TO DOWNLOAD A COPY OF THIS POSTER, VISIT WWW.WATERS.COM/POSTERS ©2012 Waters Corporation

INTRODUCTION

Economically motivated adulteration of food has emerged as a growing problem in the food industry posing potential threats

to the health of consumers. Adulteration may include

unacceptable enhancements, substitutions of ingredients, or inaccurate/misleading labeling of a product or ingredient.

In today’s food and beverage market, companies use flavors, fragrances and ingredients sourced from all over the world. To

protect a brand, food and beverage manufacturers must ensure raw materials and final products are free from

adulteration. Highly informative analytical testing methods can help

authenticate incoming raw materials and finished products. In this study, blind spiked milk samples were analyzed by high

resolution chromatography coupled with high resolution accurate mass detection. A multivariate data analysis

technique was used to reveal potential adulterations. Structural elucidation tools and database searching enabled

identification of the adulterants in the spiked milk samples.

THE USE OF MULTIVARIATE ANALYSIS FOR EARLY IDENTIFICATION OF POTENTIAL FOOD OR BEVERAGE ADULTERATIONS

Gareth Cleland, Ken Rosnack and Jennifer Burgess.

Waters Corporation, Milford, MA, USA.

METHODS

UPLC: ACQUITY UPLC®, with column manager

Column 1: ACQUITY® HSS T3 1.8 µm, 2.1 × 100 mm

Gradient: 1-99% B in 4.8 mins. A: H2O with 0.1% HCO2H; B: CH3CN with 0.1% HCO2H

Flow rate: 0.5 mL/min. Injection volume: 2 µL

Column 2: ACQUITY BEH Amide 1.7 µm, 2.1 x 150mm Gradient: 2-98% A in 4.8 mins A: 50/50 CH3CN/H2O with 10

mM NH4HCO2 and 0.125% HCO2H; B: 90/10 Acetonitrile/Water with 10 mM NH4HCO2 and 0.125% HCO2H

Flow Rate: 0.6 mL/ min Injection volume: 10 µL

MS: Waters® Xevo® G2 QTof using MSE, positive and negative

ionisation modes. Please contact the author for settings.

SAMPLE PREPARATION

Milk samples were blind spiked with melamine, cyanuric acid and a five compound mixture of human and veterinary drugs to mimic

different adulterations. Table 1 contains the formulae and masses

of melamine, cyanuric acid and the drug mix.

Table 1: Compounds used to blind spike milk samples.

Each day over a five day period a carton of milk, from the same

manufacturer, was purchased at a local store. A 5 ml aliquot was removed and stored at -60 °C. On day 6 all samples were

thawed. A 1.0 g portion of each milk sample was weighed into a scintillation vial. To mimic an adulteration an independent person

to the analysis chose to spike each sample with either 52.5 µl of water or one of the 200 µg/g standards containing melamine,

cyanuric acid or the five compound mix. This provided blind

spiked samples at 10 ppm, a level considered reasonable for economic adulteration.

To the 1.0 g of potentially spiked product:- 1 ml of formic acid in water was added (10% v/v).

Sample was vortexed for 15 seconds

1 ml of the prepared sample was added to 9 ml of 97/3 ACN/

H2O with 10 mM CH3COONH4. Sample was mixed well before filtering through a 0.2 µm PTFE

syringe filter into a 1.5 ml Waters Maximum Recovery vial.

Following generic sample preparation, data was collected in ESI+ and ESI- modes using both HILIC and reversed-phase

chromatography. The BEH Amide column has the ability to retain very polar compounds whilst the HSS T3 can retain a wide variety

of less polar compounds. Seven replicates of each milk sample provided good statistical data. Utilizing the column manager, all

samples were run in one sample list. For each polarity and mode, samples were randomized for collection and re-ordered for

processing. Results for ESI– on a HSS T3 column have been

omitted from this poster due to space constraints.

RESULTS

Information from each analysis was investigated using a PLS-

DA (Project to Latent Structures Discriminant Analysis model). Each instrument polarity and UPLC column pairing was

processed separately. MarkerLynx™ XS processing of the data in this analysis yielded

the combination of plots shown in Figure 2. The scores plot allows us to see a map of all the injections

(observations) and ascertain which observations are similar. The Exact Mass Retention Time Pairs (EMRTs) present in each

injection are responsible for this similarity or variation, and are

displayed in the Loadings plot. A variable averages plot simply plots the average intensity of

an EMRT for each group.

Component Empirical Formula

Exact Mass ([M+H]+)

Exact Mass ( [M-H]-)

Acetaminophen C8H9NO2 152.0712 150.0555

Caffeine C8H10N4O2 195.0882 193.0726

Sulfadimethoxine C12H14N4O4S 311.0814 309.0658

Terfenadine C32H41NO2 472.3216 470.3059

Reserpine C33H40N2O9 609.2812 607.2656

Melamine C3H6N6 127.0732 125.0576

Cyanuric acid C3H3N3O3 130.0253 128.0096

Figure 1: Workflow used to isolate and identify an unknown adulteration.

A: SCORES PLOT

B: LOADINGS PLOT

D:MARKERLYNX XS BROWSER RESULTS WINDOW

Figure 2: Scores (A), Loadings (B) and Variable Average (C) plots reveal potential adulterations as EMRTs. MarkerLynx XS Browser (D) shows results of proposed identifica-tions based on Elemental Composition and a database search.

CONCLUSIONS

Adulterations in three milk samples, blind spiked over five days were correctly detected and identified using the proposed workflow.

The experimental combination of ACQUITY UPLC,

ACQUITY Column Manager and Xevo G2 QTof allowed the analyst to collect a comprehensive data set in one sample list, freeing up time to work on other projects.

Information rich MSE data collection provides the

precursor ion information to isolate unknown compounds of interest and the fragment ion information to elucidate the unknown compounds of interest in a single injection.

Multivariate analysis provided swift detection of

potential adulterations. Elemental composition and MassFragment™

structural elucidation tools enabled identification of the potential adulterants.

The workflow demonstrated here can provide

reliable and highly detailed information to for the detection and identification of food adulterations.

DISCUSSION

Using the HSS T3 analysis in Figure 2 as an example, we can

see from the scores plot that something is different about Milk 1, Day 4 (M1D4) compared with the milk on the other days

(M1D1, M1D2, M1D3, M1D5). In the loadings plot, EMRTs found in all injections at the same

level reside close to the origin. The scores plot and loadings plot may be superimposed. The EMRTs responsible for causing

M1D4 differences in the scores plot are found in the same quadrant, furthest away from the swarm. Highlighting these

points gives us the best chance of identifying an unknown

adulterant. The variable averages plot corresponding to the highlighted

points in the loadings plot reveals that five compounds present in the milk on Day 4 are absent on all the other days, a very

good indication of an adulteration. Transferring the EMRTs to the MarkerLynx XS browser allows Elemental Composition and

a database search to be performed on the low energy function of the MSE data. Results from the MarkerLynx XS database

search are a perfect match for the less polar compounds of Table 1.

To complete the workflow in Figure 2, confirmation would usually be obtained by running purchased standards in MSMS

mode and comparing retention times and MSMS data to the high energy data in the MSE acquisition.

From the analysis of all the data acquired it was concluded by

the analyst that milk on day 3 had been spiked with melamine, milk on day 4 had been spiked with the five compound mix

and milk on day 5 had been spiked with cyanuric acid. The results were later confirmed with the person who had spiked

the samples.

C: VARIABLE AVERAGES PLOT