Matrix Extension with Fitness for Purpose and Stability Assessment of DHA and Additional Fatty Acids in Individual Whole Chicken Eggs

The consumption of long chain polyunsaturated fatty acids (LC-PUFA) is associated with several human health benefits. Most notable of these LC-PUFA is docosahexaenoic acid (DHA C 22:6 ) whose inclusion is considered essential for optimum human health. Biofortification of common foods such as eggs with DHA has emerged as a specific approach to increase the intake of DHA in human populations. This can be achieved by supplementing poultry rations with feeds like microalgae or fish oil that are rich in DHA, which results in an increased uptake in the egg. Gas chromatography with flame ionization detection (GC-FID) is the method of choice when analyzing food such as eggs for DHA and other fatty acids. For regulatory studies it is desirable to demonstrate that the method is specifically suitable for the analysis of DHA and fatty acids in eggs. The purpose of this paper is to further extend the scope of the AOAC 996.06 methodology examined in the paper by Dillon et al., and to demonstrate the fitness for purpose of the method by examining specific validation parameters. It is a further objective to investigate the stability of DHA and other fatty acids of short and long timepoints. A validation of the method for the determination of DHA and three other fatty acids in eggs is thus presented.


Introduction
An increased consumption of long-chain polyunsaturated fatty acids (LC-PUFA) in the human diet can be related to several health benefits including improved cognitive and cardiovascular function and immune health [1] [2] [3]. Among the group of LC-PUFA compounds, docosahexaenoic acid (DHA) is perhaps the best well known and researched and its inclusion in diets for optimum health has been well publicised [4] [5] [6]. Despite this, DHA intake remains low in many countries, particularly those where Western diets are prevalent [7] [8] [9].
The biofortification of common foods with DHA and other LC-PUFA is seen as a key approach by which to seamlessly boost consumption of these essential nutrients in human populations [3] [10]. DHA intake for example, can be increased in household foods, such as chicken eggs, by supplementing poultry dietary rations with feeds like microalgae or fish oil which are rich in DHA [11] [12] [13] [14].
Gas chromatography with flame ionization detection (GC-FID) has been established as the principle technique for analyzing fatty acids in foods, with the original method of fatty acids extraction and methylation to yield methyl ester derivatives being described by Folch et al. [15]. The method has been referenced in several publications in the analysis of DHA and fatty acids in eggs [11] [16]. In addition, variations of the method have been examined with a view to further developing the method in the analysis of eggs [17] [18]. Whilst the methodology has been widely investigated, there is a need for the method to be thoroughly validated to demonstrate to fitness for purpose in analyzing eggs for regulatory studies, for example in the analysis of fatty acids profiles, assessing egg enrichment studies and gaining an understanding of the efficacy of feed ingredients for the purpose of enriching chicken eggs.
The purpose of this paper therefore is to further extend the scope of the AOAC 996.06 methodology as examined in the paper by Dillon et al., [19] to include this additional matrix. The method was further applied to assess the stability of DHA and other fatty acids in the eggs over a range of timepoints. The parameters examined during the validation study included; linearity and range, the limit of detection (LOD) and limit of quantification (LOQ), accuracy, repeatability, inter-analyst reproducibility and specificity. Stability experiments were conducted samples of extracted egg after 2, 24 and 48 hours at room temperature; the stability of freeze-dried egg samples spiked with various concentrations of fatty acids was assessed after storage at <−16˚C for periods of 0, 2,4,8,12,16,20 and 26 weeks and a freeze/thaw study over four cycles were conducted.
To examine the acceptability of the method, two commercially available reference materials were analyzed. Reference material NIST SRM 3275-2 Anchovy Oil concentrate was analyzed with each set performed during this study to verify the acceptability of the analytical set. The reference material contained 187 ± 8 mg/g of DHA C 22:6 . A quality control reference material NIST 3290 (Dry Cat Food) was analyzed with each set performed during this study to verify the complete hydrolysis and derivatization of fatty acid bound within the chicken eggs.

Preparation of Calibration Standards, Internal Standards and Quality Control Standards
The FAME stock solutions were prepared by taking a known quantity of each of the commercial standards and dissolving in hexane and making up to 10 ml in a volumetric flask. A mixed FAME standard working solution with a final concentration of 40 mg/ml of each FAME was prepared by taking an aliquot of each FAME stock solution and making up to 10 ml with hexane. A methyl undecanoate internal standard solution was prepared by taking 1 g of methyl undecanoate, dissolving in hexane and making up to 10 ml. Calibration FAME standard solutions were prepared at 0.3, 0.75, 1.5, 3, 7.50 and 15 mg/ml by adding volumes of the mixed FAME standard working solution with 100 μl of methyl undecanoate internal standard solution and making up to 2000 μl with hexane. Standard solutions were stored at <−16˚C when not in use.

Preparation of Egg Sample
Control (blank) chicken egg samples were purchased locally and were freeze-dried prior to extraction and analysis. Any samples that could not be freeze-dried immediately were stored at ≤−16˚C until freeze-drying could be completed. Sam-

Method Validation
Linearity A standard curve that covered the range of analytes and the range of concentrations of fatty acids in the samples was prepared to demonstrate linearity on the GC-FID. Linear regression, forced through the origin and with equal weighting, was applied to the peak area ratios plot for the construction of calibration curves plotting FAME:IS peak area ratios of the calibration standards against FAME concentrations and provided information on the slope, coefficient of determination, and intercept. Standards contained C 8:0 , C 14:0 ; C 17:0 ; C 18:0 ; C 18:1T , C 22:1 cis-13 and C 22:6 FAMEs along with C11:0 internal standard.

Limit of Detection (LOD) and Limit of Quantification (LOQ)
The LOD and LOQ of the method were established for DHA C 22:6 by analyzing the endogenous quantity of DHA in a freeze-dried blank egg sample which was analyzed in ten replicates. The LOD of the method was determined as three times the standard deviation, whilst the LOD was deemed to be ten times the standard deviation.

Accuracy
The blank matrix sample, i.e. whole chicken egg composite, was spiked with C 14:0 , C 17:0 , C 18:0 and C 22:6 in FAME form at three levels: 0.1%, 0.25% and 0.4% (w/v). All four analytes have endogenous levels in the blank matrix, which was extracted and analyzed in triplicate. The average of the endogenous levels found in the blank sample was subtracted from the final concentration prior to the spike recoveries being calculated.

Repeatability
The freeze-dried whole chicken egg composite sample was analyzed in triplicate by the first analyst in three separate analysis sets producing a total of nine results. In addition, the samples were analyzed in triplicate by the second analyst, for a total of three results. The mean, standard deviation and Relative Standard Deviation (%RSD) were determined for results from both analysts, and the relative difference between the first and second analyst were compared. The acceptable criteria for %RSD was <10%.

Specificity
To determine the identity of the FAME analyzed in the study, each of the seven standards were analyzed individually and their retention times were recorded.

Reference Materials
To examine the acceptability of the method, two commercially available ref-

Inter-analyst reproducibility
To demonstrate the inter-analyst reproducibility and the inter-analyst repeatability, a second analyst repeated portions of the validation and the results were compared.

Stability
The following experiments were conducted over the course of the study to examine the stability of solutions and analytes; 1) the stability of the FAME stock solutions stored at <−16˚C over six months; 2) the calibration FAME standard solutions on the autosampler over the course of each analytical sequence; 3) the short-term stability testing was carried out on chicken egg extracts prepared during the accuracy study at room temperature to examine the shelf life of the extracted analytes during the testing procedure. Each matrix extract was spiked at three different concentration levels, 0.1% (w/v), 0.25% (w/v) and 0.4% (w/v).
Three aliquots for each of the three levels, and therefore a total of nine aliquots of each chicken matrix, were examined. The extracts were analysed at 2, 24 and 48 hours.; 4) Long term stability testing was carried out on the freeze-dried and stored egg, to evaluate the stability of the analytes and egg matrix during the an- Freezing was established at <−16˚C overnight and thawing took place at room temperature for more than two hours. Initial analysis after one freeze-thaw cycle established the baseline measurement. Sample aliquots were tested after each freeze-thaw cycle for four cycles.

Linearity
The method was linear over the calibration range of 0.3 mg/ml to 15 mg/ml for all seven FAME analytes; C 8:0 , C 14:0 ; C 17:0 ; C 18:0 ; C 18:1T , C 22:1 cis-13 and C 22:6 FAME. The coefficients of determination, R 2 , were found to range from 0.998 to 1.000 for the seven analytes within the acceptable criteria of R 2 ≥ 0.990 (R ≥ 0.995). See results in Table 1.

Limit of Detection (LOD) and Limit of Quantification (LOQ)
To determine this method's LOD and LOQ for C 22:6 the blank egg matrix samples were analyzed in ten replicates. The standard deviation was determined to be 0.0068%. The LOD, calculated as three times the Standard Deviation, is 0.02%. The LOQ, calculated as ten times the Standard Deviation, is 0.07%. See Table 2. Accuracy, repeatability and specificity The spike recoveries for the four monitored compounds were calculated and ranged from 91% to 110% which is within the acceptable range of 90% to 110%.
The mean of the recoveries ranged from 97% to 108%. See Table 3 for the summary of accuracy results.
An example calculation for Day-1-Spike-1, C 14:0 at spiked at 0.1%, is as follows:   The samples were analyzed in triplicate over three separate analysis sets by the first analyst, giving a total of nine determinations. In order to assess the repeatability, the mean and standard deviation of the recoveries were calculated to determine the %RSD. The %RSD for chicken egg samples were within the acceptable range of <10% RSD for DHA (C 22:6 ) with results of 1.0% for the first analyst and 1.9% for the second analyst. A summary of the results can be found in Table 4.
To assess the specificity, the retention times of each FAME standard were de-

Reference Materials
The results of the reference materials supported the suitability of the method in determining the fatty acid analytes. The reference material NIST SRM 3275-2 Anchovy Oil concentrate yielded results ranged from 173 to 183 mg/g. These results are within the acceptable range of 2 standard deviations from the mean, which was determined to be 172 to 185 mg/g. See Table 5. were within three standard deviations of the established mean. See Table 6.

Inter-analyst Reproducibility
To demonstrate the robustness and the inter-analyst reproducibility, a second     Table 5. The results for the following analytes on the quality control sample for the second analyst were within three standard deviations of the established mean.
The egg samples were also analyzed by the second analyst. Results for DHA (C 22:6 ) were found to be within three standard deviations of the established mean. The mean established by the first analyst for DHA was 0.504%, with a standard deviation of 0.005%. Calculating ± 3 standard deviations around the mean, this equates to an acceptable range of 0.489% to 0.519%. The second analyst had mean results of 0.513% for DHA, which is within the acceptable range ( Table 4). The recoveries on the spiked matrix samples ranged between 91% and 108% for the second analyst, within the acceptable criteria of 90% to 110%. See Table 3.

Stability of FAME Stock Solutions
New stock solutions were compared against the original stock solutions which were prepared prior to use on experiments. The change of the seven fatty acid FAME analytes between the old stock solution and new stock solutions were determined to be between −1.9% and 8% relative difference and showed stability for 207 days (over six months).

Stability of Calibration FAME Standard Solutions
Fresh calibration solutions were prepared for each analysis performed on the GC instrument autosampler. Repeated injections were performed at the start and end of the sequence to confirm stability over the course of the analysis. The calibration solutions at the end of the sequence ranged from −5.9% to 3.5% relative error compared to the calibration solutions at the beginning of the sequence. This is within the acceptable range of ±20% relative error.

Short-Term Stability-Extracts
Short term stability testing was carried out on egg extracts from the accuracy study at room temperature to account for the shelf life of the extracted analytes during the testing procedure. This involved spiking at three different concentration levels, 0.1% (w/v), 0.25% (w/v) and 0.4% (w/v). Three aliquots for each of the three levels were studied, for a total of nine aliquots. The extracts were analyzed at 2 hours, 24 hours and 48 hours. The relative error (%) was calculated as the difference between the mean recovery at time t and the mean recovery at time 0 (2 hours), divided by the mean recovery at time 0, expressed as a percentage. The calculation formula can be found as a footnote on Table 8. The relative error (%) was found to be within ±20% relative error (ranging between −6.4% to 2.9%) compared to extracts from initial extraction (T = 0 (2 hours)) for the parameters  Table 7.

Long-Term Stability
Long term stability testing was carried out on the freeze-dried and stored egg, to evaluate the stability of the analytes and egg matrix during the anticipated storage time of the samples. For the 4-and 8-week time frames, the stored samples were within ±20% relative error (ranging between −20% to 7.0%) compared to extracts from initial extraction (T = 0) for the parameters monitored: C 14:0 , C 17:0 , C 18:0 and C 22:6 . For the 12-week stability samples, however, the data for C 18

Freeze and Thaw
The freeze and thaw process consisted of repeated analysis of egg samples over the course of four freeze and thaw cycles. Freezing was established at ≤−16˚C overnight and thawing took place at room temperature for more than two hours.
Initial analysis after one freeze-thaw cycle established the baseline measurement.
Sample aliquots were tested after each freeze-thaw cycle for four cycles. The results after each freeze-thaw cycle compared to extracts from initial extraction (T = 0 (cycle 1)) for the parameters monitored: C 14:0 , C 17:0 , C 18:0 and C 22:6 ranged from −9.6% to 0.8%, within the acceptable range of ±20% relative error. The analytes were deemed to be stable over four freeze and thaw cycles (Table 9).

Conclusion
Considering that whole egg is frequently consumed, either as a stand-alone food item in the diet or in the formulation of a variety of food products, with the growing trend in omega-3 biofortification through the chicken's diet, the establishment of an accurate quantification assay for the determination of the fatty acid composition is important for nutritionists and regulatory scientists. From the results of our study, the suitability of the AOAC method 996.06 for the de-