3.1. Efficiency of the triplex qPCR system

In order to determine the suitability of the applied real-time PCRmethod for quantification of beef and pork fractions in minced meatproducts, the amplification efficiency (AE) and correlations coeffi-cient (R2) were compared for multiple runs, employing dilutions ofDNA extracted from selected meat products (data not included).The standard curves, which were linear over all six measurementpoints, showed an average AE P 91%, while the mean computed R2

was 0.999. Fig. 1 shows typical standard curves generated for thebeef (bos PDE), pork (sus) and myostatin (my) specific gene systems.In most multiplex systems, the challenge is demonstrating compara-ble run efficiencies among the different systems included in the mul-tiplex mix. Amplicon size and the amount of starting template bothplay important roles in the overall efficiency of the reaction. In thework of Bai, Xu, Huang, Cao, and Luo (2009), common disadvantagesof multiplex systems were cited as low amplification efficiencies andunequal PCR proficiencies on different templates, thus decreasingthe commercial application of such methods. The triplex qPCR in thiswork was developed with these parameters in mind. Short ampliconlength (80 bp–104 bp) for example ensures that the amplified prod-ucts are representative of the animal species present in the meatsample, even in highly processed samples. To demonstrate that nosignificant loss of sensitivity occurred, the efficiency of the triplexreaction was compared with several combinations of singleplexand duplex reactions (data not shown). The PCR reaction efficiencies,precision and sensitivities were comparable in all cases.

The results in this work also showed a typical linear correlationbetween CT and log DNA, with a PCR efficiency nearing the pre-dicted one doubling per cycle.

3.2. Specificity

With the plant species tested, exclusivity was 100% (false posi-tive rate 0%) for all employed primers and probe systems, with nocross reactivity observed. With the beef primer and probe system,a slight cross reactivity was observed with buffalo (ct values how-ever appeared very late) while the pork system could not accu-rately discriminate between pork and wild boar. In the context ofthis work where quantitative determination of minced meat frac-tions is the focus, this observed cross reactivity may not be signif-icant. Additionally, contamination by wild boar may be unlikelybecause wild boar is a less common and more expensive meatsource. However, when analysing meat products where traces ofvenison are expected, this might merit some consideration. Suchcross reactivity among animal species that are closely related havebeen previously reported (Iwobi, Huber, Hauner, Miller, & Busch,2011; Rentsch et al., 2013). When the analytical focus of the workis not compromised by such cross-reactivity, the applied multiplexsystems can be readily implemented without bias.

3.3. Validation of the triplex qPCR method

3.3.1. Limit of detection (LOD) and limit of quantification (LOQ)The limit of detection (LOD6) of the system reached 20 genome

copies for both pork and beef which meets the acceptance criterionspecified in the AFNOR XP V03-020-2 guidelines (2003). TheLOD95% which was carried out to validate the result of the LOD6

over 60 replicates, verified the 20 genome copies as limit of detec-tion of the assay (EURL-GMFF Report, 2009).

In this study, the absolute LOQ was determined at 1% for porkand 2% for beef. The presence of pork in greater proportionsappeared to compromise the detection and quantification of traceamounts of beef in a sample. Thus 0.5% pork against a 99.5% back-ground of beef could be reliably quantified, while the reverse,namely, 0.5% beef in a binary mixture containing 99.5% pork wasnot reproducibly quantified (data not shown). This could be dueto the inherent fatty nature of pork which could hamper the effi-ciency of DNA extraction procedures (Laube, Zagon, & Broll,2007). Other published work regarding quantification of beef andpork binary mixtures report similar analytical sensitivities, namely1% sensitivity for low processed meat (López-Andreo et al., 2012;Laube, Zagon, Spiegelberg, et al., 2007; Rodríguez, García,Gonzalez, Hernandez, & Martin, 2005).

3.3.2. Precision and accuracyIn order to assess the precision (relative repeatability standard

deviation RSDr) and accuracy (relative mean deviation in% againstthe true value of pork/beef) of the triplex method, various mincedmeat fractions with defined proportions of beef and pork, coveringthe dynamic range of the assay, were analysed under repeatabilityconditions. The results, summarised in Table 2 indicate good per-formance of the applied method, with the calculated precision,accuracy and trueness of the method lying well within the accep-tance criterion of 625% (ENGL 2008).

3.3.3. Robustness and reproducibilityFor assessment of robustness, the method was validated against

the background of different PCR Cyclers: CFX 384 Biorad, USA),Rotorgene-Q (Qiagen, Hilden) and Mx3005P (Agilent Technologies,USA). All real-time PCR platforms supported with 100% fidelity theapplied real-time triplex approach. In a complementary approach,the DNA extraction procedure was implemented with anotherDNA-isolation method, namely the Surefood kit (Qiagen, Hilden)for processing of meat of other matrices as analysed in this work.While the CTAB method was the best in terms of DNA yield andenabled the most representative sampling, the Surefood kit (Qia-gen, Hilden) however also generated acceptable results with goodDNA yield.

3.3.4. Measurement of uncertainty (MU)The measurement of uncertainty of the method was determined

at 1.83%. This was estimated by calculating the statistical mean ofthe measurement of uncertainty (MU) generated from three inde-pendent runs (2.18%, 1.80%, and 1.52%) (see Section 2.8.6).

3.4. Validation of the triplex qPCR assay with 50 commercial mincedmeat products

In order to demonstrate robustness of the qPCR assay, 50 com-mercially available minced meat products randomly selected bythe official food surveillance authority were analysed by themethod. The results, which are summarised in Table 3, indicategood applicability of the method on real mince samples. Generally,the proportion of quantified pork was greater in relation to thequantifiable beef in most of the samples. This was expected aspork, being the cheapest meat source of the two, would be prefer-entially used in greater amounts in the production of commercialmince. In at least one of the samples, (mm12_049), the percentageof pork was surprisingly almost a third lower than the quantifiedbeef fractions in the mince, although the product declaration indi-cated the presence of more pork than beef in the sample. Afterrepeated analysis of the sample, no significant deviation in resultswas observed. A possible explanation for this perceived discrep-ancy could be an inadvertent mislabelling of the product. Alterna-tively, the producers might have used ‘‘normal’’ beef tissues

Myostatin (FAM)

R²= 0.999

Myostain

Beef (ROX)

R²=0.999

Beef

Pork (HEX)

R²=0.996

Pork

Eff: 101.4%

Eff: 101.1%

Eff: 91.6%

Fig. 1. Amplification plots and accompanying standard curves for the quantification of beef and pork fractions in minced meat. Three standard curves are generated for thecalculation of pork and beef fractions extrapolated against the proportion of myostatin in the respective samples.

A. Iwobi et al. / Food Chemistry 169 (2015) 305–313 311

(homogeneous mix of muscle and fatty tissues) with more fattycomponents of pork. Because fatty tissues give a comparativelylower DNA yield, the presence of pork in the minced meat mightbe underestimated against the backdrop of the beef fractions pres-ent in the mix, although this would make little sense from a

commercial point of view. In the work of Laube, Zagon, and Broll(2007), various tissue types (kidney, heart, liver, sinews, muscle,brain, fatty tissue etc.) taken from a single pig were examinedand an assessment of their extracted DNA concentrations was car-ried out. The assessment uncovered significant variations in the

312 A. Iwobi et al. / Food Chemistry 169 (2015) 305–313

extracted DNA quantities, with kidney, heart, liver and sinewsyielding the highest DNA levels, while fatty tissues yielded consid-erably lower DNA concentrations compared with muscle. The diffi-culties in homogenizing fatty substances were cited as possiblefactors impacting the efficiency of the DNA extraction procedure,coupled with the fact that high quantities of fatty tissues mightinfluence the separation of phases during DNA isolation employingthe CTAB method.

Sometimes however, the difficulty in carefully cleaning meatprocessing equipment between batches might result in unavoid-able cross contaminations, which do not constitute economic fraud– this fact thus merits consideration in the analysis of results.

3.5. Comparison of the performance of the triplex qPCR with otherreal-time based quantification approaches

For the comparison, 21 minced meat products were examinedin parallel with the triplex qPCR method, the AllMeat Kit and themethod from Laube, Zagon, Spiegelberg, et al. (2007) (see Section2.9). Results are depicted in Table 4. The triplex qPCR comparedwell with the Laube method which utilised two singleplex reac-tions on a single reaction plate – differences in generated beefand pork percentages were on average 6.9%, with some mince sam-ples, for example mm12_031 exhibiting no significant percentiledifference between the two methods. One sample, namelymm12_038, which according to manufacturer’s declaration con-tains more pork than beef (P/B), however exhibited unusually highvariance between the two methods (more than 25% percentile var-iance). The Laube protocol relies on two separate singleplex reac-tions for quantification of beef relative to the universal sequencemyostatin present in the sample (our triplex PCR reaction is anadaptation of the Laube method which employs a multiplex ratherthan two singleplex reactions). While the method directly quanti-fies only the presence of beef in the sample, and the proportion ofpork is indirectly inferred (100% – % beef), the sensitivity of our tri-plex reaction might be higher because the proportion of beef andpork (using beef and pork specific detection systems) are concom-itantly computed against the backdrop of the universal sequencemyostatin. Following normalisation of results (beef and pork frac-tions must yield 100% total meat content), the triplex reactionmight offer a more realistic and accurate quantification strategybecause of its dual analytical approach. Generally, the amount ofpork in the products was considerably more than the product dec-laration – in most cases up to 10% over-quantification of pork frac-tions. This is however expected as most manufacturers of mincewould most likely incorporate more pork, which is a cheaper meatsource, than beef in the production process.

While multiplex real-time PCR systems can be sometimes lim-ited in efficiency and are more prone to variability of results com-pared with singleplex reactions, the results generated in this workindicate comparable results for both the Laube method (2 single-plex reactions) and the triplex qPCR method. No significant differ-ences were found between the two methods, with the computedbeef proportions in the samples and extended measurement ofuncertainties indicating comparable results for all analysed datasets.

In contrast, the AllMeat method although showing good compa-rability at a few measurement points, exhibited generally thegreatest variability in computed meat percentages. This illustratesa common problem inherent in quantification strategies that relypurely on a DNA dilution series. In the work of Eugster et al.(2009), the accurate measurement of meat proportions usingDNA-based methods is impaired when analysing samples with avariety of tissue types. Since different tissue types may exhibit var-iable DNA concentrations, such DNA-based analytical proceduresmay not always be inherently accurate.

3.6. Application of the triplex qPCR for quantification of othercommercially available meat products with different matrices

In order to assess the transferability of the triplex qPCR to othermatrices, the method was used for quantitative assessment ofmeat products with other composition and texture (see Table 5).The results show good performance of the triplex qPCR, with thecomputed measurement of uncertainty lying between 1.83 and5.9, indicating little dispersion of results. The commercial productsexamined in this part of the study exhibited great variability in tex-ture and composition – cheese kransky (sausages like frankfurterswith cheese), mortadella with paprika, beer sausages, salami andbeef meat loaf. It is therefore noteworthy that regardless of theseapparent differences in texture and complexity of the products,the triplex qPCR performed considerably well.

It is well known that the sensitivity of PCR reactions relies heav-ily on the quality of the initial DNA template. In cases where foodor meat products have been heavily processed, accompanying PCRreactions may not perform optimally due to decreased DNA qualityand concentration (Buntjer, Lamine, Haagsma, & Lenstra, 1999;Laube, Zagon, & Broll, 2007). In the work of Laube, Zagon, andBroll (2007) the application of a real-time PCR approach allowedfor quantification in low-processed foods at 10 times the efficiencyfor foods with a much higher processing index. López-Andero,Aldeguer, Guillén, Gabaldón, and Puyet (2012) also investigatedin their work the correlation between heat treatment and theextent of DNA degradation. They reported that cooking at 65 �C fol-lowed by sterilisation at 126 �C from 10 to 30 min led to DNA rup-ture to approximately 100 bp-long fragments, which however stillallowed for detection of 5% pork and its accurate quantification inbinary mixtures. The authors thus concluded that the capability ofshort qPCR detectors considerably enhanced the PCR efficiency. Inthis work, the PCR systems employed all target genome sequencesof about 100 bp or less, thus enabling the amplification ofsequences in trace amounts or in highly processed foods. Accord-ingly, the analysis of samples with some degree of processing aswith the commercial products like meat loaf, mortadella, or Pfälzeras listed in Table 5 was possible.

3.7. Use of a reference gene for meat species quantification

One of the biggest challenges in the quantitative assessment ofmeat with DNA-based analysis lies in the generation of accurateresults against the backdrop of the variability of tissues employedin the production of such meat samples. Equal amounts of beef andpork lean muscle may not contain the same number of copies oftarget DNA. In the study by Laube, Zagon, and Broll mentionedpreviously (2007), the extracted DNA quantities from various tis-sues showed high variations. A similar study in our laboratory alsorevealed striking variations in the DNA extractable from the sameamount of starting material from different tissue types taken froman animal. Generally, fatty tissues yielded the least amount of DNAwith higher yields observed for kidney, liver, connective tissue andmuscle. In order to address this issue, the use of matrix-adaptedreference material containing different proportions of meat specieshas been proposed. In the work of Eugster, Ruf, Rentsch, Hubner,and Köppel (2008), Eugster et al. (2009), such matrix-adapted stan-dards yielded more accurate results than the use of just DNA dilu-tions to build calibration curves. However, it is not economicallyfeasible or logistically possible to produce such matrix-adaptedstandards for each commercial product category. Additionally,the manufacturing process for the same product might vary widelyin terms of the meat and tissue types employed.

In this work, an alternative approach is presented, namely therelative quantification of meat components over the universalgene myostatin. This triplex qPCR approach was applied for the

A. Iwobi et al. / Food Chemistry 169 (2015) 305–313 313

quantification of minced meat fractions and other commercialproducts. The accuracy as revealed through validation on binarymince of known composition was between 1.36% and 13.82% onaverage for pork fractions in all tested samples, while the precisionwas between 0.3% and 11.6% (Table 2). These values comparefavourably with previous reports where matrix-adapted standardswere employed for quantification (Köppel et al., 2012; Rentschet al., 2013). In the work of López-Andreo et al. (2012) where asingle matrix reference material was used, the accuracy for allruns were 17% and 13% for non-treated and treated samplesrespectively.

While the triplex qPCR relies on the principle of relative quan-tification over the endogenous universal myostatin, it is crucial topre-screen the samples for the presence of other meat sourceswhich could interfere with reliable result generation. When thisbasic requirement is met, the use of a reference gene for quantifi-cation could successfully circumvent the rigours associated withmatrix-adapted standards.

4. Conclusion

The method presented in this work offers a reliable quantifica-tion strategy for minced meat, which was successfully extended toother commercial meat products with varying matrix and compo-sition. When compared with two other quantification strategies forminced meat analysis, the method was shown to be robust and theresults were comparable in accuracy for all measurement points.The calculated measurement of uncertainties when the methodwas applied to other commercial meat products was surprisinglylow and this shows the method can be readily transferred to otherproducts of different matrix and composition. Finally, the triplexmethod with its reliance on the universal sequence myostatinand dual quantification nature (beef and pork fractions are quanti-fied in parallel relative to the myostatin content) offers a uniquealternative to the application of matrix-adapted standards.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.foodchem.2014.07.139.

References

AFNOR XP V03–020-2. (2003). Produits alimentaires. Détection et quantificationdes organismes végétaux génétiquement modifiés et produits dérivés. Partie 2:Méthodes basées sur la réaction de polymérisation en chaîne. Normeexpérimentale.

AFNOR Standard XP-V-03-044. (2008). Critères de validation intra-laboratoire pourles méthodes de détection et quantification de séquences d’acides nucléiquesspécifiques; AFNOR: Saint-Denis La Plaine.

Ali, M. E., Hashim, U., Sabar Dhahi, Th., Mustafa, S., Bin Che Man, Y., & Abdul Latif,Md. (2012). Analysis of pork adulteration in commercial burgers targetingporcine-specific mitochondrial cytochrome b gene by Taqman probe real-timepolymerase chain reaction. Food Analytical Methods, 5, 784–794.

Bai, W., Xu, W., Huang, Y., Cao, S., & Luo, Y. (2009). A novel common primermultiplex PCR (CP-M-PCR) method for the simultaneous detection of meatspecies. Food Control, 20, 366–370.

Buntjer, J. B., Lamine, A., Haagsma, N., & Lenstra, J. A. (1999). Species identificationby oligonucleotide hybridisation: The influence of processing of meat products.Journal of the Science of Food and Agriculture, 7, 53–57.

Bustin, S. A., Benes, V., Garson, J. A., Hellemans, J., Huggett, J., Kubista, M., et al.(2009). The MIQE guidelines: Minimum information for publication ofquantitative real-time PCR experiments. Clinical Chemistry, 55, 611–622.

Drummond, M. G., Brasil, B. S. A. F., Dalsecco, L. S., Brasil, R. S. A. F., Teixeira, L. V., &Oliveira, D. A. A. (2013). A versatile real-time PCR method to quantify bovinecontamination in buffalo products. Food Control, 29, 131–137.

Eugster, A., Ruf, J., Rentsch, J., Hubner, P., & Köppel, R. (2008). Quantification of beefand pork fractions by real-time PCR analysis: Results of an interlaboratory trial.European Food Research Technology, 227, 17–20.

Eugster, A., Ruf, J., Rentsch, J., & Köppel, R. (2009). Quantification of beef, pork,chicken and turkey proportions in sausages: Use of matrix-adapted standardsand comparison of single versus multiplex PCR in an interlaboratory trial.European Food Research Technology, 230, 55–61.

European Network of Genetically Modified Organism Laboratories (ENGL)Document. (2008). Definition of minimum performance requirements foranalytical methods of GMO testing. http://gmo-crl.jrc.ec.europa.eu/doc/Min_Perf_Requirements_Analytical_methods.pdf.

European Union Reference Laboratory for GM Food and Feed (EURL-GMFF). (2009)Report on the verification of the performance of a construct-specific assay forthe detection of Flax CDC Triffid Event FP967 using real-time PCR. 15–10-2009.

Girish, P. S., Haunshi, S., Vaithiyanathan, S., Rajitha, R., & Ramakrishna, C. (2013). Arapid method for authentication of Buffalo (Bubalus bubalis) meat by AlkalineLysis method of DNA extraction and species specific polymerase chain reaction.Journal of Food Science Technology, 50, 141–146.

International Standard (ISO) 5725-3. (1994). Accuracy (Trueness and Precision) ofMeasurement Methods and Results. Genève, Switzerland: InternationalOrganisation for Standardisation.

International Standard (ISO) 21571:2005. Foodstuffs – Methods of analysis for thedetection of genetically modified organisms and derived products – Nucleicacid extraction.

Iwobi, A. N., Huber, I., Hauner, G., Miller, A., & Busch, U. (2011). Biochip technologyfor the detection of animal species in meat products. Food Analytical Methods, 4,389–398.

Laube, I., Zagon, J., Spiegelberg, A., Butschke, A., Kroh, L. W., & Broll, H. (2007).Development and design of a, ready-to-use’ reaction plate for a PCR-basedsimultaneous detection of animal species used in foods. International Journal ofFood Science Technology, 42, 9–17.

Laube, I., Zagon, J., & Broll, H. (2007). Quantitative determination of commerciallyrelevant species in foods by real-time PCR. International Journal of Food ScienceTechnology, 42, 336–341.

López-Andero, M., Aldeguer, M., Guillén, I., Gabaldón, J. A., & Puyet, A. (2012).Detection and quantification of meat species by qPCR in a heat-processed foodcontaining highly fragmented DNA. Food Chemistry, 134, 518–523.

Köppel, R., Ruf, J., Zimmerli, F., & Breitenmoser, A. (2008). Multiplex real-time pCRfor the detection and quantification of DNA from beef, pork, chicken and turkey.European Food Research Technology, 227, 1199–1203.

Köppel, R., Eugster, A., Ruf, J., & Rentsch, J. (2012). Quantification of meatproportions by measuring DNA contents in raw and boiled sausages usingmatrix-adapted calibrators and multiplex real-time PCR. Journal of AOACInternational, 95, 494–499.

Köppel, R., Daniels, M., Felderer, N., & Brünen-Nieweler, C. (2013). Multiplex real-time PCR for the detection and quantification of DNA from duck, goose, chicken,turkey and pork. European Food Research Technology, 236, 1093–1098.

Mane, B. G., Mendiratta, S. K., & Tiwari, A. K. (2012). Beef specific polymerase chainreaction assay for authentication of meat and meat products. Food Control, 28,246–249.

Rentsch, J., Weibel, S., Ruf, J., Eugster, A., Beck, K., & Köppel, R. (2013).Interlaboratory validation of two multiplex quantitative real-time PCRmethods to determine species of cow, sheep and goat as a measure of milkproportions in cheese. European Food Research Technology, 236, 217–227.

Rodríguez, M. A., García, T., Gonzalez, L., Hernandez, P. E., & Martin, R. (2005).Taqman real-time PCR for the detection and quantification of pork in meatmixtures. Meat Science, 70, 113–120.

The Economist Online. (2012). Kings of the Carnivores. http://www.economist.com/blogs/graphicdetail/2012/04/daily-chart-17.