Professional Documents
Culture Documents
To cite this article: M. J. Scotter , L. Castle , C. A. Honeybone & C. Nelson (2002) Method development and
analysis of retail foods for annatto food colouring material, Food Additives & Contaminants, 19:3, 205-222, DOI:
10.1080/02652030110085386
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”)
contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors
make no representations or warranties whatsoever as to the accuracy, completeness, or suitability
for any purpose of the Content. Any opinions and views expressed in this publication are the opinions
and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of
the Content should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings,
demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising
directly or indirectly in connection with, in relation to or arising out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any substantial
or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or
distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can
be found at http://www.tandfonline.com/page/terms-and-conditions
Food Additives and Contaminants , 2002, Vol. 19, No. 3, 205 ± 222
Table 1. Permitted uses of annatto and maximum levels rather, on the total pigment content, usually
of addition (EC 1994). expressed as the major component,, i.e. bixin or
norbixin. This is due largely to the shortcomings of
Maximum
permitted available analytical methods, which have generally
Food commodity type level (mg kg 1 ) relied on essentially qualitative techniques and/or
spectrophotometric methods that may be prone to
Margarine, minarine, other fat emulsions interference (Scotter et al. 1994). Current research on
and fats essentially free from water 10 methods of analysis for annatto have focused largely
Decorations and coatings 20
Fine bakery wares 10 on colour formulations and their thermal stability,
Edible ices 20 where developments in HPLC techniques have en-
Liqueurs, including forti®ed beverages with abled the separation and determination of the major
<15% alcohol by volume 10 and minor geometric isomers of both bixin and
Flavoured processed cheese 15 norbixin (Scotter et al. 1994, 1998, Scotter 1995).
Ripened orange, yellow and broken white
The analysis of food commodities for annatto has
cheese; un¯avoured processed cheese 15
Desserts 10 been concerned mainly with dairy produce and high-
fat foods where annatto has well-established usage,
Downloaded by [New York University] at 05:32 23 June 2015
concentration of analytes in the sample extracts and still frozen, the composite was grated again to
subsequently in the samples was determined from the produce a ®ne crumb.
calibration line regression equation. . Margarine: commodities were allowed to reach
room temperature before a 50 g subsample of
each was blended with the other subsamples.
Hand mixing was used to avoid phase separation
Sampling regimen of reduced fat spreads.
. Ice cream and edible ices: a 50 g subsample of each
A comprehensive list of annatto-containing foodstuŒs commodity was homogenized before compositing,
(sorted by product types within a speci®ed food taking necessary steps to avoid complete thawing.
group) covering the scope of those foods permitted . Yoghurt, dried dessert mixes, custard powders,
to contain annatto was obtained from the Food extruded snacks, biscuits, cake decorations, and
Additives Information System (FAIS) held by the breakfast cereals: a 50 g subsample of each com-
UK Food Standards Agency. modity was homogenized before compositing.
Dessert mix concentrates and custard powders
were analysed as purchased and not prepared
Downloaded by [New York University] at 05:32 23 June 2015
except for (1) samples that required cooking were not chloroform:acetic acid if necessary to remove any
cooked before analysis and (2) samples such as desert colour. The solvent was removed to near dryness
mix concentrates and jellies which required further using vacuum rotary evaporation (440°C) taking
preparation (according to manufacturers instruc- care not to allow the extract to dry.
tions) before consumption were analysed as concen-
trates. In all cases, the results are expressed on an `as The concentrated extract was quantitatively trans-
consumed’ basis by applying appropriate dilution ferred to a 25 ml volumetric ¯ask with methanol
factors were necessary. and diluted to the mark. After mixing, a 2-ml aliquot
was transferred to a glass syringe ®tted with a micro-
®lter (0.2 mm) and subsequently ®ltered into a vial for
analysis by HPLC.
Regime 1: cheese, cheese products and cheese-based
compound foods
The sample (3 0.01 g) was accurately weighed into Regime 2: custard powder and low-fat dessert dry
a 250 ml centrifuge bottle to which was added mixes
Downloaded by [New York University] at 05:32 23 June 2015
Regime 3: desserts, cake decorations, ®ne bakery (15 ml) was added to the centrifuge tube and the
wares (including cakes and biscuits), extruded extraction repeated until no further colour could be
snacks and breakfast cereals extracted.
To the combined extracts in the separating funnel was
The ®nely homogenized sample was accurately added 10 ml 50% acetic acid solution whereupon
weighed (10 g 0.01 g) into a 250 ml centrifuge bottle, the funnel was stoppered and the contents mixed.
except for custard powder where 3 0.01 g sample Chloroform:acetic acid solution (20 ml) was added
was taken. The samples were then extracted as and the funnel once more stoppered and shaken with
follows. venting. The chloroform layer was transferred to a
100 ml rotary evaporation ¯ask containing 1 ml BHT
About 3 g Celite 545 ®lter-aid, 1 ml BHT in methanol in methanol solution.
solution and 20 ml aqueous ammonaical ethanol
solution were added to the sample. The centrifuge Any colour remaining in the aqueous layer was
bottle was stoppered and shaken vigorously for 1 min, treated with further addition of 10 ml 50% acetic acid
removing the stopper occasionally to prevent pressure and 20 ml chloroform:acetic acid solution and the
build-up. The mixture was then centrifuged at combined chloroform extracts transferred to a rotary
Downloaded by [New York University] at 05:32 23 June 2015
2500 rpm for 10 min. evaporation ¯ask. The solvent was removed and the
remaining extract treated as in Regime 1.
For fat-containing commodities such as cake, biscuit,
extruded snack and breakfast cereals, a preliminary
fat wash was required. To the other reagents in the
centrifuge bottle was added 50 ml hexane. The bottle Regime 5: ®sh, including brine-cured or smoked
was then stoppered and shaken vigorously for 1 min, `white’ ®sh (e.g. cod and haddock) and `oily’ ®sh
removing the stopper occasionally to prevent pressure (e.g. herring and mackerel); ice cream, ice cream-
build-up. The mixture was centrifuged at 2500 rpm based confectionery, yoghurt and other dairy
for 10 min and the upper hexane layer removed by desserts
vacuum aspiration . A further 50 ml hexane was added
to the centrifuge bottle and the fat extraction re- Fish. About 5 g of sample homogenate was
peated. The extracts were then treated as in Regime accurately weighed into a beaker along with 5 g
1 following removal of the hexane layer. Celite 545 ®lter-aid. The sample was then mixed to
absorb any liquid and carefully transferred to a
mortar. The mass was intimately ground to a friable
consistency, whereupon 0.2 ml 2 m hydrochloric
Regime 4: margarine, fat-based emulsions and acid was added and the grinding continued until
spreads, butter and fat-based compound foods the mixture was homogeneous. The mixture was
transferred to a 250 ml centrifuge bottle and
200 mg ascorbyl palmitate, 50 ml hexane and
The sample (8 0.01 g) was accurately weighed into a
25 ml acetonitrile added.
50 ml centrifuge tube and 2 g Celite 545 ®lter-aid,
1 ml BHT in methanol solution and 15 ml ethanol
solution added. After mixing with a glass rod, 10 ml Ice cream, ice cream-based confectionery, yoghurt and
hexane was added, the tube stoppered and shaken other dairy desserts. The sample (5 0.01 g) was
vigorously for 1 min, removing the stopper occasion- accurately weighed into a 250 ml centrifuge bottle
ally to prevent pressure build-up. The tube was then whereupon 3 g Celite 545 ®lter-aid, 1 ml BHT
centrifuge at 2500 rpm for 10 min and the upper solution in methanol and 0.2 ml 2 m hydrochloric
hexane layer removed by vacuum aspiration. A fresh acid were added. The contents were mixed well with
10 ml portion of hexane was added and the fat a glass rod and 25 ml acetonitrile and 50 ml hexane
extraction repeated. added.
For both regimens:
The ethanolic extract was transferred with a Pasteur
pipette through a glass wool pledget contained in . the bottle was then stoppered and shaken vigor-
the neck of a glass ®lter funnel, into a 100 ml separat- ously for 1 min. The mixture was centrifuged at
ing funnel. Aqueous ammonaical ethanol solution 2500 rpm for 10 min and the upper hexane layer
Method development and analysis of retail foods for annatto food colouring material 211
removed by vacuum aspiration. A fresh 50 ml For validation purposes, a range of food commodities
portion of hexane was added to the bottle and the known not to contain annatto were spiked with either
fat extraction repeated; norbixin or bixin at appropriate levels and analysed
. the acetonitrile layer was decanted through a ®lter repeatedly to ascertain recovery. Spiking was
funnel incorporating a glass wool plug in the stem, achieved by adding the appropriate amount of bixin
into a 250 ml rotary evaporation ¯ask containing or norbixin reference standard to a homogenized
1 ml BHT in methanol solution. A fresh 25 ml subsample contained in the extraction vessel immedi-
portion of acetonitrile was added to the centrifuge ately before analysis. The annatto levels used for
bottle, whereupon it was stoppered and shaken spiking are appropriate to current usage levels as
vigorously for 1 min. The mixture was centrifuged determined from known data or from preliminary
at 2500 rpm for 10 min and the acetonitrile analyses. Validation (recovery) data for food matrices
decanted into a rotary evaporation ¯ask. A fresh spiked with annatto are given in table 2.
25 ml portion of acetonitrile was added to the
centrifuge bottle and the extraction repeated. Analysis was carried out on a group basis. A
The acetonitrile extraction stage was repeated batch analysis regimen was used in which the
until no more colour could be extracted from the analytical system was calibrated for each batch.
Downloaded by [New York University] at 05:32 23 June 2015
sample. The glass wool plug in the stem of the ®lter An appropriate in-house reference material
funnel was rinsed with a few millilitres of aceto- (IHRM) sample, a reagent blank and one sample
nitrile; and analysed in duplicate were included in each
. the volume of the pooled acetonitrile extracts batch along with a set of calibration standards.
was reduced by rotary evaporation (as above) to If the results for the analyses of the reagent blank,
ca 20 ml, initially at a vacuum setting of 200 the duplicate sample or the IHRM fell outside pre-
mbar to avoid foaming. The vacuum was then determined limits (essentially 2 SD units calculated
reduced further if required until the extract volume from repeat analysis of validation samples), the batch
had been reduced to 1±2 ml. The concentrated was repeated.
extract was then quantitatively transferred to a
Validation data for the IHRM’s, i.e. cheese, custard
25 ml volumetric ¯ask containing 1 ml BHT sol-
powder, ice cream and margarine are presented in
ution with methanol and diluted to the mark,
table 3. Each IHRM was utilized where appropriate
mixed well, micro®ltered and analysed by HPLC
as above. to the extraction method used in the analysis batches.
Table 3. In-house reference materials analysis data. A spectral bandwidth of 10 nm was applied to
extend the monitoring wavelength band to
Number of Mean annatto
IHRM replicates content (mg kg 1 ) RSD (%)
455 5 nm to ensure sensitive detection of trans-,
monocis- and dicis-isomers without compromising
Cheese 8 34.3 4.8 on selectivity. A reference wavelength of 600 4 nm
Custard powder 11 59.4 1.1 bandwidth was used to correct for background
Ice cream 7 3.2 4.5 absorbances and baseline drift. The HPLC-PDA
Margarine 9 2.6 8.9
separation of a standard mixture is given in ®gure 1
with the corresponding photodiode array spectra in
®gure 2.
A ¯ow cell of 10 mm ¯ow-path length was used in
Where there was no closely matched IHRM available
for a particular batch, a surrogate was used, e.g. ice conjunction with an optical slit width of 4 nm. This
has been shown in previous studies to give the best
cream IHRM was used for ®sh.
compromise between signal-to-noise (sensitivity) and
spectral resolution (Scotter et al. 1994). Con®rmation
Downloaded by [New York University] at 05:32 23 June 2015
Optimization of HPLC analytical selectivity and of analyte peaks was achieved through peak retention
sensitivity. The monitoring wavelength of the time matching and comparison of analyte and stan-
photodiode-array detector was optimized at 455 nm dard spectra, facilitated by the use of a spectral
to detect all bixin and norbixin isomers simul- library. Both the ¶max shift and cis-peak measurement
taneously at wavelengths close to their correspond- provided the means by which the diŒerent stereo-
ing peak maxima, thus providing good sensitivity. isomers could be identi®ed.
mAU
90 4
80
3
70
60
50
40
30 6
20
10
12
5 9
7 8
0
Figure 1. HPLC-PDA separation of annatt o colour principles. Conditions as given in text. Assignment of peaks: 1Ð
Trans-norbixin; 2ÐDicis-norbixin; 3Ð9 0 cis-norbixin; 4ÐTrans-bixin; 5 and 9ÐDicis-bixin isomers; 6Ð9 0 -cis-
bixin; 7Ð15-cis-bixin*; 8Ð13 0 -cis-bixin* (tentative).
Method development and analysis of retail foods for annatto food colouring material 213
20
0
Developments made to the Lancaster and Lawrence
procedure
300 350 400 450 500 nm
curcumin colour formulations are often used com- analysis of ice cream, ice cream-based confectionery,
mercially to colour ®sh and other commodities, a yoghurt and other dairy products.
method was required that could determine them
The method developed for ®sh analysis, was notice-
simultaneously. However, curcumin is reported to
ably more rapid than the Lancaster and Lawrence
be highly unstable under alkaline conditions
procedure and it utilized HPLC with photodiode
(Tùnessen and Karlsen 1985) hence the Lancaster
array detection. It also allowed both the 9 0 -cis-
and Lawrence (1995) procedure for annatto, which
and trans-isomers of bixin and the 9 0 -cis-isomer
requires the use of ethanolic aqueous ammonia as the
of norbixin to be determined quantitatively, and
extracting solvent, was not suitable.
other minor cis-isomers of bixin and norbixin semi-
To facilitate homogenization of the ®sh, the sample quantitatively.
was ground in a pestle and mortar in the presence of
Celite ®lter aid. A small amount of hydrochloric acid
solution was added to promote release of the colours
from the food matrix and to neutralize the eŒects of Conclusions
any bases present. Ascorbyl palmitate was added to
Downloaded by [New York University] at 05:32 23 June 2015
Data type
1
Quantitative (mg kg ) Semi-quantitative (mg kg 1 †
Table 4Ðcontinued
Data type
Table 4Ðcontinued
081 Cakes, mixed Danish, twists ND ND ND ND ND ND
and splits
082 Cakes, with custard ND ND 0.1 ND ND ND
083 Biscuits, shortbread ND ND 0.3 ND 0.3 ND
084 Cakes, cream ND ND ND ND ND ND
085 Cakes, tarts ND ND ND ND ND ND
086 Biscuits, cream 0.5 0.2 0.3 ND 0.3 ND
087 Biscuits, cream 0.2 0.2 0.1 ND ND ND
088 Cocktail snacks ND ND 1.8 ND 1.8 0.6
089 Wafers ND ND 2.3 ND 0.9 0.3
090 Brandy snap ND ND 0.1 ND ND ND
096 Fruit cake N/A
128 Gateaux, orange/lemon ND ND 0.5 ND 0.3 ND
151 Cakes, chilled/frozen ND ND 0.4 ND 0.3 ND
152 Cakes, chilled/frozen 0.2 0.2 1.3 ND 0.7 ND
153 Cakes, chilled/frozen ND ND 0.2 ND 0.3 ND
Downloaded by [New York University] at 05:32 23 June 2015
Table 4Ðconcluded
Data type
1
Quantitative (mg kg ) Semi-quantitative (mg kg 1 †
custard powders and dry dessert mixes, to allow for extruded snack, mousse dessert and Red Leicester
preparation as per the commodity package instruc- cheese are given in ®gures 3±7.
tions. The determined estimated levels found in com-
modities selected for analysis in this project are Validated extraction methods utilizing modern HPLC
summarized by commodity in table 5. These data analysis techniques coupled with spectral con®rma-
show at a glance the estimated range of results tion have been developed. These have provided accu-
obtained for each commodity and compares the rate and precise qualitative and quantitative data on
®gures with the appropriate maximum permitted the annatto content of a selected range of foods
levels of annatto. Examples of chromatograms ob- currently on sale in the UK. Isomer pro®les of the
tained from extracts of smoked ®sh, cake decoration, major bixin and norbixin colour principles have been
Method development and analysis of retail foods for annatto food colouring material 219
mAU
3
4
1 2
5 10 15 20 25
Figure 3. HPLC-PDA chromatogram of smoked ®sh extract. For peak assignments refer to ®gure 1.
220 M. J. Scotter et al.
Norm.
3
10
4
Downloaded by [New York University] at 05:32 23 June 2015
1
2
4
2 5
0
Retention time (min)
5 10 15 20 25 30
Figure 4. HPLC-PDA chromatogra m of cake decoration extract. For peak assignments refer to ®gure 1.
N orm.
6
5
4
1
2
1
3 5
2 U 8
0
Retention time (min)
5 10 15 20 25 30
Figure 5. HPLC-PD chromatogra m of extruded snack extract. For peak assignments refer to ®gure 1. U ˆ Unknown,
tentative identi®cation 13 0 -cis-norbixin.
Method development and analysis of retail foods for annatto food colouring material 221
mAU
20
3
17.5
15
12.5
10
1
7.5
Downloaded by [New York University] at 05:32 23 June 2015
2.5
4
0
5 10 15 20 25
Figure 6. HPLC-PD chromatogra m of dessert (mousse) extract. For peak assignments refer to ®gure 1.
mAU
1
40
m AU
1 .5 U
35 3 1
0 .5
0
30
40 0 50 0 nm
m AU
25 3
7 .5
5
2 .5
20
0
40 0 50 0 nm
15 m AU
1
15
10
10 5
0
U 40 0 50 0 nm
5 2
0
Retention time (min)
5 10 15 20 25
Figure 7. HPLC-PD chromatogra m of Red Leicester cheese extract with corresponding diode-array spectra. For peak
assignments refer to ®gure 1. U = Unknown, tentative identi®cation 13 0 -cis-norbixin.
222 M. J. Scotter et al.
obtained which, with the use of prepared reference Luf, W., and Brandl, E., 1988, The detection of the annatto dyestuŒ
norbixin/bixin in cheese using derivative spectroscopy and
materials, has enabled their quantitative determina- high-performance liquid chromatography. Zeitschrift fuÈr
tion at levels well below the maximum levels speci®ed Lebensmittel-Untersuchung und -Forschung, 186, 327±332.
in the Council Directive on colours for use in McNeal, J. E., 1976, Qualitative tests for added colouring matter in
foodstuŒs (EC 1994). This is particularly relevant meat products. Journa l of the Association of O cial Analytical
to the ADI, which is based on pure colouring Chemists, 59, 570.
Ministry of Agriculture, Fisheries and Food, 1987, Survey of
component. Colour Usage in Food. Food Surveillance Paper No. 19
(London: HMSO).
Ministry of Agriculture, Fisheries and Food, 1993, Dietary
Intake of Food Additives in the UK: Initial Surveillance. Food
Acknowledgement Surveillance Paper No. 37 (London: HMSO).
Najar, S. V., Bobbio, F. O., and Bobbio, P. A., 1988, EŒects of light,
air, antioxidants and pro-oxidants on annatto extracts (Bixa
orellana). Food Chemistry, 29, 283±289.
Financial support for the work was provided by the Navaz Diaz, A., and Ramos Peinado, M. C., 1992, Fluorometric
UK Ministry of Agriculture Fisheries and Food, determination of curcumin in yoghurt and mustard. Journal
Joint Food Safety and Standards Group, which in of Agricultural and Food Chemistry, 40, 56±59.
Preston, H. D., and Rickard, M. D., 1980, Extraction and chemistry
Downloaded by [New York University] at 05:32 23 June 2015
April 2000 became part of the new UK Food of annatto. Food Chemistry, 5, 47±56.
Standards Agency. Scott, K. J., 1992, Observations on some of the problems associated
with the analysis of carotenoids in foods by HPLC. Food
Chemistry, 45, 357±364.
Scotter, M., 1995, Characterisation of the coloured thermal degrada-
tion products of bixin from annatto and a revised mechanism
References for their formation. Food Chemistry, 53, 177±185.
Scotter, M. J., Thorpe, S. A., Reynolds, S. L., Wilson, L. A., and
Strutt, P. R., 1994, Characterzation of the principal colouring
Association of Official Analytical Chemists, 1980, Association components of annatto using high performance liquid chroma-
of O cial Analytical Chemists O cial Methods of Analysis, tography with photodiode array detection. Food Additives and
11th edn (Washington, DC: AOAC). Contaminants, 11, 301±315.
Brockmann, R., 1984, Lebensmittelchem Gerichtl Chem., 38, 44. Scotter, M. J., Wilson, L. A., Appleton, G. P., and Castle, L.,
Collins, P., 1992, The role of annatto in food colouring. Food 1998, Analysis of annatto (Bixa orellana) food colouring
Ingredients and Processing International, February, 23±27. formulations. 1. Determination of colouring components
Corradi, C., and Micheli, G., 1981, A rapid research and identi®ca- and coloured thermal degradation products by high-
tion method for the natural colorant E160b bixin, norbixin performance liquid chromatography with photodiode array de-
(Oriana, annatto) in food products. Industrie Alimentari, tection. Journa l of Agricultural and Food Chemistry, 46, 1031±
May, 372±375.
1038.
European Parliament and Council Directive 94/36/EC, 30 June
Scotter, M. J., Wilson, L. A., Appleton, G. P., and Castle, L.,
1994, On colours for use in foodstuŒs. O cial Journa l of the
2000, Analysis of annatto (Bixa orellana) food colouring for-
European Communities, No. L 237, 13±29.
Hammond, E. G., Chang, J., and Reinbold, G. W., 1975, mulations. 2. Determination of aromatic hydrocarbon thermal
Colorimetric method for residual annatto in dry whey. degradation products by gas chromatography. Journal of
Journal of Dairy Science, 58, 1365±1366. Agricultural and Food Chemistry, 48, 484±488.
Lancaster, F. E., and Lawrence, J. F., 1995, Determination of Smith, P. R., Blake, C. J., and Porter, D. C., 1983, Determination
annatto in high-fat dairy products, margarine and hard candy of added natural colours in foods. III. Annatto. Leatherhead
by solvent extraction followed by high-performance liquid Food R. A. Research Report No. 431.
chromatography. Food Additives and Contaminant s, 12, 9±19. Tønessen, H. J., and Karlsen, J., 1985, Studies on curcumin
Levy, L. W., and Rivadeneira, D. M., 2000, Annatto. In Natural and curcuminoids. V. Alkaline degradation of curcumin.
Food Colorants, edited by G. B. Lauro and F. J. Francis Zeitschrift fuÈr Lebensmittel-Untersuchung und -Forschung, 180,
(New York: Marcel Dekker). 132±134.