Abstract
Context
The skeletal effects of vitamin D remain controversial and it is uncertain whether variation in serum 25-hydroxyvitamin D (25OHD) levels over time influences bone mineral density (BMD).
Objective
We evaluated longitudinal stability of serum 25OHD and associations with changes in BMD in participants aged 46-70 years at baseline.
Methods
We studied 3698 Busselton Healthy Ageing Study participants (2040 female) with serum 25OHD and dual-energy x-ray absorptiometry (DXA) BMD assessments at baseline and at ∼6 years follow-up. Restricted cubic splines were used to evaluate associations between changes in 25OHD and BMD.
Results
Mean season-corrected serum 25OHD was 81.3 ± 22.7 and 78.8 ± 23.1 nmol/L at baseline and 6 years, respectively, and showed moderate correlation (intraclass correlation coefficient: 0.724). Significant predictors of change in 25OHD concentration (Δ25OHD) included baseline 25OHD, change in body mass index and vitamin D supplementation at follow-up. Greater decline in serum 25OHD over time was associated with significantly greater reduction in BMD at total hip and femoral neck, but the magnitude of the differences was small (estimated differences 0.004 g/cm2 and 0.005-0.007 g/cm2, respectively, for lowest quartile of Δ25OHD compared with higher quartiles, adjusted for sex, baseline BMD, 25OHD, and demographics). No significant associations between Δ25OHD and lumbar spine BMD were observed. Increase in 25OHD levels was not associated with change in BMD.
Conclusions
In this predominantly vitamin D–replete middle-aged cohort, serum 25OHD showed moderate longitudinal stability. Declining serum 25OHD over time was associated with greater reduction in BMD at the total hip and femoral neck.
Keywords: 25-hydroxyvitamin D, longitudinal stability, bone mineral density, middle-aged adults, Busselton Healthy Ageing Study
Despite extensive research, the skeletal effects of insufficient vitamin D status in adults are still controversial. Observational studies, based on a single 25-hydroxyvitamin D (25OHD) measurement at baseline, have shown that circulating 25OHD levels are positively associated with bone mineral density (BMD) in middle to older aged adults, up to a threshold level (which in different studies varies widely from 50 nmol/L [1-3] to 90-100 nmol/L [4, 5]), above which the relationship is attenuated or plateaus. By contrast, in Mendelian randomization studies (designed to minimize bias from confounding), genetically higher 25OHD status has not been positively associated with BMD [6, 7]. However, the Mendelian randomization studies have important limitations: only linear relationships between predictor and outcome variables were evaluated to date, and the single nucleotide polymorphisms (SNPs) included explain only a small percentage of the variance in 25OHD levels (2%-10%) [8]. Randomized controlled trials of vitamin D supplementation, recruiting mostly vitamin D–replete participants, have shown minimal effect on bone mass [9]. The recent VITAL study of 2000 IU/day supplementation showed no significant effect overall on changes in dual-energy x-ray absorptiometry (DXA)-assessed BMD after 2 years in a subcohort of 771 participants (men ≥ 50 and women ≥ 55 years of age), although improved spine BMD (0.75%) and reduced bone loss (0.56%) at total hip were observed in a subgroup with low free 25OHD concentration at baseline (<14.2 pmol/L) [10]. Furthermore, no significant reduction in fracture risk was observed over 5.3 years follow-up in the whole cohort of 25 871 participants [11]. The ViDA trial of 100 000 IU/month supplementation in 452 older adults showed ∼0.5% reduction in bone loss at total hip and femoral neck after 2 years, and in a subgroup of participants with baseline serum 25OHD ≤ 30 nmol/L (n = 46), greater reduction in bone loss of 2% was observed at the spine and femoral neck [12].
Vitamin D status in individuals is not necessarily stable over time. Only a few studies have evaluated repeated measures of 25OHD in individuals at intervals ranging from 1 to 14 years. The results show moderate stability, with higher correlation coefficients for short intervals of 1 to 3 years (0.66-0.75) [13, 14] than for longer intervals such as 5 years (0.61) or 14 years (0.39-0.52) [15, 16]. There have been no previous studies examining changes in vitamin D status and changes in BMD in large cohorts, and the influence of variation in serum 25OHD levels over time on age-related bone loss is uncertain. Therefore, studies with more than one measure of 25OHD will advance understanding of the relationship between vitamin D and skeletal health, and also aid interpretation of previous studies in which 25OHD was measured only at baseline.
In the Busselton Healthy Ageing Study, a cohort study of a representative middle-aged population in Western Australia, serum 25OHD and DXA BMD were assessed in participants at baseline and after ∼6 years. The aims of this study were to evaluate: (1) stability of vitamin D status over a 6-year period; (2) predictors of change in serum 25OHD; and (3) associations between changes in 25OHD and changes in BMD.
Methods
Participants
The Busselton Healthy Ageing Study is a prospective study of noninstitutionalized “baby boomers” (born from 1946 to 1964) living in the Shire of Busselton, a coastal town in the southwest of Western Australia (latitude −33.6°) with a predominantly White population. The rationale and design of the study have been detailed previously [17]. All eligible residents listed on the electoral roll, for which registration is compulsory in Australia, have been invited to participate. Phase 1 of the study (baseline survey) was conducted between May 2010 and December 2015, with 5107 participants recruited (comprising ∼80% of those eligible). The follow-up (at 6 years) survey was conducted between March 2016 and January 2022 with 3888 of the original cohort (76%) attending. After excluding participants with missing data and outliers (17 without baseline 25OHD, 8 without baseline BMD, 149 without 25OHD at 6 years, 7 without BMD at 6 years, and 9 with baseline serum 25OHD > 200 nmol/L), 3698 participants (2040 females) with DXA and 25OHD measures at both time points were included in this analysis. The study received ethics approval from the University of Western Australia Human Research Ethics Committee (Number RA/4/1/2203) and written informed consent was obtained from each participant.
DXA Scans
At both time points, BMD (g/cm2) of anterior-posterior lumbar spine (L1-L4), femoral neck and total hip were measured by DXA scanning using a GE Lunar Prodigy Pro densitometer (Madison, WI, USA). Scans were analyzed using enCORE Version 16 software (GE Health), with the “copy” feature used to analyze follow-up scans; manual inspection of regions of interest and adjustment where necessary were made by 2 independent reviewers (K.Z. and M.H.) [18]. The DXA machine had annual servicing and calibration according to manufacturer's specifications. Calibration using a phantom was performed prior to each scanning session and the quality assurance plot showed no obvious shift in the phantom BMD values over the study period (coefficient of variation = 0.30%). The precision error was < 2.0% for each measured site at standard speed based on repeated scans in a random sample of 30 subjects.
Serum 25OHD Assessment
Fasting blood samples were collected at baseline and 6 years, and serum 25OHD was measured using the ARCHITECT 25-OH Vitamin D immunoassay (Abbott, Cat# 5P02, RRID: AB_2924942; Abbott Laboratories, Abbott Park, Illinois, USA). The inter-assay coefficient of variation was 4.0% at 57.5 nmol/L and 2.6% at 178.3 nmol/L. A total of 117 samples from baseline (randomly selected within 3 strata of 25OHD) were also assayed using isotope-dilution liquid chromatography/tandem mass spectrometry (LC-MS/MS) according to published methodology [19]. Both assays are accredited by the National Association of Testing Authorities, Australia (NATA), use calibrators aligned to reference material NIST 972 and are included in the Vitamin D External Quality Assessment Scheme (DEQAS). There was a strong correlation between the 2 techniques (r2 = 0.88) [18], with a tendency for the immunoassay to overestimate 25OHD at higher concentrations (for immunoassay 25OHD ≤110 nmol/L (N = 3267), difference in mean value 3.4 (95% CI 0.9, 6.0) nmol/L; for immunoassay 25OHD >110 nmol/L (N = 431), difference in mean value 30.8 (95% CI 23.3, 38.3) nmol/L). For each time point, we calculated season-corrected serum 25OHD values using a sinusoidal model fitted to serum 25OHD level, with week of attendance as the predictor variable. Residual values were added to the mean serum 25OHD level to obtain predicted mean annual values for each participant [20, 21].
Other Assessments
At baseline and 6 years, standing height and body weight were measured using standard anthropometric techniques with the participants lightly clothed and shoeless. Body mass index (BMI) was calculated as weight (kg)/height (m)2. Data on health history, medication use, alcohol consumption and smoking habit (current, never or previous smokers) were collected using a questionnaire [17]. Physical activity level was assessed using the International Physical Activity Questionnaire (IPAQ), and categorized as low, medium and high according to the IPAQ scoring protocol [22].
Data Analysis
Variables are presented as means ± SD for summary statistics, means (95% CI) or means ± standard error of mean (SEM) for estimated (adjusted) values unless otherwise stated. Comparisons of characteristics between males and females were made by Student t test and chi-squared test. Stability of both raw values of 25OHD and season-corrected 25OHD were assessed using Pearson correlation coefficient [23] and intraclass correlation coefficient. Linear regression analysis was used to evaluate predictors of change in season-corrected serum 25OHD over 6 years, with sex, baseline season-corrected 25OHD, age, BMI, and lifestyle factors, change in BMI from baseline to follow-up, and vitamin D supplementation at baseline and 6 years as independent variables; the semi-partial R2 for each predictor variable was calculated to estimate the proportion of the variance associated uniquely with each predictor.
Restricted cubic spline modeling, which allows the assessment of whether the relationship is nonlinear, was used to evaluate relationships between baseline serum 25OHD and BMD as well as between changes in serum 25OHD (Δ25OHD, 6 years − baseline) and changes in BMD (ΔBMD, 6 years − baseline) using R package “rms” with 3 knots (10th, 50th and 90th percentile) [24]. Covariates adjusted for in the models included sex, race, and baseline age, BMI, physical activity, smoking, alcohol intake, osteoporosis medication, vitamin D supplement, and dairy avoidance; models for ΔBMD additionally adjusted for measurement interval, baseline BMD, and baseline 25OHD. For lumbar spine BMD, as there was significant interaction between 25OHD variables and sex, male and female participants were analyzed separately. The least square means of baseline BMD or ΔBMD of each site with 95% CIs at mid-quartile levels of each of baseline 25OHD or Δ25OHD were estimated, and comparisons between mid-quartile means were made.
In addition, we analyzed by categories of vitamin D status. We defined low vitamin D status as 25OHD below 50 nmol/L, based on the sufficient level recommended by the U.S. Institute of Medicine Committee (50 nmol/L) [25], high vitamin D status as reaching the sufficient level recommended by the Endocrine Society (75 nmol/L) [26], and medium level as between 50 and 74.9 nmol/L. Tracking patterns were defined as: low (25OHD <50 nmol/L at both time points, n = 99); decreasing (moved to a lower category over time, n = 796); increasing (moved to a higher category over time, n = 538); medium (25OHD 50-74.9 nmol/L at both time points, n = 750); and high vitamin D status (25OHD ≥75 nmol/L at both time points, n = 1515). Comparisons between the 5 tracking pattern groups with regard to change in BMD over the 6 years were performed using a general linear model, adjusted for sex, race, measurement interval and baseline BMD, age, BMI, physical activity, smoking, alcohol intake, osteoporosis medication, dairy avoidance, and vitamin D supplement use. Statistical significance level was set at P < 0.05 (two-tailed). All analyses were performed using IBM SPSS (version 27, IBM, Chicago, IL, USA) and R (version 4.0.3, R Foundation for Statistical Computing, Vienna, Austria) [27].
Results
In total, 3698 participants (2040 female) were included in the analysis. Compared with members of the original cohort not included in this analysis (n = 1409), those included were not significantly different in baseline age (57.9 ± 5.7 vs 58.2 ± 6.0 years, P = 0.145), season-corrected 25OHD (81.4 ± 23.1 vs 80.6 ± 28.0 nmol/L, P = 0.355), or proportion female (55.2% vs 54.0%, P = 0.458), but had slightly lower BMI (28.0 ± 4.8 vs 28.6 ± 5.2 kg/m2, P < 0.001).
Table 1 shows participant characteristics. The mean measurement interval between baseline and follow-up was 6.2 ± 0.9 years, and mean serum season-corrected 25OHD decreased in male participants from baseline to 6 years (85.1 ± 22.8 vs 78.6 ± 21.9 nmol/L, P < 0.001) but did not change significantly in female participants (78.4 ± 23.0 vs 78.5 ± 24.6 nmol/L, P = 0.868). At baseline, 16.6% women and 5.6% men were taking vitamin D supplement, and at 6 years, 19.2% and 5.0%, respectively.
Table 1.
Characteristics of participants
| All (n = 3698) |
Male (n = 1658) |
Female (n = 2040) |
P valuea | |
|---|---|---|---|---|
| White, % | 99.1 | 99.3 | 98.9 | 0.261 |
| Measurement interval, years | 6.2 ± 0.9 | 6.2 ± 0.9 | 6.2 ± 0.9 | 0.977 |
| Baseline characteristics | ||||
| Age, years | 57.9 ± 5.7 | 58.1 ± 5.8 | 57.8 ± 5.7 | 0.122 |
| Measured serum 25OHD, nmol/L | 81.6 ± 24.4 | 85.1 ± 24.4 | 78.7 ± 23.9 | <0.001 |
| Season-corrected serum 25OHD, nmol/L | 81.4 ± 23.1 | 85.1 ± 22.8 | 78.4 ± 23.0 | <0.001 |
| Body mass index, kg/m2 | 28.0 ± 4.8 | 28.4 ± 4.1 | 27.7 ± 5.4 | <0.001 |
| Physical activity, % Low | ||||
| Low | 22.7 | 19.7 | 25.0 | <0.001 |
| Medium | 31.9 | 26.4 | 36.5 | |
| High | 45.4 | 53.9 | 38.5 | |
| Smoking current, % | 7.8 | 8.4 | 7.3 | 0.180 |
| Alcohol intake, % | ||||
| 0-7 glasses/week | 50.1 | 33.5 | 63.5 | <0.001 |
| 7-14 glasses/week | 22.3 | 20.4 | 23.9 | |
| >14 glasses/week | 27.6 | 46.1 | 12.6 | |
| Osteoporosis medication, % | 0.5 | 0.1 | 0.9 | <0.001 |
| Avoidance of dairy products, % | 1.6 | 1.1 | 2.1 | 0.030 |
| Vitamin D supplement, % | ||||
| Baseline | 11.7 | 5.6 | 16.6 | <0.001 |
| 6 years | 12.8 | 5.0 | 19.2 | <0.001 |
| Change over 6 years | ||||
| ΔMeasured serum 25OHD, nmol/L | −3.0 ± 24.2 | −6.3 ± 22.5 | −0.4 ± 25.2 | <0.001 |
| ΔSeason-corrected serum 25OHD, nmol/L | −2.9 ± 21.7 | −6.5 ± 18.8 | −0.1 ± 23.3 | <0.001 |
| ΔBody mass index, kg/m2 | 0.3 ± 2.0 | 0.2 ± 1.7 | 0.4 ± 2.2 | 0.034 |
Values are mean ± SD unless otherwise stated. Abbreviation: 25OHD, 25-hydroxyvitamin D.
P values obtained using Student t test or chi-square test for comparisons between males and females.
Serum 25OHD at baseline and 6 years showed moderate correlation, with Pearson correlation coefficient 0.505 for the raw values and 0.567 for season-corrected 25OHD, and the corresponding intraclass correlation coefficients of 0.671 and 0.724, respectively. The correlation was stronger in male than female participants, stronger in those not taking vitamin D supplementation at either visit than those who took supplements, slightly stronger for those with BMI < 30 kg/m2 compared with BMI ≥ 30 kg/m2, but was similar for those aged < 55 or ≥ 55 years (Table 2). Linear regression models showed that baseline 25OHD, male sex, baseline BMI, and change in BMI were negatively associated with change in 25OHD concentration over 6 years, whereas vitamin D supplement use at 6 years was associated with increasing 25OHD. Baseline 25OHD uniquely accounted for 18.0% of the variation, whereas vitamin D supplement use at 6 years accounted for 6.5% (Table 3). Variables examined in the model accounted for only 30% of the variation in Δ25OHD.
Table 2.
Pearson’s correlation coefficients and intraclass correlation coefficients for raw and season-corrected serum 25-hydroxyvitamin D values at baseline and 6 years
| Pearson's correlation coefficient | Intraclass correlation coefficient (95% CI) | |||
|---|---|---|---|---|
| Raw values | Season-corrected values | Raw values | Season-corrected values | |
| All (n = 3698) | 0.505 | 0.567 | 0.671 (0.649, 0.692) | 0.724 (0.705, 0.741) |
| Sex | ||||
| Male (n = 1658) | 0.552 | 0.648 | 0.710 (0.681, 0.737) | 0.786 (0.765, 0.806) |
| Female (n = 2040) | 0.477 | 0.520 | 0.645 (0.613, 0.675) | 0.683 (0.655, 0.710) |
| BMI category | ||||
| <30 kg/m2 (n = 2633) | 0.496 | 0.563 | 0.663 (0.637, 0.688) | 0.720 (0.698, 0.741) |
| ≥30 kg/m2 (n = 1065) | 0.480 | 0.526 | 0.648 (0.603, 0.688) | 0.688 (0.648, 0.723) |
| Age group | ||||
| <55 years (n = 1261) | 0.521 | 0.576 | 0.685 (0.648, 0.718) | 0.731 (0.700, 0.759) |
| ≥55 years (n = 2437) | 0.496 | 0.563 | 0.663 (0.636, 0.689) | 0.720 (0.697, 0.741) |
| Vitamin D supplementation at either time point | ||||
| No (n = 2935) | 0.566 | 0.644 | 0.723 (0.702, 0.742) | 0.784 (0.767, 0.799) |
| Yes (n = 763) | 0.350 | 0.385 | 0.517 (0.444, 0.581) | 0.555 (0.487, 0.614) |
Abbreviation: BMI, body mass index.
Table 3.
Predictors of change in season-corrected serum 25-hydroxyvitamin D levels over 6 years
| Regression coefficients (95% CI) | P value | R2a | |
|---|---|---|---|
| Season-correct serum 25OHD at baseline, nmol/L | −0.423 (−0.449, −0.396) | <0.001 | 0.180 |
| Sex—male | −2.021 (−3.352, −0.691) | 0.003 | 0.002 |
| Caucasian | −0.194 (−6.341, 5.953) | 0.951 | — |
| Baseline age, years | 0.039 (−0.064, 0.143) | 0.457 | — |
| Baseline BMI, kg/m2 | −0.295 (−0.423, −0.167) | <0.001 | 0.004 |
| Change in BMI over 6 years, kg/m2 | −1.256 (−1.558, −0.954) | <0.001 | 0.013 |
| Current smoking | −1.110 (−3.315, 1.096) | 0.324 | — |
| Alcohol intake 7-14 glasses/week | 0.424 (−1.081, 1.928) | 0.581 | — |
| Alcohol intake ≥14 glasses/week | 1.892 (0.382, 3.402) | 0.014 | 0.001 |
| Moderate level of physical activity | −0.590 (−2.207, 1.027) | 0.474 | — |
| High level of physical activity | −0.058 (−1.603, 1.487) | 0.941 | — |
| Vitamin D supplement use at baseline | −0.643 (−2.541, 1.255) | 0.507 | — |
| Vitamin D supplement use at 6 years | 17.276 (15.440, 19.112) | <0.001 | 0.065 |
| Total variance explained | 0.300 |
Abbreviation: 25OHD, 25-hydroxyvitamin D; BMI, body mass index.
R2 for individual predictor variable refers to semi-partial R2, which is the proportion of the variance associated uniquely with the predictor.
At baseline, restricted cubic spline analyses showed significant positive associations between serum 25OHD and BMD of total hip and femoral neck in all participants and with lumbar spine BMD in males (Fig. 1). When participants were divided into quartiles of baseline 25OHD, estimated total hip and femoral neck BMD were significantly higher in Q3 and Q4 compared with Q1 (by 0.019-0.028 g/cm2) and with Q2 (by 0.008-0.017 g/cm2); participants in Q2 also had significantly higher total hip and femoral neck BMD than those in Q1 (by 0.011-0.015 g/cm2). In males, estimated lumbar spine BMD was significantly higher for the highest quartile of baseline 25OHD compared with the lowest quartile (by 0.029 g/cm2) (Table 4).
Figure 1.
Associations between baseline serum 25-hydroxyvitamin D (25OHD) levels and bone mineral density (BMD) based on fitted restricted cubic spline regression with 3 knots, adjusted for sex, race, and baseline age, BMI, physical activity, smoking, alcohol intake, osteoporosis medication, vitamin D supplement, and dairy avoidance. Analysis for lumbar spine BMD was made for male and female participants separately due to significant interaction with sex. Gray shadow represents 95% CI.
Table 4.
Estimated mean baseline bone mineral density at mid-quartile levels of baseline serum 25-hydroxyvitamin D
| Q1 | Q2 | Q3 | Q4 | |
|---|---|---|---|---|
| All (n = 3698), mid-quartile 25OHD | 56.6 nmol/L | 72.1 nmol/L | 86.0 nmol/L | 107.6 nmol/L |
| Total hip BMD, g/cm2 | 1.083 (1.038, 1.127) | 1.094 (1.050, 1.138)a | 1.102 (1.057, 1.146)a,b | 1.111 (1.066, 1.155)a,b |
| Femoral neck BMD, g/cm2 | 0.982 (0.939, 1.025) | 0.997 (0.954, 1.039)a | 1.007 (0.964, 1.050)a,b | 1.010 (0.966, 1.053)a,b |
| Male (n = 1658), mid-quartile 25OHD | 60.8 nmol/L | 76.0 nmol/L | 89.4 nmol/L | 110.8 nmol/L |
| Lumbar spine BMD, g/cm2 | 1.161 (1.051, 1.271) | 1.175 (1.066, 1.285) | 1.175 (1.065, 1.284) | 1.190 (1.079, 1.301)a |
| Female (n = 2040), mid-quartile 25OHD | 53.9 nmol/L | 69.2 nmol/L | 83.1 nmol/L | 103.7 nmol/L |
| Lumbar spine BMD, g/cm2 | 1.187 (1.118, 1.257) | 1.196 (1.126, 1.265) | 1.209 (1.138, 1.280) | 1.199 (1.128, 1.271) |
Values are restricted cubic spline estimated least square mean (95% CI) for the mid-quartile levels of baseline serum 25-hydroxyvitamin D (25OHD); analysis for lumbar spine BMD was made for males and females separately due to significant interaction with sex.
Abbreviations: 25OHD, 25-hydroxyvitamin D; BMD, bone mineral density; BMI, body mass index.
P < 0.05 vs Q1.
P < 0.05 vs Q2, adjusted for sex, race, and baseline age, BMI, physical activity, smoking, alcohol intake, osteoporosis medication, vitamin D supplement, and dairy avoidance.
For changes over 6 years, restricted cubic spline analyses showed that greater decline in vitamin D status over time was associated with greater bone loss at total hip and femoral neck in all participants, whereas increase in 25OHD levels did not appear to be associated with change in BMD (Fig. 2). When analyzed by quartiles of Δ25OHD, participants in Q1 (with the greatest reduction in serum 25OHD) had significantly greater loss of BMD at the total hip (by 0.004 g/cm2) compared with Q2 and Q3 and femoral neck (by 0.005-0.007 g/cm2) compared with the 3 higher quartiles (Table 5). No significant associations with Δlumbar spine were observed (Fig. 2 and Table 5).
Figure 2.
Associations between changes in serum 25-hydroxyvitamin D (25OHD) levels and changes in bone mineral density (BMD) over 6 years based on fitted restricted cubic spline regression with 3 knots, adjusted for sex, race, measurement interval and baseline BMD, 25OHD, age, BMI, physical activity, smoking, alcohol intake, osteoporosis medication, vitamin D supplement, and dairy avoidance. Analysis for lumbar spine BMD was made for males and females separately due to significant interaction with sex. Gray shadow represents 95% CI.
Table 5.
Estimated mean changes in bone mineral density over 6 years at mid-quartile levels of changes in serum 25-hydroxyvitamin D
| Q1 | Q2 | Q3 | Q4 | |
|---|---|---|---|---|
| All (n = 3698), mid-quartile Δ25OHD | −23.7 nmol/L | −9.2 nmol/L | 1.4 nmol/L | 18.6 nmol/L |
| ΔTotal hip BMD, g/cm2 | −0.025 (−0.041, −0.010) | −0.021 (−0.037, −0.006)a | −0.021 (−0.036, −0.005)a | −0.023 (−0.039, −0.007) |
| ΔFemoral neck BMD, g/cm2 | −0.028 (−0.045, −0.011) | −0.023 (−0.040, −0.007)a | −0.021 (−0.038, −0.005)a | −0.023 (−0.040, −0.006)a |
| Male (n = 1658), mid-quartile Δ25OHD | −25.4 nmol/L | −11.0 nmol/L | −1.8 nmol/L | 11.6 nmol/L |
| ΔLumbar spine BMD, g/cm2 | 0.005 (−0.032, 0.042) | 0.008 (−0.029, 0.046) | 0.010 (−0.027, 0.047) | 0.007 (−0.030, 0.045) |
| Female (n = 2040), mid-quartile Δ25OHD | −22.3 nmol/L | −7.0 nmol/L | 4.7 nmol/L | 23.7 nmol/L |
| ΔLumbar spine BMD, g/cm2 | −0.044 (−0.071, −0.016) | −0.041 (−0.068, −0.014) | −0.042 (−0.069, −0.015) | −0.043 (−0.070, −0.015) |
Values are restricted cubic spline estimated least square mean (95% CI) for the mid-quartile levels of change in serum 25-hydroxyvitamin D over 6 years (Δ25OHD); analysis for lumbar spine BMD was made for males and females separately due to significant interaction with sex.
Abbreviations: 25OHD, 25-hydroxyvitamin D; BMD, bone mineral density; BMI, body mass index.
P < 0.05 vs Q1, adjusted for sex, race, measurement interval, and baseline BMD, 25OHD, age, BMI, physical activity, smoking, alcohol intake, osteoporosis medication, vitamin D supplement, and dairy avoidance.
Analyses by vitamin D status tracking pattern groups showed that participants in the decreasing vitamin D status group had significantly greater decline in femoral neck BMD compared with the increasing, medium, and consistently high vitamin D status tracking patterns (by 0.004-0.006 g/cm2) after adjustment for covariates (sex, measurement interval, and baseline BMD, age, BMI, vitamin D supplementation and lifestyle factors) (Table 6). No significant associations with Δlumbar spine or Δtotal hip BMD were observed (Table 6).
Table 6.
Estimated mean changes in bone mineral density over 6 years by vitamin D status tracking groups
| Low n = 99 |
Decreasing n = 796 |
Increasing n = 538 |
Medium n = 750 |
High n = 1515 |
P value | |
|---|---|---|---|---|---|---|
| ΔTotal hip BMD, g/cm2 | −0.027 ± 0.004 | −0.032 ± 0.002 | −0.028 ± 0.002 | −0.028 ± 0.002 | −0.027 ± 0.001 | 0.076 |
| ΔFemoral neck BMD, g/cm2 | −0.037 ± 0.005 | −0.036 ± 0.002a,b,c | −0.031 ± 0.002 | −0.030 ± 0.002 | −0.032 ± 0.001 | 0.052 |
| ΔLumbar spine BMD, g/cm2 | 0.001 ± 0.006 | −0.005 ± 0.002 | −0.006 ± 0.003 | −0.004 ± 0.002 | −0.004 ± 0.002 | 0.883 |
Values are estimated mean ± SEM. Vitamin D status tracking groups defined as: low (both serum 25-hydroxyvitamin D measures <50 nmol/L), decreasing (moved to lower category), increasing (moved to higher category), medium (both measures 50-74.9 nmol/L) or high (both measures ≥75 nmol/L).
Abbreviation: BMD, bone mineral density.
P < 0.05 vs Increasing.
P < 0.05 vs Medium.
P < 0.05 vs High, general linear model adjusted for sex, race, measurement interval, and baseline BMD, age, BMI, physical activity, smoking, alcohol intake, osteoporosis medication, vitamin D supplement, and dairy avoidance.
Discussion
In this community-based, middle-aged, predominantly vitamin D–replete Australian cohort, serum 25OHD measured 6 years apart showed moderate correlation, with an intraclass correlation coefficient of 0.724. At baseline, 25OHD positively associated with BMD of total hip and femoral neck in the cohort as a whole and with lumbar spine BMD in males. Over 6 years, greater decline in serum 25OHD was associated with higher bone loss at total hip and femoral neck, whereas increasing 25OHD levels were not associated with changes in BMD. To our knowledge, this is the first published study to examine relationships between 25OHD and BMD measured on more than one occasion in a large cohort.
With regard to longitudinal stability of 25OHD status, the correlation coefficients observed in the present study of samples collected ∼6 years apart (0.51 raw values, 0.57 seasonally-adjusted) are slightly lower than previously reported for samples collected 1 to 3 years apart (0.66-0.75) [13, 14] or at a 5-year interval (0.61) [15], but higher than reported for samples collected 14 years apart (0.39 for all participants, 0.52 for samples collected in the same season) [16], suggesting variability in circulating 25OHD concentrations is greater over longer intervals.
Significant predictors of change in serum 25OHD concentration over 6 years included baseline 25OHD, sex, baseline and change in BMI, and vitamin D supplement use at 6 years. With regard to BMI, there is a well-established inverse relationship between serum 25OHD and body weight and BMI, thought to be due to volume dilution and/or sequestration effect [28, 29]; in the present study, change in BMI uniquely explained more of the variation of change in 25OHD than baseline BMI. Vitamin D supplement at follow-up, but not at baseline, was a significant predictor of the change in 25OHD levels over 6 years, associated with 17 nmol/L increase in 25OHD concentration and uniquely explained 6.5% of the variation. The greater variability observed in women could reflect their greater use of vitamin D supplements than men. These results extend knowledge regarding the longitudinal stability of vitamin D status and may assist interpretation of previous studies in which 25OHD was measured only at baseline.
The positive relationship at baseline between serum 25OHD and BMD in this middle-aged cohort is consistent with positive associations reported in cross-sectional studies of other age groups and populations [1-5]. In the longitudinal analysis, greater decline in 25OHD levels over a 6-year period was associated with greater reductions in BMD at total hip and femoral neck but not lumbar spine. This could reflect a causal relationship in which low 25OHD results in increased bone turnover and reduced mineralization [30, 31] (which is reported to occur preferentially in cortical bone over trabecular bone [32]), or it could reflect other factors, such as poorer general health, affecting both 25OHD and BMD through other mechanisms. The magnitude of the observed BMD differences between quartiles of Δ25OHD was relatively small, and in line with the modest effects observed in vitamin D supplementation studies [9, 12]. Increasing 25OHD levels were not associated with changes in BMD, probably reflecting the vitamin D–replete status of the majority of participants. Overall, the data are consistent with the view that in vitamin D–replete individuals, variation in vitamin D status has a relatively minor influence on BMD.
Strengths of our study include the large sample size, community-based design, and the measurements of serum 25OHD and BMD at 2 time points to allow longitudinal analysis of vitamin D status and associations with bone loss in this representative population. The study also has limitations. First, it is observational in nature and although important confounding variables, including lifestyle factors, were accounted for in the restricted cubic spline analyses, it remains possible that the greater bone loss observed with greater decline in 25OHD was due to uncontrolled or residual confounding, as those with higher vitamin D status are more likely to be physically active and have healthier lifestyles [33], whereas inactivity is associated to both reduced vitamin D status and greater bone loss. Second, most participants were White, therefore generalizing the study findings to other ethnic groups should be exercised with caution. Third, the small number of participants with serum 25OHD below 50 nmol/L at either visit limited the statistical power of the analysis by vitamin D status tracking pattern groups and made it impractical to study the impact of overt vitamin D deficiency.
In conclusion, in this middle-aged, community-based cohort, serum 25OHD measured 6 years apart showed moderate longitudinal stability. Participants with greater decrease in circulating 25OHD levels had significantly greater decline in BMD than those with stable or increasing levels, but the magnitude of the BMD changes was small. We conclude that the impact of variation in vitamin D status on BMD in this predominantly vitamin D–replete population is relatively minor.
Acknowledgments
We thank the Western Australian Country Health Service—South West for core infrastructure support. In-kind equipment and consumable support for the Busselton Healthy Ageing Study is gratefully received from ResMed Science Centre, Stallergens, and BD Biosciences. Financial support has also been provided by School of Physiotherapy, Curtin University; Ear Science Institute Australia Inc.; Lions Eye Institute and Lions Hearing Foundation of Western Australia Inc. We thank the operational team in Busselton for participant recruitment and data collection, laboratory staff in Perth for specimen handling, processing and storage, and the community of Busselton for their ongoing support and participation.
Abbreviations
- 25OHD
25-hydroxyvitamin D
- BMD
bone mineral density
- BMI
body mass index
- DXA
dual-energy x-ray absorptiometry
- IPAQ
International Physical Activity Questionnaire
- SEM
standard error of the mean
Contributor Information
Kun Zhu, Email: kun.zhu@uwa.edu.au, Department of Endocrinology and Diabetes, Sir Charles Gairdner Hospital, Nedlands, WA 6009, Australia; Medical School, University of Western Australia, Crawley, WA 6009, Australia.
Michael Hunter, School of Population and Global Health, University of Western Australia, Crawley, WA 6009, Australia; Busselton Population Medical Research Institute, Busselton, WA 6280, Australia.
Jennie Hui, School of Population and Global Health, University of Western Australia, Crawley, WA 6009, Australia; Busselton Population Medical Research Institute, Busselton, WA 6280, Australia; Department of Clinical Biochemistry, PathWest Laboratory Medicine, Queen Elizabeth II Medical Centre, Nedlands, WA 6009, Australia.
Kevin Murray, School of Population and Global Health, University of Western Australia, Crawley, WA 6009, Australia.
Alan James, Medical School, University of Western Australia, Crawley, WA 6009, Australia; Department of Pulmonary Physiology and Sleep Medicine, Sir Charles Gairdner Hospital, Nedlands, WA 6009, Australia.
Ee Mun Lim, Department of Endocrinology and Diabetes, Sir Charles Gairdner Hospital, Nedlands, WA 6009, Australia; Department of Clinical Biochemistry, PathWest Laboratory Medicine, Queen Elizabeth II Medical Centre, Nedlands, WA 6009, Australia.
Brian R Cooke, Department of Clinical Biochemistry, PathWest Laboratory Medicine, Fiona Stanley Hospital, Murdoch, WA 6150, Australia.
John P Walsh, Department of Endocrinology and Diabetes, Sir Charles Gairdner Hospital, Nedlands, WA 6009, Australia; Medical School, University of Western Australia, Crawley, WA 6009, Australia.
Funding
Core funding of the Busselton Healthy Ageing Study is provided by the Federal Government of Australia, the Government of Western Australia (Department of Health and Department of Jobs, Tourism, Science and Innovation), the National Health and Medical Research Council (NHMRC) Equipment Grant scheme, the City of Busselton, and private donations to the Busselton Population Medical Research Institute. Charlies Foundation for Research for Research funded this project and the assay kits for 25-OH Vitamin D was donated by Abbott Australasia Pty Ltd.
Conflicts of Interest
The authors declare no conflicts of interest.
Data Availability
Restrictions apply to the availability of some or all data generated or analyzed during this study to preserve patient confidentiality or because they were used under license. The corresponding author will on request detail the restrictions and any conditions under which access to some data may be provided.
References
- 1. Michaelsson K, Wolk A, Byberg L, Mitchell A, Mallmin H, Melhus H. The seasonal importance of serum 25-hydroxyvitamin D for bone mineral density in older women. J Intern Med. 2017;281(2):167–178. [DOI] [PubMed] [Google Scholar]
- 2. Kuchuk NO, Pluijm SM, van Schoor NM, Looman CW, Smit JH, Lips P. Relationships of serum 25-hydroxyvitamin D to bone mineral density and serum parathyroid hormone and markers of bone turnover in older persons. J Clin Endocrinol Metab. 2009;94(4):1244–1250. [DOI] [PubMed] [Google Scholar]
- 3. Kuchuk NO, van Schoor NM, Pluijm SM, Chines A, Lips P. Vitamin D status, parathyroid function, bone turnover, and BMD in postmenopausal women with osteoporosis: global perspective. J Bone Miner Res. 2009;24(4):693–701. [DOI] [PubMed] [Google Scholar]
- 4. Bischoff-Ferrari HA, Dietrich T, Orav EJ, Dawson-Hughes B. Positive association between 25-hydroxy vitamin D levels and bone mineral density: a population-based study of younger and older adults. Am J Med. 2004;116(9):634–639. [DOI] [PubMed] [Google Scholar]
- 5. Zhu K, Lewis JR, Sim M, Prince RL. Low vitamin D status is associated with impaired bone quality and increased risk of fracture-related hospitalization in older Australian women. J Bone Miner Res. 2019;34(11):2019–2027. [DOI] [PubMed] [Google Scholar]
- 6. Larsson SC, Melhus H, Michaelsson K. Circulating serum 25-hydroxyvitamin D levels and bone mineral density: Mendelian randomization study. J Bone Miner Res. 2018;33(5):840–844. [DOI] [PubMed] [Google Scholar]
- 7. Li SS, Gao LH, Zhang XY, et al. Genetically low vitamin D levels, bone mineral density, and bone metabolism markers: a Mendelian randomisation study. Sci Rep. 2016;6:33202. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Bouillon R, Manousaki D, Rosen C, Trajanoska K, Rivadeneira F, Richards JB. The health effects of vitamin D supplementation: evidence from human studies. Nat Rev Endocrinol. 2022;18(2):96–110. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Reid IR, Bolland MJ, Grey A. Effects of vitamin D supplements on bone mineral density: a systematic review and meta-analysis. Lancet. 2014;383(9912):146–155. [DOI] [PubMed] [Google Scholar]
- 10. LeBoff MS, Chou SH, Murata EM, et al. Effects of supplemental vitamin D on bone health outcomes in women and men in the VITamin D and OmegA-3 TriaL (VITAL). J Bone Miner Res. 2020;35(5):883–893. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. LeBoff MS, Chou SH, Ratliff KA, et al. Supplemental vitamin D and incident fractures in midlife and older adults. N Engl J Med. 2022;387(4):299–309. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Reid IR, Horne AM, Mihov B, et al. Effect of monthly high-dose vitamin D on bone density in community-dwelling older adults substudy of a randomized controlled trial. J Intern Med. 2017;282(5):452–460. [DOI] [PubMed] [Google Scholar]
- 13. Major JM, Graubard BI, Dodd KW, et al. Variability and reproducibility of circulating vitamin D in a nationwide U. S. population. J Clin Endocrinol Metab. 2013;98(1):97–104. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. McKibben RA, Zhao D, Lutsey PL, et al. Factors associated with change in 25-hydroxyvitamin D levels over longitudinal follow-up in the ARIC study. J Clin Endocrinol Metab. 2016;101(1):33–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Meng JE, Hovey KM, Wactawski-Wende J, et al. Intraindividual variation in plasma 25-hydroxyvitamin D measures 5 years apart among postmenopausal women. Cancer Epidemiol Biomarkers Prev. 2012;21(6):916–924. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Jorde R, Sneve M, Hutchinson M, Emaus N, Figenschau Y, Grimnes G. Tracking of serum 25-hydroxyvitamin D levels during 14 years in a population-based study and during 12 months in an intervention study. Am J Epidemiol. 2010;171(8):903–908. [DOI] [PubMed] [Google Scholar]
- 17. James A, Hunter M, Straker L, et al. Rationale, design and methods for a community-based study of clustering and cumulative effects of chronic disease processes and their effects on ageing: the Busselton Healthy Ageing study. BMC Public Health. 2013;13(1):936. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Zhu K, Hunter M, James A, Lim EM, Cooke BR, Walsh JP. Relationship between visceral adipose tissue and bone mineral density in Australian baby boomers. Osteoporos Int. 2020;31(12):2439–2448. [DOI] [PubMed] [Google Scholar]
- 19. Cooke DJ, Cooke BR, Bell DA, Vasikaran SD, Glendenning P. 25-Hydroxyvitamin D C3-epimer is universally present in neonatal Western Australian samples but is unlikely to contribute to diagnostic misclassification. Ann Clin Biochem. 2016;53(Pt 5):593–598. [DOI] [PubMed] [Google Scholar]
- 20. Shoben AB, Kestenbaum B, Levin G, et al. Seasonal variation in 25-hydroxyvitamin D concentrations in the cardiovascular health study. Am J Epidemiol. 2011;174(12):1363–1372. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Sachs MC, Shoben A, Levin GP, et al. Estimating mean annual 25-hydroxyvitamin D concentrations from single measurements: the Multi-Ethnic Study of Atherosclerosis. Am J Clin Nutr. 2013;97(6):1243–1251. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. IPAQ . IPAQ scoring protocol. Accessed February 1, 2022. https://sites.google.com/site/theipaq/scoring-protocol
- 23. Twisk JW, Kemper HC, Mellenbergh GJ. Mathematical and analytical aspects of tracking. Epidemiol Rev. 1994;16(2):165–183. [DOI] [PubMed] [Google Scholar]
- 24. Harrell FE Jr. R Package ‘rms’. Vienna, Austria: R Foundation for Statistical Computing; 2022. https://cran.r-project.org/web/packages/rms/rms.pdf
- 25. Institute of Medicine . Dietary Reference Intakes for Calcium and Vitamin D. National Academy of Sciences, Institute of Medicine, 2010. [Google Scholar]
- 26. Holick MF, Binkley NC, Bischoff-Ferrari HA, et al. Evaluation, treatment, and prevention of vitamin D deficiency: an Endocrine Society clinical practice guideline. J Clin Endocrinol Metab. 2011;96(7):1911–1930. [DOI] [PubMed] [Google Scholar]
- 27. R Core Team . R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2020. https://www.R-project.org/ [Google Scholar]
- 28. Drincic AT, Armas LA, Van Diest EE, Heaney RP. Volumetric dilution, rather than sequestration best explains the low vitamin D status of obesity. Obesity (Silver Spring). 2012;20(7):1444–1448. [DOI] [PubMed] [Google Scholar]
- 29. Earthman CP, Beckman LM, Masodkar K, Sibley SD. The link between obesity and low circulating 25-hydroxyvitamin D concentrations: considerations and implications. Int J Obes (Lond). 2012;36(3):387–396. [DOI] [PubMed] [Google Scholar]
- 30. Lips P. Vitamin D deficiency and secondary hyperparathyroidism in the elderly: consequences for bone loss and fractures and therapeutic implications. Endocr Rev. 2001;22(4):477–501. [DOI] [PubMed] [Google Scholar]
- 31. Priemel M, von Domarus C, Klatte TO, et al. Bone mineralization defects and vitamin D deficiency: histomorphometric analysis of iliac crest bone biopsies and circulating 25-hydroxyvitamin D in 675 patients. J Bone Miner Res. 2010;25(2):305–312. [DOI] [PubMed] [Google Scholar]
- 32. Lauretani F, Bandinelli S, Russo CR, et al. Correlates of bone quality in older persons. Bone. 2006;39(4):915–921. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Malacova E, Cheang PR, Dunlop E, et al. Prevalence and predictors of vitamin D deficiency in a nationally representative sample of adults participating in the 2011-2013 Australian Health Survey. Br J Nutr. 2019;121(8):894–904. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Restrictions apply to the availability of some or all data generated or analyzed during this study to preserve patient confidentiality or because they were used under license. The corresponding author will on request detail the restrictions and any conditions under which access to some data may be provided.