Journal of Chromatography A, 1101 (2006) 6368

Determination of Vitamin B12 in food products and in premixes by reversed-phase high performance liquid chromatography and immunoafnity extraction
O. Heudi a, , T. Kilinc a , P. Fontannaz a , E. Marley b
b

Quality and Safety Department, Nestl e Research Center, Nestec Ltd., Vers-Chez-les-Blanc, 1000 Lausanne 26, Switzerland R-Biopharm Rhone Ltd., Unit 3.06 Kelvin Campus, West of Scotland Science Park, Maryhill Road, Glasgow G20 0SP, Scotland, UK Received 18 August 2005; received in revised form 22 September 2005; accepted 26 September 2005 Available online 10 October 2005

Abstract A new, faster and simple method to quantify Vitamin B12 , both in foods and in premixes, by reversed-phase liquid chromatography with UV detection has been developed. Vitamin B12 was extracted from food products with 50 mM sodium acetate buffer pH 4.0 (at 100 C for 35 min) in the presence of sodium cyanide, followed by a purication step on an immunoafnity column prior to the LC analysis. An enzymatic hydrolysis (pepsin at 37 C and pH 4 for 3 h) prior to the purication step efciently released the bound Vitamin B12 , and thus, allowed obtaining total Vitamin B12 content in food products. Vitamin B12 was monitored by UV at 361 nm after its separation on a reversed-phase narrow-bore column with a gradient of mobile phase made of water/acetonitrile and triuoroacetic acid (TFA) 0.025%. The specicity of the method was demonstrated by the retention characteristics, UV spectra and by comparing the peak purity with the Vitamin B12 standard. The calibration graphs plotted with six concentrations of Vitamin B12 was linear with a regression coefcient R2 > 0.9997. The repeatability of the method was evaluated at different levels of concentration on six fortied products and the relative standard deviation (RSDr ) was below 3.2%. The value of the relative standard deviation of the intermediate precision was below 5.6% (n = 4). The method was successfully applied to several food products and consistent results were obtained in comparison with microbiological assay (MBA). Our data demonstrate that the immunoafnity columns are highly efcient for the purication of Vitamin B12 and that our HPLC could be used as an alternative method to the microbiological assay for the determination of Vitamin B12 in food products. 2005 Elsevier B.V. All rights reserved.
Keywords: Vitamin B12 ; Food analysis; Liquid chromatography; Narrow-bore column; Immunoafnity column

1. Introduction Vitamin B12 is a family of compounds called cobalamins amongst which cyanocobalamin, hydroxycobalamin, adenosylcobalamin (with an adenosine group) and methylcobalamin are the major forms. Each form has its own potential biological activity in terms of absorption and potency. Vitamin B12 is an essential nutrient linked to human growth and cell development. Also, Vitamin B12 is an important component of several enzymes and is involved in the metabolism of certain amino acids [1]. The

Corresponding author. Present address: Novartis Institute for Tropical Diseases Pte Ltd., 10 Biopolis Road, #05-01 Chromos, Singapore 138670, Singapore. Tel.: +65 67330233. E-mail address: heudio@hotmail.com (O. Heudi).

deciency of Vitamin B12 in humans is manifested by anemia and neuropathy. Naturally occurring forms of Vitamin B12 are found predominantly in meat and dairy products, but only one form (cyanocobalamin) is used for the fortication of food products. Typical levels in foods range from the low ng g1 for cheese and sh to hundreds of ng g1 for liver and fortied cereals [2]. Vitamin B12 is routinely determined in food products by microbiological assay (MBA), which uses Lactobacillus leishmanii as the test organism [3]. Although the assay is highly sensitive, it lacks specicity since in some food matrices inactive cobalamins could interfere with the growth of the microorganism [1]. Another drawback is that the method is time consuming and exhibits low precision. Radioisotopic dilution has been applied to the determination of Vitamin B12 in food samples [4,5] but the method lacks selectivity as the intrinsic factor used for the assay could also bind other cobalamins or analogues. A method

0021-9673/$ see front matter 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2005.09.059

64

O. Heudi et al. / J. Chromatogr. A 1101 (2006) 6368

based on protein-binding assay associated with surface plasma resonance spectroscopy detection was recently developed for the determination of Vitamin B12 in milk products [6]. The sensitivity of this method (2 ng g1 ) allows the detection of trace amounts of Vitamin B12 in food products; however, specicity remains a problem as the binding protein used in the assay could also react with cobalamin analogues. Several publications report the determination of Vitamin B12 by reversed-phase HPLC using either isocratic or gradient elution with UV visible detection mode [713]. However, UV detection is not sensitive enough to detect Vitamin B12 in non-fortied food products. Solid phase extraction (SPE) technique as clean-up and concentrating procedures have also investigated, prior to the analysis of Vitamin B12 by LCUV in biological uids [14] and in selected foods [15], but this approach was not applicable to a wide range of food samples, due to insufcient selectivity. Other specic detection methods based on the dosage of cobalt either by atomic absorption spectrometry [16] or by inductively coupled argon plasma emission spectrometric detection [17] were proposed. However, the limit of detection (LOD) of these methods (150 mg g1 ) only allows the quantication of high amounts of Vitamin B12 in food samples. Recent methods for quantifying cobalamins include capillary electrophoresis [18], and micro-HPLC [19], combined with a detection by ICP-MS. Although this hybrid method allows the quick detection and measurement of specic cobalamins in food product, the low levels of Vitamin B12 (1500600 ng g1 ) measured by these methods are still above the contents of Vitamin B12 to be found in food products, which are typically between 3 and 20 ng g1 . Although the uorimetric detection method is better adapted for the determination of low contents of Vitamin B12 (310 ng g1 ) in food products, it is necessary to release, by a chemical reaction and enzymatic hydrolysis a characteristic uorescent compound of Vitamin B12 , as this vitamin is not naturally uorescent [20]. The purpose of the present work was to investigate LC coupled with UV detection for the determination of Vitamin B12 in food products. Because Vitamin B12 exists in different forms and at very low concentrations in food products, the sensitivity of the method and the sample preparation are crucial steps in the method development. In this study, a better selectivity was obtained by introducing a purication step using immunoafnity columns, and sensitivity of the method was enhanced by using a narrow bore LC column combined with UV detection at 361 nm. 2. Experimental 2.1. Reagents Vitamin B12 (cyanocobalamin), sodium acetate, triuoroacetic acid (TFA) and sodium cyanide were obtained from Merck (Geneva, Switzerland). Polyvitaminated premixes were obtained from DSM Nutritional Products (Saint-Louis, France). Water was puried using a Milli-Q system from Millipore (Le Mont-sur-Lausanne, Switzerland). The following enzymes were used: pepsin (EC 3.4.23.1, Sigma, catalogue No. P 7125) and -amylase (EC 3.2.1.1, Sigma, catalogue No. A 2771). All other reagents used were analytical grade.

2.2. Samples Several samples including milk-based infant formula powder used as Nestl e reference materials (NRM), milk-based infant formula powder, infant cereals with milk and fruits, infant cereals without milk, breakfast cereals with wheat, infant cereals without milk, polyvitaminated mixes, petfood, hypoallergenic milkbased infant formula and health care products with or without chocolate, were used for the present study. Two NIST reference samples (milk powder, SRM/RM 1846 and meat homogenate SRM 1546) from National Institute of Standard and Technology (NIST, Gaithersburg, MD, USA) and one milk powder CRM 421 (Community Bureau of Reference, BCR Belgium) were also included. 2.3. Free Vitamin B12 extraction The sample (525 g, depending on the content of Vitamin B12 in the food product) was accurately weighed into a 250 ml Erlenmeyer ask. Sixty millilitres of 50 mM sodium acetate buffer pH 4, 1 ml of a stock solution of sodium cyanide 1% and 0.25 g of -amylase were added under agitation and the solution was incubated at 42 C for 30 min. The pH of the solution was adjusted to 4.8 and then heated at 100 C for 35 min under nitrogen steam reux and agitation. After cooling at room temperature, the solution was quantitatively transferred into a volumetric 100 ml ask and lled to the mark with distilled water. The resulting solution was shaken well and ltered through a folded lter paper S&S 597. For the premix samples, 1 g was accurately weighed into a 250 ml Erlenmeyer ask. A 60 ml of 50 mM sodium acetate buffer pH 4, 1 ml of a stock solution of sodium cyanide 1% was added under agitation and the solution was incubated at 42 C for 30 min. After cooling at room temperature, the solution was handled as described earlier. 2.4. Total Vitamin B12 extraction The sample (525 g, depending on the content of Vitamin B12 in the food product) was accurately weighed into a 250 ml Erlenmeyer ask. A 60 ml of 50 mM sodium acetate buffer pH 4, 1 ml of a stock solution of sodium cyanide 1%, 0.25 g of -amylase and 1 g of pepsin were added under agitation and the solution was incubated at 37 C for 3 h. After cooling the solution was handled as described earlier (Section 2.3). 2.5. Purication of the extract on immunoafnity column Five millilitres of the extract obtained in Section 2.3 or Section 2.4 was loaded onto an immunoafnity column (6 mm 8 mm i.d.; prototype, R-Biopharm, Saint Didier au Mont dOr France) previously washed with 10 ml of distilled water. The column was then washed with 5 ml of water and dried by passing through 10 ml of air with a syringe. The Vitamin B12 was then eluted with 2 ml of methanol. The eluate was concentrated to dryness under nitrogen steam and then reconstituted in 300 l solvent A before the LCUV analysis.

O. Heudi et al. / J. Chromatogr. A 1101 (2006) 6368

65

2.6. LCUV The analysis of sample was performed by LC using an Agilent 1100 HPLC system (Agilent Technology, Urdorf, Switzerland) equipped with a DAD UV detector. Sample injections of 100 l were made from an Agilent 1100 Series auto-sampler. The chromatographic separations were performed on C18 , Ace 3 AQ, 150 mm 3.0 mm column (Hichrom, UK), with a ow rate of 250 l min1 . The mobile phase was a series of gradient steps consisting of water and 0.025% of TFA pH 2.6 (solvent A) and acetonitrile (solvent B) that follows: range 03.5 min 100:0, range 3.511 min 75:25, range 1119 min 65:35, range 1920 min 90:10 and range 2026 min 100:0 (v/v). The column was further equilibrated for 4 min before the next injection of samples. The column efuent was monitored by UV at 361 nm. 2.7. MBA for Vitamin B12 determination in food products The MBA used for the analysis of Vitamin B12 was based on the assay described by the British Analytical Methods Committee [21]. This method uses L. leishmanii as test organism and the extraction of Vitamin B12 is performed in a boiling water bath in the presence of cyanide. 2.8. Statistics The precision and the trueness data were evaluated with an in-house statistical program making use of the robust-statistics concept of Rousseew and Croux [22]. 3. Results and discussion The objective of the present work was to propose a sensitive and selective HPLC method with UV detection for the quantication of Vitamin B12 in food products. The complexity of Vitamin B12 analysis in food products is due to several parameters: (i) the concentration of Vitamin B12 in food products is very low (in the range of 350 ng g1 of products) in comparison to other vitamins. (ii) Vitamin B12 is fortied in food products as cyanocobalamin but other natural forms such as hydroxycobalamin also occur in food products. (iii) Vitamin B12 exists in free or bound forms in food products. In the present work, the purication of Vitamin B12 was achieved on immunoafnity columns in order to improve the selectivity and the quantication limit of our method, and sensitivity was enhanced with the use of LC narrow-bore columns at a low ow rate. 3.1. Sample preparation In the MBA method used for the determination of Vitamin B12 in food products, the cobalamins are converted into the cyanocobalamin or Vitamin B12 [3,21], which is the form that is declared on the label. This conversion is achieved by boiling the samples for at least 30 min at 98 C in the presence of cyanide [23]. This treatment was applied in our study, as we were interested in the total Vitamin B12 content of the food product for

the purposes of labeling and also for comparing our LC method with MBA. 3.2. Enrichment of Vitamin B12 on Immunoafnity column Based on the results of a recent publication that shows the inefcacy of SPE C18 for Vitamin B12 purication in food sample [20], and taking into account the poor selectivity that we obtained on Oasis HLB (Waters, Milford, MA, USA) or conventional SPE cartridges such as C18 (data not shown), it was decided to use immunoafnity columns. With this approach, under our experimental conditions, which were slightly different from the one recently described [20], we obtained a recovery of Vitamin B12 of about 100%. It should be noted that under the column supplier recommended protocol, i.e. phosphate buffer as washing solution and pH 7 for the loading of samples onto the column, similar recovery was obtained. The amylase treatment was introduced to help the sample ltration particularly for foods containing starch. Centrifugation was not adopted in the sample preparation, as with high-fat-containing samples, it was not possible to obtain clear samples. This problem has been extensively studied by Iwase and Ono [15] who suggested use of a special membrane lter for the cleaning of oily particles. 3.3. Method validation for Vitamin B12 quantication in food products Endogenous Vitamin B12 in different food samples was quantied by means of an external calibration curve. Under the optimum analysis conditions, linearity was studied over the concentration range of 10200 ng ml1 for Vitamin B12 (Vitamin B12 stock solution was prepared in water and the nal dilution was performed in solvent A). Six concentration points were used to generate the calibration curve that was linear (r2 = 0.9997). The limit of detection of the method was calculated as 3 SD, where SD is the standard deviation of background noise of the standard diluted in the same buffer as the food sample. The LOD was estimated to be 3 ng ml1 (Table 1). It should be pointed out that the better signal-to-noise ratio obtained with a narrow bore column at optimum ow rate of 250 l min1 enhanced the sensitivity of our method by a factor of 300 in comparison to the results of Wongyai [13]. The limit of quantication (LOQ) was calculated as 10 SD and the value found was 10 ng ml1 (Table 1), which enables the quantication of 2.5 ng g1 of Vitamin B12 in food products. Quantication of
Table 1 Calibration curve data and quantication parameters for the analysis of Vitamin B12 in food products by LCUV Slope Intercept Regression, r2 Range (ng ml1 ) Limit of detection (ng ml1 ) Limit of quantication (ng ml1 ) Injection volume (l) 0.1949 0.7346 0.9997 10200 3 (S/N = 3) 10 (S/N = 10) 100

66

O. Heudi et al. / J. Chromatogr. A 1101 (2006) 6368

Table 2 Recovery percentages for three spiked infant formula samples and precision data Products Initial concentration (ng g1 ) 17.3 15.6 18.8 Added concentration (ng g1 ) 20 36 32 Concentration found (ng g1 ) 36 51.7 49 Recovery (%) RSDr (%)a RSDI (%)b

Infant formula (NRM 2/2003) Infant formula powder Infant formula powder HA
a b

94 100 94

3.2

5.6

RSDr represents the repeatability data which were obtained by analyzing in duplicate six different food products. Relative standard deviation represents the intermediate precision which was obtained by analyzing Vitamin B12 in NRM 2/2003 during four consecutive days.

Vitamin B12 in fortied food products was performed by the external standard method with a single calibration point due to the linearity of method and because the intercept of the calibration curve includes the zero. The results found with this approach were not signicantly different from those obtained with the calibration graph (data not shown). The calibration point used for the quantication was set at 100 ng ml1 . The recovery of the method, was determined by spiking three different food products with known amounts of Vitamin B12 and the samples were submitted to the purication procedure. The recovery results ranged between 94 and 100%. (Table 2). It was also found that the pH of the hydrolysate applied onto the column has no effect in the range 4.87.0 recovery percentage of Vitamin B12 from food products. The repeatability of the method was evaluated at different concentration levels by analysing six different food products in duplicate. The relative standard deviation of repeatability (RSDr ) was below 3.2% (Table 2). The intermediate precision was determined by analysing one NRM during four consecutive days; the relative standard deviation of the intermediate precision value was estimated at 5.6% (Table 2). The use of reference materials is invaluable for assessing the accuracy of the method. Thus the trueness study was further established by analyzing several certied reference samples as well as NRM. The values found in Table 3 show that there is no signicant difference between the results obtained by our HPLC method and the reference values. Such data also conrms the efcacy of the extraction procedure in recovering Vitamin B12 from the various food matrices.

3.4. Quantication of free and total Vitamin B12 in food products Since Vitamin B12 exists in free and bound form in food products, it is necessary to be able to quantify the total amount of this vitamin. Most of the products tested in this study are supplemented with Vitamin B12 . However, when comparing the results that we obtained with those found for Vitamin B12 by MBA some differences were observed. The HPLC results were consistently lower than those of the MBA (Table 4) except for the premix samples. Therefore, we made the assumption that the raw materials used for manufacturing these products contain bound forms of Vitamin B12 . To release this form we applied a protocol described by Pakin et al. [20], which involved pepsin digestion of the sample. With this protocol it was possible to unambiguously identify the Vitamin B12 peak in the food products as depicted in the chromatograms of some selected foods (Fig. 1). The results obtained with this approach show the efcacy of pepsin in releasing the bound forms of Vitamin B12 for most of the tested products (Table 4). In addition, a t-test performed on the different data obtained after the determination of Vitamin B12 in different food products by the developed HPLC method and MBA demonstrated that the two sets values were comparable and that the bias was not signicantly different from zero at 95% condence. Therefore, we decided to apply the pepsin treatment to all samples, as it is difcult to know in advance whether or not the raw materials contain bound forms of Vitamin B12 . The problem of precision with the petfood products is due to the inhomogeneity of the sample as the MBA results varied between 50 and 100% according to the sample size (see Table 4). Consequently, to reduce the effect of heterogeneity, a larger test portion of 50 g was used for these particular products. For some samples such as the infant formula powder hypoallergenic product (where proteins have been enzymatically hydrolysed to peptides), the values obtained with the pepsin treatment were still lower compared to that from the MBA. The recovery data found in this product for Vitamin B12 were around 100% (Table 2). In addition, the amount of Vitamin B12 did not change when different concentrations of pepsin (0.25, 1 and 5 g) were tested (data not shown). Consequently, the discrepancy observed between our HPLC method and the MBA is not due to the bound forms of Vitamin B12 . This discrepancy is not only observed with our HPLC method but also with other assays such as chemiluminescence [24]. Since L. leishmanii can utilise deoxyribosides and deoxynucleotides as well as Vitamin B12 [25], the higher values by the MBA may be due to occurrence

Table 3 Determination of Vitamin B12 in certied reference materials Products Milk-based infant formula NRM 2/2003 Milk-based infant formula NRM 2/2004 Milk-based infant formula NRM 3/2004 Milk-based infant formula NRM 1/2005 Milk-based infant formula NRM 2/2005 Milk powder NIST 1846 Meat homogenate NIST 1546 Milk powder CRM 421
a

Founda (ng g1 ) 17 1 12.6 0.6 13.2 1 12.3 0.9 7.7 34.6 1.8 71 30.1

Reference (ng g1 ) 21 3 15 2 12 2 12.5 1.9 7.1 1.1 39 3 61 34 5

Mean standard deviation, n = 3 NRM 2/2003, NRM 2/2004 and NRM 2/2004 are infant formula powders; NRM 1/2005 is a clinical nutritional product and NRM 2/2005 is infant formula containing starch.

O. Heudi et al. / J. Chromatogr. A 1101 (2006) 6368 Table 4 Liquid chromatographic determination of Vitamin B12 concentration in various food products Products HPLC (ng g1 ) Free Vitamin B12 Milk-based infant formula powder Milk-based infant formula powder Milk-based infant formula powder Infant cereals with milk and fruits Infant cereals without milk 1 Infant cereals without milk 2 Breakfast cereals with wheat Polyvitamin mix 1 Polyvitamin mix 2 Petfood (1) Petfood (2) Hypoallergenic milk-based infant formulab Healthcare product (liquid) Healthcare product with chocolate 13 (0.6) Total Vitamin B12 17.3 (1) 13 13 12 9.5 12 27.2

67

MBA (ng g1 )

810 (28) 390 (31) 264.2 (6.6) 34.7 (2.1) 14

301 41.3 (2) 19.7 (1.9) 48.1 45

19.3 (1.3) 18 18 14 9.7 19.7 35.9 850 333a 269.1 (76.3) 73.9 27.4 51.5 46

Average of three determinations (standard deviation in parentheses). a Indicative value. b This compound was not included in the statistical analysis for the trueness evaluation.

Fig. 1. Chromatograms obtained from determination of Vitamin in premix sample (850 ng g1 ) (the values in brackets represent the certied or the values obtained by MBA) (a); NIST milk powder, SRM/RM 1846 sample (39 ng g1 ) (b); meat homogenate SRM 1546 sample (6 ng g1 ) (c) and powdered milk (12 ng g1 ) (d), UV detection 361 nm. Peak (1) Vitamin B12 . The sample size and the nal dilution used for the Vitamin B12 determination varied according to the Vitamin B12 content in products.

68

O. Heudi et al. / J. Chromatogr. A 1101 (2006) 6368 [4] P. Casey, K. Speckman, F. Ebert, W. Hobbs, J. Assoc. Off. Anal. Chem. 65 (1982) 85. [5] K. Muhammad, D. Briggs, G. Jones, Food Chem. 48 (1993) 431. [6] H. Indyk, B. Persson, M. Caselunghe, A. Moberg, E. Filonzi, D. Woollard, J. AOAC. Int. 85 (2002) 72. [7] S. Albala-Hurtado, M.T. Veciana-Nogues, M. Izquierdo-Pulido, A. Marine-Font, J. Chromatogr. A 778 (1997) 247. [8] B. Klejdus, J. Petrlova, D. Potesil, V. Adam, R. Mikelova, J. Vacek, R. Kizek, V. Kuban, Anal. Chim. Acta 520 (2004) 57. [9] L.A. Kozhanova, G.A. Fedorova, G.I. Baram, J. Anal. Chem. 57 (2002) 40. [10] H.B. Li, F. Chen, Chromatographia 54 (2001) 270. [11] P. Moreno, V. Salvado, J. Chromatogr. A 870 (2000) 207. [12] P. Vinas, C. Lopez-Erroz, N. Balsalobre, M. Hernandez-Cordoba, J. Chromatogr. A 1007 (2003) 77. [13] S. Wongyai, J. Chromatogr. A 870 (2000) 217. [14] P.F. Chatzimichalakis, V.F. Samanidou, R. Verpoorte, I.N. Papadoyannis, J. Sep. Sci. 27 (2004) 1181. [15] H. Iwase, I. Ono, J. Chromatogr. A 771 (1997) 127. [16] P. Vinas, N. Campillo, I. Lopez-Garcia, M. Hernandez-Cordoba, Anal. Chim. Acta 318 (1996) 319. [17] M. Morita, T. Uehiro, K. Fuwa, Anal. Chem. 52 (1980) 351. [18] S. Baker, N. Miller-Ihli, Spectrochim, Acta Part B 55 (2000) 1823. [19] H. Chassaigne, R. Lobinski, Anal. Chim. Acta 359 (1998) 227. [20] C. Pakin, M. Bergaentzle, D. Aoude-Werner, C. Hasselmann, J. Chromatogr. A 1081 (2005) 182. [21] Analytical Methods Committee, Analyst 81 (1956) 132. [22] P. Rousseew, C. Croux, J. Am. Stat. Assoc. (1993) 1273. [23] E.P. Frenkel, R. Prough, R.L. Kitchens, Methods Enzymol. 67 (1980) 31. [24] F. Watanabe, H. Katsura, S. Takenaka, T. Enomoto, E. Miyamoto, T. Nakatsuka, Y. Nakano, Int. J. Food Sci. Nutr. 52 (2001) 263. [25] Z. Schneider, A. Stroinski, in: Z. Schneider, A. Stroinski (Eds.), Comprehensive B12 , Walter de Gruyter, Berlin, 1987, p. 157.

of Vitamin B12 analogue in the hypoallergenic product tested in this study. 4. Conclusions The column prototype used in the present study, which will very soon be commercially available, appears to be an appropriate choice for the selective extraction and purication of Vitamin B12 from food products. The simplicity of the developed HPLC method together with better selectivity, repeatability and intermediate precision over the routine MBA method should make the HPLC approach highly desirable for the quantication of Vitamin B12 in food products. Acknowledgements The authors are grateful to R-Biopharm, France, products for providing them with the immunoafnity columns. The authors would also like to thank Dr. C.J. Blake and Dr. M. Bonny for their helpful discussion and their critical review of the manuscript. References
[1] G.F.M. Ball, Bioavailability and Analysis of Vitamins in Foods, Chapmann & Hall, London, 1998, p. 497. [2] S.W. Souci, N. Fachmann, H. Kraut, Food Composition and Nutrition Tables, sixth ed., CRC Press, Boca Raton, 1994. [3] Anonymous, in: K., Cunniff (Ed.), AOAC Ofcial Methods of Analysis, Association of Ofcial Analytical Chemist Inc., Gaithersburg, 2000, p. 47.