The Effect of Obesity on the Relationship Between Serum Parathyroid Hormone and 25-Hydroxyvitamin D in Women

Abstract

Context:

Obesity is associated with lower serum concentrations of 25-hydroxyvitamin D (25OHD) and higher intact PTH. The threshold of 25OHD needed to maximally suppress intact PTH has been suggested as a marker of optimal vitamin D status.

Objective:

In this study, we hypothesized that whereas the obese have a higher serum PTH and lower 25OHD, suppression of serum PTH by 25OHD would be independent of body weight.

Design, Setting, and Participants:

We performed a retrospective analysis on 383 women (ages 24–75 y) with a wide range of body weights (43–185 kg) who were stabilized to 1–1.2 g calcium/d for 1 month before blood draw. Body composition, serum PTH, 25OHD, calcium, and creatinine were measured. Locally weighted regression and smoothing scatterplots were used to depict the association between serum PTH and 25OHD. A nonlinear exponential model determined the point for near maximal suppression of PTH by 25OHD.

Results:

The point for near maximal suppression of PTH by 25OHD for all women (body mass index, 31.4 ± 7.7 kg/m²) occurred at a 25OHD concentration of 21.7 ng/mL (95% confidence interval, 28–48 ng/mL). No point of maximal suppression was found for nonobese women, yet in the obese women (n = 207; body mass index, >30 kg/m²) suppression of PTH occurred at a 25OHD concentration of 11.1 ng/mL (95% confidence interval, 4.7–17.5 ng/mL).

Conclusions:

These results suggest that if PTH is suppressed at a lower serum 25OHD in the obese compared to the entire population, the lower average 25OHD concentrations in the obese may not have the same physiological significance as in the general population.

There are a large number of studies suggesting that there is an altered vitamin D endocrine system in obesity. Studies show that obesity, and specifically body fat content, is inversely associated with 25-hydroxyvitamin D (25OHD) and is positively associated with PTH concentrations (1–3). Vitamin D insufficiency in obesity may be due to the decreased bioavailability of vitamin D₃ stored in cutaneous tissues and because of its deposition in the adipose tissue (4–6). Lower concentrations of 25OHD have been shown to be associated with multiple health outcomes such as increased bone loss, risk of fracture, muscle weakness and falls, and insulin resistance (7, 8); although the implications in the obese population may be the same, they are not clear.

The level at which serum concentrations of PTH are suppressed by serum 25OHD is used as a marker for defining vitamin D sufficiency, especially for bone-related outcomes. The classic inverse relationship between PTH and 25OHD concentrations suggests that the point for maximal PTH suppression by 25OHD occurs at approximately 30 ng/mL (75 nmol/L) (9). However, more recently the suppression value has been variable in different studies and is attributed to many factors (10, 11). A 3-phase model suggests this might be explained by 2 thresholds of 25OHD rather than a single inflection point (12) and is complicated by at least a few factors. Given the altered vitamin D system in obesity together with high serum PTH concentrations, our goal was to examine whether there is a different relationship between intact PTH and serum 25OHD in obese women, and to examine the point of maximal suppression of PTH by 25OHD in obese women compared to normal/overweight women.

Subjects and Methods

Participants

We performed a retrospective review of 383 healthy women recruited for various clinical trials at Rutgers University between 1994 and 2010. Subjects were given calcium citrate (200 mg elemental calcium per tablet [Citracal]; Bayer Healthcare LLC, Morristown, New Jersey) as needed to achieve an intake of 1–1.2 g/d for approximately 1 month before blood draw. Premenopausal women who were menstruating monthly were included, and postmenopausal women were at least 2 years since their last menstruation. Exclusion criteria were history of prior osteoporotic fracture, medications known to affect bone metabolism, malabsorption, untreated thyroid disease, history of metabolic bone disease, diabetes, end-stage organ disease, myocardial infarction within 6 months of screening visit, uncontrolled hypertension, cancer within the past 5 years, and use of any illicit drugs. The study was conducted in accordance with the guidelines in The Declaration of Helsinki and was approved by the Rutgers University Institutional Review Board, and all participants signed an informed consent form.

Measurements

Weight and height were measured using a balance beam scale and stadiometer, respectively (Detecto, Webb City, Missouri). Dual-energy x-ray absorptiometry (Lunar Prodigy Advanced; GE Lunar Corp, Madison, Wisconsin; coefficient of variation, <1% for all sites) scans using enCORE 2004 software (version 8.10.027; GE Lunar Corp) were performed to assess total soft tissues. Calcium intake was estimated using 3-day food records and analyzed with nutrient analysis software (FoodWorks version 10.1; FoodWorks, Long Valley, New Jersey).

Serum analysis

Blood samples were collected after an overnight fast, and serum was stored at −80°C until further analysis. Concentrations of intact PTH and 25OHD were analyzed by RIA, using both internal and external standards. The coefficient of variation was < 6.8% for PTH (Diagnostic Systems Laboratories, Webster, Texas; Scantibodies Laboratory, Inc, Santee, California) and < 12.5% for 25OHD (DiaSorin, Stillwater, Minnesota). Our laboratory participates in vitamin D external quality assessment scheme (DEQAS) to monitor the performance of the RIA used for assessment of 25OHD. Estimated glomerular filtration rate (eGFR) was calculated based on serum creatinine and calculated body surface area (BSA): eGFR = GFR (modification of diet in renal disease [MDRD]) * BSA/1.73 m². Serum was analyzed for calcium using Arsenazo reagent (Pointe Scientific, Canton, Michigan).

Statistical methods

Locally weighted regression and smoothing scatterplots (LOESS) (13) were used to depict the association between PTH and 25OHD. A nonlinear exponential model was used to model the relationship between PTH and 25OHD with the equation: PTH = A + B × exp (C × 25OHD). The near maximal PTH suppression by 25OHD, or the value after which the exponential curve reaches a plateau, was estimated by −3/C (14). This exponential regression model is able to control for the asymptotic behavior of intact PTH in response to a wide range of 25OHD concentrations and estimates confidence intervals (CIs) (11). Standard error and normal distribution-based CIs were computed by the delta method, and STATA 12.1 statistical software (StataCorp, College Station, Texas) was used to carry out these statistical analyses. Multiple linear regression was used to assess the relative influence of independent variables on PTH and 25OHD. One-way ANOVA and simple linear correlations were used to assess the relationship between PTH and 25OHD using SAS Statistical Package (version 9.2; SAS Institute, Cary, North Carolina). Women were categorized as obese if body mass index (BMI) was greater than 30 kg/m². Data in tables are expressed as mean ± SD. P value < .05 was considered significant.

Results

Characteristics of the women are shown in Table 1. Fifty-nine percent of women were postmenopausal. Women were primarily Caucasian, with 92% Caucasian in the obese group and 87% in the nonobese group. As expected, there were higher serum concentrations of PTH and lower 25OHD in the obese subjects compared to leaner subjects (P < .001). Although age differed between the obese and nonobese groups (P < .05), the mean age differed by only 2 years. The eGFR was higher (P < .001) in the obese group than the nonobese group (Table 1).

Table 1.

Characteristics of All Women, Nonobese Women, and Obese Women

Variable	All Women	BMI < 30 kg/m2	BMI > 30 kg/m2	P Valuea
n	383	207	176
Age, y	53 ± 10.6	54.4 ± 10.2	52.2 ± 11.1	.037
BMI, kg/m2	31.4 ± 7.7	26.4 ± 2.9	37.8 ± 7.3	<.001
PTH, pg/mL	39.9 ± 23.7	34.3 ± 16.3	46.4 ± 28.9	<.001
25OHD, ng/mL	25.6 ± 10.2	28.9 ± 9.0	22.6 ± 10.0	<.001
% Population, 25OHD < 20 ng/mLb	34%	14%	43%	<.001
Fat mass, kg	34.8 ± 11.4	27.8 ± 7.2	44.3 ± 8.9	<.001
Serum calcium, mg/dL	9.5 ± 0.5	9.5 ± 0.6	9.6 ± 0.5	.424
GFR (MDRD), mL/min	67.0 ± 13.2	65.5 ± 12.9	68.9 ± 15.6	.014
GFR adjusted to BSA, mL/min	71.3 ± 18.5	64.8 ± 15.5	79.7 ± 18.7	<.001

Data are expressed as mean ± SD, unless otherwise specified.

^aOne-way ANOVA used to compare women with BMI < 30 kg/m² to obese women with BMI > 30 kg/m².

^bχ² analysis.

The scatterplot of serum PTH and 25OHD is depicted with LOESS graphs for obese women and the total population (Figure 1). The results of the unadjusted nonlinear models for the obese women are:

$$\begin{array}{c}\text{Entire}\hspace{0.17em}\text{population}:\hspace{0.17em}\text{PTH}=33.5+108.9\times {\text{e}}^{-0.138\hspace{0.17em}\times \hspace{0.17em}25\text{OHD}}\\ \text{Obese}:\hspace{0.17em}\text{PTH}=42.2+232.6\times {\text{e}}^{-0.271\hspace{0.17em}\times \hspace{0.17em}25\text{OHD}}\end{array}$$

Figure 1.

The point for near maximal suppression for serum intact PTH by 25OHD by obesity status. This value is 11.1 ng/mL for obese, none for the nonobese, and 21.7 ng/mL for the total population.

Nonlinear exponential analysis showed that for the entire population, the point for near maximal suppression of PTH occurred at a 25OHD concentration of 21.7 ng/mL with SE of 4.75 ng/mL and corresponding 95% CI of 12.4–31.1 ng/mL. The point for near maximal suppression of PTH in obese women occurred at a 25OHD concentration of 11.1 ng/mL with SE of 3.25 ng/mL and corresponding 95% CI of 4.7–17.5 ng/mL. Few nonobese women had an elevated PTH (approximately 3%), and not surprisingly, they showed no point of maximal suppression.

Relationships and predictors of 25OHD

Serum concentrations of 25OHD were negatively associated with BMI (r = −0.30; P < .0001) and positively with serum PTH (r = 0.25; P < .0001). The slopes of linear correlation using analysis of covariance between PTH and 25OHD also tended to differ (P < .08) in the nonobese vs obese subpopulations. Multiple linear regression analysis was performed with age, GFR, and serum calcium as independent variables, and with menopausal status, season, hormones (PTH or 25OHD), and BMI on the dependent variable (PTH or 25OHD). For PTH as the dependent variable, the model R² was 0.21, and BMI was the most significant predictor (P < .001), followed by 25OHD (P < .001). Similarly for 25OHD as the dependent variable, the model R² was 0.24, with season (P < .0001) as the most significant predictor, followed by PTH (P < .001). Other explanatory variables analyzed were age, GFR, serum calcium, and menopausal status, and these did not explain any significant variance in the model.

Discussion

The exact concentration for suppression of PTH by serum 25OHD has been a controversial topic due to the variable findings in different studies (10). This is likely related to numerous sources of variation and/or the different methods to determine the point of suppression. We hypothesized that BMI is a source of variation in this relationship that goes beyond well-known factors, such as age and renal function. Despite the significant influence of BMI on both serum concentrations of 25OHD and PTH, the influence of BMI on the relationship between 25OHD and PTH has not been examined in other studies. Factors contributing to lower 25OHD in obesity include greater adiposity serving as a depot for 25OHD (4–6), that may be exacerbated by reduced UVB exposure in this population. The higher serum PTH with greater adiposity (2, 3) may be at least partially related to the lower serum 25OHD. Our findings show that there is a near maximal suppression of PTH at a lower 25OHD concentration in the obese compared to the entire population, and the slope differs so that once PTH rises, it does so more dramatically in the obese.

The etiology of the lower serum 25OHD threshold associated with a rise in PTH in the obese is not clear. It is possible that there may be a different set-point for the calcium-PTH relationship in the obese, as demonstrated in a calcium-citrate clamp study that showed an exaggerated PTH response to hypocalcemia as compared to normal subjects (15). It is interesting that polymorphisms of the vitamin D binding protein (DBP) have been shown to predict serum 25OHD concentrations (16) and body fat (17). It is possible that DBP plays a role in the relationship because other binding proteins are low in obesity (ie, SHBG), but others suggest that serum DBP shows no relationship with BMI (18). The roles of 1,25-dihydroxyvitamin D or hyperlipidemia were not examined in this study and may have independent effects on the PTH threshold. The higher bone mineral density in the obese may be due to their higher calcium absorption, circulating estradiol, and/or greater weight bearing (19). In addition, whereas the obese have lower serum 25OHD, their lower threshold for a PTH response may protect their bones. However, the more dramatic rise in PTH in the obese once 25OHD is < 11 ng/mL is a concern because high PTH is associated with extraskeletal diseases (20). Several factors need to be considered in a vitamin D recommendation that is specific for the obese population.

It is also interesting to note that there are several means by which the point of maximal suppression is calculated and reported in the literature, with thresholds ranging between 12 and 50 ng/mL or no threshold (10, 11). In a 3-phase model used to estimate maximal suppression in persons 65 and older (12), 2 thresholds of 12 and 28 ng/mL were found that could also explain differences between studies. Different statistical methodologies used to estimate this relationship may contribute to the variability (11) as well as different demographics of the populations. The absence of controlling for calcium intake, which suppresses PTH, may have also affected different findings in previous studies (10, 11). In addition, variability in assays used to measure 25OHD (21) and PTH (22) also complicates the issue.

Limitations of this study include a potential ability to generalize the data because this is primarily a Caucasian population and we only studied women. There is also a concern because we were limited to subjects recruited for studies within our laboratory, which may also have introduced a selection bias. The absence of high PTH in the nonobese may be due to the healthy women recruited in these studies who were given adequate calcium intake. Because data were collected from 1994 to 2010, there has been variability in laboratory assays for PTH, but we have shown good correlation with these assays previously (21). In addition, the same 25OHD assay was used, and subjects were stabilized to 1–1.2 g calcium/d before the blood draw to avoid variability in PTH due to a calcium-deficient diet.

In obese women who are characterized by a lower 25OHD and higher PTH serum concentrations, we found that the point of near maximal suppression for PTH occurs at a lower level of 25OHD compared to the general population. If this relationship is a marker of vitamin D status, then the lower 25OHD in the obese has different implications for deficiency or is a poor marker of vitamin D status if it differs for subsets of the population. In addition, the absence of an inflection point in the nonobese may indicate that serum 25OHD does not influence PTH in nonobese women who consume at least 1 g calcium/d, but this requires further study with more individuals who have low serum 25OHD. In summary, the relationship of 25OHD and PTH is complicated by obesity in a healthy female outpatient population.