Optimization and Validation of a UV–HPLC Method for Vitamin C Determination in Strawberries (Fragaria ananassa Duch.), Using Experimental Designs

Abstract

Vitamin C or total ascorbic acid (TAA) in fruits can be assumed as ascorbic acid (AA) plus dehydroascorbic acid (DHA) content. The aim of this work was to optimize and validate, using experimental designs, a new highperformance liquid chromatographic method for vitamin C determination in strawberries. The mobile phase (MP) consisted of a 0.03 M sodium acetate/acetic acid buffer, 5% methanol. For optimization, a Box-Behnken design was used (three factors at three levels: (a) pH of MP, 3.8-5.8; (b) wavelength, 240-270 nm; and (c) flow rate, 0.5-1.2 ml min −1 ). Responses were: AA and TAA areas, peak widths, and retention times. A global optimization was performed using the Derringer desirability function, and a value of 0.84 was reached for the combination of design factors: A05.8, B0251 nm, and C01.15 ml min −1 . Method validation, using AA standard solutions, included: linearity study, limits of detection and quantification, and calibration and analytical sensitivity quantifications. Precision and accuracy were studied in strawberry extracts. The coefficients of variation (percent) were: AA, 1.5%; TAA, 1.8%, and DHA, 4.9%. Accuracy was evaluated with AA standard spiked in 30-150% range of the expected amount of analyte in real samples. The joint confidence elliptical region test and t test were employed for the study of the difference between recoveries (percent) and the ideal 100%. The robustness was analyzed using a fractional factorial design (3 4−2 ), and an AA recovery study after slight changes in operative variables was performed. The results indicate that the optimized method was linear, sensible, precise, accurate, and robust.

Determination of vitamin C in foods: Current state of method validation

“…The revised methods [36,37,55,62,63,68] proved to be generally unaffected by small variations in the chromatographic conditions assayed. Robustness is generally not considered in most validation guidelines [37,55,62,63,68], so it is rarely used. Only six of the revised methods studied this parameter.…”

“…However, this evaluation is recommended in non-consecutive days, to assess the variation over a longer period of time. The intermediate precision was not evaluated in many works [20,22,[32][33][34]39,44,48,50,53,57,62,63,65,66,98], contrary to what is recommended. Given the poor stability of L-AA, the study of the intermediate precision is imperative to evaluate the method ability to provide the same results in different days.…”

“…Unfortunately, few of the revised publications reported the use of statistical methods for linearity evaluation [27,42,62,68], indicating that the validation of this parameter should definitely be improved. Among the reported works in this review, Chebrolu et al [27], after constructing the calibration curve for L-AA determination in fruits and vegetables, confirmed the linearity by residual plot and normal probability plot (P-P plot) (Fig.…”

“…Valente et al used the ANOVA one-way test to prove variance homogeneity in their calibration working range [68]. The analysis of variance of the regression and a residual plot were performed by Odriozola-Serrano et al [42] for the analysis of fruits, whereas an F-test was carried out by Van de Velde et al [62]. Supplementary Fig.…”

“…However, a limited number of methods included statistical data in the validation of the accuracy, which improved the reliability of the presented results. On the one hand, Ciulu et al [54], Odriozola-Serrano et al [42], and Van de Velde et al [62] applied a statistical analysis to their recoveries results, using the Student's t-test. On the other hand, Valente et al [68] calculated the standard uncertainty from the method accuracy, at three concentration levels.…”