DOI: https://doi.org/10.21203/rs.3.rs-546848/v1
Background: In recent years, it was demonstrated that high-density lipoprotein cholesterol (HDL-C), a critical lipid for human lipid metabolism, was not completely beneficial to human health, implying that extremely high HDL-C levels may also affect human health and contribute to various diseases. The correlation between HDL-C and bone metabolism was uncertain and controversial. This study aimed to explore the correlation between HDL-C level and bone mineral density (BMD), investigating whether this relationship is different in diverse populations by stratifying age and gender.
Method: The data utilized were extracted from 2005-2010 National Health and Nutrition Examination Survey (NHANES). We reviewed the data to exclude the participants aged over or equal to 20 years old or with missing core data. Multivariate linear regression analyses were conducted to estimate the association between HDL-C and BMD. A subgroup analysis was also utilized to estimate the difference in diverse populations by stratifying age and gender. Moreover, fitted smoothing curves and generalized additive models were also performed to address the nonlinear relationship between HDL-C levels and BMD.
Result: Multivariable-adjusted linear regression models demonstrated that HDL-C was negatively associated with BMD, especially in females. Meanwhile, smooth curve fittings and generalized additive models also suggested an inverted U-shaped curve among females aged 30-40 or over 60. A U-shaped curve was observed for the relationship between HDL-C and BMD in femoral regions in females aged 20 to 30 or 50 to 60. Besides, female participants aged over 40 at a higher than or equal to 71 mg/dL HDL-C level were more likely to have a high risk of osteopenia or osteoporosis.
Conclusion: HDL-C and BMD exhibited a negative correlation among females and different associations in diverse age groups. In addition, HDL-C can serve as a marker for osteopenia or osteoporosis.
High-density lipoprotein cholesterol (HDL-C) is a type of cholesterol contained in or bound to high-density lipoproteins (HDL) [1]. HDL-C was believed to possess beneficial impacts on human health and was inversely associated with cardiovascular disease over a long time [2, 3]. For instance, Gordon et al. exhibited an independent inverse association of HDL-C levels and coronary heart disease event rates [4]. Rosenson et al. observed that low HDL-C levels below target may be beneficial in cardiovascular disease reduction [5]. However, over the past few years, some different voices increase. Madsen et al. reported that men and women with extremely high HDL cholesterol paradoxically have high all-cause mortality [6]. Hamer et al. observed a U-shaped association between HDL-C and mortality in a large general population sample [7]. These findings may indicate that we should reconsider our perspective on HDL-C.
Osteoporosis is a worldwide public health problem characterized by low bone mineral density (BMD) and a high risk of osteoporotic fracture [8]. According to International Osteoporosis Foundation, one-third of women and one-fifth of men aged over 50 years old have osteoporosis or low bone mass and are at risk of osteoporotic fracture [9]. Simultaneously, as the population ages and grows, the prevalence of osteoporosis continues to rise [10]. At present, apart from genetic factors, age, or sex, the impact of other factors like lipid metabolism or lifestyle for bone metabolism has recently attracted considerable concern [11–13]. Meanwhile, researchers hope to discover novel modalities for osteoporosis prevention and treatment.
The correlation between HDL-C and BMD was uncertain and controversial. Some previous studies indicated that HDL-C level was elevated in post-menopausal women, negatively associated with bone mineral density (BMD). Maghbooli et al. found a negative correlation between HDL-C and BMD in post-menopausal Iranian women with vitamin D deficiency [14]. Zhang et al. observed that HDL-C was negatively associated with lumbar spine BMD in Chinese women [15]. Conversely, Cui et al. suggested that HDL-C level was not associated with BMD values at any of the sites in pre- and post-menopausal subjects [16]. Apart from the above, Jeong et al. observed that HDL-C was positively associated with BMD at the lumbar spine in Korean post-menopausal women [17]. Overall, the findings from these studies are contradictory. Meanwhile, since the participants in all the studies are usually from the same country or region, and most studies mainly focus on women, especially post-menopausal ones, it is difficult to say whether the relationship between HDL-C and BMD is different in diverse populations, like males or young adults. In addition, the relationship between HDL-C and BMD may be nonlinear, but the specific results require further investigation.
Accordingly, this study used a representative sample of adults aged over 20 years old from the National Health and Nutrition Examination Survey (NHANES), tried to explore linear or nonlinear relationship between HDL-C level and BMD, and investigated whether the relationship between them is different in diverse populations by stratifying age and gender.
The data analyzed in this study was extracted from National Health and Nutrition Examination Survey (NHANES), an ongoing study to assess the health and nutritional status of noninstitutionalized U.S. populations. We extracted the data for all participants from 2005–2010 [18–20]. The study was approved by the ethics review board of the National Center for Health Statistics, and written consent was obtained from each participant.
Data Extraction
We extracted the following information:
Demographic data (age, gender, race/ethnicity, education level, and income to poverty ratio)
Examination data (total femur BMD, femur neck BMD, trochanter BMD, intertrochanter BMD, total spine BMD, L1 BMD, L2 BMD, L3 BMD, and L4 BMD)
Laboratory data [HDL-C level (mg/dL), total cholesterol level (mg/dL), alanine aminotransferase (ALT) (U/L), asparate aminotransferase (AST) (U/L), and total calcium (mg/dL)]
Questionnaire data [drinking status (had at least 12 alcohol drinks past one year), smoking status (smoked at least 100 cigarettes in life), BMI (height and weight); diabetes (has a doctor told that you have diabetes), and hypertension (ever told you had high blood pressure)]
In addition, we selected the “Full Sample 2 Year MEC Exam Weight (WTMEC2YR)” to represent the weight value. Because we combine three two-year cycles of the continuous NHANES, the final weight we used was equal to one-third of the“Full Sample 2 Year MEC Exam Weight (WTMEC2YR)” according to the rule of constructing weights when combining survey cycles on the NHANES website [21].
The subjects aged over or equal to 20 with available BMD and HDL-C data were included in this study. The participants missing other variables data (data missing, answered "do not know" or refused to answer were also considered missing) are excluded.
The Measurement of HDL-C was performed using Lipid Laboratory Johns Hopkins. Detailed information is accessible at NHANES website [22]. Based on the information provided at NHANES website, briefly, HDL-C is measured directly in serum. The basic principle of the method is as follows. The apolipoprotein-B (apoB) containing lipoproteins in the specimen are reacted with a blocking reagent that renders them non-reactive with the enzymatic cholesterol reagent under the assay conditions. The apoB containing lipoproteins are thus effectively excluded from the assay, and only HDL-C-cholesterol is detected under the assay conditions.
The femur scans provide bone measurements for total femur, femoral neck, trochanter, and intertrochanter based on information provided on NHANES website. The DXA examinations were performed using Hologic QDR-4500A fan-beam densitometers (Hologic, Inc., Bedford, MA, USA) and software version Apex 3.2 by trained technologists. Further details of DXA examination protocol are documented in Body Composition Procedures Manual located on NHANES website [23].
Mean femoral BMD of 20-29-year-old non-Hispanic white women from NHANES III was selected as the reference value. According to the research of Looker et al. [24].
(1) osteopenia: BMD value in any femoral regions between 1 and 2.5 SD below the mean of reference value [males (total femur BMD: 0.68–0.90 g/cm2, femur neck BMD: 0.59–0.79 g/cm2, trochanter BMD: 0.49–0.66 g/cm2, or intertrochanter BMD: 0.78–1.03 g/cm2); females (total femur BMD: 0.64–0.82 g/cm2, femur neck BMD: 0.56–0.74 g/cm2, trochanter BMD: 0.46–0.61 g/cm2, or intertrochanter BMD: 0.74–0.95 g/cm2)];
(2) osteoporosis: BMD value in any femoral regions > 2.5 SD below mean BMD of reference value [males (total femur BMD: < 0.68 g/cm2, femur neck BMD: < 0.59 g/cm2, trochanter BMD: < 0.49 g/cm2, or intertrochanter BMD: < 0.78 g/cm2); females (total femur BMD: < 0.64 g/cm2, femur neck BMD: < 0.56 g/cm2, trochanter BMD: < 0.46 g/cm2, or intertrochanter BMD: < 0.74 g/cm2)].
We used mean (continuity variable) or proportion (categorical variable) to describe the baseline characteristics of participants. A weighted multivariate linear regression model was used to evaluate the association between HDL-C and BMD. A subgroup analysis was performed by stratified multivariate regression analysis. Furthermore, smooth curve fittings and generalized additive models were used to address the nonlinear relationship between HDL-C and BMD. For nonlinear models, the inflection point in this relationship was calculated using a recursive algorithm. A two-piecewise linear regression model was conducted on both sides of the inflection point when nonlinearity was detected. Multiple logistic regression analyses were performed to investigate the odds ratios (ORs) of osteopenia and osteoporosis. All analyses were performed using software R, V.4.0.3 [R: a language and environment for statistical computing (program). Vienna, Austria: R Foundation for Statistical Computing, 2016] and EmpowerStats (http://www.empowerstats. com), with a P-value < 0.05 considered statistically significant.
A total of 31,034 participants were included in NHANES dataset from 2005 to 2010. Firstly, we excluded the participants without available BMD data (n = 14344). Secondly, we excluded the participants without available HDL-C data (n = 1080). Thirdly, the participants aged below 20 years old (n = 5516) and missing data on other variables (n = 1424, education level: 9, income to poverty ratio: 706, current BMI: 265, ALT: 60, AST: 1, diabetes: 5, hypertension: 9, smoking status: 2, drinking status: 367) were excluded. Finally, 8670 participants aged 20 years and over with complete data were analyzed. The detailed selection process is presented in Fig. 1.
The basic demographics of sample subjects are summarized in Table 1. A total of 8670 participants, 20–85 years of age, were included in our analysis, with weighted characteristics of participants subclassified based on gender. In this sample, subjects had a mean age of 44.39 ± 15.26; mean income to poverty ratio of 3.15 ± 1.62; mean ALT of 25.95 ± 18.18; mean AST of 25.66 ± 13.65; mean total calcium of 9.46 ± 0.35; mean total cholesterol of 197.61 ± 40.64. Most subjects were non-Hispanic whites (72.16%), received education above high school (60.07%), had at least 12 alcohol drinks past one year (77.97%), smoked less than 100 cigarettes in life (53.07%), had a BMI less than 25 kg/cm2 (37.69%). Diabetes and hypertension cases accounted for 6.02% and 25.45%, respectively.
| Characteristics | Means or proportions |
|---|---|
| Age (years, mean ± SD) | 44.39 ± 15.26 |
| Sex, n (%) Male Female | 4477 (50.53%) 4193 (49.47%) |
| Race/ethnicity, n (%) | |
| Mexican American | 1590 (7.92%) |
| Other Hispanic | 710 (4.29%) |
| Non-Hispanic White | 4347 (72.16%) |
| Non-Hispanic Black | 1633 (9.84%) |
| Other Race | 390 (5.79%) |
| Education level, n (%) | |
| Under high school | 2195 (16.50%) |
| High school or equivalent | 2026 (23.43%) |
| Above high school | 4449 (60.07%) |
| Income to poverty ratio (mean ± SD) | 3.15 ± 1.62 |
| BMI, n (%) | |
| < 25 | 3024 (37.69%) |
| 25–30 | 3248 (36.48%) |
| >=30 | 2398 (25.83%) |
| Diabetes, n (%) | |
| Yes | 744 (6.02%) |
| No | 7791 (92.67%) |
| Broadline | 135 (1.31%) |
| Hypertension, n (%) | |
| Yes | 2513 (25.45%) |
| No | 6157 (74.55%) |
| Smoked at least 100 cigarettes in life, n (%) | |
| Yes | 4095 (46.93%) |
| No | 4575 (53.07%) |
| Had at least 12 alcohol drinks past one year? n (%) | |
| Yes | 6408 (77.97%) |
| No | 2262 (22.03%) |
| ALT (U/L, mean ± SD) | 25.95 ± 18.18 |
| AST (U/L, mean ± SD) | 25.66 ± 13.65 |
| Total calcium (mg/dL, mean ± SD) | 9.46 ± 0.35 |
| Total cholesterol (mg/dL, mean ± SD) | 197.61 ± 40.64 |
| HDL-C (mg/dL, mean ± SD) | 53.39 ± 16.17 |
| Total femur BMD (g/cm2, mean ± SD) | 0.98 ± 0.15 |
| Femur neck BMD (g/cm2, mean ± SD) | 0.84 ± 0.15 |
| Trochanter BMD (g/cm2, mean ± SD) | 0.74 ± 0.13 |
| Intertrochanter BMD (g/cm2, mean ± SD) | 1.16 ± 0.18 |
| Total spine BMD (g/cm2, mean ± SD) | 1.04 ± 0.14 |
| L1 BMD (g/cm2, mean ± SD) | 0.96 ± 0.15 |
| L2 BMD (g/cm2, mean ± SD) | 1.05 ± 0.15 |
| L3 BMD (g/cm2, mean ± SD) | 1.07 ± 0.15 |
| L4 BMD (g/cm2, mean ± SD) | 1.07 ± 0.15 |
| BMI, body mass index; SD, standard deviation; n, numbers of subjects; %, weighted percentage. | |
The results of the multivariate regression analyses are presented in Table 2. In the unadjusted model, HDL-C was negatively correlated with BMD. After adjustment for age, gender, and race/ethnicity, this negative association was still present in model 2. After adjustment for age, sex, race/ethnicity, education level, income to poverty ratio, smoking status, drinking status, BMI, diabetes, hypertension, ALT, AST, total calcium, and total cholesterol, the negative association remained statistically significant. The nonlinear relationship using smooth curve fittings and generalized additive models, employed to characterize between HDL-C and BMD, was displayed in Fig. 2.
| Model 1 β (95% CI) P value | Model 2 β (95% CI) P value | Model 3 β (95% CI) P value | |
|---|---|---|---|
| Total femur BMD | -0.0025 (-0.0027, -0.0023) < 0.000001 | -0.0013 (-0.0015, -0.0011) < 0.000001 | -0.0004 (-0.0005, -0.0002) 0.000239 |
| Femur neck BMD | -0.0019 (-0.0021, -0.0017) < 0.000001 | -0.0011 (-0.0013, -0.0009) < 0.000001 | -0.0003 (-0.0004, -0.0001) 0.006194 |
| Trochanter BMD | -0.0018 (-0.0020, -0.0016) < 0.000001 | -0.0008 (-0.0010, -0.0007) < 0.000001 | -0.0002 (-0.0004, -0.0000) 0.037159 |
| Intertrochanter BMD | -0.0030 (-0.0032, -0.0027) < 0.000001 | -0.0016 (-0.0018, -0.0014) < 0.000001 | -0.0005 (-0.0007, -0.0003) 0.000027 |
| Total spine BMD | -0.0012 (-0.0014, -0.0010) < 0.000001 | -0.0010 (-0.0012, -0.0008) < 0.000001 | -0.0003 (-0.0005, -0.0001) 0.002281 |
| L1 BMD | -0.0019 (-0.0021, -0.0017) < 0.000001 | -0.0012 (-0.0014, -0.0010) < 0.000001 | -0.0005 (-0.0007, -0.0002) 0.000016 |
| L2 BMD | -0.0015 (-0.0017, -0.0013) < 0.000001 | -0.0011 (-0.0013, -0.0009) < 0.000001 | -0.0004 (-0.0006, -0.0002) 0.000079 |
| L3 BMD | -0.0009 (-0.0011, -0.0007) < 0.000001 | -0.0009 (-0.0011, -0.0007) < 0.000001 | -0.0003 (-0.0005, -0.0001) 0.011329 |
| L4 BMD | -0.0008 (-0.0010, -0.0006) < 0.000001 | -0.0008 (-0.0010, -0.0006) < 0.000001 | -0.0002 (-0.0004, 0.0001) 0.167819 |
| Model 1: no covariates were adjusted. Model 2: age (40–49; 50–59; 60–69; ≥70), sex (male; female), race/ethnicity (Mexican American; other Hispanic; non-Hispanic white; non-Hispanic black; other Races) were adjusted. Model 3: age (40–49; 50–59; 60–69; ≥70), sex (male; female), race/ethnicity (Mexican American; other Hispanic; non-Hispanic white; non-Hispanic black; other Races), Education (under high school; high school or equivalent; above high school), income to poverty ratio (quartile groups), BMI (obese, overweight, normal), drink status (had at least 12 alcohol drinks past one year; don not have at least 12 alcohol drinks past one year), smoking status (less than 100 cigarettes; greater than or equal to 100 cigarettes), diabetes (yes; no), hypertension (yes; no), ALT (quartile groups), AST (quartile groups), total calcium (quartile groups) and total cholesterol (quartile groups) were adjusted. | |||
The subgroup analyses, stratified by age (20 ≤ Aged < 30; 30 ≤ Aged < 40; 40 ≤ Aged < 50; 50 ≤ Aged < 60; 60 ≤ Aged) and gender (male or female), were reported in Tables 3–4. After adjusting for confounders, the negative correlation of HDL-C with BMD remained present, especially the participants aged 30 to 40 or over 50 female participants.
| Model 1 β (95% CI) P value | Model 2 β (95% CI) P value | Model 3 β (95% CI) P value | ||
|---|---|---|---|---|
| Total femur BMD | 20 ≤ Aged < 30 | -0.0019 (-0.0023, -0.0014) < 0.000001 | -0.0008 (-0.0013, -0.0004) 0.000358 | 0.0001 (-0.0004, 0.0006) 0.604404 |
| 30 ≤ Aged < 40 | -0.0023 (-0.0027, -0.0019) < 0.000001 | -0.0016 (-0.0021, -0.0012) < 0.000001 | -0.0008 (-0.0012, -0.0003) 0.000564 | |
| 40 ≤ Aged < 50 | -0.0020 (-0.0024, -0.0017) < 0.000001 | -0.0012 (-0.0016, -0.0009) < 0.000001 | -0.0003 (-0.0007, 0.0001) 0.119456 | |
| 50 ≤ Aged < 60 | -0.0027 (-0.0031, -0.0022) < 0.000001 | -0.0016 (-0.0020, -0.0011) < 0.000001 | -0.0006 (-0.0010, -0.0001) 0.015563 | |
| 60 ≤ Aged | -0.0026 (-0.0031, -0.0022) < 0.000001 | -0.0012 (-0.0016, -0.0008) < 0.000001 | -0.0004 (-0.0007, 0.0000) 0.063049 | |
| Femur neck BMD | 20 ≤ Aged < 30 | -0.0015 (-0.0019, -0.0010) < 0.000001 | -0.0008 (-0.0013, -0.0003) 0.000638 | 0.0002 (-0.0003, 0.0006) 0.508709 |
| 30 ≤ Aged < 40 | -0.0016 (-0.0020, -0.0012) < 0.000001 | -0.0013 (-0.0018, -0.0009) < 0.000001 | -0.0005 (-0.0010, -0.0001) 0.018517 | |
| 40 ≤ Aged < 50 | -0.0014 (-0.0018, -0.0011) < 0.000001 | -0.0012 (-0.0015, -0.0008) < 0.000001 | -0.0004 (-0.0007, 0.0000) 0.066283 | |
| 50 ≤ Aged < 60 | -0.0018 (-0.0022, -0.0014) < 0.000001 | -0.0013 (-0.0017, -0.0009) < 0.000001 | -0.0004 (-0.0009, -0.0000) 0.041834 | |
| 60 ≤ Aged | -0.0017 (-0.0021, -0.0014) < 0.000001 | -0.0010 (-0.0013, -0.0006) < 0.000001 | -0.0002 (-0.0005, 0.0002) 0.308122 | |
| Trochanter BMD | 20 ≤ Aged < 30 | -0.0014 (-0.0018, -0.0010) < 0.000001 | -0.0005 (-0.0009, -0.0001) 0.007112 | 0.0000 (-0.0004, 0.0005) 0.868720 |
| 30 ≤ Aged < 40 | -0.0016 (-0.0020, -0.0012) < 0.000001 | -0.0011 (-0.0014, -0.0007) < 0.000001 | -0.0005 (-0.0009, -0.0001) 0.013278 | |
| 40 ≤ Aged < 50 | -0.0015 (-0.0018, -0.0011) < 0.000001 | -0.0008 (-0.0011, -0.0005) 0.000001 | -0.0002 (-0.0006, 0.0001) 0.198229 | |
| 50 ≤ Aged < 60 | -0.0020 (-0.0024, -0.0016) < 0.000001 | -0.0010 (-0.0014, -0.0006) < 0.000001 | -0.0003 (-0.0007, 0.0001) 0.166214 | |
| 60 ≤ Aged | -0.0019 (-0.0023, -0.0016) < 0.000001 | -0.0006 (-0.0010, -0.0003) 0.000189 | -0.0001 (-0.0004, 0.0003) 0.721758 | |
| Intertrochanter BMD | 20 ≤ Aged < 30 | -0.0022 (-0.0027, -0.0016) < 0.000001 | -0.0010 (-0.0015, -0.0005) 0.000181 | 0.0001 (-0.0005, 0.0007) 0.729035 |
| 30 ≤ Aged < 40 | -0.0028 (-0.0033, -0.0023) < 0.000001 | -0.0020 (-0.0025, -0.0015) < 0.000001 | -0.0010 (-0.0015, -0.0005) 0.000263 | |
| 40 ≤ Aged < 50 | -0.0024 (-0.0028, -0.0020) < 0.000001 | -0.0014 (-0.0019, -0.0010) < 0.000001 | -0.0003 (-0.0008, 0.0002) 0.207675 | |
| 50 ≤ Aged < 60 | -0.0032 (-0.0037, -0.0026) < 0.000001 | -0.0019 (-0.0024, -0.0013) < 0.000001 | -0.0008 (-0.0013, -0.0002) 0.006320 | |
| 60 ≤ Aged | -0.0033 (-0.0037, -0.0028) < 0.000001 | -0.0017 (-0.0021, -0.0012) < 0.000001 | -0.0006 (-0.0011, -0.0001) 0.009357 | |
| Total spine BMD | 20 ≤ Aged < 30 | -0.0006 (-0.0010, -0.0002) 0.004010 | -0.0007 (-0.0011, -0.0003) 0.000428 | -0.0002 (-0.0006, 0.0003) 0.414071 |
| 30 ≤ Aged < 40 | -0.0005 (-0.0008, -0.0001) 0.022318 | -0.0009 (-0.0013, -0.0005) 0.000025 | -0.0005 (-0.0009, -0.0000) 0.033293 | |
| 40 ≤ Aged < 50 | -0.0005 (-0.0009, -0.0002) 0.004028 | -0.0007 (-0.0011, -0.0004) 0.000105 | -0.0001 (-0.0005, 0.0003) 0.560152 | |
| 50 ≤ Aged < 60 | -0.0020 (-0.0025, -0.0016) < 0.000001 | -0.0016 (-0.0020, -0.0011) < 0.000001 | -0.0008 (-0.0013, -0.0003) 0.001722 | |
| 60 ≤ Aged | -0.0021 (-0.0026, -0.0017) < 0.000001 | -0.0010 (-0.0014, -0.0005) 0.000011 | -0.0002 (-0.0007, 0.0002) 0.333068 | |
| L1 BMD | 20 ≤ Aged < 30 | -0.0009 (-0.0013, -0.0005) 0.000020 | -0.0007 (-0.0011, -0.0002) 0.002833 | 0.0000 (-0.0004, 0.0005) 0.953560 |
| 30 ≤ Aged < 40 | -0.0011 (-0.0015, -0.0007) < 0.000001 | -0.0010 (-0.0014, -0.0006) 0.000007 | -0.0005 (-0.0009, 0.0000) 0.053660 | |
| 40 ≤ Aged < 50 | -0.0013 (-0.0017, -0.0009) < 0.000001 | -0.0010 (-0.0014, -0.0006) < 0.000001 | -0.0003 (-0.0007, 0.0001) 0.157263 | |
| 50 ≤ Aged < 60 | -0.0028 (-0.0032, -0.0023) < 0.000001 | -0.0019 (-0.0023, -0.0014) < 0.000001 | -0.0011 (-0.0016, -0.0005) 0.000053 | |
| 60 ≤ Aged | -0.0030 (-0.0035, -0.0026) < 0.000001 | -0.0015 (-0.0019, -0.0011) < 0.000001 | -0.0006 (-0.0011, -0.0002) 0.004883 | |
| L2 BMD | 20 ≤ Aged < 30 | -0.0007 (-0.0011, -0.0003) 0.000763 | -0.0008 (-0.0012, -0.0003) 0.000504 | -0.0001 (-0.0006, 0.0003) 0.570538 |
| 30 ≤ Aged < 40 | -0.0007 (-0.0011, -0.0003) 0.001000 | -0.0010 (-0.0014, -0.0006) 0.000007 | -0.0006 (-0.0010, -0.0001) 0.018724 | |
| 40 ≤ Aged < 50 | -0.0007 (-0.0011, -0.0004) 0.000098 | -0.0009 (-0.0013, -0.0005) 0.000023 | -0.0003 (-0.0007, 0.0002) 0.230465 | |
| 50 ≤ Aged < 60 | -0.0023 (-0.0027, -0.0018) < 0.000001 | -0.0017 (-0.0022, -0.0012) < 0.000001 | -0.0010 (-0.0016, -0.0005) 0.000320 | |
| 60 ≤ Aged | -0.0024 (-0.0028, -0.0019) < 0.000001 | -0.0011 (-0.0015, -0.0006) 0.000001 | -0.0004 (-0.0008, 0.0001) 0.126408 | |
| L3 BMD | 20 ≤ Aged < 30 | -0.0004 (-0.0008, 0.0000) 0.069294 | -0.0007 (-0.0012, -0.0003) 0.000635 | -0.0003 (-0.0007, 0.0002) 0.266685 |
| 30 ≤ Aged < 40 | -0.0001 (-0.0005, 0.0003) 0.651313 | -0.0008 (-0.0013, -0.0004) 0.000221 | -0.0005 (-0.0010, -0.0000) 0.030862 | |
| 40 ≤ Aged < 50 | -0.0001 (-0.0004, 0.0003) 0.730709 | -0.0005 (-0.0009, -0.0001) 0.007935 | -0.0000 (-0.0005, 0.0004) 0.954597 | |
| 50 ≤ Aged < 60 | -0.0018 (-0.0023, -0.0013) < 0.000001 | -0.0016 (-0.0021, -0.0011) < 0.000001 | -0.0009 (-0.0015, -0.0004) 0.001325 | |
| 60 ≤ Aged | -0.0018 (-0.0023, -0.0014) < 0.000001 | -0.0008 (-0.0013, -0.0004) 0.000389 | -0.0001 (-0.0005, 0.0004) 0.822908 | |
| L4 BMD | 20 ≤ Aged < 30 | -0.0004 (-0.0008, 0.0000) 0.057092 | -0.0007 (-0.0012, -0.0003) 0.000929 | -0.0003 (-0.0008, 0.0002) 0.217005 |
| 30 ≤ Aged < 40 | -0.0001 (-0.0005, 0.0003) 0.740294 | -0.0007 (-0.0011, -0.0003) 0.001378 | -0.0004 (-0.0008, 0.0001) 0.126720 | |
| 40 ≤ Aged < 50 | -0.0002 (-0.0006, 0.0002) 0.275874 | -0.0007 (-0.0011, -0.0003) 0.001433 | 0.0000 (-0.0004, 0.0005) 0.925507 | |
| 50 ≤ Aged < 60 | -0.0015 (-0.0020, -0.0011) < 0.000001 | -0.0012 (-0.0017, -0.0007) 0.000003 | -0.0004 (-0.0010, 0.0001) 0.113592 | |
| 60 ≤ Aged | -0.0016 (-0.0020, -0.0011) < 0.000001 | -0.0006 (-0.0011, -0.0001) 0.010860 | 0.0000 (-0.0005, 0.0005) 0.866702 | |
| Model 1: no covariates were adjusted. Model 2: sex (male; female) and race/ethnicity (Mexican American; other Hispanic; non-Hispanic white; non-Hispanic black; other Races) were adjusted. Model 3: sex (male; female), race/ethnicity (Mexican American; other Hispanic; non-Hispanic white; non-Hispanic black; other Races), Education (under high school; high school or equivalent; above high school), income to poverty ratio (quartile groups), BMI (obese, overweight, normal), drink status (had at least 12 alcohol drinks past one year; don not have at least 12 alcohol drinks past one year), smoking status (less than 100 cigarettes; greater than or equal to 100 cigarettes), diabetes (yes; no), hypertension (yes; no), ALT (quartile groups), AST (quartile groups), total calcium (quartile groups) and total cholesterol (quartile groups) were adjusted. | ||||
| Model 1 β (95% CI) P value | Model 2 β (95% CI) P value | Model 3 β (95% CI) P value | ||
|---|---|---|---|---|
| Total femur BMD | Male | -0.0011 (-0.0014, -0.0008) < 0.000001 | -0.0012 (-0.0015, -0.0009) < 0.000001 | -0.0002 (-0.0005, 0.0001) 0.233455 |
| Female | -0.0016 (-0.0018, -0.0013) < 0.000001 | -0.0014 (-0.0016, -0.0011) < 0.000001 | -0.0005 (-0.0008, -0.0003) 0.000029 | |
| Femur neck BMD | Male | -0.0009 (-0.0012, -0.0006) < 0.000001 | -0.0010 (-0.0012, -0.0007) < 0.000001 | -0.0001 (-0.0003, 0.0002) 0.715642 |
| Female | -0.0015 (-0.0018, -0.0013) < 0.000001 | -0.0013 (-0.0015, -0.0010) < 0.000001 | -0.0004 (-0.0007, -0.0002) 0.000249 | |
| Trochanter BMD | Male | -0.0006 (-0.0008, -0.0003) 0.000011 | -0.0007 (-0.0010, -0.0005) < 0.000001 | -0.0000 (-0.0003, 0.0003) 0.894142 |
| Female | -0.0010 (-0.0013, -0.0008) < 0.000001 | -0.0009 (-0.0011, -0.0007) < 0.000001 | -0.0003 (-0.0006, -0.0001) 0.001941 | |
| Intertrochanter BMD | Male | -0.0014 (-0.0018, -0.0011) < 0.000001 | -0.0015 (-0.0019, -0.0012) < 0.000001 | -0.0003 (-0.0006, 0.0001) 0.093156 |
| Female | -0.0019 (-0.0022, -0.0016) < 0.000001 | -0.0016 (-0.0019, -0.0014) < 0.000001 | -0.0006 (-0.0009, -0.0004) 0.000012 | |
| Total spine BMD | Male | -0.0005 (-0.0007, -0.0002) 0.002124 | -0.0007 (-0.0010, -0.0004) 0.000003 | -0.0001 (-0.0004, 0.0002) 0.648678 |
| Female | -0.0013 (-0.0016, -0.0010) < 0.000001 | -0.0012 (-0.0014, -0.0009) < 0.000001 | -0.0005 (-0.0008, -0.0003) 0.000096 | |
| L1 BMD | Male | -0.0006 (-0.0009, -0.0004) 0.000014 | -0.0008 (-0.0011, -0.0005) < 0.000001 | -0.0001 (-0.0004, 0.0002) 0.698403 |
| Female | -0.0016 (-0.0019, -0.0014) < 0.000001 | -0.0015 (-0.0017, -0.0012) < 0.000001 | -0.0008 (-0.0010, -0.0005) < 0.000001 | |
| L2 BMD | Male | -0.0006 (-0.0009, -0.0003) 0.000236 | -0.0008 (-0.0011, -0.0005) < 0.000001 | -0.0002 (-0.0005, 0.0001) 0.252641 |
| Female | -0.0015 (-0.0018, -0.0012) < 0.000001 | -0.0013 (-0.0016, -0.0011) < 0.000001 | -0.0006 (-0.0009, -0.0003) 0.000011 | |
| L3 BMD | Male | -0.0003 (-0.0006, -0.0000) 0.037652 | -0.0006 (-0.0009, -0.0003) 0.000206 | -0.0000 (-0.0004, 0.0003) 0.864132 |
| Female | -0.0013 (-0.0015, -0.0010) < 0.000001 | -0.0011 (-0.0014, -0.0009) < 0.000001 | -0.0005 (-0.0008, -0.0002) 0.000399 | |
| L4 BMD | Male | -0.0003 (-0.0006, 0.0000) 0.071875 | -0.0006 (-0.0009, -0.0003) 0.000176 | 0.0000 (-0.0003, 0.0003) 0.973395 |
| Female | -0.0010 (-0.0013, -0.0007) < 0.000001 | -0.0009 (-0.0012, -0.0007) < 0.000001 | -0.0003 (-0.0006, -0.0000) 0.029269 | |
| Model 1: no covariates were adjusted. Model 2: age (40–49; 50–59; 60–69; ≥70) and race/ethnicity (Mexican American; other Hispanic; non-Hispanic white; non-Hispanic black; other Races) were adjusted. Model 3: age (40–49; 50–59; 60–69; ≥70), race/ethnicity (Mexican American; other Hispanic; non-Hispanic white; non-Hispanic black; other Races), Education (under high school; high school or equivalent; above high school), income to poverty ratio (quartile groups), BMI (obese, overweight, normal), drink status (had at least 12 alcohol drinks past one year; don not have at least 12 alcohol drinks past one year), smoking status (less than 100 cigarettes; greater than or equal to 100 cigarettes), diabetes (yes; no), hypertension (yes; no), ALT (quartile groups), AST (quartile groups), total calcium (quartile groups) and total cholesterol (quartile groups) were adjusted. | ||||
In male participants, after adjustment for confounders, except for the subjects aged 30–40 years, where the results indicated that HDL-C level was negatively associated with BMD in the femoral region of intertrochanter (β = -0.0008, 95% CI: -0.0015 to -0.0000, P = 0.040346), no evidence demonstrated that HDL-C was associated with BMD. The detailed results are listed in Table 5. The smooth curve fittings and generalized additive models were also used to characterize the nonlinear relationship between HDL-C and BMD in male participants, Fig. 3.
| Model 1 β (95% CI) P value | Model 2 β (95% CI) P value | Model 3 β (95% CI) P value | ||
|---|---|---|---|---|
| Total femur BMD | 20 ≤ Aged < 30 | -0.0006 (-0.0013, 0.0001) 0.115524 | -0.0009 (-0.0016, -0.0002) 0.017402 | 0.0002 (-0.0006, 0.0010) 0.615178 |
| 30 ≤ Aged < 40 | -0.0014 (-0.0021, -0.0008) 0.000014 | -0.0015 (-0.0022, -0.0009) 0.000002 | -0.0008 (-0.0014, -0.0001) 0.018263 | |
| 40 ≤ Aged < 50 | -0.0007 (-0.0013, -0.0001) 0.021800 | -0.0009 (-0.0015, -0.0003) 0.002213 | 0.0001 (-0.0006, 0.0007) 0.839256 | |
| 50 ≤ Aged < 60 | -0.0012 (-0.0020, -0.0005) 0.001535 | -0.0015 (-0.0023, -0.0008) 0.000073 | -0.0003 (-0.0011, 0.0005) 0.447713 | |
| 60 ≤ Aged | -0.0013 (-0.0019, -0.0006) 0.000056 | -0.0014 (-0.0020, -0.0008) 0.000011 | -0.0005 (-0.0011, 0.0001) 0.113393 | |
| Femur neck BMD | 20 ≤ Aged < 30 | -0.0006 (-0.0013, 0.0002) 0.138530 | -0.0008 (-0.0015, -0.0001) 0.026261 | 0.0002 (-0.0006, 0.0009) 0.649731 |
| 30 ≤ Aged < 40 | -0.0010 (-0.0017, -0.0004) 0.002045 | -0.0011 (-0.0017, -0.0005) 0.000692 | -0.0004 (-0.0011, 0.0002) 0.208086 | |
| 40 ≤ Aged < 50 | -0.0006 (-0.0012, -0.0001) 0.023136 | -0.0009 (-0.0015, -0.0004) 0.001203 | -0.0001 (-0.0007, 0.0005) 0.690638 | |
| 50 ≤ Aged < 60 | -0.0006 (-0.0012, 0.0001) 0.075347 | -0.0009 (-0.0015, -0.0002) 0.007817 | 0.0003 (-0.0004, 0.0009) 0.466506 | |
| 60 ≤ Aged | -0.0010 (-0.0016, -0.0004) 0.000462 | -0.0011 (-0.0017, -0.0006) 0.000065 | -0.0004 (-0.0009, 0.0002) 0.211866 | |
| Trochanter BMD | 20 ≤ Aged < 30 | -0.0003 (-0.0010, 0.0003) 0.330555 | -0.0005 (-0.0011, 0.0001) 0.105673 | 0.0001 (-0.0006, 0.0008) 0.865177 |
| 30 ≤ Aged < 40 | -0.0008 (-0.0014, -0.0003) 0.004085 | -0.0009 (-0.0014, -0.0004) 0.001311 | -0.0004 (-0.0010, 0.0002) 0.188158 | |
| 40 ≤ Aged < 50 | -0.0003 (-0.0008, 0.0002) 0.216489 | -0.0005 (-0.0010, 0.0000) 0.074508 | 0.0002 (-0.0004, 0.0007) 0.600983 | |
| 50 ≤ Aged < 60 | -0.0008 (-0.0015, -0.0002) 0.016474 | -0.0011 (-0.0018, -0.0004) 0.001514 | -0.0002 (-0.0009, 0.0006) 0.620892 | |
| 60 ≤ Aged | -0.0005 (-0.0010, 0.0000) 0.071928 | -0.0006 (-0.0011, -0.0000) 0.032413 | -0.0001 (-0.0007, 0.0004) 0.709963 | |
| Intertrochanter BMD | 20 ≤ Aged < 30 | -0.0007 (-0.0016, 0.0001) 0.096844 | -0.0011 (-0.0019, -0.0002) 0.011457 | 0.0002 (-0.0007, 0.0011) 0.618908 |
| 30 ≤ Aged < 40 | -0.0018 (-0.0026, -0.0011) 0.000003 | -0.0019 (-0.0027, -0.0012) < 0.000001 | -0.0011 (-0.0019, -0.0003) 0.005125 | |
| 40 ≤ Aged < 50 | -0.0008 (-0.0015, -0.0001) 0.018369 | -0.0011 (-0.0018, -0.0004) 0.001913 | 0.0001 (-0.0007, 0.0008) 0.853152 | |
| 50 ≤ Aged < 60 | -0.0015 (-0.0024, -0.0006) 0.000849 | -0.0019 (-0.0028, -0.0010) 0.000050 | -0.0005 (-0.0014, 0.0005) 0.330337 | |
| 60 ≤ Aged | -0.0018 (-0.0025, -0.0011) 0.000001 | -0.0019 (-0.0027, -0.0012) < 0.000001 | -0.0008 (-0.0015, -0.0001) 0.028094 | |
| Total spine BMD | 20 ≤ Aged < 30 | -0.0002 (-0.0009, 0.0004) 0.430869 | -0.0005 (-0.0011, 0.0002) 0.140474 | 0.0000 (-0.0007, 0.0007) 0.991680 |
| 30 ≤ Aged < 40 | -0.0006 (-0.0012, 0.0000) 0.051732 | -0.0007 (-0.0013, -0.0001) 0.025491 | -0.0003 (-0.0009, 0.0004) 0.384966 | |
| 40 ≤ Aged < 50 | 0.0000 (-0.0006, 0.0006) 0.956595 | -0.0002 (-0.0008, 0.0004) 0.470854 | 0.0006 (-0.0001, 0.0012) 0.073739 | |
| 50 ≤ Aged < 60 | -0.0011 (-0.0019, -0.0003) 0.006953 | -0.0015 (-0.0023, -0.0007) 0.000173 | -0.0006 (-0.0014, 0.0003) 0.195890 | |
| 60 ≤ Aged | -0.0007 (-0.0014, -0.0000) 0.041374 | -0.0009 (-0.0015, -0.0002) 0.011893 | -0.0004 (-0.0011, 0.0003) 0.255188 | |
| L1 BMD | 20 ≤ Aged < 30 | -0.0002 (-0.0008, 0.0005) 0.611215 | -0.0003 (-0.0010, 0.0003) 0.288404 | 0.0003 (-0.0004, 0.0010) 0.421029 |
| 30 ≤ Aged < 40 | -0.0005 (-0.0012, 0.0001) 0.091467 | -0.0006 (-0.0012, 0.0000) 0.056638 | -0.0001 (-0.0008, 0.0005) 0.729167 | |
| 40 ≤ Aged < 50 | -0.0002 (-0.0008, 0.0005) 0.622394 | -0.0004 (-0.0010, 0.0003) 0.247077 | 0.0005 (-0.0001, 0.0012) 0.126636 | |
| 50 ≤ Aged < 60 | -0.0014 (-0.0021, -0.0006) 0.000389 | -0.0018 (-0.0025, -0.0010) 0.000006 | -0.0007 (-0.0015, 0.0001) 0.098942 | |
| 60 ≤ Aged | -0.0012 (-0.0018, -0.0006) 0.000266 | -0.0013 (-0.0020, -0.0007) 0.000078 | -0.0006 (-0.0013, 0.0001) 0.071599 | |
| L2 BMD | 20 ≤ Aged < 30 | -0.0003 (-0.0010, 0.0003) 0.355879 | -0.0005 (-0.0012, 0.0001) 0.119616 | -0.0000 (-0.0007, 0.0007) 0.951247 |
| 30 ≤ Aged < 40 | -0.0006 (-0.0013, -0.0000) 0.049876 | -0.0007 (-0.0014, -0.0001) 0.025488 | -0.0003 (-0.0010, 0.0004) 0.377686 | |
| 40 ≤ Aged < 50 | -0.0001 (-0.0007, 0.0005) 0.775438 | -0.0003 (-0.0009, 0.0003) 0.312485 | 0.0004 (-0.0003, 0.0011) 0.251953 | |
| 50 ≤ Aged < 60 | -0.0013 (-0.0021, -0.0005) 0.002333 | -0.0017 (-0.0025, -0.0008) 0.000067 | -0.0009 (-0.0018, 0.0000) 0.057990 | |
| 60 ≤ Aged | -0.0006 (-0.0013, 0.0000) 0.064998 | -0.0008 (-0.0015, -0.0001) 0.021833 | -0.0004 (-0.0012, 0.0003) 0.230982 | |
| L3 BMD | 20 ≤ Aged < 30 | -0.0002 (-0.0009, 0.0004) 0.471605 | -0.0005 (-0.0011, 0.0002) 0.167605 | -0.0001 (-0.0008, 0.0006) 0.740059 |
| 30 ≤ Aged < 40 | -0.0006 (-0.0013, 0.0001) 0.072596 | -0.0007 (-0.0014, -0.0000) 0.036172 | -0.0004 (-0.0011, 0.0003) 0.239139 | |
| 40 ≤ Aged < 50 | 0.0002 (-0.0005, 0.0008) 0.596499 | -0.0001 (-0.0007, 0.0006) 0.838056 | 0.0007 (-0.0000, 0.0014) 0.062945 | |
| 50 ≤ Aged < 60 | -0.0010 (-0.0018, -0.0001) 0.026701 | -0.0014 (-0.0022, -0.0005) 0.001307 | -0.0005 (-0.0014, 0.0004) 0.293799 | |
| 60 ≤ Aged | -0.0004 (-0.0011, 0.0003) 0.272922 | -0.0006 (-0.0013, 0.0001) 0.117742 | -0.0001 (-0.0008, 0.0007) 0.873756 | |
| L4 BMD | 20 ≤ Aged < 30 | -0.0003 (-0.0009, 0.0004) 0.423373 | -0.0005 (-0.0012, 0.0001) 0.128665 | -0.0001 (-0.0008, 0.0007) 0.834089 |
| 30 ≤ Aged < 40 | -0.0005 (-0.0012, 0.0001) 0.097663 | -0.0006 (-0.0013, 0.0000) 0.053291 | -0.0002 (-0.0009, 0.0005) 0.535861 | |
| 40 ≤ Aged < 50 | 0.0001 (-0.0006, 0.0007) 0.778714 | -0.0002 (-0.0008, 0.0005) 0.625760 | 0.0007 (0.0000, 0.0014) 0.047933 | |
| 50 ≤ Aged < 60 | -0.0008 (-0.0017, 0.0000) 0.060838 | -0.0013 (-0.0021, -0.0004) 0.003298 | -0.0003 (-0.0012, 0.0007) 0.555758 | |
| 60 ≤ Aged | -0.0005 (-0.0013, 0.0003) 0.194689 | -0.0007 (-0.0015, 0.0001) 0.071464 | -0.0004 (-0.0012, 0.0004) 0.310063 | |
| Model 1: no covariates were adjusted. Model 2: race/ethnicity (Mexican American; other Hispanic; non-Hispanic white; non-Hispanic black; other Races) were adjusted. Model 3: race/ethnicity (Mexican American; other Hispanic; non-Hispanic white; non-Hispanic black; other Races), Education (under high school; high school or equivalent; above high school), income to poverty ratio (quartile groups), BMI (obese, overweight, normal), drink status (had at least 12 alcohol drinks past one year; don not have at least 12 alcohol drinks past one year), smoking status (less than 100 cigarettes; greater than or equal to 100 cigarettes), diabetes (yes; no), hypertension (yes; no), ALT (quartile groups), AST (quartile groups), total calcium (quartile groups) and total cholesterol (quartile groups) were adjusted. | ||||
In female participants, after adjustment for confounders, multivariate linear regression results indicated that HDL-C level was negatively correlated with BMD in all age groups. Especially in females aged 30 to 40 or 50 to 60, the results revealed that HDL-C levels were negatively correlated with BMD in both femoral regions and spinal areas, Table 6.
| Model 1 β (95% CI) P value | Model 2 β (95% CI) P value | Model 3 β (95% CI) P value | ||
|---|---|---|---|---|
| Total femur BMD | 20 ≤ Aged < 30 | -0.0009 (-0.0015, -0.0003) 0.002740 | -0.0009 (-0.0014, -0.0003) 0.002334 | -0.0000 (-0.0006, 0.0006) 0.956497 |
| 30 ≤ Aged < 40 | -0.0017 (-0.0023, -0.0011) < 0.000001 | -0.0017 (-0.0023, -0.0012) < 0.000001 | -0.0010 (-0.0016, -0.0003) 0.002547 | |
| 40 ≤ Aged < 50 | -0.0014 (-0.0019, -0.0009) < 0.000001 | -0.0014 (-0.0019, -0.0009) < 0.000001 | -0.0005 (-0.0010, -0.0000) 0.043755 | |
| 50 ≤ Aged < 60 | -0.0016 (-0.0022, -0.0011) < 0.000001 | -0.0016 (-0.0021, -0.0010) < 0.000001 | -0.0006 (-0.0012, -0.0000) 0.033738 | |
| 60 ≤ Aged | -0.0010 (-0.0015, -0.0005) 0.000071 | -0.0011 (-0.0016, -0.0006) 0.000011 | -0.0004 (-0.0009, 0.0001) 0.098618 | |
| Femur neck BMD | 20 ≤ Aged < 30 | -0.0009 (-0.0014, -0.0003) 0.004488 | -0.0009 (-0.0014, -0.0003) 0.003061 | 0.0001 (-0.0005, 0.0007) 0.721703 |
| 30 ≤ Aged < 40 | -0.0015 (-0.0020, -0.0009) < 0.000001 | -0.0015 (-0.0021, -0.0010) < 0.000001 | -0.0008 (-0.0015, -0.0002) 0.007730 | |
| 40 ≤ Aged < 50 | -0.0013 (-0.0018, -0.0009) < 0.000001 | -0.0013 (-0.0018, -0.0008) < 0.000001 | -0.0005 (-0.0010, 0.0000) 0.054536 | |
| 50 ≤ Aged < 60 | -0.0016 (-0.0021, -0.0011) < 0.000001 | -0.0016 (-0.0021, -0.0011) < 0.000001 | -0.0007 (-0.0012, -0.0001) 0.016750 | |
| 60 ≤ Aged | -0.0007 (-0.0012, -0.0003) 0.001014 | -0.0009 (-0.0013, -0.0004) 0.000074 | -0.0001 (-0.0006, 0.0003) 0.527758 | |
| Trochanter BMD | 20 ≤ Aged < 30 | -0.0006 (-0.0011, -0.0001) 0.011943 | -0.0006 (-0.0011, -0.0002) 0.010425 | -0.0001 (-0.0006, 0.0005) 0.802236 |
| 30 ≤ Aged < 40 | -0.0012 (-0.0017, -0.0007) 0.000004 | -0.0012 (-0.0017, -0.0007) 0.000001 | -0.0007 (-0.0013, -0.0002) 0.010806 | |
| 40 ≤ Aged < 50 | -0.0010 (-0.0014, -0.0005) 0.000009 | -0.0010 (-0.0014, -0.0006) 0.000003 | -0.0004 (-0.0009, 0.0000) 0.080617 | |
| 50 ≤ Aged < 60 | -0.0010 (-0.0015, -0.0006) 0.000022 | -0.0010 (-0.0015, -0.0005) 0.000030 | -0.0003 (-0.0008, 0.0002) 0.234918 | |
| 60 ≤ Aged | -0.0006 (-0.0010, -0.0001) 0.008169 | -0.0007 (-0.0011, -0.0002) 0.002373 | -0.0001 (-0.0006, 0.0003) 0.520647 | |
| Intertrochanter BMD | 20 ≤ Aged < 30 | -0.0011 (-0.0017, -0.0004) 0.001673 | -0.0010 (-0.0017, -0.0004) 0.001610 | -0.0001 (-0.0008, 0.0006) 0.702295 |
| 30 ≤ Aged < 40 | -0.0019 (-0.0026, -0.0012) < 0.000001 | -0.0020 (-0.0027, -0.0013) < 0.000001 | -0.0011 (-0.0018, -0.0004) 0.002608 | |
| 40 ≤ Aged < 50 | -0.0016 (-0.0022, -0.0011) < 0.000001 | -0.0016 (-0.0022, -0.0010) < 0.000001 | -0.0005 (-0.0011, 0.0001) 0.102820 | |
| 50 ≤ Aged < 60 | -0.0020 (-0.0027, -0.0013) < 0.000001 | -0.0019 (-0.0026, -0.0013) < 0.000001 | -0.0009 (-0.0016, -0.0002) 0.016607 | |
| 60 ≤ Aged | -0.0014 (-0.0020, -0.0008) 0.000009 | -0.0015 (-0.0021, -0.0009) 0.000002 | -0.0007 (-0.0013, -0.0001) 0.033965 | |
| Total spine BMD | 20 ≤ Aged < 30 | -0.0009 (-0.0015, -0.0004) 0.000836 | -0.0010 (-0.0015, -0.0005) 0.000254 | -0.0004 (-0.0010, 0.0002) 0.190767 |
| 30 ≤ Aged < 40 | -0.0009 (-0.0015, -0.0004) 0.000962 | -0.0011 (-0.0016, -0.0005) 0.000166 | -0.0008 (-0.0014, -0.0002) 0.011145 | |
| 40 ≤ Aged < 50 | -0.0010 (-0.0015, -0.0005) 0.000099 | -0.0011 (-0.0016, -0.0006) 0.000011 | -0.0005 (-0.0010, 0.0001) 0.085278 | |
| 50 ≤ Aged < 60 | -0.0017 (-0.0023, -0.0011) < 0.000001 | -0.0017 (-0.0023, -0.0011) < 0.000001 | -0.0009 (-0.0015, -0.0002) 0.009989 | |
| 60 ≤ Aged | -0.0009 (-0.0014, -0.0003) 0.002841 | -0.0010 (-0.0016, -0.0005) 0.000342 | -0.0002 (-0.0008, 0.0004) 0.541538 | |
| L1 BMD | 20 ≤ Aged < 30 | -0.0009 (-0.0015, -0.0003) 0.002901 | -0.0010 (-0.0015, -0.0004) 0.001059 | -0.0002 (-0.0008, 0.0004) 0.547791 |
| 30 ≤ Aged < 40 | -0.0013 (-0.0019, -0.0007) 0.000047 | -0.0014 (-0.0020, -0.0008) 0.000010 | -0.0010 (-0.0016, -0.0003) 0.004785 | |
| 40 ≤ Aged < 50 | -0.0013 (-0.0019, -0.0008) < 0.000001 | -0.0014 (-0.0019, -0.0009) < 0.000001 | -0.0008 (-0.0014, -0.0002) 0.006333 | |
| 50 ≤ Aged < 60 | -0.0020 (-0.0026, -0.0014) < 0.000001 | -0.0020 (-0.0026, -0.0014) < 0.000001 | -0.0011 (-0.0018, -0.0005) 0.000860 | |
| 60 ≤ Aged | -0.0014 (-0.0020, -0.0009) < 0.000001 | -0.0016 (-0.0021, -0.0010) < 0.000001 | -0.0007 (-0.0013, -0.0001) 0.015418 | |
| L2 BMD | 20 ≤ Aged < 30 | -0.0010 (-0.0015, -0.0004) 0.001366 | -0.0010 (-0.0016, -0.0004) 0.000528 | -0.0003 (-0.0009, 0.0003) 0.395872 |
| 30 ≤ Aged < 40 | -0.0011 (-0.0017, -0.0005) 0.000200 | -0.0013 (-0.0018, -0.0007) 0.000032 | -0.0009 (-0.0016, -0.0003) 0.006432 | |
| 40 ≤ Aged < 50 | -0.0011 (-0.0016, -0.0006) 0.000042 | -0.0012 (-0.0018, -0.0007) 0.000004 | -0.0005 (-0.0011, 0.0000) 0.069252 | |
| 50 ≤ Aged < 60 | -0.0018 (-0.0024, -0.0011) < 0.000001 | -0.0018 (-0.0024, -0.0011) < 0.000001 | -0.0010 (-0.0017, -0.0003) 0.007973 | |
| 60 ≤ Aged | -0.0011 (-0.0017, -0.0005) 0.000249 | -0.0013 (-0.0018, -0.0007) 0.000023 | -0.0004 (-0.0010, 0.0002) 0.213659 | |
| L3 BMD | 20 ≤ Aged < 30 | -0.0009 (-0.0015, -0.0004) 0.001420 | -0.0010 (-0.0016, -0.0005) 0.000365 | -0.0004 (-0.0010, 0.0002) 0.197957 |
| 30 ≤ Aged < 40 | -0.0008 (-0.0014, -0.0002) 0.006788 | -0.0009 (-0.0015, -0.0004) 0.001369 | -0.0008 (-0.0014, -0.0001) 0.018031 | |
| 40 ≤ Aged < 50 | -0.0007 (-0.0013, -0.0002) 0.005685 | -0.0009 (-0.0014, -0.0004) 0.001043 | -0.0003 (-0.0009, 0.0002) 0.254036 | |
| 50 ≤ Aged < 60 | -0.0018 (-0.0025, -0.0012) < 0.000001 | -0.0019 (-0.0025, -0.0012) < 0.000001 | -0.0011 (-0.0018, -0.0004) 0.003002 | |
| 60 ≤ Aged | -0.0008 (-0.0014, -0.0002) 0.007618 | -0.0010 (-0.0016, -0.0004) 0.001248 | -0.0001 (-0.0008, 0.0005) 0.693671 | |
| L4 BMD | 20 ≤ Aged < 30 | -0.0009 (-0.0015, -0.0003) 0.001687 | -0.0010 (-0.0015, -0.0004) 0.000783 | -0.0006 (-0.0012, -0.0000) 0.048496 |
| 30 ≤ Aged < 40 | -0.0006 (-0.0012, -0.0001) 0.025961 | -0.0008 (-0.0013, -0.0002) 0.008134 | -0.0006 (-0.0012, 0.0001) 0.078155 | |
| 40 ≤ Aged < 50 | -0.0009 (-0.0014, -0.0004) 0.001001 | -0.0010 (-0.0015, -0.0005) 0.000156 | -0.0003 (-0.0009, 0.0002) 0.250199 | |
| 50 ≤ Aged < 60 | -0.0013 (-0.0019, -0.0006) 0.000081 | -0.0013 (-0.0019, -0.0007) 0.000068 | -0.0004 (-0.0011, 0.0003) 0.239878 | |
| 60 ≤ Aged | -0.0004 (-0.0010, 0.0002) 0.202832 | -0.0006 (-0.0011, 0.0000) 0.065270 | 0.0002 (-0.0004, 0.0009) 0.482683 | |
| Model 1: no covariates were adjusted. Model 2: race/ethnicity (Mexican American; other Hispanic; non-Hispanic white; non-Hispanic black; other Races) were adjusted. Model 3: race/ethnicity (Mexican American; other Hispanic; non-Hispanic white; non-Hispanic black; other Races), Education (under high school; high school or equivalent; above high school), income to poverty ratio (quartile groups), BMI (obese, overweight, normal), drink status (had at least 12 alcohol drinks past one year; don not have at least 12 alcohol drinks past one year), smoking status (less than 100 cigarettes; greater than or equal to 100 cigarettes), diabetes (yes; no), hypertension (yes; no), ALT (quartile groups), AST (quartile groups), total calcium (quartile groups) and total cholesterol (quartile groups) were adjusted. | ||||
The smooth curve fittings and generalized additive models were also utilized to characterize the nonlinear relationship between HDL-C and BMD in female participants. After adjustment for confounders, this association was different among different age groups. (i) An inverted U-shaped curve of the relationship between HDL-C and BMD in femoral regions was observed among participants aged 30 to 40 or over 60. When HDL-C was less than the inflection point, BMD increased with increasing HDL-C; when HDL-C was greater than the inflection point, BMD decreased with increasing HDL-C. (ii) A U-shaped curve of the relationship between HDL-C and BMD in femoral regions in females aged 20 to 30 or 50 to 60. When HDL-C was less than the inflection point, BMD decreased with increasing HDL-C; when HDL-C was greater than the inflection point, BMD increased with increasing HDL-C. Besides, the inflection point was identified using a two-piecewise linear regression model. The inflection points for each group (20–30, 30–40, 50–60, ≥ 60) were about 65–67, 35–37, 71–83, 54-61mg/dL, respectively. The specific values can be found in Fig. 4 and Table 7.
| 20 ≤ Aged < 30 | Index | Total femur BMD | Femur neck BMD | Trochanter BMD | Intertrochanter BMD |
| Fitting by the standard linear model | -0.0007 (-0.0013, -0.0002) 0.0140 | -0.0007 (-0.0013, -0.0001) 0.0320 | -0.0006 (-0.0011, -0.0000) 0.0330 | -0.0009 (-0.0016, -0.0003) 0.0075 | |
| Fitting by the two-piecewise linear model | |||||
| Inflection point (mg/dL) | 66 | 65 | 67 | 66 | |
| HDL-C < Infection point | -0.0020 (-0.0029, -0.0011) < 0.0001 | -0.0017 (-0.0027, -0.0008) 0.0004 | -0.0016 (-0.0023, -0.0008) < 0.0001 | -0.0024 (-0.0034, -0.0013) < 0.0001 | |
| HDL-C > Infection point | 0.0015 (0.0001, 0.0028) 0.0314 | 0.0010 (-0.0003, 0.0024) 0.1241 | 0.0014 (0.0001, 0.0026) 0.0297 | 0.0016 (0.0000, 0.0031) 0.0500 | |
| Log likelihood ratio | < 0.001 | 0.004 | < 0.001 | < 0.001 | |
| 30 ≤ Aged < 40 | Index | Total femur BMD | Femur neck BMD | Trochanter BMD | Intertrochanter BMD |
| Fitting by the standard linear model | -0.0018 (-0.0024, -0.0012) < 0.0001 | -0.0016 (-0.0022, -0.0010) < 0.0001 | -0.0013 (-0.0019, -0.0008) < 0.0001 | -0.0021 (-0.0028, -0.0014) < 0.0001 | |
| Fitting by the two-piecewise linear model | |||||
| Inflection point (mg/dL) | 36 | 35 | 37 | 37 | |
| HDL-C < Infection point | 0.0073 (0.0004, 0.0143) 0.0383 | 0.0073 (-0.0006, 0.0152) 0.0703 | 0.0065 (0.0012, 0.0119) 0.0162 | 0.0074 (0.0003, 0.0145) 0.0412 | |
| HDL-C > Infection point | -0.0021 (-0.0028, -0.0015) < 0.0001 | -0.0019 (-0.0025, -0.0012) < 0.0001 | -0.0016 (-0.0022, -0.0011) < 0.0001 | -0.0025 (-0.0033, -0.0017) < 0.0001 | |
| Log likelihood ratio | 0.008 | 0.024 | 0.003 | 0.007 | |
| 50 ≤ Aged < 60 | Index | L1 BMD | L2 BMD | L3 BMD | L4 BMD |
| Fitting by the standard linear model | -0.0015 (-0.0021, -0.0008) < 0.0001 | -0.0013 (-0.0020, -0.0006) 0.0003 | -0.0014 (-0.0021, -0.0007) < 0.0001 | -0.0008 (-0.0015, -0.0001) 0.0215 | |
| Fitting by the two-piecewise linear model | |||||
| Inflection point (mg/dL) | 83 | 72 | 72 | 71 | |
| HDL-C < Infection point | -0.0024 (-0.0032, -0.0015) < 0.0001 | -0.0025 (-0.0036, -0.0013) < 0.0001 | -0.0026 (-0.0038, -0.0015) < 0.0001 | -0.0019 (-0.0031, -0.0008) 0.0009 | |
| HDL-C > Infection point | 0.0013 (-0.0005, 0.0032) 0.1569 | 0.0002 (-0.0012, 0.0017) 0.7431 | 0.0003 (-0.0011, 0.0017) 0.6573 | 0.0006 (-0.0007, 0.0020) 0.3425 | |
| Log likelihood ratio | 0.001 | 0.013 | 0.005 | 0.012 | |
| 60 ≤ Aged | Index | Total femur BMD | Femur neck BMD | Trochanter BMD | Intertrochanter BMD |
| Fitting by the standard linear model | -0.0013 (-0.0018, -0.0008) < 0.0001 | -0.0009 (-0.0014, -0.0005) < 0.0001 | -0.0008 (-0.0012, -0.0004) 0.0002 | -0.0017 (-0.0023, -0.0011) < 0.0001 | |
| Fitting by the two-piecewise linear model | |||||
| Inflection point (mg/dL) | 55 | 60 | 61 | 54 | |
| HDL-C < Infection point | 0.0007 (-0.0007, 0.0021) 0.3442 | 0.0001 (-0.0008, 0.0011) 0.7926 | 0.0004 (-0.0005, 0.0013) 0.3638 | 0.0007 (-0.0011, 0.0025) 0.4399 | |
| HDL-C > Infection point | -0.0021 (-0.0029, -0.0014) < 0.0001 | -0.0016 (-0.0024, -0.0009) < 0.0001 | -0.0018 (-0.0025, -0.0010) < 0.0001 | -0.0026 (-0.0035, -0.0017) < 0.0001 | |
| Log likelihood ratio | 0.002 | 0.017 | 0.002 | 0.004 | |
| Race/ethnicity (Mexican American; other Hispanic; non-Hispanic white; non-Hispanic black; other Races), Education (under high school; high school or equivalent; above high school), income to poverty ratio (quartile groups), BMI (obese, overweight, normal), drink status (had at least 12 alcohol drinks past one year; don not have at least 12 alcohol drinks past one year), smoking status (less than 100 cigarettes; greater than or equal to 100 cigarettes), diabetes (yes; no), hypertension (yes; no), ALT (quartile groups), AST (quartile groups), total calcium (quartile groups) and total cholesterol (quartile groups) were adjusted. | |||||
According to the lowest threshold of HDL-C in subjects aged 50 to 60, we used a threshold of 71 mg/dL to investigate whether a high HDL-C level could increase the bone loss risk. After adjustment for confounders. The results of multiple logistic regression models displayed that participants with an equal to or higher than 71 mg/dL HDL-C levels had a significantly elevated prevalence of osteoporosis or osteopenia, especially in subjects aged over 40 (since the sample size of osteoporosis or osteopenia group were much smaller than those of the normal BMD group after weighting, the OR value and 95%CI could not be calculated, so the sample numbers are not weighted in this analysis). No statistically significant associations were found in other age groups. The detailed results are displayed in Table 8.
| 20 ≤ Aged < 30 | 30 ≤ Aged < 40 | 40 ≤ Aged < 50 | 50 ≤ Aged < 60 | 60 ≤ Aged | |
|---|---|---|---|---|---|
| Non-adjusted | |||||
| HDL-C < 71 mg/dL | Reference | Reference | Reference | Reference | Reference |
| HDL-C > = 71 mg/dL | 0.9961 (0.6172, 1.6078) 0.987290 | 1.3181 (0.8361, 2.0780) 0.234326 | 2.0106 (1.4439, 2.7997) 0.000036 | 2.1462 (1.4648, 3.1445) 0.000089 | 1.5311 (1.1042, 2.1230) 0.010636 |
| Adjust | |||||
| HDL-C < 71 mg/dL | Reference | Reference | Reference | Reference | Reference |
| HDL-C > = 71 mg/dL | 1.0532 (0.6220, 1.7832) 0.847032 | 1.3787 (0.8409, 2.2605) 0.203053 | 1.9732 (1.3614, 2.8599) 0.000332 | 2.1305 (1.4041, 3.2327) 0.000377 | 1.7778 (1.2381, 2.5529) 0.001828 |
| Race/ethnicity (Mexican American; other Hispanic; non-Hispanic white; non-Hispanic black; other Races), Education (under high school; high school or equivalent; above high school), income to poverty ratio (quartile groups), BMI (obese, overweight, normal), drink status (had at least 12 alcohol drinks past one year; don not have at least 12 alcohol drinks past one year), smoking status (less than 100 cigarettes; greater than or equal to 100 cigarettes), diabetes (yes; no), hypertension (yes; no), ALT (quartile groups), AST (quartile groups), total calcium (quartile groups) and total cholesterol (quartile groups) were adjusted. | |||||
Osteoporosis is a worldwide public health problem characterized by low BMD and a high risk of osteoporotic fracture [8]. At present, apart from genetic factors, age, or sex, the impact of other factors like lipid metabolism or lifestyle for bone metabolism has recently garnered increasing attention [11–13]. Meanwhile, HDL-C, a critical lipid for human lipid metabolism, has recently been demonstrated to be detrimental to human health, implying that extremely high HDL-C levels can also affect human health and contribute to several diseases [6, 7]. This study demonstrated that HDL-C was negatively associated with BMD, especially in females. Additionally, HDL-C might have a potential predictive value for osteopenia or osteoporosis in females. As a result, we conclude that our findings complement existing research and provide guidance for future studies.
This study results exhibited a negative correlation between HDL-C and BMD, mainly among females but not males. This finding may suggest that correlation is affected by hormone levels, and we found some evidence for this hypothesis. Jirapinyo et al. observed that combined oral estrogen/progestogen increased BMD of spine and hip in post-menopausal women but decreased HDL-C [25]. Han et al. found that tanshinol exerted a bone-protective function by modulating bone turnover markers via blocking NF-κB pathway and decreased HDL-C levels [26]. However, although this phenomenon was observed in some studies, the specific mechanism is yet unclear. Meanwhile, the relationships between HDL-C and BMD were different in diverse age groups. Especially in females aged 20 to 30, although the results of the multivariate regression analyses manifested a negative correlation between HDL-C and BMD, a nonlinear relationship by smooth curve fittings and generalized additive models suggested that BMD increased with increasing HDL-C as HDL-C exceeded the inflection point. The specific reason for such a discrepancy is unclear, and future mechanism studies are required. Besides, we stated that elevated HDL-C levels might be predictive of osteoporosis or osteopenia, implying that patients with a high HDL-C level needed to monitor BMD, especially among females aged over 40. Meanwhile, because of different inflections in diverse age groups, the threshold choices may require adjustments, and More studies will be required to further investigate this aspect.
Some previous studies also explored the association between HDL-C and BMD [14–17]. For example, Maghbooli et al. found in Iranian women that HDL-C levels were negatively correlated with BMD in post-menopausal women with vitamin D deficiency [14]. Zhang et al. demonstrated in Chinese women a negative correlation between HDL-C and BMD in the population above 50 [15]. Makovey et al. observed a modest inverse relationship between hip BMD and HDL-C in post-menopausal women [27]. Jeong et al. found that HDL-C was positively associated with BMD at the lumbar spine in post-menopausal women, but a positive correlation was too weak (β < 0.001) [17]. Cui et al. demonstrated that HDL-C levels were not linked to BMD values at any of the sites in pre- and post-menopausal women [16]. In summary, because the conclusions remain controversial, and we considered some limitations in these studies, like the small sample size, selected population, or adjusted variables, we improved these shortcomings. First, we used a nationally representative sample of NHANES, with huge sample size. Second, since previous studies usually considered the relationship between HDL-C and BMD in females, especially post-menopausal females, this study also considered the potential impact of gender and age. Third, this study adjusted more variables that might potentially influence BMD. As expected, we demonstrate not only a correlation between HDL-C and BMD except in post-menopausal females but also a potential predictive value of HDL-C for osteoporosis or osteopenia.
For a long time, numerous researchers and studies have believed that HDL-C is beneficial to health [28, 29]. Especially in the field of cardiovascular disease [2, 4], HDL-C is considered to be negatively correlated with adverse cardiovascular events [2–5]. However, numerous research results indicated that HDL-C contribution to human health might be highly overestimated. Several years ago, it was demonstrated that drugs that increased HDL-C did not prevent adverse cardiovascular events [30]. Other recent studies reported an inverted U-shaped relationship between HDL-C level and all-cause mortality [7, 31]. All of this indicates that elevated HDL-C levels may detrimental to health and may even cause some adverse events. This study established an inverse relationship between HDL-C and BMD in adult females, corroborating this view. Besides, it is worth mentioning that most basic studies usually focus on the impact of low HDL-C but not high HDL-C on bone metabolism [32, 33]. Other studies published in the last few years have revealed an inverted U-shaped association between HDL-C levels and all-cause mortality. All of this indicates that elevated HDL-C levels can be detrimental to health and may even be a source of some adverse events. This study established an inverse relationship between HDL-C and BMD in adult females, corroborating this view. Additionally, it is worth noting that most basic research focuses on low HDL-C effect on bone metabolism but not on high HDL-C impact. As a result, future research may focus on the specific mechanism underlying elevated HDL-C levels.
Several limitations of this study should be noted:
Our research is based on American participants, so it is hard to say whether our conclusion applies to other countries or races. In subgroup analysis for race, we can also observe that HDL-C and BMD of the relevance between different races afforded different results, implying that our conclusion has some unavoidable limitations.
Our study excluded participants who did not have BMD or HDL-C data, introducing potential bias even though it was inevitable for subsequent analysis.
Part of the data we collected was questionnaire data. The subjects may refuse to answer for personal reasons or make wrong judgments due to recall bias. All these are factors that may potentially affect the research conclusion.
There will inevitably be missing values in the data set. When we deal with these missing values, we choose to exclude rather than fill them indirectly. Although both methods produce certain biases, the accuracy of the conclusion will inevitably be affected.
There remains the possibility of bias caused by other potential confounding factors that were not adjusted.
We concluded that HDL-C and BMD were negatively correlated among females and were different in diverse age groups. Moreover, HDL-C might be predictive for osteopenia or osteoporosis.
Ethics approval and consent to participate
All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. All analyses were based on data of the National Health and Nutrition Examination Survey (NHANES). The study was approved by the ethics review board of the National Center for Health Statistics. The detailed information located on the NHANES website.
Consent for publication
Not applicable.
Availability of data and materials
The datasets obtained and analysed during the current study are available in the NHANES [https://www.cdc.gov/nchs/nhanes/index.htm].
Competing interests
The authors declare that they have no conflict of interest.
Funding
This study was supported by the National Natural Science Foundation of China (81874017, 81960403 and 82060405); Natural Science Foundation of Gansu Province of China (20JR5RA320); Cuiying Scientific and Technological Innovation Program of Lanzhou University Second Hospital (CY2017-ZD02).
Authors’ contribution
YT and SW contributed equally to this work. YT and BG contributed the central idea, YT and SW analyzed most of the data. YT wrote the initial draft of the paper. The remaining authors contributed to refining the ideas, carrying out additional analyses and finalizing this paper.
Acknowledgements
This study was supported by the National Natural Science Foundation of China (81874017, 81960403 and 82060405); Natural Science Foundation of Gansu Province of China (20JR5RA320); Cuiying Scientific and Technological Innovation Program of Lanzhou University Second Hospital (CY2017-ZD02).