Extension of an Ultra-High Performance Liquid Chromatography–MS/MS Method for the Determination of C3 Epimers of 25-Hydroxyl Derivatives of Vitamin D in Human Plasma

Abstract

Current vitamin D quantification methods do not account for 25-hydroxyl epimers, which can falsely increase concentrations and mask actual deficiencies. Previously, we developed an ultra-high performance liquid chromatography–tandem mass spectrometry method to measure 25(OH)D₃, 3-epi-25(OH)D₃ and 25(OH)D₂; here, we extended this method to include 3-epi-25(OH)D₂. Analytes were separated using a Shimadzu UPLC with a Kinetex F5 column (100 × 2.1 mm, 2.6 μm). The mobile phase contained 0.1% formic acid in methanol and water (70:30, v/v). The internal standard, deuterated 25(OH)D₃ and analytes were extracted with hexane. Detection was performed by a mass spectrometer equipped with a triple quadrupole after prior electrospray ionization. It demonstrated sufficient precision and spike recovery within and between days, with a coefficient of variation ≤15% and an error of determination ≤18%. The method exhibited linearity in the 2–100-ng/mL concentration range. The limits of quantification and limits of detection were 2 and 1 ng/mL, respectively. Extraction recoveries ranged from 70.05% to 97.13%. The matrix effect, carryover and dilution integrity were evaluated and met the FDA acceptance criteria. The stability of all metabolites in plasma was confirmed after 3 h of storage at room temperature and after three cycles of freezing at −80°C and thawing. Applying the method to clinical samples showed a high 25-hydroxyl epimer derived from vitamin D.

Introduction

Vitamin D plays a significant role in the human body, maintaining the homeostasis of calcium and phosphate ions (1). Its deficiency can contribute to the occurrence of many diseases, including those related to the cardiovascular system (2, 3), mental health (4) or the skeletal system (1). As a result of metabolic transformations of the prohormone vitamin D, hydroxyl derivatives are formed, including 25-hydroxyvitamin D (25(OH)D), the concentration of which is used to determine the vitamin D status in the body, and 1,25-dihydroxyvitamin D, which is biologically active (5). According to 25(OH)D₃ levels, individuals might be categorized according to vitamin D status (sufficient, ≥20 ng/mL; insufficient, 12 to <20; deficient, <12 ng/mL) (6).

Vitamin D metabolites can also undergo epimerization by 3-epimerase. Its activity has been detected in keratinocytes, parathyroid gland cells, osteoblasts, liver cells and colon cancer cells (7). Notably, 3-epimerase does not belong to the cytochrome P450 family, and while the gene encoding this enzyme is still unidentified, it seems to require NADPH as a cofactor (8). 25(OH)D metabolites differ from 3-epi-25(OH)D in the hydroxyl group position at the third carbon atom of ring A (Figure 1) (9, 10). Epimers have a lower binding affinity to vitamin D binding protein (DBP) compared to 25(OH)D and a lower affinity to the vitamin D nuclear receptor than 1,25(OH)₂D (10). This indicates their lower transcriptional and biological activity (10).

Figure 1.

The chemical structure of the analytes 25-hydroxyvitamin D₃ [25(OH)D₃], 25-hydroxyvitamin D₂ [25(OH)D₂], 3-epi-25-hydroxyvitamin D₃ [3-epi-25(OH)D₃] and 3-epi-25-hydroxyvitamin D₂ [3-epi-25(OH)D₂].

The metabolite 3-epi-25(OH)D₃ is the product of calcitriol epimerization via 3-epimerase (the orientation of the C-3 hydroxy group is changed from α to β) (11, 12) that may be accidentally measured by routine laboratory methods when determining 25(OH)D levels (13). According to Torugsa et al. (14), 3-epi-25(OH)D₃ could account for up to 32.15% of total serum 25(OH)D in humans (14). We observed in our previous study that this percentage could reach up to 50% in cardiac patients (2). These observations support Strathmann et al.’s (15) findings as they reported that 3% of adults would be misclassified as sufficient if 3-epi-25(OH)D₃ was included in the quantification of 25(OH)D₃.

Several studies have shown a positive correlation between circulating C-3 epimers and total 25-hydroxyvitamin D (2, 16–18). Because most routine assays do not chromatographically resolve these epimers, they can overestimate vitamin-D status by reporting combined rather than individual concentrations (2, 19, 20). Among available techniques, high performance liquid chromatography–tandem mass spectrometry (HPLC-MS/MS) uniquely separates each parent metabolite from its epimer and quantifies them in the low-nanogram range. Supplementary Table I summarizes published HPLC-MS/MS protocols for vitamin-D profiling in biological fluids (2, 19–28). These methods typically use solid-phase or liquid–liquid extraction with hexane or heptane, occasional derivatization to boost sensitivity and protein precipitation for clean-up; their lower limits of quantification (LLOQs) vary with the specific analyte panel.

Despite growing interest in 3-epi-25(OH)D₃, information on 3-epi-25(OH)D₂ and its plasma recovery remains scarce. Most assays that quantify the D₃ epimer do not include the D₂ analogue, and the only report that does so relies on a high-resolution quadrupole-Orbitrap platform that is seldom available in routine clinical laboratories (29). Moreover, adding 3-epi-25(OH)D₂ to an existing method would require complete re-validation and adjustment of HPLC-MS/MS parameters, limiting inter-laboratory transferability.

Building on our earlier work (2), the present study incorporates 3-epi-25(OH)D₂ into the previously validated ultra-high performance liquid chromatography–tandem mass spectrometry (UPLC-MS/MS) assay for 25(OH)D₂, 25(OH)D₃ and 3-epi-25(OH)D₃, in the presence of their precursors in plasma. We used the pentafluorophenyl (Kinetex F5, 100 × 2.1 mm) column, which delivered baseline resolution of these closely related isobars without derivatization or solid-phase clean-up required in other methods (20, 21, 25). Although some specialized protocols achieve sub-nanogram LLOQs (e.g., 0.013 ng/mL for 25(OH)D₃ (23)), they do so at the expense of additional metabolites, multi-step sample preparation or high-end Orbitrap detection (23, 24, 29).

We confirmed the reliability of the developed method through full validation according to FDA guidelines (30) and demonstrated its applicability to clinical plasma samples.

Materials and methods

Chemicals and reagents

The metabolites 25(OH)D₂, 3-epi-25(OH)D₂ and 3-epi-25(OH)D₃ (50 μg/mL in ethanol) were obtained from Sigma-Aldrich (Steinheim, Germany), while 25(OH)D₃ (as pure substance in solid state) and the internal standard d₆–25(OH)D₃ (50 μg/mL in ethanol) were obtained from Santa Cruz Biotechnology, Inc. (Dallas, TX, USA). Formic acid, human albumin and HPLC-grade hexane and isopropanol were obtained from Merck (Darmstadt, Germany). The LC-MS-grade methanol and water were purchased from J.T. Baker (New Jersey, USA) and Witko (Łodz, Poland), respectively, and they were used to prepare the mobile phase. Active charcoal was purchased from POCH (Gliwice, Poland). Blank plasma was obtained from the Regional Blood Donation and Hemotherapy Center, Poznan, Poland.

Chromatographic parameters for UPLC-MS/MS

The research utilized a Shimadzu UPLC Nexera system (Shimadzu Co., Kyoto, Japan) comprising a five-channel degasser (DGU-20A5) and a temperature-controlled autosampler (SIL-30 AC). The UPLC system was coupled with the LCMS-8030, a triple-quadrupole mass spectrometer, and the data were processed using the Labsolutions Series Workstation system (Shimadzu, Kyoto, Japan).

The separation of four analytes, 25(OH)D₃, 25(OH)D₂, 3-epi-25(OH)D₃ and 3-epi-25(OH)D₂, and the internal standard d6–25(OH)D₃, was performed using UPLC column Kinetex F5 (100 × 2.1 mm; 2.6 μm) attached to Phenomenex security guard cartridge (Torrance, CA, USA). The column temperature was kept at 25°C by Shimadzu model CTO-2 AC column oven. The mobile phase was a 0.1% formic acid solution in methanol and water (70:30, v/v) with flow rate adjustment (0.3 mL/min). The injected sample volume was 10 μL, and the autosampler temperature was 10°C.

Positive electrospray ionization mode introduced the eluent directly from the UPLC column to the MS interface. The electrospray needle voltage was set at 4.5 kV. The MS interface was adjusted with the following parameters: desolvation line temperature at 240°C, heat block temperature at 400°C and interface temperature at 350°C. Nitrogen served as both the drying and nebulizing gas, with 12- and 2-L/min flow rates, respectively. To minimize the carryover effect, a methanol–water mixture (80:20, v/v) was used to rinse the autosampler needle before and after sample aspiration. Analytes’ specific transitions were observed using the multiple reaction monitoring (MRM) mode. The most sensitive mass transitions were detected from m/z 401.5 to 383.25 for 25(OH)D₃ and 3-epi-25(OH)D₃, from m/z 413.5 to 395.3 for 25(OH)D₂ and 3-epi-25(OH)D₂, and from m/z 407.2 to 158.90 for d₆–25(OH)D₃. Table I shows the ionic transitions of the analyzed compounds in MRM mode.

Table I.

The Ionic Transitions of the Analyzed Compounds in MRM Mode

Metabolite	Transitions, m/z		Q1 [V]	CE [V]	Q3 [V]
	Precursor ion	Product ion
25(OH)D3	401.50	383.25 365.30 159.00	−15 −15 −15	−9 −11 −31	−26 −25 −27
3-epi-25(OH)D3	401.50	383.30 365.30 107.10	−15 −15 −12	−10 −11 −35	−27 −26 −17
25(OH)D2	413.50	395.30 54.95 83.10	−10 −16 −12	−10 −52 −26	−18 −20 −30
3-epi-25(OH)D2	413.50	395.30 54.95 83.10	−10 −10 −10	−10 −50 −25	−28 −21 −30
d6–25(OH)D3	407.30	389.10 371.20 158.90	−10 −10 −10	−9 −11 −25	−27 −18 −16

Preparation of matrix

Two matrices were prepared for the study: vitamin D–depleted charcoal-stripped plasma and a 5% albumin solution (discussed later in Method optimization subsection). According to Carter’s procedure (31), for each 10 mL of plasma, we added 1.4 g of activated charcoal, stirred for 24 h and centrifugated twice (15 min, 3000×g). The supernatant was filtered by syringe filters (pore size 0.22 μm, diameter 33 mm; Millipore, USA). Half a gram of albumin was dissolved in 10 mL of water to obtain a 5% albumin solution.

Stock and standard solutions

First, we used the manufacturer’s stock solution of 50 μg/mL internal standard (IS), and we prepared stock solutions of 10 μg/mL for the four analytes: 25(OH)D₃, 25(OH)D₂, 3-epi-25(OH)D₃ and 3-epi-25(OH)D₂. Next, the IS was diluted with methanol to obtain a concentration of 0.5 μg/mL. Similarly, the other analytes were diluted to achieve concentrations of 10, 20, 50, 75, 100, 200, 350, 500, 750 and 1000 ng/mL. Both stock and standard solutions were stored at −80°C until they were needed for use.

Preparation of samples for method validation and healthy volunteers analysis

In a 4-mL vial, 200 μL of 5% albumin solution, 20 μL of a standard solution of vitamin D metabolites at a specific concentration and 20 μL of a 0.5-μg/mL IS solution were added. The blank sample contained 200 μL of 5% albumin solution and 40 μL of methanol. For clinical samples, 200 μL of plasma, 20 μL of 0.5 μg/mL IS solution and 20 μL of methanol were transferred into a 4-mL vial. Then, 200 μL of water was added to all samples and vortexed for 10 s. The next step involved adding 400 μL of a methanol–isopropanol solution (80:20, v/v) to precipitate proteins and vortexing for 15 s. To this mixture, 1000 μL of hexane was added and shaken for 3 min. The mixture was centrifuged for 10 min at 3000×g and 20°C, resulting in two layers. The upper organic layer (~800 μL) was transferred to glass vials, and the remaining residue was treated with 1000 μL of hexane, repeating the extraction process. After transferring the second portion of the extract to a vial, the solvent was evaporated to dryness in a vacuum concentrator (Concentrator Plus; Eppendorf, USA) at 45°C for 30 min.

To the obtained dry residue, 200 μL of a methanol–water solution (80:20, v/v) was added, and the entire volume was transferred to UPLC vials with inserts. Subsequently, 10 μL was injected into the chromatographic column. The concentrations of vitamin D metabolites in the samples for the standard curve were as follows: 1, 2, 5, 7.5, 10, 20, 35, 50, 75 and 100 ng/mL.

Method validation

Linearity

Calibration curves for the concentrations of analytes, 25(OH)D₃, 25(OH)D₂, 3-epi-25(OH)D₃ and 3-epi-25(OH)D₂, in 5% albumin were established within the range of 1–100 ng/mL. The linearity of the calibration curve was plotted, and the correlation coefficient (r) was calculated. Student’s t-test measured the significance of the intercept, and the equations derived from the calibration curve (y = ax+b) were utilized to calculate the concentrations of analytes in the plasma samples from healthy individuals.

Limit of detection and lower limit of quantification

We identified these limits by observing the peak signal and corresponding signal-to-noise ratio (S/N). The lower limit of quantification (LLOQ), denoting the lowest quantifiable concentration of the analytes, was defined based on method recovery within the range of 80%–120% and precision not exceeding 20%. Similarly, the limit of detection, representing the smallest detectable analyte concentration with an S/N ratio >3:1, was established using this method.

Precision and spike recovery

Inter- and intra-day precision (percent coefficient of variation, CV%) was calculated as

The inter-day precision was calculated for the following concentrations: 1, 2, 5, 7.5, 10, 20, 35, 50, 75 and 100 ng/mL, and for the intra-day precision, the concentrations were 2, 5, 7.5, 20 and 75 ng/mL. The obtained CV values ≤20% for the LLOQ and ≤ 15% for other concentrations indicate the method’s precision.

The spike recovery data were used to estimate accuracy at the same concentrations as the method’s precision. The percent recovery was calculated from the following equation:

Extraction efficiency

Extraction efficiency is defined as the percentage of the analyte that can be isolated from the matrix using a specific extraction method. The extraction efficiency of vitamin D metabolites at concentrations of 7.5 and 75 ng/mL from plasma was investigated. For this purpose, two sets of samples were prepared. The first set (P1), consisting of six samples, was prepared using the standard procedure involving the extraction process described in Preparation of samples subsection.

The second set of samples (P2) was prepared by adding 20 μL of methanol and 20 μL of IS solution to 200 μL of plasma. The samples were subjected to extraction following the procedure described in Preparation of samples subsection, and the dried residue was dissolved in a solution containing 20 μL of a standard solution of vitamin D metabolites at concentrations of 7.5 or 75 ng/mL, along with 140 μL of methanol and 40 μL of water. In this way, six samples were prepared for each concentration. Recovery was calculated as (P1)/(P2)·100%.

Matrix effect

The matrix effect refers to the impact of endogenous compounds originating from the tested sample on the ionization of analytes in the MS detector. It was determined for concentrations of vitamin D metabolites equal to 7.5 and 75 ng/mL. For this purpose, plasma samples were collected from six different individuals and prepared according to the procedure described in Preparation of samples subsection to determine the endogenous concentrations of vitamin D derivatives. Then, for the P2 series, plasma samples from the same individual were prepared following the procedure described in Preparation of samples subsection. Additionally, three samples of the P3 series were prepared, which consisted of methanolic solutions of vitamin D metabolites at concentrations of 7.5 and 75 ng/mL. The matrix effect is measured by the normalized matrix factor (MF), which can be described using the formula (P2)/(P3)·100%. In the calculations for P2, the difference in peak surface areas of vitamin D metabolite peaks in individual samples injected with analytes and samples without analyte loading was considered due to their presence in the pure matrix.

The value of MF should fall within the range of 0.85 to 1.15. Values below unity suggest the presence of compounds inhibiting the analytical signal, while values above unity indicate the presence of compounds enhancing the signal.

Carryover

The carryover, or the carryover effect, is determined by injecting a sample without analyte after a sample containing analytes at the highest concentration from the standard curve. According to FDA criteria, the carryover effect should not exceed 20% of the LOQ.

Stability

The stability of vitamin D derivatives in samples was investigated after repeated freezing and thawing, as well as their short-term stability. Samples containing vitamin D derivatives at concentrations of 7.5 and 75 ng/mL were prepared in plasma.

The short-term stability was determined by storing the samples for the time necessary for their preparation before analysis. For this purpose, 20 μL of standard solutions of vitamin D metabolites at concentrations of 7.5 and 75 ng/mL were added to 200 μL of the matrix, and they were left at room temperature for 3 h. After this time, 20 μL of IS solution was added to the samples, and they were prepared according to the procedure described in Preparation of samples subsection. Each sample series was prepared in three replicates.

To determine the stability of analytes after repeated freezing and thawing, standard solutions of vitamin D metabolites at concentrations of 75 and 750 ng/mL were added to the matrix to achieve concentrations of 7.5 and 75 ng/mL. The samples were then frozen at −80°C. After thawing, three samples of 220 μL volume were taken for each concentration, and the remaining portion was frozen again. To the collected samples, 20 μL of IS solution was added, and they were prepared according to the procedure described in Preparation of samples subsection. The freezing and thawing process was repeated three times. The stability of the analyte was determined as the relative difference between the nominal concentration of the analyte and the determined concentration in samples stored under specific conditions. According to FDA guidelines, this difference should not exceed ±15% of the nominal concentration.

Effect of dilution

The impact of sample dilution on the concentration of the analyzed analytes was determined for the concentration of vitamin D derivatives in the matrix exceeding the highest concentration from the standard curve. For this purpose, a standard solution with a concentration of 200 ng/mL was added to a 5% albumin solution, and then, the sample was diluted twice with the matrix to achieve a 100-ng/mL concentration. In the subsequent step, the samples were extracted following the procedure described in Preparation of samples subsection. The accuracy and precision of these measurements should fall within the range of ±15%.

The application of the method to clinical samples

The utility of the developed method was assessed by analyzing human plasma samples obtained from healthy volunteers. The study included 35 humans, comprising 33 females and 2 males. Following established protocols, a 5-mL blood sample was taken from each volunteer in the morning and preserved in a tube with ethylenediaminetetraacetic acid. Plasma was separated by centrifuging the blood samples at 1620×g for 10 min. The resulting plasma samples were then stored at −80°C until analyzed, employing the UPLC-MS/MS method.

Results

Method optimization

The best separation of the analytes, 25(OH)D₃, 3-epi-25(OH)D₃, 25(OH)D₂, 3-epi-25(OH)D₂ and d₆–25(OH)D₃, was achieved using a Kinetex F5 column (100 × 2.1 mm; 2.6 μm) as the stationary phase and 0.1% formic acid in methanol and water as the mobile phase in an isocratic flow (70:30, v/v) at a rate of 0.3 mL/min. The column temperature was adjusted to 25°C upon various testing (Figure 2), and a 5% albumin solution was used to validate the method and prepare samples for the calibration curve and QCS.

Figure 2.

The impact of column temperature on metabolite peak separation.

UPLC-MS/MS analysis

Figure 3 shows chromatograms for a blank sample, LOQ concentration from the standard curve and clinical samples. The determined concentrations of analytes in the healthy volunteer are 12 ng/mL, 2.9 ng/mL, 7 ng/mL and 3.8 ng/mL for 25(OH)D₃, 3-epi-25(OH)D₃, 25(OH)D₂ and 3-epi-25(OH)D₂, respectively (Figure 3).

Figure 3.

Chromatograms for a blank sample, LOQ and volunteer’s samples for the concentrations of 12 ng/mL, 2.9 ng/mL, 7 ng/mL and 3.8 ng/mL for 25(OH)D₃, 3-epi-25(OH)D₃, 25(OH)D₂ and 3-epi-25(OH)D₂, respectively.

Method validation

The method was validated according to FDA criteria (32). The linearity of the method was tested in the concentration range of 1–100 ng/mL. A correlation coefficient “r” was close to unity (>0.998). However, for 1 ng/mL, precision was >20%, and thus, the linearity was determined to be for the concentration range of 2–100 ng/mL for each metabolite. The LOQ value of the validated method is 2 ng/mL for each analyzed vitamin D derivative with precision <20% (Table II).

Table II.

Intra-Day Precision of the Measurements (n = 5)

Nominal concentration [ng/mL]	Metabolite	Measured concentration Mean ± SD [ng/mL]	Precision [CV%]
2	25(OH)D3	1.9 ± 0.23	12.4
	3-epi-25(OH)D3	1.6 ± 0.20	12.4
	25(OH)D2	1.9 ± 0.19	10.0
	3-epi-25(OH)D2	1.7 ± 0.05	2.9
5	25(OH)D3	4.5 ± 0.50	11.1
	3-epi-25(OH)D3	5.5 ± 0.58	10.7
	25(OH)D2	5.2 ± 0.54	10.4
	3-epi-25(OH)D2	4.4 ± 0.09	2.1
7.5	25(OH)D3	6.8 ± 0.90	13.1
	3-epi-25(OH)D3	7.8 ± 0.93	12.0
	25(OH)D2	7.4 ± 1.05	14.2
	3-epi-25(OH)D2	6.9 ± 0.94	13.7
20	25(OH)D3	22.5 ± 3.16	14.0
	3-epi-25(OH)D3	18.9 ± 1.49	7.9
	25(OH)D2	18.9 ± 1.87	9.9
	3-epi-25(OH)D2	22.3 ± 1.53	6.9
75	25(OH)D3	68.1 ± 9.47	13.9
	3-epi-25(OH)D3	86 ± 4.83	5.6
	25(OH)D2	79.9 ± 6.72	8.4
	3-epi-25(OH)D2	83.7 ± 7.67	9.2

This method demonstrates appropriate precision for all tested vitamin D derivatives. In intra-day measurements, the lowest concentration (2 ng/mL) showed a precision, expressed as CV, of 12.4%. For higher concentrations (5, 7.5, 20 and 75 ng/mL), the CV ranged from 2.1% to 14.2% (Table II). Inter-day precision (evaluated over 12 days) was also acceptable, with CV values <15% for all metabolites (Table III).

Table III.

Inter-Day Precision of the Measurements (n = 12)

Nominal concentration [ng/mL]	Measured concentration Mean ± SD [ng/mL]	Precision [CV%]
25(OH)D3
2	1.9 ± 0.28	14.5
5	5.0 ± 0.51	10.1
7.5	7.8 ± 0.79	10.1
10	10.5 ± 0.72	6.9
20	19.9 ± 1.6	8.1
35	34.4 ± 2.3	6.8
50	51.4 ± 2.6	5.0
75	72.4 ± 4.2	5.8
100	101.2 ± 3.3	3.3
3-epi-25(OH)D3
2	2.2 ± 0.18	8.3
5	5.1 ± 0.51	10.1
7.5	7.5 ± 0.57	7.5
10	9.8 ± 1.1	10.8
20	20.9 ± 1.9	9.1
35	34.6 ± 2.2	6.5
50	50.2 ± 3.3	6.5
75	71.1 ± 5.1	7.2
100	102.4 ± 3.6	3.5
25(OH)D2
2	1.8 ± 0.10	5.5
5	5.0 ± 0.52	10.4
7.5	7.8 ± 0.84	10.7
10	9.9 ± 1.1	10.8
20	19.8 ± 1.6	8.0
35	32.9 ± 3.5	10.5
50	50.3 ± 4.1	8.1
75	73.8 ± 5.0	6.8
100	101.0 ± 3.4	3.4
3-epi-25(OH)D2
2	2.2 ± 0.27	12.6
5	5.1 ± 0.44	8.7
7.5	7.5 ± 0.96	12.9
10	9.9 ± 0.97	9.8
20	20.1 ± 1.6	8.1
35	33.5 ± 1.9	5.7
50	47.9 ± 3.7	7.7
75	76.1 ± 4.9	6.4
100	99.6 ± 5.4	5.4

Regarding recovery, intra-day results were as follows: 25(OH)D₃: 9.7% (7.3%–12.5%), 3-epi-25(OH)D₃: 10.2% (3.4%–18.1%), 25(OH)D₂: 4.0% (0.9%–6.5%) and 3-epi-25(OH)D₂: 12.3% (8.6%–17.0%). Inter-day recoveries were lower: 25(OH)D₃: 2.2% (0.3%–4.7%), 3-epi-25(OH)D₃: 2.7% (0.1%–7.8%), 25(OH)D₂: 2.8% (0.2%–9.3%) and 3-epi-25(OH)D₂: 2.4% (0.4%–8.3%), reported as mean (range).

The recovery of the analytes tested from plasma ranges from 83.8% to 97.1%, 70.1%–91.1%, 93.4%–97.1% and 79.4%–80.5% for 25(OH)D₃, 3-epi-25(OH)D₃, 25(OH)D₂ and 3-epi-25(OH)D₂, respectively. These values were repeatable, confirmed by the CV% in the range of 8.4%–14.5% (Table IV).

Table IV.

Extraction Efficiency of the Analyzed Analytes from Plasma and Influence of the Matrix on the Ionization of Vitamin D Metabolites

Nominal concentration [ng/mL]	Extraction efficiency			Matrix factor
	Metabolite	Mean ± SD [%]	Precision [CV%]	Matrix factor ± SD	Precision [CV%]
7.5	25(OH)D3	97.1 ± 12.9	13.3	0.94 ± 0.08	8.1
	3-epi-25(OH)D3	70.1 ± 8.3	11.9	0.87 ± 0.10	12.0
	25(OH)D2	97.1 ± 14.1	14.5	0.86 ± 0.09	10.9
	3-epi-25(OH)D2	79.4 ± 6.9	8.7	1.0 ± 0.15	14.9
75	25(OH)D3	83.8 ± 9.7	11.6	0.93 ± 0.09	9.6
	3-epi-25(OH)D3	91.1 ± 8.2	9.0	0.98 ± 0.02	1.5
	25(OH)D2	93.4 ± 7.9	8.4	0.99 ± 0.005	0.5
	3-epi-25(OH)D2	80.5 ± 6.9	8.5	0.91 ± 0.10	11.0

The carryover effect was also checked. The concentration of a non-analyte-laden sample injected after a 100-ng/mL sample is <20% of the LOQ. According to FDA requirements, there is no carryover effect; thus, there is no risk of overestimating low concentrations of vitamin D derivatives in samples injected after samples with high concentrations of analytes.

The next stage of the study included checking the short-term stability of vitamin D metabolites (Table V). The concentrations of the analytes in the plasma matrix were within the limits of a tolerable error of <15%. The stability of the analytes after three freezing and thawing processes was also investigated. After both the first and second cycles, the obtained concentrations of all vitamin D derivatives in plasma remained within the acceptable error range of <15%. After the third cycle of freezing and thawing of the samples, the accuracy of the determinations was in accordance with FDA guidelines and was ≤15% nominal concentration for all metabolites (Table V).

Table V.

Concentrations of Vitamin D Metabolites after 3 h of Storage at Room Temperature and during the Three Freezing and Thawing Cycles in Plasma^a

Nominal concentration [ng/mL]	Metabolite	Room temperature			Freeze and thaw cycles
					1st			2nd			3rd
		Mean ± SD [ng/mL]	Precision [CV%]	% nominal conc.	Mean ± SD [ng/mL]	Precision [CV%]	% nominal conc.	Mean ± SD [ng/mL]	Precision [CV%]	% nominal conc.	Mean ± SD [ng/mL]	Precision [CV%]	% nominal conc.
7.5	25(OH)D3	7.7 ± 0.36	4.6	102.3	8.3 ± 0.18	2.1	110.7	7.6 ± 0.7	8.9	101.6	6.7 ± 0.40	6.0	89.7
	3-epi-25(OH)D3	8.6 ± 0.80	9.3	114.7	8.3 ± 0.42	5.0	110.2	8.2 ± 0.56	6.9	109.4	8.6 ± 0.09	1.0	114.8
	25(OH)D2	7.5 ± 0.71	9.5	100.0	8.1 ± 0.66	8.1	108.4	8.5 ± 0.76	8.9	113.8	6.9 ± 0.75	10.8	92.3
	3-epi-25(OH)D2	8.33 ± 0.38	4.5	111.0	8.6 ± 0.13	1.5	114.4	8.5 ± 0.46	5.5	113.4	7.5 ± 1.01	13.5	100.4
75	25(OH)D3	81.3 ± 1.48	1.8	108.3	78.3 ± 2.2	2.8	104.4	76.3 ± 9.7	12.8	101.7	82.7 ± 6.0	7.3	110.3
	3-epi-25(OH)D3	66.6 ± 3.73	5.6	88.8	72.5 ± 3.6	4.9	96.7	80.4 ± 4.2	5.2	107.2	85.3 ± 10.7	12.5	113.8
	25(OH)D2	66.0 ± 5.07	7.7	88.0	65.1 ± 3.2	4.9	86.8	68.6 ± 6.0	8.8	91.4	79.1 ± 1.9	2.5	105.5
	3-epi-25(OH)D2	66.5 ± 3.35	5.0	88.6	83.5 ± 3.8	4.5	111.3	80.9 ± 4.9	6.0	107.9	72.7 ± 2.1	2.8	97.0

^a

Discussion

Method optimization

The first stage of the study focused on optimizing and refining the chromatographic separation and MS detection conditions outlined in our previous study (2). This included selecting suitable stationary and mobile phases. Samples were analyzed on columns: Kinetex C18 (100 × 2.1 mm; 2.6 μm), Kinetex F5 (100 × 2.1 mm; 2.6 μm) and Kinetex F5 (50 × 2.1 mm; 2.6 μm). Ultimately, Kinetex F5 (100 × 2.1 mm; 2.6 μm) was selected because it allowed the separation of all analytes. Other columns were ruled out due to poor peak shape or poor chromatographic separation. The utility of formic acid solutions in water and methanol at concentrations ranging from 0.1% to 0.3% and in water and acetonitrile at 0.1% concentration was investigated. Additionally, lithium acetate solutions in water and methanol at concentrations of 0.2 mM to 2 M were examined to enhance the method’s sensitivity. This compound forms conjugates with vitamin D derivatives characterized by high intensity in the MS detector (33). However, the inclusion of lithium acetate resulted in adduct formation with 25(OH)D₃ and 3-epi-25(OH)D₃, which led to increased signal noise and reduced the overall sensitivity of the method. The impact of column temperature on peak separation was also evaluated. Differences in chromatogram appearance are presented in Figure 2. It was observed that a lower temperature resulted in better separation of the analytes. A temperature of 25°C was applied since further lowering was impossible—in our case—due to extremely high pressure on the column.

Concerning the matrix, human plasma contains endogenous amounts of vitamin D metabolites. Our previous study proved that vitamin D–depleted charcoal-stripped plasma is suitable for the method validation (2). However, the matrix preparation procedure is time consuming, and some endogenous vitamin D metabolites may still be present, which could affect the precision and accuracy of the method. Therefore, the usefulness of water and 5% albumin solution was checked as surrogate matrices. However, water was not a suitable matrix for validating the method intended to analyze vitamin D concentrations due to the lack of effect on the ionization of analytes and the inability to bind vitamin D to protein, which occurs under physiological conditions. For further studies, a 5% albumin solution was used to validate the method and to prepare samples for the calibration curve and QCS. Such a solution was also proposed by Zhang et al. (22). Deuterated 25(OH)D₃ was used as an internal standard due to its similarity of structure and properties to the analytes to be determined.

UPLC-MS/MS analysis

In the developed UPLC-MS/MS method, the retention time of the internal standard is ~8 min. Due to the significant similarity of epimers to their precursors, 25(OH)D₃ and 3-epi-25(OH)D₃ were analyzed using the same characteristic mass transition. Similarly, 25(OH)D₂ and 3-epi-25(OH)D₂ were analyzed together, but the compounds had to be chromatographically separated in both cases. The retention time for 25(OH)D₃ is ~9 min, and its epimer elutes ~10 min. The retention time for 25(OH)D₂ is 9 min, while its epimer elutes ~10.5 min.

Method validation

LOQ value reported in this method is comparable to the 2.5-ng/mL value reported by Zhang et al. (22). In our previous work (2), we reported a LOQ value of 1 ng/mL for 25(OH)D₂, 25(OH)D₃ and 3-epi-25(OH)D₃ when used as a matrix of activated carbon shaken plasma. In this developed method, introducing an additional 3-epi-25(OH)D₂ analyte required a decrease in the column temperature and an increase in the analysis time, adversely affecting the peak height and resulting in worse sensitivity.

The extraction efficiency of 25(OH)D₃, 3-epi-25(OH)D₃ and 25(OH)D₂ were comparable to those described in the literature (2, 23). However, there is no literature data on the extraction efficiency of 3-epi-25(OH)D₂. The values obtained in this study indicate a good efficiency of the extraction process. There was no significant effect of endogenous compounds in the samples on the ionization of 25(OH)D₃, 3-epi-25(OH)D₃, 25(OH)D₂ and 3-epi-25(OH)D₂. This is confirmed by the matrix factor, which is in the range of 0.86–1.02 and has been determined with a precision of <15% (Table IV). Similar values of matrix factor were presented by Abouzid et al. (2) and Usoltseva et al. (27).

Concerning the stability, the concentrations of the analytes in the plasma matrix were within the limits of a tolerable error of <15%. This proves the stability of the studied compounds and the lack of their decomposition during their preparation for analysis. Shelf life at room temperature was also confirmed by Zhang et al. (22) for plasma 25(OH)D₂ and 25(OH)D₃ after 24 h of storage and the same metabolites in 5% bovine serum albumin for 6 h. 25(OH)D₃ was stable in samples stored for 4 h at room temperature and 4°C (Table V). The stability of the analytes after three freezing and thawing processes was also investigated. After the first and second cycles, the obtained concentrations of all vitamin D derivatives in plasma remained within the acceptable error range of <15%. Other authors reported similar results using different cycles, as described in the literature (2, 19, 21–23). Shelf life has also been confirmed for 25(OH)D₃, 3-epi-25(OH)D₃ and 25(OH)D₂ when stored for 3 months at −80°C and 24 h in an autosampler (2) and for 25(OH)D₃ and 25(OH)D₂ for 265 days at −20°C (23). It was found that the dilution of the sample did not significantly affect the concentration of the analytes tested. When the sample is diluted twice at a concentration of 200 ng/mL, the obtained concentrations are in the range of 100.2–106.7 ng/mL. CV and RE% are within the range of 5.3%–9.4% and 0.24%–6.7%, respectively (Table VI). The available literature also describes the possibility of a 5-fold dilution of samples containing 25(OH)D₂ and 25(OH)D₃ while maintaining appropriate precision and accuracy (22). Notably, we aimed to validate our method under conditions where plasma vitamin D levels may reach ~200 ng/mL. Such concentrations can occur in cases of hypervitaminosis D, granulomatous diseases such as sarcoidosis or tuberculosis, high-dose supplementation or therapeutic interventions, impaired metabolism or excretion, or certain pathological conditions that lead to elevated endogenous production of vitamin D metabolites (34–37). Although less common, these scenarios necessitate that the assay reliably quantify vitamin D metabolites even at higher concentrations.

Table VI.

Influence of Diluting the Sample Twice on Concentrations of Vitamin D Metabolites

Metabolite	Average concentration ± SD [ng/mL]	Precision [CV%]	% nominal conc.
25(OH)D3	106.7 ± 9.1	8.5	6.7
3-epi-25(OH)D3	102.2 ± 9.6	9.4	2.2
25(OH)D2	103.9 ± 5.5	5.3	3.9
3-epi-25(OH)D2	100.2 ± 6.6	6.5	0.2

Comparing our method with the previously published HPLC-MS/MS assays (Supplementary Table I), our workflow offers the broadest epimer-oriented panel attainable on a routine triple-quadrupole platform. Specifically, it quantifies 25(OH)D₂, 25(OH)D₃ and both C-3 epimers—3-epi-25(OH)D₃ and, uniquely among the methods compared, 3-epi-25(OH)D₂—after low-cost hexane extraction.

Only one method by Liebisch and Matysik reported the analysis of 3-epi-25(OH)D₂ (29). The authors utilize a Q Exactive Orbitrap mass spectrometer, which offers high mass resolution and accuracy. This configuration is advantageous for distinguishing structurally similar compounds and applications requiring high selectivity. In contrast, our assay achieves a broader analytical scope on far more accessible hardware, the Shimadzu LCMS-8030 triple-quadrupole system optimized for targeted quantification using MRM. Despite the difference in instrumentation, our method achieves a lower and uniform LLOQ of 2 ng/mL for all vitamin D metabolites, including 3-epi-25(OH)D₂. Their method reports slightly higher and variable LLOQs, ranging from 2.24 to 3 ng/mL.

Another distinction lies in the use of internal standards. While our method uses a single deuterated compound (d₆–25(OH)D₃), their protocol includes both d₆–25(OH)D₃ and d₆–25(OH)D₂. Using two internal standards may improve quantification accuracy for both D₂ and D₃ forms, but also increases procedural complexity.

The entire sample work-up has been simplified accordingly. Our method uses protein precipitation with methanol–isopropanol followed by dual hexane extraction, whereas their method applies iso-octane–chloroform extraction with butylated hydroxytoluene as an antioxidant. Our solvent choice avoids chlorinated waste, reduces per-sample volume by ~30% and eliminates the regulatory burden associated with chloroform disposal. Despite using a safer, greener solvent system, we inject only 10 μL of re-dissolved extract and reach the same low-nanogram sensitivity that the Orbitrap achieves with high-resolution targeted MS.

Chromatographically, a 100-mm pentafluorophenyl (F5) column operated isocratically at 70% methanol resolves the four metabolites with baseline separation of both epimer pairs. This fixed-composition mobile phase delivers excellent retention-time stability, avoids divert-valve plumbing and solvent-programming steps, and eases instrument maintenance. The Orbitrap method, in contrast, uses a 50-mm PFP column with a steep 5-min gradient; the shorter run time is attractive, but it comes at the price of more complex pump operation, higher flow rates and the need to divert the first and last portions of the gradient away from the expensive mass analyzer.

Calibration strategies have also differed; the Orbitrap workflow employs NIST-traceable serum calibrators that raise consumable costs but facilitate formal proficiency testing. In our method, despite the good accuracy observed during the spike recovery analysis, the evaluation was based on spiked plasma samples rather than certified reference materials due to regional unavailability (human plasma/serum containing certified levels of the metabolites), which should be considered in similar work in the future.

Finally, although their 96-well plate format delivers ready-made automation, our extraction can be transferred to solid-supported-liquid-extraction plates or robotic liquid handlers without changing the core chemistry, preserving throughput potential while keeping capital demands modest.

Analysis of clinical samples

The utility of the validated method was assessed among healthy volunteers (Figure 4). Concentrations of 25(OH)D₃ were within the range considered to be sufficient levels of vitamin D (20–30 ng/mL), with values of 26.7 ± 13 ng/mL. Epimers of 25(OH)D₃ accounted for a median of 10.8 (7.1–15.9)% of the total sum of 25(OH)D₃ and 3-epi-25(OH)D₃ metabolites. However, the content of 25(OH)D₂ epimers was higher, constituting 39.4 ± 18.5%. Literature data on epimer concentration analysis in the body are limited. According to available studies, the percentage content of 3-epi-25(OH)D₃ in maternal serum was, on average, 3.6%, and in umbilical cord serum, it was 4.2% (19). It was <5% in children and adults, and in infants, it was <10% (20). The values obtained using this method suggest a significant contribution of epimers in the determined concentrations of 25(OH)D, confirming the necessity of chromatographic separation of these compounds.

Figure 4.

Average concentrations of vitamin D hydroxy metabolites and their epimers in all healthy volunteers, with the percentage contribution of C3 epimers calculated only from the subset of subjects where both the main metabolite and its corresponding epimer were detected.

Conclusion

We extended our previously developed and validated UPLC-MS/MS method—originally used for quantifying 25(OH)D₃, 3-epi-25(OH)D₃ and 25(OH)D₂—to additionally include 3-epi-25(OH)D₂. This extended method has been thoroughly validated, demonstrating excellent linearity, accuracy, recovery and precision, along with low detection and quantitation limits. The approach provides rapid, sensitive, accurate and reproducible measurements, meeting the stringent requirements of bioanalytical assay validation.

The method was successfully applied to analyze 25(OH)D₃, 3-epi-25(OH)D₃, 25(OH)D₂ and 3-epi-25(OH)D₂ in healthy volunteers. The results revealed the significant contribution of epimers to the measured concentrations of 25(OH)D, shedding light on their presence despite the ambiguity surrounding their clinical significance.

Author contributions

Mohamed Abouzid (Conceptualization [equal], Data curation [equal], Formal analysis [equal], Funding acquisition [equal], Investigation [equal], Methodology [equal], Visualization [equal], Writing—original draft [equal], Writing—review & editing [equal], Validation), Julia Kerner (Data curation [equal], Formal analysis [equal], Investigation [equal], Software [equal], Validation [equal], Visualization, Validation), Aniceta Mikulska-Sauermann (Resources [equal], Writing—review & editing [equal]), Dorota Filipowicz (Resources [equal], Writing—review & editing [equal]), Matylda Resztak (Resources [equal], Writing—review & editing [equal]), Franciszek Glówka (Resources [equal], Writing—review & editing [equal]), Leonid Kagan (Supervision [equal], Writing—review & editing [equal]), and Marta Karazniewicz-Lada (Conceptualization [equal], Data curation [equal], Investigation [equal], Methodology [equal], Project administration [equal], Resources [equal], Software [equal], Supervision [equal], Validation [equal], Writing—review & editing [equal]). All the authors have accepted responsibility for the entire content of this submitted manuscript and approved the submission. All authors have read and agreed to the published version of the manuscript.

Conflict of interest. None declared.

Source of funding

The current research was financed from the statutory subsidy for young scientists awarded to Mohamed Abouzid by Doctoral School of the Poznan University of Medical Sciences, grant number (502-14-33064130-41319).

Ethical approval

The study obtained approval from the Bioethics Committee at the Poznan University of Medical Sciences (Approval No. 510/21). Every participant was granted their written consent.