AOAC OMA# 2017.03 (Final Action)-OMB

D RAHER & W HITE : J OURNAL OF AOAC I NTERNATIONAL V OL . 101, N O . 3, 2018 829

fluorescence as the means of detection significantly benefits method specificity by limiting matrix response.

BSA spike solution weight and the milligrams per gram Trp content determined from five triplicated analyses were used to calculate the theoretical spike. Sample preparations were hydrolyzed and prepared for analysis according to the method procedure described. Accuracy was further assessed by analysis of SRM 1849a and SRM 927e (BSA solution).

Results and Discussion

Precision

Repeatability requirements were met for all SPIFAN matrixes, with the exception of the partially hydrolyzed soy-based powder (Lot No. 410457651Z) at 5.0%. This sample also showed clumping during reconstitution. Other repeatability results ranged for the RSDs, from 0.2 to 2.0%. The overall average RSD r was 1.0% and the RSD iR was 2.3%. See Table 2017.03A for precision results.

Linearity

Three independent experiments were run with eight levels of Trp, ranging from approximately 0.2 to 30 mg/L. An internal standard linear regression constructed from the standards was used to back-calculate the concentration of each working standard. The error at each level was determined, and the average computed for each level over the three runs.

Accuracy

Spike recovery requirements were met for all SPIFAN matrixes. Mean spike recovery data ranged from 93.8 to 104.9%. See Table 2017.03B for accuracy data. The SRM 1849a mean result of 181.3 mg/100 g is 98.5% of the 184 mg/100 g reference value.

Analytical Range

The analytical range for the method was determined using the calculated LOQ to establish the low end. The high working standard concentration was used to calculate an upper limit based on a reasonable sample weight for a high-protein sample.

Linearity

The method demonstrated acceptable linearity over a standard range of approximately 0.2 – 30 mg/L Trp, with eight dilution levels having an average deviation from the theoretical value of 0.3%. No single point in the experiments exceeded a 5.0% deviation from the theoretical. See Table 2017.03C for linearity data.

LOQ

The Trp LOQ was determined based on the ability to distinguish sample signals from blank variability. The quantified Trp milligrams-per-liter amounts from enzyme blank pairs were calculated from each of the 11 calibration curves. The within-run SD was determined for the blank paired data, which was considered to constitute the noise for this analysis. The LOQ was determined as 10 times the blank RSD in milligrams per liter as injected and subsequently extrapolated to RTF and powder samples using the typical sample preparation weight and dilution.

Analytical Range

The analytical range was determined to be 0.18 – 300 mg/100 g. The LOQ was calculated to be 0.18 mg/100 g. The upper limit was calculated by multiplying the concentration of the high working standard (30 mg/L) by the typical sample dilution of 50 mL, dividing this by the lowest sample weight used for an RTF sample (0.5 g), and then multiplying this result by 100 g to give units of milligram/100 grams. In practice, for extremely high- concentration samples, the sample preparation weight or sample dilution can be altered to extend the effective range. Unusually high Trp levels would not be encountered in an intact protein sample, but could occur in a highly fortified nutritional product. For these reasons, the analytical range of 0.18 – 300 mg/100 g was acceptable for all 14 SPIFAN matrixes and would likely be acceptable for the vast majority of samples tested by this method. The LOQ requirement of ≤ 0.4 mg/100 g reconstituted final product was met. The within-run SD of the enzyme blank was calculated as 0.007 mg/L, as injected. The LOQ concentration, as injected, was 0.07 mg/L (10 times the within-run SD) or 0.0035 mg per 50 mL sample preparation volume. Given the reconstituted powder sample addition of 2.0 g and RTF formula addition of 2.0 g, the calculated powder and RTF sample LOQ was 0.18 mg/100 g. LOQ

Ruggedness

Parameters varied during the validation to establish method ruggedness included precision runs prepared by two analysts; samples run on two HPLC systems; two analytical columns from different stationary phase batches; new working standards and mobile phases, which were prepared for each precision run; and two different stock standard solutions, which were used throughout the validation.

Specificity

All of the unfortified matrixes were expected to contain Trp from the protein component and, therefore, could not be used as placebo samples to evaluate method specificity. To address this lack, a protein-free infant base powder product, ProPhree, was tested by quantitative analysis in duplicate in a single run as a true placebo sample. In addition, specificity for the internal standard, 5- methyl- DL -tryptophan, was evaluated by analysis of ProPhree and all SPIFAN matrixes without internal standard addition. These chromatograms were overlaid with a standard chromatogram to determine any coeluting background peaks. Any coeluting peaks would be quantified as a percent of the average internal standard area from the standard injections. The method ’ s use of Trp native

Ruggedness

The method performance achieved with the variables enumerated above is the demonstration of the method ’ s ruggedness under typical long-term variables.