BMI but Not Race Contributes to Vitamin D–Parathyroid Hormone Axis in Peripubertal Girls

Abstract

Currently, there is widespread interest in establishing 25-hydroxy vitamin D (25OHD) level preventing a secondary elevation in parathyroid hormone (PTH). The aim of this study was to identify the 25OHD nadir resulting in a rise of PTH and to determine if this inflection point is weight- or race-specific during growth and development in peripubertal girls. A total of 104 normal (n = 61) and overweight (n = 43) African American (AA) and European American (EA) girls, 5 to 14 years of age, were included. Though AAs had lower 25OHD levels, there was no difference in PTH compared with EAs. A 25OHD concentration of 27.2 ng/mL (P < .01) was indicated to increase PTH in normal-weight girls, although a statistically significant level was not established in overweight girls. No race difference in inflection point was observed. These data suggest a potential influence of weight status on the 25OHD-PTH inflection point in peripubertal girls. Accordingly, on determination of 25OHD level reflecting optimal health, consideration of weight status appears to be important during this critical period of growth and development.

Introduction

For reasons not entirely clear, there has been a resurgence in the prevalence of vitamin D deficiency/insufficiency over the past decade,^1-6 which parallels an increase in fracture risk. Because of its implication in skeletal health, identification of optimal vitamin D status for maximum parathyroid hormone (PTH) suppression has gained considerable attention.^4,7,8 Perturbation of calcium regulatory mechanisms (ie, vitamin D and PTH) is particularly impactful during periods of rapid growth in childhood, likely influencing long-term health outcomes.⁹

Although disagreement persists regarding a level signifying insufficiency, the physiological relevance in establishing a threshold of 25-hydroxyvitamin D (25OHD) concentration (the conventional indicator of vitamin D status) to prevent a secondary elevation in serum PTH has gained widespread interest, with a number of cut-points having been reported.^10-16 Additionally, many^5,17,18 but not all^7,19 studies report an inverse relationship between 25OHD and PTH. This discrepancy in inflection point and the degree of reciprocity of the 25OHDPTH axis may be influenced by differences in populations studied, such as variation in age/stage of development, race, body habitus, and sex.^6,19-24 The dynamic physiological processes occurring during growth and development (eg, rapid skeletal remodeling coupled with dynamic fluctuations in the reproductive and metabolic hormonal milieu) further complicate calcium regulatory mechanisms and might be explained by a 25OHD-independent secondary increase in serum PTH, most likely as a result of the higher calcium requirements for growth prior to the onset of puberty.

“Perturbation of calcium regulatory mechanisms (ie, vitamin D and PTH) is particularly impactful during periods of rapid growth in childhood, likely influencing long-term health outcomes.”

The increase in prevalence of pediatric overweight/obesity has been suggested to contribute to vitamin D insufficiency as well as to skeletal health. Vitamin D is hypothesized to be sequestered in the adipose tissue in obese states, thus limiting availability in circulation necessary for regulatory mechanisms.^3,25-30 Thus, interactions between obesity and vitamin D metabolism may result in increased PTH concentrations. Although the extent to which racial variation contributes to the 25OHD-PTH relationship is unclear, there are also ethnic differences in individual concentrations of 25OHD and PTH. African Americans (AAs) generally present with lower 25OHD concentrations and higher PTH.^23,31,32 Despite these differences in these calcium-regulating hormones, AAs have greater bone mineral density, suggesting a skeletal resistance to the actions of PTH and/or differences in vitamin D metabolism. Thus, investigations focusing on European American (EA) and AA girls, with varying weight status during the peripubertal period may provide some insight into factors influencing the 25OHD-PTH axis. Changes in the hormonal microenvironment unique to pubertal onset and progression differ between males and females,²³ and as such, homogeneity in selection of study participants in terms of developmental stage and sex is necessary to limit confounding when investigating metabolic pathway characteristics. For this reason, this study focused solely on early pubertal girls. The main objective of the study was to investigate the existence of specific thresholds of 25OHD for PTH suppression in peripubertal EA and AA girls between the ages of 5 and 14 years and evaluate the influence of weight status and race.

Methods

Study Design

Participants were drawn from various studies, observational in nature, conducted at University of Alabama at Birmingham (UAB) between the years 2005 and 2010. Inclusion criteria were female and AA or EA race categorization, age between 5 and 14 years, and complete data, including weight, height, BMI, pubertal status, race, and measured serum 25OHD and PTH. Studies had distinct yet overlapping hypotheses centered on metabolic outcomes associated with body habitus across racial groups in the pediatric population and did not include an intervention that could have affected vitamin D levels. All participants were healthy and not on medication known to influence body composition (ie, antihyperactivity, antiasthmatic, steroidal drugs) nor receiving other medications such as over-the-counter vitamin D, calcium, or other micronutrients (including multivitamins). Parents and children provided consent/assent. Each protocol was approved by the institutional review board for human participants at UAB.

Anthropometric Assessment

Participants were weighed to the nearest 0.1 kg in minimal clothing without shoes (Scale-tronix 6702W; Scale-tronix, Carol Stream, IL). Height was recorded without shoes using a digital stadiometer (Heightronic 235; Measurement Concepts, Snoqualmie, WA). BMI percentile was calculated using sex- and age-specific Centers for Disease Control growth charts.³³

Pubertal Status

The 1-to-5 staging of Marshall and Tanner,³⁴ based on examination of both breast and pubic hair development, was used for pubertal stage assessment, with 1 composite number representing the higher of the 2 assigned values.

Laboratory Analyses

Serum 25OHD and PTH were assayed either in the UAB Core Laboratory or commercially (Quest Diagnostics, San Juan Capistrano, CA). In the UAB Core Laboratory, serum 25OHD was assessed by enzyme immunoassay (Immunodiagnostics Systems Inc, Fountain Hills, AZ) and PTH by a 2-site immunoradiometric assay. The intra-assay coefficients of variation (CV) for the analysis of PTH and 25OHD were 7.76% and 4.83%, and the mean interassay CVs were 2.07% and 4.94%, respectively. Commercial assays assessed 25OHD using liquid chromatography–tandem mass spectrometry (LC-MS/MS) methodology³⁵ and serum PTH using a 2-site immunoradiometric assay that detects intact PTH (1-84) and the amino-terminally truncated PTH (7-84) fragments (normal range 10-65 pg/mL). The assay has an intra-assay CV of 2.2% to 2.3% and an interassay CV of 5.6% to 8.6%. There were no differences between the studies in laboratory measures, so data were pooled for analysis.

Statistical Analyses

General linear models were used to evaluate demographic characteristics as well as 25OHD and PTH concentrations, by weight status and race. Weight- and race-specific multiple linear regression modeling was next used to evaluate the relationship between 25OHD and PTH, controlling for pubertal stage. For this analysis, participants were divided based on weight classification and race, such that those above and below the 85th BMI percentile, and AA and EA, were analyzed separately.

To obtain initial estimates for the 25OHD-PTH inflection point, we visually inspected scatterplots and fit a locally weighted (LOESS) regression³⁶ to characterize the relationship between 25OHD and PTH for both groups using SAS PROC LOESS. The LOESS smoothed lines supported the nonlinear shape for both groups of participants. Next, we used SAS PROC REG to estimate initial values for the regression coefficients for the relationship of 25OHD to PTH on each side of the thresholds for each group. A piecewise change point model was developed to determine and estimate the 25OHD threshold for each grou.³⁷ SAS PROC NLIN was used to simultaneously estimate the parameters in the 2-group piecewise regression solution through an iterative process. This 8-parameter change-point model can be expressed as follows:

$$\begin{array}{cc}\hfill \mathrm{PTH}=& b_{0\mathrm{B}}+b_{1\mathrm{B}}X\text{for}X\le c_{\mathrm{B}},\hfill \\ \hfill b_{0\mathrm{B}}+c_{\mathrm{B}}(b_{1\mathrm{B}}-b_{2\mathrm{B}})& +b_{2\mathrm{B}}X\text{for}X>c_{\mathrm{B}},{}_{+e,}\hfill \\ \hfill & b_{0\mathrm{A}}+b_{1\mathrm{A}}X\text{for}X\le c_{\mathrm{A}},\hfill \\ \hfill b_{0\mathrm{A}}+c_A(b_{1\mathrm{A}}-b_{2\mathrm{A}})& +b_{2\mathrm{A}}X\text{for}X>c_{\mathrm{A}},\hfill \end{array}$$

where b_0A and b_0B are the PTH intercepts, c_A and c_B are the 25OHD threshold levels, b_1A and b_1B are the slope estimates before the thresholds, and b_2A and b_2B are the slope estimates after the thresholds for the EA and AA and above and below the 85th percentile of BMI, respectively.

Results

Table 1 illustrates the descriptive characteristics for the entire sample as well as stratified according to race and weight status. AA girls were older, taller, heavier, and reproductively more mature compared with EA girls. Girls categorized as overweight were older and taller, yet reproductively less mature relative to those categorized as normal weight.

Table 1.

Descriptive Sample Characteristics for Overall Sample and by Weight Status and Race

	Overall (n = 104)	BMI ≤85th Percentile (n = 61)	BMI >85th Percentile (n = 43)	AA (n = 66)	EA (n = 38)
Age (years)	11.3 ± 0.2	10.9 ± 0.3a	11.8 ± 0.3a	11.7 ± 0.3a	10.6 ± 0.4a
Pubertal stageb	2.8 ± 0.2	2.8 ± 0.2c	1.7 ± 0.3c	3.6 ± 0.2a	1.4 ± 0.1a
Height (cm)	152.2 ± 1.4	148.7 ± 1.8a	156.5 ± 1.9a	155.6 ± 1.4a	146.4 ± 2.5a
Weight (kg)	79.6 ± 3.7	66.1 ± 4.8a	93.7 ± 5.0a	82.7 ± 4.2	72.0 ± 7.6
BMI percentile	57.6 ± 1.5	49.3 ± 4.6a	91.7 ± 1.6a	54.6 ± 7.8	59.1 ± 5.9

Abbreviations: AA, African American; EA, European American.

^aP < .05.

^bOn a scale of 1 to 5, according to the criteria of Marshall and Tanner.³⁴

^c.05 < P < .10.

Table 2 illustrates serum 25OHD and PTH concentrations based on weight status and race. PTH concentrations were lower in overweight girls relative to normal weight girls, whereas there was no difference according to weight status for 25OHD. AA girls had lower 25OHD, yet PTH concentration did not differ from those of EA girls.

Table 2.

The Association Between 25OHD and PTH Measures by BMI Percentile and Race

	BMI ≤85th Percentile (n = 61)	BMI >85th Percentile (n = 43)	EA (n = 36)	AA (n = 63)
Serum 25OHD (ng/mL)	23.0 ± 1.1	22.2 ± 1.4	29.9 ± 1.3a	18.5 ± 0.8a
Serum PTH (pg/mL)	42.1 ± 2.2a	35.5 ± 2.2a	36.6 ± 2.5	40.9 ± 2.1
Associationsa	β = −1.00b	NS	NS	β = −0.93b

Abbreviations: 25OHD, 25-hydroxy vitamin D; PTH, parathyroid hormone; EA, European American; AA, African American; NS, not significant.

^aP < .05.

^bAll models controlled for pubertal status; weight-specific model controlled for race; race-specific model controlled for BMI percentile.

Multiple linear regression analysis for the 25OHD-PTH associations (Table 2) revealed no relationship among EAs (n = 38), whereas among AA girls (n = 68), there was an inverse relationship between 25OHD and PTH (β = −0.929, P = .004). The inclusion of BMI percentile in each of the models did not influence the results. There was also variation in associations by weight status, such that a relationship between 25OHD and PTH was not apparent in overweight girls, yet an inverse relationship was indicated among those classified as normal weight (β = −1.004, P = .001). Additionally, these findings were not influenced by inclusion of race in the model.

The relationship between serum 25OHD and PTH in normal (<85th BMI percentile, solid dots) and overweight (≥85th BMI percentile, open squares) girls is illustrated in Figure 1. Using piecewise regression models, the data suggested that for the normal-weight girls, a threshold of 27.22 ng/mL 25OHD was estimated for the inflection point. As illustrated, an inverse relationship of 25OHD and PTH (P = .002) represents the change in the slope from negative to positive. For the overweight girls, an inflection point of 14.44 ng/mL was identified; however, this trend and the postthreshold relationship were not significant. The solid and dashed lines illustrate the predicted values of serum 25OHD and serum PTH based on the point estimates for each individual. There was no statistically significant difference in the 25OHD-PTH relationship by race (data not shown).

Figure 1.

Predicted Values of Change-Point Model for Normal and Overweight Adolescent Girls.^a

^a Solid dots are the parathyroid hormone (PTH) values, and the solid line is the piecewise-regression-predicted values with a threshold at c_B = 27.22 ng/mL for the <85th BMI percentile group. Open squares are the PTH values, and the dashed line is the piecewise-regression-predicted values with a threshold at c_A = 14.44 ng/mL for the ≥85th BMI percentile group.

Discussion

The magnitude of vitamin D insufficiency, which elicits adverse health outcomes is not clear, and data in growing children are limited; however, many but not all studies indicate <20 ng/mL (<50 nmol/L) as that threshold. It is speculated that the 25OHD level associated with a rise in PTH may be important during critical periods in growth and development, particularly in peripubertal females, given the crucial involvement in skeletal health. Secondary hyperparathyroidism is a detrimental outcome of hypovitaminosis D, and elevated serum PTH is known to induce bone mineral resorption. Findings from this study indicate that the 25OHD concentration resulting in PTH elevation differs by weight status. In individuals of normal weight, a 25OHD concentration of 27.22 ng/mL was the threshold, whereas in overweight individuals, 14.4 ng/mL was identified as the threshold, but statistical modeling did not detect this level to be significant.

Our findings may explain some of the inconsistencies in identifying a clinically optimal 25OHD concentration. For example, Hill et al⁷ evaluated a wide age range (7-18 years) in a multiethnic cohort of healthy children and were unable to detect a threshold of 25OHD at which PTH was suppressed.⁷ However, consideration of body weight was not included in the analysis. In contrast, among EA boys and girls before and after weight loss, Reinehr et al³⁸ reported normalization in 25OHD and PTH concentrations after weight loss, indicating that alterations may be consequences of rather than causes of overweight. In adults, equivocal results and variation in the study population have also precluded a definitive effect of weight on the 25OHD-PTH relationship. If in the obese state adipose tissue sequesters vitamin D, then the counterargument may suggest greater bioavailability among lean counterparts.³⁹ There is evidence that the body adapts by making the fat depot stores of 25OHD more bioavailable to sustain normocalcemia, altering the normal metabolic functioning of adipose tissue³⁹; however, further investigation is required to substantiate this postulation. Though further investigation is warranted, it is likely that weight status plays a role in the 25OHD-PTH axis, though the extent may differ based on the age of the patient and the stage of sexual maturation. Indeed, further exploration using methods to differentiate the body tissue compartments may reveal the true relationship between 25OHD and PTH with adipose tissue compartments.

The 25OHD-PTH relationship has also been reported to differ between racial groups. Despite lower 25OHD levels in our AA girls, the PTH concentration was similar to that in EA girls. Although multiple linear regression analysis revealed an inverse relationship between the 2 hormones in AAs, we were unable to identify an inflection point in either racial/ethnic group. Similar discrepancies have been observed in AA adults; thus, various explanations have been considered. For example, AAs reportedly have reduced urinary excretion of calcium, lending a greater capacity for calcium conservation, which in theory would prevent secondary hyperparathyroidism in response to vitamin D deficiency. It is also possible that AAs have enhanced 1-α hydroxylase activity and thus better synthesis of 1,25OH₂D (active vitamin D, which enhances calcium absorption), thereby limiting the potential of secondary hyperparathyroidism. Notwithstanding, it should not be misinterpreted that AAs have a lower requirement for 25OHD; rather, inherent mechanistic differences appear to add complexity to race-specific determination of the 25OHD-PTH inflection point, thereby making further investigations essential.⁴⁰

The strengths of the study include a unique, well-defined participant population of children aged 5 to 14 years. To our knowledge, this is the first study to demonstrate that 25OHD and PTH inflection point may be specific to weight status classification. Though not generalizable, this has clinical implications when considering treatment options of vitamin D insufficiency in overweight children. One of the limitations of the study is its cross-sectional nature. Additionally, the use of BMI, though informative, does not provide a measure of body composition. Future studies that include indices of adipose tissue deposition and distribution may elucidate the extent of mediation of the 25OHD-PTH relationship. Additionally, we did not have measures of bone mass, which would have provided insight into implications of the 25OHD-PTH relationship. As the 25OHD-PTH relationship is mainly relevant from a skeletal health stand point, these implications may not translate when considering an optimal vitamin D concentration for nonskeletal outcomes (eg, insulin sensitivity).

Conclusions

In conclusion, the intricate interrelationship between vitamin D and PTH is influenced by BMI status during growth and development. Our findings suggest that the observed specificity of the 25OHD-PTH relationship among differing populations may be particularly apparent according to weight status, with potential confounding by racial variation during periods of rapid growth. Because obesity is becoming an increasing feature among the pediatric population along with prevalent vitamin D insufficiency, ongoing investigations of the 25OHD-PTH relationship should take these findings into consideration.