A Sensitive Quantitative Method for Fat Soluble Vitamins in Dietary Supplements by LC/MS/MS



Fat soluble vitamins (FSV) are required for a wide variety of physiological functions. Deficiencies in FSV have been associated with increased risk of cancer type II diabetes mellitus, and a number of immune system disorders. Therefore, quantitative FSV assays are critical for addressing these issues. Currently, 59 different quantitative FSV assays have been published by AOAC and USP. Most of these assays use liquid-liquid extraction (LLE) coupled with saponification, which is time and cost consuming. We present a non-LLE assay for the analysis of FSV (see Table 1), and compare this non-LLE method with LLE assay using samples with different matrices.


Standard Preparation:

Standard is prepared in DMSO:water:isopropanol with the range of 5.00-500 ng/mL. About 2.00 grams of sample was extracted by DMSO:water:isopropanol. The extracted sample was cleaned by 0.20 μM filtration before analysis.

UPLC-MS Conditions:

UPLC system: Nexera UPLC system including SIL-30AC auto-sampler, controller, column heater and binary pump (SHIMADZU)

Column: 100×2.1 mm, 2.6 μm Kinetex F5 (Phenomenex)

Mobile Phase A: Formic acid and water

Mobile Phase B: Acetonitrile

Flow rate: 1.00 mL/min

Pump Gradient Cycle time: 10.0 minutes

MS detector: Triple Quadrupole 5500 MS (AB Sciex)

Results and Discussion

During the method development, different columns and mobile phases were investigated to separate D2 from D3 and K2-MQ4 from K2-MQ7. The Kinetex F5 column with acidic mobile phases provided sufficient resolution for these FSV. The method was successfully developed over the range of 5.00-500 ng/mL. The specificity experiment showed that there was no significant contribution between analytes/IS and no visible interference peaks showed in blank diluent at the expected retention time ( Fig. 2). The LLOQ has sufficient sensitivity S/N> 10) Fig. 2). System suitability consisted of six replicate injections of the middle standard solution and was injected before sample analysis, and RSD was ≤7.55% ( Table 2). The response linearity study revealed that quadratic regression with 1/X weighing factor provides the best fit, and the correlation coefficient r2 is ≥0.995 Fig. 3 and Table 3). In order to develop a non-LLE extraction method, different extraction solvents have been investigated and the result indicated that DMSO:water:IPA provided highest recovery. Comparison to the LLE method revealed that most matrices can be efficiently extracted by non-LLE method Fig. 4). However, several matrices like some vitamin mineral tablet and protein shake still need a LLE/saponification to minimize potential interferences ( Fig. 5). Dyad Labs analyzes samples with a non-LLE method unless matrix interference is observed, in which the LLE extraction will be performed.


The specific, fast and high-throughput non-LLE LC/MS/MS assay for quantitation of FSV has been successfully developed and applied into different matrices.