2
Introduction Recent advances in instrumentation allow for automation and small scale esterification of lipids, specifically by the Agilent 7696A Sample Prep Workbench (Figure 1). While the Workbench works well for oils and extracted lipids, transferring direct methods for use with the workbench is challenging. In this investigation, an automated method using the 7696A Sample Prep Workbench was developed based on a method by Golay, et al. 1 Three different infant formulas were used for method investigation; results were compared to conventional direct methods for accuracy. Optimization of the automated method included evaluation of initial sample weights, aliquot size used for reaction, and solvent volumes. Data shows that the automated method has many advantages; precise control of reaction times, analyst labor savings, and reduction in solvent usage by 90%. Results Initial testing resulted in lower than expected results (Table 1). All fatty acids were quantitated and evaluated, however palmitic, oleic, and linoleic were chosen as representative saturated, monounsaturated, and polyunsaturated fatty acids. Further testing was performed, using various sample weights, internal standard in MTBE volumes, and sample aliquots (Table 2). Reducing the sample aliquot improved method performance, particularly for sample IF1 (Table 3). Increasing the volume of internal standard in MTBE to 400 µL yielded, in general, higher results than 300 µL (Table 4). Based on the results of trial 3, a 150 mg aliquot and 400 µL of internal standard in MTBE was used to analyze three infant formulas (IF1, IF2, and IF3). All fatty acids were quantitated and evaluated, however linoleic, α linolenic, arachidonic, and docosahexaenoic acid were chosen for display due to their importance in infant nutrition. Comparable results were generated using both methods (Figures 2-5). Note: Results for IF2 appear low, indicating further need for method optimization. Materials and Methods Infant Formula n Powdered formula, approximately 30% lipid (IF1) n Powdered formula, approximately 15% lipid (IF2) n Ready to feed formula, approximately 3.5% lipid (IF3) Automated Method Reagents n Internal standard solution: C11:0 methyl undecanoate + C13:0 tridecanoin in methyl tert-butyl ether (MTBE) (1.667 mg/mL each) n 5% sodium methoxide in methanol n 0.1 g/mL disodium hydrogen citrate plus 0.15 g/mL sodium chloride in water Methods n Covance method: Direct methylation using 0.5 M NaOH in methanol, and 14% BF3 in methanol n Final Automated Workbench method: Preparation prior to Workbench: 1. Adjust sample weights for powders to obtain approximately 5 mg lipid in the final aliquot, record weight. (~0.24 g for IF1, ~0.45 g for IF2) 2. Add 2 mL H 2 O, record total weight. Mix well. 3. Allow to stand for at least 15 minutes. 4. Using a pipette, aliquot approximately150 µL into a 2 mL injection vial and record weight of aliquot. Cap vial. 5. Aliquot liquid samples directly into the 2 mL vial. Cap vial. 7696A Sample Prep Workbench was programmed to do the following steps: 1. Add 400 µL internal standard in MTBE (methyl tert-butyl ether). 2. Mix at 2500 rpm for 20 seconds 3. Add 500 µL 5% sodium methoxide 4. Mix at 2000 rpm for 40 seconds 5. Wait 10 seconds 6. Mix at 2000 rpm for 20 seconds 7. Add 100 µL hexane 8. Add 600 µL aqueous NaCl/disodium citrate 9. Mix at 2000 rpm for 40 seconds The reaction was timed so that 185 seconds elapsed between the addition of sodium methoxide and hexane. A total of 240 seconds elapsed until the addition of NaCl/Citrate solution. All final solutions were analyzed by gas chromatography using the same conditions (column SP-2560, 100 m x 0.25 mm x 0.20 micron film thickness) and external calibration standards. Comparison of an Automated Method to Conventional Methods for the Determination of Fatty Acids in Infant Formula Barb Mitchell, Sarah Kanable, Christopher Leuenberger, Erin Meinholz, Meagan Dallman, Erika Vacha, John Richard, Chad Volkmann and Wayne Ellefson Covance Laboratories Inc., Madison, Wisconsin Future Work Results for IF2 (15% lipid) remained lower than the target values, further testing is needed to improve overall recoveries. Additional testing will focus on determining the most efficient sample to reagent ratios. In addition, other types of formula will be investigated. Conclusions Automation of direct transesterification of triacylglycerols in infant formulae is possible using the Agilent 7696A Sample Prep Workbench. This automation reduces solvent use, frees analyst time, reduces exposure to hazardous chemicals, and allows for consistent timing of a critical reaction. Since the method tolerates water, small sample aliquots needed for analysis on the Workbench are possible without compromising the advantage of a larger sample size. Acknowledgements The authors would like to first thank Agilent Technologies for the use of a 7697A Sample Prep Workbench for this investigation. We would also like to thank Chad Scheuerell for helping with continuing our ongoing investigation and Brent Rozema for his overall support of our research. References 1. Golay, Pierre-Alain, Giuffrida, Franscesca, Dionisi, Fabiola, Destaillats, Frédéric (2009) Journal of AOAC International Vol. 92, No. 5: 1301-1309. 2. Suter, Bea, Grob, Konrad, Pacciarelli, Bruno (1997) Z. Lebensm Unters Forsch A 204: 252-258. Figure 1. Agilent 7696A Sample Prep Workbench. Table 1. Trial 1 Results - Initial Testing Target (mg/g) Results (mg/g) % Deviation IF1 Palmitic 52.7 36.1 -31.5 Oleic 104.3 64.8 -37.9 Linoleic 53.1 35.4 -33.3 IF2 Palmitic 22.8 13.0 -43.0 Oleic 51.8 27.2 -47.5 Linoleic 41.5 23.6 -43.1 Table 2. Workbench Method Experiments Trial # Sample size Volume of H 2 0 added Aliquot size Volume of IS solution 1 To obtain 50 mg lipid 2.0 mls 250 mg 300 µL 2 To obtain 50 mg lipid 1.0 mls 100 mg 200 mg 300 µL 3 To obtain 5 mg lipid in the aliquot 1.0 mls 100 mg 150 mg 200 mg 300 µL 400 µL Table 3. Trial 2 Results - Reduced Sample Aliquots 100 mg aliquot 200 mg aliquot Target (mg/g) Result (mg/g) % Deviation Result (mg/g) % Deviation IF1 Palmitic 52.7 50.7 -3.8 31.4 -40.4 Oleic 104.3 100.7 -3.5 58.8 -43.6 Linoleic 53.1 51.0 -4.0 31.2 -41.2 IF2 Palmitic 22.8 15.1 -33.8 9.14 -59.9 Oleic 51.8 33.9 -34.6 19.7 -62.0 Linoleic 41.5 27.6 -33.5 16.5 -60.2 Table 4. Trial 3 Results - Increased Internal Standard in MTBE Volume IF1 300 µL ether 400 µL ether 100 mg Target (mg/g) Result (mg/g) % Deviation Result (mg/g) % Deviation Palmitic 52.7 51.2 -2.8 51.7 -1.9 Oleic 104.3 101.2 -3.0 103.4 -0.9 Linoleic 53.1 52.0 -2.1 52.6 -0.9 150 mg Palmitic 52.7 42.7 -19.0 53.9 2.3 Oleic 104.3 81.3 -22.1 108.0 3.5 Linoleic 53.1 43.0 -19.0 54.7 3.0 200 mg Palmitic 52.7 33.6 -36.2 47.7 -9.5 Oleic 104.3 61.4 -41.1 93.7 -10.2 Linoleic 53.1 33.3 -37.3 48.4 -8.9 IF2 300 µL ether 400 µL ether 100 mg Target (mg/g) Result (mg/g) % Deviation Result (mg/g) % Deviation Palmitic 22.8 13.8 -39.5 16.5 -27.6 Oleic 51.8 30.5 -41.1 36.9 -28.8 Linoleic 41.5 25.1 -39.5 30.0 -27.7 150 mg Palmitic 22.8 15.9 -30.3 19.6 -14.0 Oleic 51.8 34.9 -32.6 44.3 -14.5 Linoleic 41.5 29.1 -29.9 35.9 -13.5 200 mg Palmitic 22.8 15.4 -32.5 20.7 -9.2 Oleic 51.8 33.4 -35.5 47.0 -9.3 Linoleic 41.5 28.4 -31.6 38.1 -8.2 Table 5. Overall Precision Sample Type Automated Base Transesterification Method % RSD 18:2 Linoleic Acid IF1 1.69 IF2 3.12 IF3 2.00 18:3 αLinolenic Acid IF1 1.55 IF2 2.93 IF3 4.59 20:4 Arachidonic Acid IF1 3.39 IF2 7.03 IF3 3.55 22:6 Docosahexaenoic Acid IF1 3.85 IF2 11.2 IF3 4.59 Figure 2. Linoleic Acid mean results for IF1, IF2, and IF3 comparing methods. Figure 4. Arachidonic Acid mean results for IF1, IF2, and IF3 comparing methods. Figure 3. α Linolenic Acid mean results for IF1, IF2, and IF3 comparing methods. Figure 6. Comparison of organic solvent use per sample. Figure 5. Docosahexaenoic Acid mean results for IF1, IF2, and IF3 comparing methods. The automated method reduced manual steps for the analyst, allowed for precise control of reaction times, and reduced the amount of reagents used. (Figure 6). Not only does the Workbench reduce the amount of solvent used, it also reduces the chemists’ exposure to hazardous chemicals. Overall precision was evaluated based on results over 3 days of analysis producing a total of 15 data points, with the exception of the IF2 which is based on 3 days of data producing a total of 13 data points (Table 5).
