Choline is an essential nutrient for which the Food and. Determination of Choline in Infant Formula by Ion Chromatography

Transcription

1 1156 LAIKHTMAN & ROHRER: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 FOOD COMPOSITION AND ADDITIVES Determination of Choline in Infant Formula by Ion Chromatography LAIKHTMAN & ROHRER: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 MARK LAIKHTMAN and JEFFREY S. ROHRER 1 Dionex Corporation, 500 Mercury Dr, Sunnyvale, CA Choline was determined in infant formula by ion chromatography with suppressed conductivity detection. Samples were digested with 1M hydrochloric acid, filtered, diluted, and injected into the chromatographic system. Choline and the alkali and alkaline earth metals were separated on a high-resolution cation-exchange column and detected by suppressed conductivity. The method was linear between 2 and 200 mg/l (r 2 = ), the concentration range of the diluted samples. This method accurately determined choline in powdered, concentrated, and ready-to-feed infant formulas. Recoveries of choline spikes into powdered infant formula at approximately 1, 0.8, 0.5, and 0.2 times the labeled value ranged from 85 to 114%. This method had good agreement for 8 blind duplicates. The values determined for these samples, which were used in an AOAC collaborative study of an enzymatic method, were consistent with the values determined by the enzymatic method. Choline is an essential nutrient for which the Food and Nutrition Board is considering recommending a dietary intake level (1). Dietary choline is derived predominantly from the phospholipid phosphatidylcholine, as well as free choline, choline esters, and phosphates. The choline compounds in human breast milk are important for neonatal development. For this reason, all infant formulas approved by the U.S. Food and Drug Administration must contain choline at concentrations comparable with those in human milk (2). Many chromatographic assays are reported for choline and acetylcholine. Most determine choline and acetylcholine in plasma or tissue homogenates through some variation of a method first reported by Potter (3). In this method, choline and acetylcholine are separated by cation-exchange chromatography and treated with acetylcholine esterase and choline oxidase in a postcolumn reactor to convert both compounds to hydrogen peroxide, which is then detected by oxidation on a Received September 25, Accepted by JL April 7, Author to whom correspondence should be addressed. platinum electrode. Modifications of this method include addition of a precolumn for sample cleanup (4), changes in the separator column (5 8), optimization of the postcolumn reactor (9, 10), and substitution of chemiluminescence for amperometric detection (11). Choline also can be derivatized with 3,5-dinitrobenzoyl chloride, separated by reversed-phase ion-pair chromatography, and detected by absorbance at 254 nm (12). Milk and infant formula have been assayed for choline by spectrophotometry (13). Samples are digested with acid and then treated with phospholipase D to ensure that all the choline was liberated from phosphatidylcholine. Samples are then treated with choline oxidase, and peroxide evolution is monitored at 505 nm after treatment with peroxidase and 4-aminoantipyrine. In this study, we used the same acid hydrolysis and then separated choline from alkali and alkaline earth metals on a high-performance cation-exchange column (14). Choline and the other cations were detected by suppressed conductivity (15) which has been shown to be a sensitive detection technique for choline (16). Experimental Materials High-purity HCl (Ultrex II, 36.9%) and sulfuric acid (ACS reagent grade) were purchased from J.T. Baker (Phillipsburg, NJ). Choline bitartrate, 99% reagent grade or better, was purchased from Sigma Chemical Company (St. Louis, MO). Tris(hydroxymethyl)aminomethane (Tris) was obtained from Gibco BRL Life Technologies (Gaithersburg, MD). Filter paper (quantitative grade 494) was obtained from VWR (San Francisco, CA). Phospholipase D, type II from peanuts, and choline oxidase were purchased from Sigma. Infant formula (Similac, Ross Products Division Abbott Laboratories, Columbus, OH) was purchased in the powdered, concentrated, and ready-to-feed preparations from a local grocery store. Dried milk also was purchased from a local grocery store. Sixteen powdered infant formula and dried milk samples (8 blind duplicates) that were used in an AOAC collaborative study of an enzymatic method for determining choline in powdered infant formula were obtained from Harvey Indyk, Anchor Products, Waitoa, New Zealand. A cation standard (Six-Cation Standard II) was obtained from Dionex Corporation (Sunnyvale, CA). Type I reagent-grade deionized water

2 LAIKHTMAN & ROHRER: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, Figure 1. Determination of choline in infant formula and dried milk. Panel A shows the separation of 10 L ofa 100 mg/l choline standard, panel B shows the separation of 10 L of a powdered infant formula sample, and panel C shows the separation of 10 L of a dried milk sample. The choline concentrations calculated from the chromatograms depicted in panels B and C are 9.0 and 14.4 mg/l, respectively. Both samples were prepared as described in the Sample Preparation section. Peaks 1 6 are sodium, ammonium, potassium, choline, magnesium, and calcium, respectively. Choline and the other cations were separated on an IonPac CS12A cation-exchange column set with an 18 mn sulfuric acid mobile phase flowing at 1.0 ml/min. Detection was by suppressed conductivity with the CSRS-Ultra operating in the Autosuppression mode at a power of 300 ma. (18 M -cm or better) was used for all sample and mobile phase preparations. Equipment A DX-500 chromatographic system (Dionex) consisting of a GP40 gradient pump, a CD20 conductivity detector, an AS40 automated sampler, and an LC20 chromatographic enclosure equipped with a rear-loading Rheodyne (Cotati, CA) injection valve was used for all chromatography. A Cation Self-Regenerating Suppressor (CSRS-Ultra, Dionex) was used in the AutoSuppression mode at a power setting of 300 ma. A personal computer equipped with PeakNet chromatography software (Dionex) was used for data acquisition and instrument control.

3 1158 LAIKHTMAN & ROHRER: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 Figure 2. Identification of choline in infant formula. Panel A shows the separation of 10 L of a powdered infant formula sample. Panel B shows the separation of 10 L of a powdered infant formula sample after treatment with choline oxidase. Panel C shows the separation of 10 L of a blank preparation treated with choline oxidase. Panel D shows the separation of 10 µl of a choline standard treated in the same manner as the samples in panels B and C except that the enyzme was omitted. Peak labels and chromatographic conditions were the same as those described for Figure 1. The injected concentration of choline for panels A and D were 8.3 and 40 mg/l, respectively. Preparation of Standards In this report, the term choline means choline hydroxide. A 1 g/l choline hydroxide standard was prepared by dissolving g dry choline bitartrate in 1 L water. The appropriate working standards were prepared by diluting the 1 g/l standard with water. To determine linearity and method detection limit, we analyzed triplicate injections of 0.2, 0.5, 1, 2, 5, 10, 25, 50, 100, and 200 mg/l choline hydroxide solutions. Noise was measured between 10.5 and 11.5 min in the chromatograms of water injections by the PeakNet chromatography software. For recovery experiments, we prepared a

4 LAIKHTMAN & ROHRER: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, Table 1. Analysis of powdered, concentrated, and ready-to-feed infant formulas for choline Sample Replicates a mg/l Spike, b Amount, mg/l RSD, % Recovery, % Amount, mg/100 cal % of labeled value c Powdered infant formula Powdered infant formula Powdered infant formula Powdered infant formula Powdered infant formula Concentrated infant formula Concentrated infant formula Ready-to-feed infant formula Ready-to-feed infant formula a b c Each sample was injected 3 times. This value is the concentration of the spike in the sample (i.e., the sample was prepared and diluted for injection). The labeled choline value for all 3 infant formula preparations is 16 mg/100 cal. The label states that 5 oz of prepared formula equals 100 cal. For the powdered formula, 1 scoop is required to prepare 2 oz. The weight of 1 scoop of formula is 8.6 g. 10 g/l choline hydroxide solution by dissolving g dry choline bitartrate in 100 ml water. Both the 10 and 1 g/l standards are stable for 1 week when stored at 4 C. To determine linearity, 0.10, 0.25, 0.50, and 1.0 ml of the 6-cation standard, which contains lithium, sodium, ammonium, potassium, magnesium, and calcium, were diluted to 5 ml. These dilutions yielded the following concentration ranges: lithium, 1 10 mg/l; sodium, 4 40 mg/l; ammonium, 5 50 mg/l; potassium, mg/l; magnesium, 5 50 mg/l; and calcium, mg/l. Each solution was analyzed 3 times. Preparation of Samples Five grams of powdered sample (dried milk or infant formula) were placed into a 50 ml conical plastic tube (with cap), 30 ml 1M HCl were added, the tube was capped, and the sample was shaken until well dispersed. The tube was placed in a water bath at 70 C for 3 h and shaken once per hour. The cap was occasionally loosened or temporarily removed during the early heating stage to avoid excessive pressure buildup. After 3 h, the tube was cooled to room temperature, and the contents were filtered into a 100 ml volumetric flask. The filter was rinsed with water, and the total volume of the filtrate was adjusted to 100 ml with additional water. This solution may be stored at 4 C for 3 days. Concentrated infant formula (10 ml) and ready-to-feed infant formula (17 ml) were prepared in the same manner. Prior to analysis, a final dilution of the prepared sample should be made with water so that the amount of choline is in the calibrated linear range (2 200 mg/l). For powdered infant formula, 1 part of the prepared sample was diluted with 4 parts water. (Note: A 1 part to 3 parts water dilution can also be used.) A 2 parts sample to 3 parts water dilution was made for concentrated infant formula and ready-to-feed infant formula. Table 2. Choline concentrations (as hydroxide) of various infant formula and dried milk samples analyzed as part of a collaborative study a Infant formula sample No. Choline concentration, mg/100 g sample Sample A Sample B % Difference: (A B)/larger value a These samples were used in an AOAC collaborative study of an enzymatic method for determining choline in powdered infant formula (17).

5 1160 LAIKHTMAN & ROHRER: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 Table 3. Analysis of powdered, concentrated, and ready-to-feed infant formulas for alkali and alkaline earth metals Sample Replicates a Cation Amount, mg/l RSD, % Amount, mg/100 cal % of labeled value b Powdered infant formula 30 Na Powdered infant formula 30 + NH Not labeled Powdered infant formula 30 K Powdered infant formula 30 Mg Powdered infant formula 30 Ca Concentrated infant formula 8 Na Concentrated infant formula 8 + NH Not labeled Concentrated infant formula 8 K Concentrated infant formula 8 Mg Concentrated infant formula 8 Ca Ready-to-feed infant formula 8 Na Ready-to-feed infant formula 8 + NH Not labeled Ready-to-feed infant formula 8 K Ready-to-feed infant formula 8 Mg Ready-to-feed infant formula 8 Ca a b Each sample was injected 3 times. The labeled sodium, potassium, magnesium, and calcium values for all 3 infant formula preparations were 24, 105, 6, and 78 mg/100 cal, respectively. The label states that 5 oz prepared formula equals 100 cal. For the powdered formula, 1 scoop is required to prepare 2 oz. The weight of 1 scoop of formula is 8.6 g. The dried milk and collaborative study samples were diluted 1 part to 3 parts water. To determine recovery from the powdered infant formula, 400, 320, 200, and 100 µl of10g/l choline was added prior to the addition of acid. Concentrated infant formula, ready-to-feed infant formula, and dried milk were spiked with 400 µl of 10 g/l choline. Phospholipase D Treatment A phospholipase D solution was prepared by dissolving 150 U of phospholipase D in 200 ml 50 mm Tris HCl, ph 8.0. Two powdered infant formula samples and 2 dried milk samples that had been hydrolyzed with acid, filtered, and adjusted to 100 ml were treated with phospholipase D. One milliliter of each sample was treated with 1 ml of the phospholipase D solution and incubated at 37 C for 15 min. Choline Oxidase Treatment A choline oxidase solution was prepared by dissolving 50.6 U choline oxidase in ml 100 mm Tris (no ph adjustment). A powdered infant formula sample that had been hydrolyzed with acid, filtered, and adjusted to 100 ml was treated with choline oxidase. One milliliter of prepared sample was diluted with 4.0 ml 100 mm Tris and then treated with 20 µl of the choline oxidase solution. A method blank, a 200 mg/l choline standard, and water (1 ml of each sample) were treated in the same manner. All samples were incubated at 37 C for 2 h. As an additional control, 1 ml of a 200 mg/l choline standard was diluted with 4.0 ml 100 mm Tris and then analyzed. Chromatography A 1N sulfuric acid solution was prepared by diluting 27.2 ml concentrated sulfuric acid to 1 L with water. An 18 mn sulfuric acid mobile phase was prepared by diluting 18 ml of the 1N sulfuric acid solution to 1 L. The mobile phase was installed on the instrument and pressurized with helium (8 psi, 55.2 kpa). Standards and samples (10 µl) were separated on a cation-exchange guard (5 0.4 cm, IonPac CG12A, Dionex) and separator ( cm, IonPac CS12A, Dionex) column set at a flow rate of 1 ml/min. The chromatographic system was calibrated by analyzing triplicate injections of 2, 5, 10, 20, 50, 100, and 200 mg/l choline. Results and Discussion Figure 1, panel A, shows the separation of 10 µl ofa 100 mg/l choline standard. Choline was separated on a cation-exchange column designed to isocratically separate lithium, sodium, ammonium, potassium, magnesium, and calcium with a strong-acid mobile phase (13). Here we used an 18 mn sulfuric acid mobile phase. Choline and the other cations were detected by suppressed conductivity (14). This method was linear (r 2 = ) for choline from 2 to 200 mg/l (the range tested). The 2 mg/l choline peak has a signal-to-noise ratio (S/N) ratio of 4.4, which is greater than 3, a value often used to define method detection limit. A peak in the chromatograms of the 1 mg/l standard could not be identified. If greater sensitivity is required, a larger injection volume can be used. In Figure 1, panels B and C, show the determina-

6 LAIKHTMAN & ROHRER: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, tion of choline in powdered infant formula and dried milk, respectively. These samples were diluted 1 part to 3 parts water prior to injection. Choline was separated from the other cations and eluted between potassium and magnesium. To confirm the identity and purity of the choline peak, a powdered infant formula sample was treated with choline oxidase. Figure 2, panel A, shows the chromatogram for a powdered infant formula (diluted 1 part to 4 parts water) and panel B shows a chromatogram of the same sample after choline oxidase treatment. Treatment with choline oxidase caused the disappearance of the choline peak and the appearance of a new peak at 6 min. A choline standard treated in the same manner (chromatogram not shown) also showed the disappearance of the choline peak and the appearance of a new peak at 6 min. Panel C shows a method blank treated in the same manner. Dilution of a choline standard with the Tris solution without addition of the choline oxidase does not cause the disappearance of the choline peak or a change in its retention time (panel D). Table 1 summarizes determinations of choline in powdered, concentrated, and ready-to-feed infant formulas. The powdered infant formula samples were diluted 1 part to 4 parts water prior to injection, whereas the concentrated and ready-to-feed samples were diluted 2 parts to 3 parts water. The concentrations of the injected samples were in the linear range and well above 10 times the S/N ratio, a value often cited as a method s minimum quantitation limit. The low relative standard deviations (RSDs; 1.3 to 4.5%) suggest that the chosen dilutions are appropriate for routine analyses. Recoveries of choline spikes into powdered infant formula at approximately 1, 0.8, 0.5, and 0.2 times the expected concentration ranged from 85 to 114%. Concentrated and ready-to-feed infant formulas were spiked at approximately 2 times the expected level, and recoveries from both samples were 96%. The choline contents of the powdered, concentrated, and ready-to-feed infant formulas were 112, 117, and 118% of the labeled value, respectively. These values are consistent with the expected overages used in fortification. We also determined the choline content of duplicate dried milk samples. The choline concentrations were 1.11 and 1.15 mg/g. There was a 95% recovery for each of the duplicate samples spiked at approximately the sample s choline content. Therefore, this method had good precision and good recovery of choline from the matrix for all the samples analyzed in this study. To further assess method precision and accuracy, we analyzed the same samples used in an AOAC collaborative study of an enzymatic method for determining choline in powdered infant formula and dried milk powder (17). We received 16 samples (8 blind duplicates), and each was analyzed for choline. Table 2 shows choline concentrations for each sample and its duplicate, which were identified at the study s conclusion. The agreement between duplicate samples further demonstrates the good precision of this method. There was also agreement with the reported choline concentrations of these samples (17). Our values were an average of 14.9% lower than the values determined by the enzymatic method. The values determined by the enzymatic method are an average of values from over 20 laboratories, and each of these laboratories treated the samples with phospholipase D. We omitted this treatment and did not have enough sample to repeat the determination with phospholipase D treatment. Using the enzymatic method, Woollard and Indyk (13) found that phospholipase D treatment increased the amount of choline determined in 4 dried milk powder samples by an average of 13.6%. To determine if the omission of the phospholipase D treatment could cause incomplete choline recovery, we treated 4 samples (2 powdered infant formula and 2 dried milk samples) that had been hydrolyzed for 4 h with phospholipase D (13). We also extended the acid hydrolysis of 4 other samples to 5 h. When Woollard and Indyk analyzed a variety of food and pharmaceutical samples for choline, they found that no sample required a hydrolysis longer than 5 h (13). Neither treatment yielded a higher concentration for any of the samples. Although the omission of the enzymatic treatment might explain why our values were low compared to the enzymatic method, the values we determined may be accurate because we are directly determining choline rather than a product of a series of enzymatic and chemical reactions. We also investigated the possibility of using this method to simultaneously determine choline and alkali and alkaline earth metals in infant formula. Table 3 summarizes the determination of alkali and alkaline earth metals in the same samples in which choline was quantitated. The alkali and alkaline earth metal concentrations of the diluted samples were all within the linear ranges (sodium, 4 40 mg/l, r 2 = ; ammonium, 5 50 mg/l, r 2 = ; potassium, mg/l, r 2 = ; magnesium, 5 50 mg/l, r 2 = ; and calcium, mg/l, r 2 = ). The cation contents ranged from 111 to 152% of the labeled values, consistent with over-fortification. Reproducibility was good except for calcium in the powdered infant formula sample, which had an RSD of 8.6%. This high RSD for calcium may be due to calcium leeching from the filter paper used for sample preparation. Of 13 method blanks prepared during this work, 2 contained detectable levels of magnesium and calcium. Those 2 blanks had magnesium concentrations of 0.2 and 0.3 mg/l and calcium concentrations of 15.0 and 11.8 mg/l. The calcium contents of these blanks are a significant percentage of the amount of calcium in the analyzed infant formulas. To determine calcium accurately and precisely with this method, more work is required to find a technique for obtaining a low, reproducible calcium method blank. During the development of our assay and the analysis of the collaborative study samples, no changes were observed in either column backpressure or choline retention time. A 10 mg/l choline check standard analyzed in triplicate 9 times during 5 days of continuous sample analysis (3 mobile phase additions) had an area RSD of 1.9% and a retention time RSD of 0.4%. These results suggests that the sample preparation method is appropriate for rugged chromatographic analysis. In conclusion, we have shown that choline can be determined directly in filtered acid hydrolysates of powdered infant formula, concentrated infant formula, ready-to-feed infant formula, and dried milk by cation-exchange chromatography

7 1162 LAIKHTMAN & ROHRER: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 and suppressed conductivity detection. This method is simple, precise, and accurate. References (1) Canty, D. (1997) Nutr. Sci. News October, 1 6 (2) U.S. Food and Drug Administration Code of Federal Regulations 21 (1991) Chapter 1, pp (3) Potter, P.E., Meek, J.L., & Neff, N.H. (1984) J. Neurochem. 41, (4) Ikarashi, Y., Iwatsuki, H., Blank, C.L., & Maruyama, Y. (1992) J. Chromatogr. 575, (5) Tyrefors, N., & Carlsson, A. (1990) J. Chromatogr. 502, (6) Haen, E., Hagenmaier, H., & Remien, J. (1991) J. Chromatogr. 537, (7) Salamoun, J., Nguyen, P.T., & Remien, J. (1992) J. Chromatogr. 596, (8) Kanenda, N., Asano, M., & Nagatsu, T. (1986) J. Chromatogr. 360, (9) Stadler, H., & Nesselhut, T. (1986) Neurochem. Int. 9, (10) Flentge, F., Venema, K., Koch, T., & Korf, J. (1992) Anal. Biochem. 204, (11) Honda, K., Miyaguchi, K., Nishion, H., Tanaka, H., Yao, T., & Imai, K. (1986) Anal. Biochem. 153, (12) Buchanan, D.N., Fucek, F.R., & Domino, E.F. (1980) J. Chromatogr. 181, (13) Woollard, D.C., & Indyk, H.E. (1990) J. Micronutr. Anal. 7, 1 14 (14) Rey, M.A., & Pohl, C.A. (1996) J. Chromatogr. A 739, (15) Rabin, S., Stillian, J., Barreto, V., Friedman, K., & Toofan, M. (1992) J. Chromatogr. 640, (16) Chem, S., Soneji, V., & Webster, J. (1996) J. Chromatogr. A 739, (17) Woollard, D.C., & Indyk, H.E. (1999) J. AOAC Int. (in press)