Improved Reliability of Vitamin Analysis in Foodstuffs Using a Microbiological Assay Method: Construction of Calibration Curves Using Logistic Modeling

Vitamins are essential nutrients for human beings. Therefore, accurate determination of vitamin levels in foodstuffs is vital to confirm the proper intake of vitamins. The microbiological assay (MBA), which is used worldwide for the determination of several vitamins in foodstuffs, is very sensitive and can determine ng/mL levels of vitamins. However, the correlation between vitamin concentrations in a sample solution and the plotted growth is usually shown as a sigmoid curve. Therefore, a calibration curve derived from a linear regression may lead to error. In this study, we evaluated the effects of various models (linear, quadratic, and cubic regression models and a four-parameter logistic model (4PLM)) for calibration curve construction on the determination of vitamin B6 in infant formula. Among the four models, the calibration curve constructed with 4PLM was the most reliable for vitamin B6 determination. Moreover, the calibration curve based on 4PLM showed robustness for extrapolation; even if the vitamin concentration in the sample solution deviated from the range of the standard solution, a reasonable result could be obtained. Similarly, the 4PLM calibration curve was the most reliable for niacin determination. We conclude that 4PLM should be used for calibration curve construction to improve the reliability of vitamin determination in foodstuffs using MBA.