Introduction

Osteoporosis is an important public health problem because osteoporotic fractures threaten quality of life, and life expectancy, and increase medical costs worldwide [1–3]. Worldwide, more than 200 million people have osteoporosis, and 1.5 million fractures are reported annually in the USA [4]. The increasing trend in the number of osteoporotic fractures is more evident in Asian countries than in Western countries [5–7]. Osteoporosis is a silent disease but may give lead to serious results in elderly populations.

Osteoporosis is caused by weakening of bone strength, which is determined by bone density and bone quality [8]. Bone density in the elderly is determined by both peak bone mass and subsequent bone loss [9, 10]. Therefore, both achievement of adequate peak bone mass early in life and prevention of accelerated bone loss during aging are important preventive strategies for maintaining bone health throughout one’s life span [8]. The former strategy, which is achieving adequate peak bone mass, requires knowledge about the major determinants and those influences on peak bone mass.

Many genetic and epigenetic factors that affect bone mineral density (BMD) have been investigated [10, 11]. Among these factors, heritability provides the important information for predicting parameters related to bone strength, including peak bone mass. Several studies of mother–daughter pairs have shown that the mother’s BMD is strongly associated with the daughter’s bone mass [12–15]. A meta-analysis reported that parental history of fractures increases the risks of any fracture, osteoporotic fracture, and hip fracture in their offspring [16]. On the basis of this finding, the parents’ hip fracture history is included as one of the major predictors of future osteoporotic fractures in the 10-year fracture risk assessment model (FRAX) [17]. In other words, inheritance primarily affects parameters related to bone strength in a given individual.

Various epigenetic factors related to anthropometric characteristics, environmental effects, and nutritional support that could affect bone mass or fractures have been examined widely [18–20]. Among these factors, body mass index (BMI) has been identified as a major determinant of bone density; that is, a higher BMI increases BMD by inducing greater mechanical loading on skeletal sites [21, 22]. Therefore, in women, a low BMI at the age range at which the bone mass peaks may negatively affect future bone health.

Both hereditary transmission and epigenetic modulation are important for determining peak bone mass and bone strength. However, we do not know which factors are more dominant in determining peak bone mass and bone strength, or whether BMI, an epigenetic factor, can compensate for the effects of inherited determination. This study aimed to investigate the maternal inheritance and BMI on peak bone mass and to assess whether the hereditary influence on peak bone mass is modulated according to BMI.

Materials and methods

Subjects and data collection

This study was based on the data obtained from the first year of the Korean National Health and Nutrition Examination Survey (KNHANES) V, conducted in 2010. KNHANES is a nationwide cross-sectional survey of nationally representative non-institutionalized civilians in the Republic of Korea. This study comprises a health interview, health examination, and nutrition survey. A stratified, multistage probability sampling design was used to select household units, and the civilian, non-institutionalized population of Korea was included in the survey. Details of this survey were recently reported [23]. A total of 8958 participants of 3479 households completed the 2010 KNHANES V. Among those households, 260 households that comprised daughter participants in an age range of 20–35-year-old in which the bone mass peaks were included in this study. Finally, 187 households that had BMD data for both mothers and daughters were analyzed. Any daughters or mothers with renal insufficiency (serum creatinine concentration >1.4 mg/dL) or a history of malignancy were excluded. The subjects who used antiresorptive agents such as raloxifene, bisphosphonate, or hormone-replacement therapy were also excluded in both groups. Natural menopause in mothers group was defined as at least 12 consecutive months of amenorrhea. The daughters were stratified into two groups by their mean BMI: lower BMI group (≤21.4 kg/m2) and group with higher BMI (>21.4 kg/m2). The KNHANES was reviewed and approved by the ethics committee of the Korea Centers for Disease Control and Prevention (KCDC) (2010-02CON-21-C).

BMD measurement

BMD (g/cm2) was measured at the lumbar spine (L1–L4, LS), femoral neck (FN), and total hip (TH) using dual-energy X-ray absorptiometry (DXA, QDR4500A; Hologic Inc., Bedford, MA, USA). The DXA equipment was housed in mobile examination centers. The left hip was scanned routinely, but in participants with a left hip fracture or device, the right hip was scanned. The T-scores in the mothers group were calculated using the mean value and standard deviation (SD) for young normal individuals, and the Z-scores in the mothers and daughters groups were calculated using the mean values and standard deviation (SD)s for each decade from the recently published Korean reference data [24] by the following formula:

$$ \mathrm{Z}\hbox{-} \mathrm{score}=\left(\mathrm{actual}\ \mathrm{B}\mathrm{M}\mathrm{D}\ \mathrm{value}-\mathrm{mean}\ \mathrm{value}\ \mathrm{of}\ \mathrm{B}\mathrm{M}\mathrm{D}\ \mathrm{in}\ \mathrm{each}\ \mathrm{decade}\right)/\mathrm{S}\mathrm{D}\ \mathrm{of}\ \mathrm{B}\mathrm{M}\mathrm{D}\ \mathrm{in}\ \mathrm{each}\ \mathrm{decade} $$

The DXA data were analyzed using Hologic Discovery software version 12.1 in its default configuration.

Statistical analysis

Statistical analyses were performed using SPSS version 19 (IBM SPSS Statistics; IBM Corp., Armonk, NY, USA). The ages of mothers ranged from 45 to 67 years, and their menopausal status comprised both with premenopause and menopause. Therefore, we used Z-scores instead of T-scores to eliminate the effects of the contribution of the older ages of mothers to their lower BMD values. Pearson’s correlation coefficient was used to examine the correlation of between the BMD of daughters and the Z-scores of mothers. The mothers group was further stratified into three groups, according to their Z-score, Z-score ≤ −1.0, −1.0 < Z-score ≤ +1.0, and Z-score > +1.0. Linear regression analysis was used to evaluate the relationships between BMD and BMI of the daughters and Z-score of the mothers. The t value was used in regression analysis to estimate the contributiveness of BMI of the daughters or Z-scores of the mothers on daughters’ BMD. A multiple logistic regression analysis for the independent factor of a daughters’ Z-score of <−1.0 was performed to estimate the Exp(β) of mothers’ Z-score in the lower and higher BMI groups. The daughters group was divided into four groups according to their mean BMI or mother Z-score: BMI ≤ 21.4 kg/m2 and mother Z-score < −1.0, BMI ≤ 21.4 kg/m2 and mother Z-score ≥ −1.0, BMI > 21.4 kg/m2 and mother Z-score < −1.0, and BMI > 21.4 kg/m2 and mother Z-score ≥ −1.0. ANOVA was used to compare BMDs at each skeletal site in those four groups. Tests for interaction of daughters’ BMI between mothers’ Z-scores and daughters’ BMD were performed using interaction terms as well as main effects of those two factors in the regression models. P < 0.05 was considered significant.

Results

Baseline characteristics of daughters and their mothers

We analyzed the data for 187 daughter–mother pairs selected from the 2010 KNHANES V. The clinical characteristics of the daughters and mothers groups are shown in Table 1. The mean ages were 26.4 ± 3.9 years in the daughters and 53.1 ± 4.7 years in the mothers. The mean BMI of the daughters was 21.4 kg/m2 (21.4 ± 3.4 kg/m2). The 25-hydroxy vitamin D levels were 14.6 ± 5.4 ng/mL in daughters group and 17.4 ± 6.5 ng/mL in mothers group. Twenty-six (13.9 %) subjects continued regular exercise and 16 (8.6 %) subjects were current smokers in the daughters group. There were no subjects who had experienced fractures in daughters group, whereas 13 (7.0 %) subjects in mothers group had histories of previous fractures. One hundred and thirty-six subjects (73.3 %) in mothers group were postmenopausal, whereas the remaining 50 mothers were premenopausal. The T-scores of mothers group were −0.94 at the LS, −0.94 at the FN, and 0.03 at the TH.

Table 1 Baseline clinical parameters of daughters and their mothers groups
Full size table

Relationships between the BMD and BMI of daughters or BMD of daughter and the Z-score of mothers

Body size, represented by height, weight, or BMI, is strongly affected by hereditary factors, thus weight and height should be adjusted when comparing variables that are associated with height or weight in parent-offspring pairs [25]. The weight and height of mothers and daughters were also positively correlated in the present study (Supplementary Table 1). The results of the BMI-adjusted model and weight- and height-adjusted model were not different in the present study (data not shown), thus we adjusted for BMI when analyzing the data. To determine the factors the affected the BMD of daughters significantly, we performed a multiple regression analysis for this parameter. We found significant associations between the LS BMD of daughters and age, BMI, and mother Z-score and between the FN and TH BMD of daughters and BMI and mother Z-score (data not shown). Table 2 shows the correlations between the BMD of daughters and their BMI or the Z-score of mothers in each skeletal site, with or without adjustment for the age and BMI of the daughters or the BMI of the mothers. The BMD of daughter was strongly and positively associated with their BMI in each of the three skeletal sites (r = 0.492 for the LS, r = 0.444 for the FN, and r = 0.523 for the TH, p < 0.001). The BMD of daughter was also correlated positively with the Z-score of mothers at each skeletal site; however, this correlation was not as strong as those observed with BMI (r = 0.429 between mother LS Z-score and the LS BMD of daughter; r = 0.315 between mother FN Z-score and the FN BMD of daughter; r = 0.417 between mother TH Z-score and the TH BMD of daughters; p < 0.001) (Table 2). The significant associations observed between the BMD of daughter and mothers’ Z-score were similarly present after adjusting for the age and BMI of daughters, and the BMI of mothers and the relationships were strongest between the same skeletal sites (Table 2).

Table 2 Correlations between daughters’ BMD and BMI or their mothers’ Z-score with or without adjustment for daughters’ age and BMI and mothers’ BMI in each bone sites
Full size table

Relationships between BMD of daughters and the Z-score of mothers in each skeletal site according to the BMI of daughter

Tables 3 shows the correlations between the daughters’ BMD and their mothers’ Z-score in the same skeletal sites in the lower (BMI ≤ 21.4 kg/m2) and higher (BMI > 21.4 kg/m2) BMI groups. In the lower BMI group, daughters’ BMI was significantly positively correlated with their BMD. However, the relationships were weaker than those of the mothers’ Z-scores in the same skeletal site. These patterns were similarly observed in the LS, FN, and TH. The findings of weaker relationships with BMI but stronger associations with mothers’ Z-score were more prominent in the FN and TH. In the higher BMI group, the mothers’ Z-scores in the same skeletal sites also correlated positively with the daughters’ BMD. The correlations between the daughters’ BMI and their BMD were stronger in the higher BMI group compared with the lower BMI group and compared with the mothers’ Z-scores in the FN and TH (Table 3).

Table 3 Correlations between daughters’ BMD and BMI or their mothers’ Z-score in the lower or higher BMI groups
Full size table

The degree of contribution of the BMI of daughters or the Z-score of mothers to the BMD of daughters was compared using the t values obtained from regression analysis (Table 4). In the lower BMI group, the BMD of daughters in the LS was affected more by the LS Z-score of mothers than by their BMI (β = 0.017, t = 2.443, p = 0.017 for the BMI of daughter, β = 0.039, t = 4.665, p < 0.001 for the LS Z-score of mothers). In contrast, the effect of BMI was increased and that of the LS Z-score of mothers was decreased in the daughters with a higher BMI (BMI > 21.4 kg/m2). Furthermore, the effects of BMI and of the LS Z-score of mothers on the BMD of daughters became almost similar in this group (β = 0.015, t = 3.579, p < 0.001 for the BMI of daughter and β = 0.161, t = 3.958, p < 0.001 for the LS Z-score of mothers).

Table 4 Regression analysis of BMI and mother’s Z-score for daughters’ BMD in the lower or higher BMI groups
Full size table

At the FN and TH of the lower BMI group, the BMD of daughter was not significantly related to their BMI and was significantly associated only with their mothers’ Z-score (β = 0.010, t = 1.603, p = 0.012 for the BMI of daughters, and β = 0.026, t = 3.813, p < 0.001 for the FN Z-score of mothers, and β = 0.010, t = 1.751, p = 0.084 for the BMI of daughters, and β = 0.035, t = 5.426, p < 0.001 for the TH Z-score of mothers). However, in the higher BMI group, BMI was significantly related to BMD, and the effect of BMI was much stronger than that the Z-score of mothers at each skeletal site (β = 0.015, t = 3.872, p < 0.001 for the BMI of daughters and β = 0.038, t = 2.363, p = 0.022 for the FN Z-score of mothers, and β = 0.015, t = 4.244, p < 0.001 for the BMI of daughters and β = 0.044, t = 2.981, p < 0.004 for the TH Z-score of mothers) (Table 4).

Associations between the mothers’ Z-scores and daughters’ risk of having Z-scores < −1.0

In the daughters with a lower BMI, the mothers’ Z-scores were significantly associated with the daughters’ low Z-scores of <−1.0 at all three skeletal sites (Exp(β) = 0.574, 95 % confidence interval [CI] = 0.356–0.942 for the LS, Exp(β) = 0.476, 95 % CI = 0.2880–0.786 for the FN, and Exp(β) = 0.462, 95 % CI = 0.275–0.778 for the TH, p < 0.05, respectively) (Table 5). However, in the daughters with a higher BMI, the significant association observed between the mothers’ Z-scores and daughters’ lower Z-scores at the level <−1.0 disappeared in the LS, FN, or TH (Table 5). After further adjustment for 25-hydroxyvitamin D levels, smoking status, and regular exercise, the results were unchanged in both groups.

Table 5 Multiple logistic regression analyses of the mother’s Z-scores for the daughters’ having the Z-score < −1.0 in lower and higher BMI groups
Full size table

Comparisons of the BMD of daughters in the LS, FN, and TH according to their BMI and mothers’ Z-score in the same skeletal sites

We compared the daughters’ BMD in the LS, FN, and TH according to the daughters’ BMI (BMI ≤ 21.4 kg/m2 or BMI > 21.4 kg/m2) and the mothers’ Z-score (Mothers’ Z-score < −1.0 or Mothers’ Z-score ≥ −1.0). The daughters group with lower BMI and mothers’ Z-score of less than −1.0 showed significantly lowest BMD values in the LS, FN, and TH (Fig. 1). Whereas, the daughters group with higher BMI and mothers Z-score of higher than −1.0 provided significant highest BMD values in three skeletal sites. There were no significant differences between the daughters group with lower BMI and mothers Z-score of higher than −1.0 and group with higher BMI but mothers Z-score of less than −1.0 in the LS, FN, and TH (Fig. 1). We examined the interaction term of daughters’ BMI with the mothers’ Z-score on daughters’ BMD in three skeletal sites, and we found significant interactions in all sites (p < 0.05, respectively) (Fig. 1).

Fig. 1
figure 1

The BMD of the daughters in the LS (a), FN (b), and TH (c) according to BMI of the daughters and the Z-scores of the mothers in the same skeletal sites. Values with different superscripts are significantly different (p < 0.05 with Bonferroni correction). Interaction terms were provided as β values with 95 % confidence intervals (*p < 0.05)

Full size image

Discussion

In the present study, we found an evidence of a hereditary relationship for BMD between mothers and daughters, which reflects the strong genetic influence on the determination of BMD. However, this hereditary relationship was modulated by daughter’s BMI. That is, the hereditary influence was significant and strong in the daughters with lower BMI but was diminished in the daughters with higher BMI. In the daughters with higher BMI, the impact of their BMI on BMD was greater than that of their mothers’ Z-scores. In other words, heritability had greater influence in the lower BMI group, whereas BMI was more influential than hereditary transmission in the higher BMI group in determining bone mass.

The hereditary nature of BMD has been previously reported. Several studies have shown significant associations for BMD between mothers and their daughters both at the lumbar spine and femur [12, 26, 27]. In one study, premenopausal daughters of women with hip fractures showed significantly lower femoral BMD, indicating that the daughters were at higher risk of future hip fractures [28]. The genetic influence on BMD has also been observed in mothers and their prepubertal children, and the relationship between mothers and their offspring was significantly higher in mother–daughter pairs than in mother–son pairs [29]. Recently, the genetic bases of specific diseases have been studied widely by using analysis of single nucleotide polymorphisms (SNPs) [30] whose existence can be used to demonstrate the genetic background of a specific disease [31, 32]. Several SNPs related to bone strength have been reported in previous studies [33–35]. These findings suggest that genetic factors are major determinants of bone health-related parameters.

Body composition such as body weight and BMI correlated strongly with BMD. Higher BMI is known to have a beneficial effect on BMD by inducing mechanical loading on the skeleton [21, 22, 36, 37]. One recent study reported that people with a BMI of 18 kg/m2 had a rapider bone loss rate at femoral site than those with a BMI of 30 kg/m2 [19]. Another study reported that the mean LS BMD increased with increasing body weight quartiles and that the influence of lifestyle factors was stronger in postmenarcheal daughters than in premenarcheal daughters [14]. Birth weight was reported as an independent determinant of adult bone mass [38]. Consistent with previous data, our data indicate that the daughters’ BMI was positively correlated with their BMD.

On the other hand, the effects of the two important determining factors of daughters’ BMD, namely the daughters’ BMI and the mothers’ Z-score, differ according to the daughters’ BMI. BMD in the lower BMI group was more affected by hereditary factors, as represented by the mother’s BMD, than by the daughters’ BMI. The influence of BMI increased, but the inherited influence decreased in the daughter group with higher BMI, and the effect of BMI on BMD was stronger that of hereditary factor in this group. This suggests that the inherited influences on bone mass may vary according to the BMI of specific individuals.

In the present study, the associations of the daughters’ BMI and the mothers’ Z-score with the daughters’ BMD varied according to the skeletal sites. The hereditary influence represent differently according to the skeletal site [39]. Moreover, the effects of body compositional parameters on BMD also provide skeletal site specificity regarding factors related to weight-bearing [40]. Based on these findings, it is conceivable that the relative degree of contribution of BMI and heredity to BMD might be distinguished from the properties of the skeletal sites.

The present study has several strengths. To our knowledge, this is the first study to demonstrate that BMI modulates hereditary influences on BMD. We used data from the 2010 KNHANES, which is a representative nationwide survey, and we included a relatively large number of mother–daughter pairs. We found evidence of a compensatory role of BMI on genetic influences on BMD. However, there are also several limitations in this study. First, consistent with the body composition data for this population [41], the percentage of individuals in the obese population (with a BMI of >25.0 kg/m2) and in the low-weight population (with a BMI of <18.5 kg/m2) was around 15 % in each case. Because of the low percentage of individuals in the obese or low-weight groups, we were not able to analyze the manner in which obesity or low-weight affects hereditary effects on BMD in this population. Instead, we used the mean BMI level as a cut-off for the lower and higher BMI groups. Second, this was a subgroup analysis that included only 5 % of the households from among the whole study population. Therefore, selected participation may have introduced some bias in determining the familial interactions that affecting BMD. The generalizability of the results should be further validated in other study populations. Third, this was a cross-sectional study, and we could not investigate how weight change in the daughters might affect the achievement of peak bone mass and the inherited influences on BMD. Finally, we did not compare the daughters’ BMD according to their mothers’ fracture history because the mothers were relatively young and their rate of fractures was <5 %.

In conclusion, although maternal inheritance is well known as an important determinant of bone mass, this hereditary factors can be somewhat overcome by having a higher BMI. The maintenance of a healthy body weight during the period of attainment of peak bone mass may be effective for preventing later osteoporosis or fractures later in life.