Association between lipid biomarkers and osteoporosis: a cross-sectional study

Worldwide, osteoporosis and cardiovascular diseases (CVDs) are 2 major public health problems [1, 2]. Up to 49 million individuals have osteoporosis in North America, Europe, Japan, and Australia [3]. In 2017, CVDs lead to approximately 17.8 million deaths worldwide [4, 5]. Osteoporosis and CVDs both significantly increase risk of morbidity and mortality [6, 7].

Osteoporosis and CVDs may be linked [8, 9]. Osteoporosis and CVDs share many risk factors, such as advancing age, premature menopause, genetics, a sedentary lifestyle and smoking, etc. [10, 11]. There is evidence suggesting that individuals with CVDs have 1.69 folds higher risk of osteoporosis than those without CVDs [12]. In addition, osteoporotic patients are 1.2 to 1.4 times more susceptible to cardiovascular events than those without osteoporosis [10, 13].

Lipid profile is useful for assessing risk of CVDs [14]. There is also vitro evidence suggesting lipids are involved in the development of osteoporosis [15].However, the association between lipid profile and osteoporosis in human studies is still controversial [16‐25]. Majority of human studies suggested that there were negative associations between lipid biomarkers and bone mineral density (BMD) [16, 17, 19, 22‐25]. However, Ersoy et al. [18] and Lahon et al. [20] found that low-density lipoprotein cholesterol (LDL-C) was positively correlated with BMD. Ghadiri-Anari et al. [21] suggested a null association between lipids and BMD. Because of the conflicting associations of lipids with BMD and/or osteoporosis, we examined the associations of total cholesterol (TC), LDL-C, high-density lipoprotein cholesterol (HDL-C) and triglyceride (TG) with osteoporosis in a Chinese hospital-based population.

After excluding ineligible individuals (Fig. 1), we included 481 individuals for this study. As compared to excluded individuals, included individuals were older (62.7 ± 8.6 vs. 45.2 ± 10.1 years; P < 0.001) and more likely to be female (60.1 vs. 49.0 %; P < 0.001); BMI was not significantly different between included and excluded individuals (25.7 ± 3.5 vs. 25.7 ± 4.3 kg/m²; P = 0.886). We only had menopausal age for 234 women (81.0 % of all women included in this study). Based on the women with complete data on menopausal age, we found that women with osteoporosis had younger menopausal age than those without osteoporosis (47.7 ± 3.5 vs. 49.4 ± 3.2 years; P < 0.001).

Among included individuals, 357 (74.2 %), 228 (47.4 %) and 84 (17.5 %) individuals had type 2 diabetes, hypertension and ischemic heart disease diagnoses, respectively. The mean age for included individuals was 62.7 years (SD = 8.6 years); 60.1 % of individuals were female; the average BMI for included individuals was 25.7 kg/m² (SD = 3.5 kg/m²). As shown in Table 1, osteoporotic individuals tended to be older, more likely to be female and had higher proportion of calcium channel blockers medication, angiotensin converting enzyme inhibitors medication, higher levels of TC, alkaline phosphatase and a lower proportion of type 2 diabetes, biguanide medication, lower levels of BMI, alanine transaminase, albumin and total bilirubin.

As shown in Table 2, among all 481 individuals, in the adjusted model, each SD increase in TC (Odds Ratio [OR]: 1.48; 95 % Confidence Interval [CI]: 1.06–2.07) and TG (OR: 1.67; 95 % CI: 1.16–2.39) were associated with increased risk of osteoporosis; LDL-C and HDL-C levels were not associated with osteoporosis. Age, sex and BMI did notmodify the relationships of TC and TG with osteoporosis (all P > 0.10). When T-score of BMD was treated as continuous variable, we observed similar trends of associations of TC and TG with BMD T-score though these associations did not reach statistical significance (Table S1). After classifying lipid biomarkers according to their clinical cut-off values, we found abnormal higher TG levels in serum were associated with higher risk of osteoporosis (Table S2). In women, after further adjusting menopausal age, we found that TG was still significantly associated with osteoporosis (OR: 1.98; 95 % CI: 1.24–3.17; Table S3). In the subgroup analysis of individuals with any chronic disorders, we also found elevated TC and TG levels were associated with higher risk of osteoporosis (Table S4).

In this study of individuals aged ≥ 50 years, we demonstrated that higher TC and TG levels were associated with greater risk of osteoporosis. Serum LDL-C and HDL-C levels were not associated with osteoporosis.

The positive relationships of TC and TG levels with osteoporosis in our study are consistent with some previous studies [19, 23, 25, 30], in which there were positive correlations of TC and TG levels with osteoporosis. We found no significant associations of HDL-C and LDL-C with osteoporosis. These results are also in line with the findings reported by Ghadiri-Anari A et al. [21], those found insignificant associations between lipids and BMD. However, existing epidemiological studies suggested that there are still conflicting associations of HDL-C and LDL-C with osteoporosis [17, 18, 20]. These conflicting results are likely attributed to the different study sample sizes, study populations and covariates adjusted in the study. For example, the insignificant association between LDL-C and osteoporosis in our study may be a false negative finding due to the limited sample size. In addition, numerous studies suggested that different study populations may lead to different results [16, 22, 23]. Majority of included individuals (82.5 %) in our study had chronic diseases (i.e., type 2 diabetes, hypertension and ischemic heart disease). These diseases were suggested to be related to lipids and osteoporosis [31‐33]. However, most of the studies on the association between lipids and osteoporosis were based on generally healthy populations [22, 25, 34]. This may lead to conflicting results. Lastly, as compared to previous studies [16, 21, 24], they used different covariates to adjusting the association between lipids and osteoporosis. This may also cause inconsistent results between studies.

Regardless of consistent or conflicting findings, the positive associations of TC and TG levels with BMD can be explained by several biological mechanisms. First, the nuclear hormone receptor peroxisome proliferator activated receptor γ (PPAR γ) may play a role in the relationship of lipid biomarkers and BMD. PPAR γ can be activated by lipid metabolites. When PPAR γ level increases, osteogenesis is inhibited [35]; this leads to increased bone loss. Second, higher lipid levels are associated with an increase in oxidized lipids and higher oxidative stress levels. Higher oxidative stress level could inhibit osteoblast differentiation as well as to promote adipocyte differentiation [36, 37]. Third, higher serum TG levels are positively associated with higher bone marrow fat [38], which leads to lower trabecular BMD [39].

Our study has some strengths. DXA was used to diagnose osteoporosis. In addition, the lipid biomarkers (e.g., TC, LDL-C, HDL-C, TG) were measured using established methods in the clinical setting [40]. This ensured the reliability and accuracy of osteoporosis ascertainment and lipid measurements, which allowed us to analyzing the associations between lipids and osteoporosis with less bias. Several limitations for this study are acknowledged. First, the findings were derived from a cross-sectional study. Therefore, we cannot make causal inference about the association between lipid biomarkers and BMD. Second, we only analyzed some inpatients and outpatients in The Second Hospital of Jilin University. For example, 74.2 % of individuals in our study were type 2 diabetic patients, while the prevalence of type 2 diabetes in Jilin Province is 17.5 % [41]. Thus, the findings from this selected population may not be able to generalize to general population or the population in the hospital. Last, due to the absent data, we did not consider individuals’ lifestyle factors (i.e., smoking and alcohol use and physical activity) and vitamin D levels. However, in a previous study, adjustment of vitamin D levels did not appreciably change the associations of TC and TG with osteoporosis [42].

In conclusion, there were associations between serum TC and TG and osteoporosis in this study. These findings extended our understanding about the lipid profile and BMD. High quality prospective studies on relationship between lipid biomarkers and BMD are still warranted to confirm or oppose our results.