Sex and body mass index dependent associations between serum 25-hydroxyvitamin D and pulse pressure in middle-aged and older US adults

High pulse pressure (PP) is a valid indicator of arterial stiffness. Many studies have reported that vitamin D concentration is inversely associated with vascular stiffening. This association may differ depending on sex and body mass index (BMI). This study investigated the associations between vitamin D and PP and evaluated whether these associations differ according to sex and BMI, using data for individuals aged ≥ 50 years from the National Health and Nutrition Examination Survey (NHANES) 2007–2010. Serum 25-hydroxyvitamin D (25(OH)D) concentrations were used as biomarkers of vitamin D levels. High PP was defined as ≥ 60 mmHg. Total 25(OH)D concentrations were dose-dependently associated with lower odds ratios (ORs) for high PP (p-trend = 0.01), after controlling for sociodemographic, behavioral, and dietary factors. When stratified by sex, there was a dose-dependent association between total 25(OH)D concentrations and lower risk of high PP (p-trend < 0.001) in females, but not in males. When stratified by BMI, there was a dose-dependent association between total 25(OH)D concentrations and lower risk of high PP (p-trend < 0.001) in non-overweight subjects, but not in overweight subjects. Improving the vitamin D status could delay elevation of PP and vascular stiffening in female and non-overweight subjects.

. Participants characteristics by sex. Survey t-test for continuous variables and survey (Rao-Scott) χ 2 test for categorical variables were used. a Geometric mean (95% confidence intervals) are presented. www.nature.com/scientificreports/ Vitamin D and blood pressure in all subjects. Table 2 presents the results of multiple linear regression analyses for the associations of vitamin D with blood pressure. In all sequential models, higher total 25(OH)D and 25(OH)D 3 concentrations were dose-dependently associated with lower mean SBP and PP (p-trend < 0.05).
In the fully adjusted model (model B), subjects with sufficient total 25(OH)D (≥ 75 nmol/L) had a significant 2.95 (95% CIs 1.39, 4.50) mmHg lower SBP and 2.52 (95% CIs 1.10, 3.95) mmHg lower PP than did those with deficient total 25(OH)D (< 50 nmol/L). The results for the associations between 25(OH)D 3 concentrations and blood pressure were similar to the associations between total 25(OH)D concentrations and blood pressure.
Vitamin D and high PP in all subjects. Table 3 presents the results of multiple logistic regression analyses of the associations of vitamin D with high PP. In all sequential models, higher total 25(OH)D and 25(OH) D 3 concentrations were dose-dependently associated with lower odd ratios (ORs) for high PP (p-trend < 0.05).
In the fully adjusted models (model B), the ORs for high PP were 0.76 (95% CIs 0.61, 0.95) in the higher total 25(OH)D concentrations (≥ 75 nmol/L) group compared to the group with lower total 25(OH)D concentrations (< 50 nmol/L). The results for the associations between 25(OH)D 3 concentrations and high PP were similar to the associations between total 25(OH)D concentrations and high PP. These associations were similar after  Total 25(OH)D concentration and blood pressure stratified by sex and BMI. Figure 1 shows the results of multiple linear regression analyses for the association of total 25(OH)D levels with blood pressure stratified by sex and BMI in the fully adjusted models. When stratified by sex, the SBP decreased significantly in a dose-dependent manner with an increase in the total 25(OH)D concentration in males (p-trend = 0.02), and the SBP and PP decreased significantly in a dosedependent manner with an increase in the total 25(OH)D concentration in females (p-trend = 0.003 and 0.001). The males with sufficient total 25(OH)D concentrations had a significant 2.39 mmHg (95% CIs 0.05, 4.74) lower SBP than those with deficient total 25(OH)D concentrations, while the females with sufficient total 25(OH)D concentrations had a significant 3.19 mmHg (95% CIs 0.79, 5.59) lower SBP and 3.26 mmHg (95% CIs 1.13, 5.38) lower PP than those with deficient total 25(OH)D concentrations.
When stratified by BMI, significant dose-dependent decreases in SBP and PP (p-trend = 0.002 and 0.005) with total 25(OH)D concentrations were observed in non-overweight subjects, and significant dose-dependent decreases in SBP and PP (p-trend = 0.006 and 0.003) with total 25(OH)D concentrations were observed in overweight subjects. Non-overweight subjects with sufficient total 25(OH)D concentrations had a significant 6.16 mmHg (95% CIs 2.22, 10.1) lower SBP and 4.22 mmHg (95% CIs 1.33, 7.12) lower PP than those with deficient total 25(OH)D concentrations. Overweight subjects with sufficient total 25(OH)D concentrations had a significant 2.01mmHg (95% CIs 0.41, 3.60) lower SBP and 2.04 mmHg (95% CIs 0.64, 3.44) lower PP than those with deficient total 25(OH)D concentrations. Figure 2 shows the results of the multiple logistic regression analyses for the associations of total 25(OH)D concentration with high PP stratified by sex and BMI in the fully adjusted models. Dose-dependent associations between total 25(OH)D concentration and high PP (p-trend < 0.001 and < 0.001) were observed in females and in non-overweight subjects.

Total 25(OH)D concentration and high PP stratified by sex and BMI.
When stratified by sex, females had a significantly lower OR for high PP were 0.58 (95% CIs 0.42, 0.79) in subjects with sufficient total 25(OH)D concentrations (versus deficient subjects), whereas males had a higher OR for high PP were 1.10 (95% CIs 0.84, 1.45) in subjects with sufficient total 25(OH)D concentrations. The associations between total 25(OH)D concentration and high PP significantly differed according to sex (p for interaction = 0.005).
When stratified by BMI, non-overweight subjects had a significantly lower OR for high PP were 0.38 (95% CIs 0.22, 0.67) in subjects with sufficient total 25(OH)D concentrations (versus deficient subjects), whereas overweight subjects showed no significant difference in the risk for high PP according to the total 25(OH) Figure 1. Multiple linear regression of blood pressure with vitamin D concentrations in the subgroups by sex and BMI. p-trend < 0.05 considered significant. Adjusted for age, sex, race/ethnicity, education, season of examination, physical activity, alcohol consumption, smoking status, dietary covariates (intakes of total energy, potassium, calcium, magnesium, and sodium), height, weight, and diabetes.

Discussion
In a nationally representative sample of middle-aged and older adults, we found that vitamin D sufficiency is associated with a lower risk of high PP. When stratified by sex and BMI, females and non-overweight subjects presented significant associations between vitamin D sufficiency and lower risk of high PP, whereas the risk for high PP did not vary significantly with the 25(OH)D concentrations in males and overweight subjects. Aging is associated with a progressive increase in vascular stiffness and PP 2 . High PP has been attributed to elastic fiber degeneration and aortic dilation, which transfers the hemodynamic load to stiffer collagen 14 ; this condition is associated with an increased risk for cardiovascular disease and overall mortality 15,16 . Therefore, effective management of PP is very important to prevent cardiovascular disease and to ensure a healthy life in the elderly 6,7 . Several studies have reported that serum vitamin D concentrations are inversely associated with vascular stiffening 9,10 . It is possible that vitamin D improves general arterial health and increases arterial elasticity, which results in improvement in arterial compliance and decrease in PP 11 . Additionally, vitamin D may improve endothelial function by modulating inflammation-related cytokines and by increasing the expression of vascular endothelial growth factors 17 . One experimental study conducted by Pelham et al. investigated whether vitamin D deficiency leads to impaired function in the resistance artery by altering the phenotype of perivascular adipose tissue (PVAT). They reported that vitamin D deficiency enhances angiotensin II-induced vasoconstriction and impairs the normal ability of PVAT, whereas, the intake of supplementary vitamin D can protect against vascular dysfunction caused by impaired PVAT function due to hypoxia 18 . www.nature.com/scientificreports/ Although epidemiological evidence consistently shows that vitamin D affects musculoskeletal health 19 , there has been a recent debate regarding the role of vitamin D deficiency in blood pressure based on conflicting observational and intervention studies [20][21][22] . Inconsistent results of the association between vitamin D and PP, particularly in clinical studies, may have been caused by differences in the study population and design, i.e., differences in the definition of vitamin D deficiency, supplementary vitamin D dose, and/or serum 25(OH) D concentrations, along with differences in the study period [23][24][25][26] . Nevertheless, a clinical study by Raed et al. reported that high doses of vitamin D (> 18,000 IU) for women with vitamin D deficiency (< 20 ng/mL) were effective in reducing PP 27 , which may, in part, support our hypothesis.
The current study suggests that the role of vitamin D deficiency in PP may differ depending on sex and BMI. Accordingly, the following interpretations could be made. The differences in results between males and females observed in the current study may be due to the fact that males have a higher visceral fat percentage than females. An increase in the visceral fat component of the body leads to an increase in the secretion of free fatty acids 28 , which degrades vascular function 29 . Males also have a higher level of insulin resistance than females 28 . Increased insulin resistance is a major risk factor that lowers vascular elasticity 30 . In addition, the androgen levels in males are higher than that in females. A study reported that androgens may upregulate the renin-angiotensin system and may play an important role in the pathogenesis of hypertension, which could cause endothelial injury 31 . Moreover, a recent study conducted by Varakantham et al. suggested the sex-specific role of CYP24A1 genetic variants related to vitamin D metabolism and reported that the risk of hypertension is increased in men with the rs2762939 CC variant; however, this risk of hypertension is decreased in women with the rs2762939 CC variant. This difference is thought to be due to the different effects of CYP24A1 genetic variants on 25(OH)D and renin concentrations in a sex-dependent manner 32 . Overall, males are metabolically vulnerable to lower vascular elasticity compared to females; thus, the vascular elasticity status may not reflect the vitamin D activity.
Different findings according to weight may be explained by the fact that vitamin D is a fat-soluble vitamin, and is deposited in fat cells; this leads to a decrease in the bioavailability of vitamin D in obese individuals 33 . Obesity may reduce the elasticity of blood vessels by increasing oxidative stress and inflammation 34 . A recent study suggested that the effect variations of vitamin D are due to the differences in volume of distribution 35 . Obese individuals with a high volume of distribution require a higher dose of vitamin D compared to non-obese individuals 36 . Differences in the volume of distribution and in the bioavailability of vitamin D depending on body weight, and decrease in vascular elasticity due to obesity status, may possibly explain why an association between vitamin D concentrations and blood pressure parameters was not observed in overweight subjects in our study.
Several major strengths of this study should be noted. The primary strength was the use of a nationally representative survey of middle-aged and older adults in the US, and the inclusion of a large sample size. Vitamin D concentration was assessed using serum 25(OH)D and 25(OH)D 3 concentrations, which are accepted as biomarkers of the vitamin D status and vitamin D intake over a period of 4 years 37,38 ; the concentrations were measured using UHPLC-MS/MS, which is an accurate and precise method. In addition, our models included the adjustment for various potential confounders, including sociodemographic, behavioral, and dietary factors, and also included interaction terms to reflect participants' heterogeneity.
Several limitations of the results must also be considered. First, because of the cross-sectional design, we cannot infer temporal causality. Second, findings from the large proportion of non-Hispanic whites in the study population should be taken with caution when generalizing to other ethnic populations. Third, although we adjusted our model for a variety of covariates, there is a possibility of residual confounding. For example, parathyroid hormone is thought to be a factor that can affect the association between serum vitamin D concentration and blood pressure 39 , although we did not adjust for serum parathyroid hormone due to the lack of availability of pertinent data in the NHANES 2007-2010. Additionally, nitric oxide may be generated in the skin through ultraviolet (UV)-A radiation exposure 40 and regulated by physical exercise 41,42 , which may affect the reduction in vascular stiffness. However, we did not adjust for subcutaneous nitrogen compounds due to no available data in the NHANES. Fourth, it might be argued that our observed associations may differ depending on race/ethnicities and/or season of examination. Indeed, in our sensitivity analyses, the associations between total 25(OH)D and high PP were stronger in non-Hispanic Whites (see Supplemental Table 3) and during the warm season (see Supplemental Table 4). As solar UV radiation doses are high during the summer and there is an effect of solar UV radiation on blood pressure due to the release of nitric oxide, one needs to consider different seasons while interpreting our findings. Finally, although the NHANES includes dietary data for the first and the second day, we used dietary data for the first day alone because the method used to obtain dietary data on the second day (phone interview) differed from that used on the first day (face-to-face interview).
In summary, this study shows that vitamin D sufficiency is associated with a lower risk of high PP, possibly due to decreased arterial stiffness, in middle-aged and older adults. In particular, the association between vitamin D status and a lower risk of high PP was stronger among females and non-overweight subjects. Adequate sunlight exposure (5-30 min, 2-3 time/week) 43 , intake of foods naturally high in vitamin D (sun-dried mushrooms; egg yolks; oily fish such as salmon, sardines, mackerel; and meat) 44 and foods fortified with vitamin D (milk and cereals) 45 , and vitamin D supplementation are likely to improve the vitamin D status and delay the elevation of PP and vascular stiffening in middle-aged and older adults. www.nature.com/scientificreports/ populations, as well as those aged ≥ 80 years. The survey includes interviews based on demographic, socioeconomic, and dietary questionnaires, as well as medical interviews and physical examinations including medical, dental, and physiological measurements 46,47 . The current study used the two most recent NHANES cycles (2007-2010), which included data on serum vitamin D concentration. We did not include prior NHANES cycles because their vitamin D concentration data were obtained using a method (i.e., radio-immunoassay) that was different from the one used in the 2007-2010 NHANES cycles. Among the initially enrolled subjects (n=20686) in the NHANES 2007-2010, complete data on both interviews and examinations were available for 20015 participants. For the present study, we selected adults aged ≥ 50 years, which resulted in 5644 participants. We additionally excluded subjects lacking information on SBP, DBP (n=45), serum vitamin D concentration (n=601), and other covariates (n=433); a total of 4565 subjects (2316 males and 2249 females) were included in our analysis. Details are illustrated in Figure 3. The study protocol was approved by the National Centers for Health Statistics Research Ethics Review board (Continuation of Protocol #2005-06). Written informed consent was obtained from all the NHANES participants. The NHANES is a publicly available dataset. Additional details on study procedures, data documentation, and questionnaires are available elsewhere 46 Blood pressure. SBP and DBP were measured at the MECs using the Omron HEM 907 XL apparatus (Omron Healthcare, Kyoto, Japan). Blood pressure was measured after the participant had rested for 5 min in a seated position; the maximum inflation level was recorded. We computed the mean SBP and DBP of three consecutive SBP and DBP values. PP was calculated as the difference between SBP and DBP, and high PP was defined as a PP value of ≥ 60 mmHg 49 . Body mass index. BMI was computed as weight in kilograms divided by height in meters squared (kg/m 2 ).

Methods
We defined "overweight" as a BMI of ≥ 25 kg/m 2 and "non-overweight" as a BMI of < 25 kg/m 2 in main analysis.
Covariates. We considered participants' age, height, weight, race/ethnicity, education, season of examination, physical activity, alcohol consumption, smoking status, diabetes, and dietary intake (total energy, potas-  Statistical analysis. Data analyses were performed using the SAS survey procedure (version 9.4; SAS Institute Inc., Cary, NC) to account for the complex survey design and sample weights of the NHANES 2007-2010. All p-values were two-sided, and a p-value of < 0.05 was considered statistically significant.
In the descriptive analysis, we presented weighted mean and SE for continuous variables and weighted percentages (%) for categorical variables. We further used the survey t-test for evaluating continuous variables and the Rao-Scott χ 2 test for evaluating categorical variables. concentration and blood pressure were examined using multiple linear regression models (PROC SURVEYREG). We evaluated two sequential models: model A adjusted for age, sex, and race/ethnicity; and model B adjusted for model A covariates plus season of examination, physical activity, alcohol consumption, smoking status, dietary covariates (intakes of total energy, potassium, calcium, magnesium, and sodium), height, weight, and diabetes. In addition, the associations between 25(OH)D concentration and high PP were examined using multiple logistic regression models (PROC SURVEYLOGISTIC).
All regression and logistic analyses were evaluated in subgroups stratified by sex (males vs. females) or BMI (≥ 25 vs. < 25 kg/m 2 ). In addition, we estimated multiplicative interactions between serum 25(OH)D concentration and sex or BMI and computed the interaction p value using the Wald test.

Sensitivity analyses.
First, when we modeled subgroups stratified by BMI, we used an alternative classification of three levels of BMI (obese; BMI ≥ 30 kg/m 2 , overweight; 25-29.99 kg/m 2 , and non-overweight; BMI < 25 kg/m 2 ). Second, we performed sensitivity analysis on the association between total vitamin D levels and PP in subgroups stratified by race/ethnicity (non-Hispanic White, non-Hispanic Black, Mexican American, and other ethnicity) and the season of examination (November-April and May-October). Third, we conducted another sensitivity analysis after adjusting for additional dietary factors such as total fat, saturated fat, and cholesterol as potential confounders because these dietary factors are linked to blood pressure and are associated with elevated cardiovascular disease 52-54 .

Data availability
The datasets generated during and/or analyzed during the current study are available in the NHANES repository [https:// wwwn. cdc. gov/ nchs/ nhanes/ Defau lt. aspx].