Effect of Vitamin D Supplementation on Free and Total Vitamin D

A Comparison of Asians and Caucasians

Jaya Sujatha Gopal-Kothandapani; Lucy Faith Evans; Jennifer S. Walsh; Fatma Gossiel; Alan S. Rigby; Richard Eastell; Nick J. Bishop


Clin Endocrinol. 2019;90(1):222-231. 

In This Article

Materials and Methods

Study Design

This was an exploratory study to determine the size of the effect on serum free and total 25OHD of a 150 000 unit dose of vitamin D3, given to young adults from different ethnic groups (Figure 1).

Figure 1.

Study design

Study Participants

We recruited a cohort of sixty apparently healthy young adult men from two ethnic groups—White Caucasian (n = 30) or South-East Asian (n = 30) origin. Eligibility criteria included men aged between 18 and 25 years who were free from any condition affecting bone health, general nutrition, growth and glucose metabolism. Subjects with any chronic illness involving the liver and kidney, causing malabsorption, who used steroids, anticonvulsants, or vitamin D treatment as well supplementation or any medication that might affect calcium and vitamin D metabolism, were excluded. Recruitment took place within the University of Sheffield using a mixture of email, posters/leaflets and prelecture 2-minute talks. Subjects were students of the University of Sheffield, mostly medical and dental students. None of the study participants travelled to destinations where cutaneous vitamin D synthesis could have taken place during the study period. Whilst it was anticipated that gender should not impact on the outcome, the administration of a large dose of vitamin D3 might have had unexpected effects on an early-stage pregnancy, and thus, we excluded young women.

Demography and Auxology

At baseline, height (without shoes to next succeeding 1 mm by wall-mounted stadiometer [Holtain, Crymych]), weight (wearing vest and pants to nearest 0.1 kg by Marsden portable weighing scales, body mass index [BMI (kg/m2) and waist:hip circumference ratio (paper tape measure) were recorded in all subjects.

Dietary Calcium and Vitamin D Intake

A validated 101-item food frequency questionnaire (DIET-Q) was used to ascertain the dietary intake of calcium and vitamin D at baseline. Mean daily calcium (grams/d) and vitamin D (IU/d) intake were analysed using Q Builder (V4.0) nutritional software (Tinuviel software, Llanfechell, Anglesey UK).

Skin Type, Season and sun Exposure

Skin type was assessed using a 6-point Fitzpatrick scale in all subjects at baseline.[9] Fitzpatrick one skin type is fair and freckled; type six is very dark/black. The study was conducted in the UK during January and February to avoid sun exposure.


Baseline. Fasting blood samples were collected for serum total 25OHD, free 25OHD, calcium, phosphate, magnesium, albumin, alkaline phosphatase, PTH, DBP and DBP genotype.

Four weeks. All blood investigations (except DBP genotype) were repeated.

Urinary calcium: creatinine ratio. A fasting second void urine sample for calcium:creatinine ratio was collected at baseline, 1 and 4 weeks after vitamin D3 administration.

Vitamin D3 Dosing

A single dose of 150 000 IU of Vitamin D3 [6 mL of Invita D3 (Consiliant) 25 000 IU/mL oral solution] was administered under direct supervision. We chose this dose based on the work done by Oliveri B et al[10] where the authors demonstrated the safety of a single dose of 150 000 IU of vitamin D to maintain appropriate levels of 25OHD without causing hypercalcaemia or hypercalciuria.

Laboratory Methods

Serum total 25OHD. Serum total 25OHD levels were determined using an UPLC/Mass Spectrometer Semi-automated hexane extraction in the Acquity Ultra Performance LC/Quattro MS (Waters) analyser. Lower limit of detection for 25OHD2 was 6 nmol/L and for 25OHD3 3.5 nmol/L. The interassay coefficient of variation (CV) for 25OHD2 and 25OHD3 was 5.7% and 5.4%, respectively.

Serum free 25OHD. Free 25OHD levels were measured using an ELISA from Future Diagnostics Solutions. The interassay CV was 4.8%.

Serum vitamin D binding protein. D binding protein levels were measured using an ELISA from Genways Biotech Inc. The interassay CV was 5.8%.

Serum calcium, phosphate, albumin and alkaline phosphatase; urine calcium, creatinine. Measured using Micro Slide Technology Colorimetric/Rate by Reflectance Spectrophotometry in the Vitros 5, 1 FS System (Ortho Clinical Diagnostics) analyser. The interassay CVs were as follows: calcium (1.4%), phosphate (1.6%), albumin (2.9%), alkaline phosphatase (2.4%), urine calcium (1.7%) and urine creatinine (4.4%).

Intact Parathyroid hormone (PTH). Parathyroid hormone was measured using Immunoassay (Chemiluminescent Microparticle Immunoassay) in the Architect i 1000 System (Abbot) [PTH analytical sensitivity ≤1 ng/L].

DBP genotyping. A pyrosequencing assay was developed in house, using PSQ assay design software version 1.0.6 (Qiagen), to detect two single-nucleotide polymorphisms (SNPs), rs4588 and rs7041 in the DBP gene, that give rise to three common variants of DBP (a) Gc1f (b) Gc1s and (c) Gc2. All subjects were genotyped for six different haplotypes—Gc1f-1f, Gc1f-1s, Gc1f-2, Gc1s-1s, Gc1s-2 and Gc2-2, ranked here in order of 25OHD binding affinity, highest to lowest.

PCR and sequencing primers were as follows:




To ensure distinct pyrosequencing signals, ~10 ng of human genomic DNA was amplified by 47 × PCR cycles using OneTaq® 2X Master Mix with Standard Buffer (NEB). For each genotype determination, single-stranded DNA was purified from 5 μL of PCR products using PyroMark Q96 Vacuum Prep Workstation (Qiagen). PCR products were denatured to single-stranded DNA (ssDNA) and annealed with sequencing primers. Pyrosequencing was then performed on a PyroMark Q96 MD Instrument according to manufacturer's instructions (Qiagen). Nucleotide dispensation order was as follows: DBP rs4588 + rs7041 CATGTCACACACTG. SNP analysis was carried out using the SNP analysis software provided (Qiagen).

Calculated free and bioavailable 25OHD. Free and bioavailable 25OHD levels based on fixed affinity constant for the DBP genotype (Gc1f-1f) were calculated using the mathematical model described by Chun et al[11]

Sample Size

This was an exploratory study to determine effect size and variance; hence, no formal calculation of sample size was undertaken. We sought advice from the Yorkshire and Humber Research Design Service who suggested that a sample size of 30 per group was sufficient to undertake exploratory work of this nature.

Statistical Methods

Continuously distributed data were summarized by the median (25th/75th centiles), categorical data by n(%). Missing values are tabulated but not considered otherwise in our analysis.

Normality checks were carried out; data were generally normally or near-normally distributed, and hence, parametric tests were used to assess differences between groups.

To compare the groups prior to dosing, two sample t tests of baseline characteristics were performed and are reported in Table 1.

The main question of interest was whether ethnicity impacted on change in total and free 25OHD following dosing. Two sample t tests were carried out to compare the changes in total and free 25OHD between the groups. Analysis of variance was also used to determine any other statistical differences between the groups and adjust for covariates. Generalized linear models were used to determine statistical differences between response variables that had error distribution models other than normal, and to assess the interactive effect of ethnicity and DBP on change in free 25OHD.

Fisher's exact test was used to compare the categorical data. Graphical presentation was made by Box and Whisker plot. P values were used sparingly with an arbitrary threshold of 0.05 (two-tailed).

We performed all our analyses using Statistical Package for the Social Sciences version 22 (SPSS by IBM), Data Desk™ v6.2.1 and Stata v14.[12] For method comparison between directly measured serum free 25OHD and calculated free 25OHD, we used Medcalc for Windows, version 8.0 (Medcalc Software, Mariakerke, Belgium) to perform Bland-Altman analysis[13] and Passing-Bablok regression.[14]