Sample collections and laboratory analyses
Blood samples were collected into pre-cooled lithium–heparin (LH) or EDTA tubes (S-Monovette, Sarstedt Ltd, Leicester, UK). The EDTA sample was placed on ice immediately. The LH whole blood sample was measured for ionized calcium (iCa; pH 7.4 corrected values), haemoglobin (Hb) and pH within 10 min of collection (ABL77 blood gas analyser, Radiometer, Brønshøj, Denmark), and the remaining sample was then placed on ice. Plasma was separated within 1 h of collection in a refrigerated centrifuge at 1,800 g for 20 min, and aliquots were stored at −70 °C.
Urine was collected in acid-washed containers, mixed thoroughly. Non-acidified and acidified (concentrated hydrochloric acid (HCl), 10 ml/l, laboratory reagent grade, SG 1.18, Fisher Scientific) aliquots were taken and stored at −20 °C.
After completion of the study, plasma and urine samples were packed and shipped on dry ice to MRC Human Nutrition Research, Cambridge and subsequently stored at −80 °C until analysis.
LH plasma was used for the measurement of 1,25(OH)2D (radioimmunoassay IDS Ltd., Tyne and Wear, UK), 25-hydroxyvitamin D (25(OH)D), bone-specific alkaline phosphatase (BALP), osteocalcin (OC) (all chemiluminescent immunometric automated assays, CLIA; DiaSorin, Stillwater, MN, USA), β C-terminal cross-linked telopeptide of type 1 collagen (βCTX) (ELISA, IDS Ltd., Tyne & Wear, UK), cAMP (ELISA, R&D Systems, Abington, UK), total calcium (tCa), phosphate (P), creatinine (Cr) and albumin (Alb) (colorimetric methods, Kone Lab 20i clinical chemistry analyser platform, Kone Espoo, Finland). EDTA plasma was used for the measurement of PTH by immunoassay (Immulite, Siemens Healthcare Diagnostics Ltd, Camberley, UK).
Urinary (u) calcium (uCa), phosphate (uP) and creatinine (uCr) were measured in acidified urine (colorimetric methods, Kone Lab 20i, as above). Concentrations of uCa and uP were expressed as a ratio relative to uCr to adjust for urinary volume. Urinary cAMP was measured in non-acidified urine (ELISA, R&D Systems, as above).
All assays except PTH (between-assay coefficient of variation (CV), 4.7 %) were performed in duplicate. Assay performance was monitored using kit and in-house controls and under strict standardisation according to ISO 9001:2000. Quality assurance of 25(OH)D and 1,25(OH)
2D assays were performed as part of the Vitamin D External Quality Assessment Scheme (
www.deqas.org) and PTH assays as part of the National External Quality Assessment Scheme (
www.ukneqas.org.uk), and all were within accepted limits.
Within- and between-assay CVs for 1,25(OH)2D were 7.5 and 9.0 %. Cross-reactivity of the assay is 100 and 91 % for 1,25(OH)2D3 and 1,25(OH)2D2, respectively. Cross-reactivity of the 25(OH)D assay is 100 and 104 % for 25(OH)D3 and 25(OH)D2, respectively. Within- and between-assay CVs were 3.7 and 2.9, 1.6 and 3.6, and 3.8 and 4.0 % for 25(OH)D, BALP and OC, respectively. The within- and between-assay CVs for βCTX were 2.9 and 1.4 %. Within- and between-assay CVs for all Kone assays were <2 and <4 %, respectively. Within- and between-assay CVs for pcAMP and ucAMP were 6.0 and 4.8 %.
Statistics, data handling and derived variables
Data processing and statistics were performed using Microsoft Excel 2010 (Microsoft Corp., Seattle, WA, USA) and Linear Model Software in Data Desk 6.0 (Data Description, Inc., Ithaca, NY, USA). Non-normally distributed data were transformed to natural logs. Within- and between-group differences were analysed with ANOVA or ANCOVA as appropriate, with Scheffé post-hoc tests. The absolute change in each analyte was examined to investigate the response to calcium loading. The level of significance was set at P ≤ 0.05. Because of the small numbers of participants, P values ≤0.10 are also reported to indicate possible trends in the data.
The following variables were derived:
For the calculation of albumin-corrected calcium, different equations [
13,
16,
17] and group-specific equations, as based on regression analyses, were used because the albumin–calcium relationships may differ between populations and reproductive stages. Bland–Altman analyses [
18] showed no significant differences between the values calculated according to different methods. Further, regression analyses of the calcium–albumin relationship showed no significant group interaction (
P = 0.4). Therefore, the Payne equation [
13,
16] was used for further analyses. The dataset contained one outlier in Ca
e in the pregnant group as detected by standard procedures (Data Desk 6.0), and this value was excluded from analyses, but its inclusion made no material difference to the conclusions drawn.
We aimed to be able to detect a difference of 1.5 SD between groups with a sample size of n = 10 per group. A formal power calculation could not be performed for this study as the mean and distribution of most of the measured biochemical parameters are known to be markedly different from those in Western populations, and no data for the response to calcium loading are available in this population.