Decreased Bone Mineral Density Is an Independent Predictor for the Development of Atherosclerosis: A Systematic Review and Meta-Analysis

Background There is conflicting evidence regarding the association between decreased bone mineral density (BMD) and atherosclerosis. To this end, we performed a systematic review and meta-analysis to clarify the association. Methods To identify relevant studies, PubMed, Embase, and the Cochrane Library were systematically searched up to November 2015. All observational and comparative studies directly investigating the relationship between decreased BMD and clinical consequences of atherosclerotic vascular abnormalities, including carotid artery calcification (CAC), cardiovascular disease (CAD), and coronary artery disease (CAD) were obtained, without limitation of language or publication year. Results A total of 25 studies involving 10,299 patients were included. The incidence of atherosclerotic vascular abnormalities was significantly increased in low BMD patients, compared to patients with normal BMD (OR, 1.81, 95% CI [1.01, 2.19], p<0.00001)). Similar results were also observed for postmenopausal women (OR, 2.23, 95% CI [1.72, 2.89], p<0.00001). Subgroup analyses of osteopenia, osteoporosis, and normal BMD also revealed that the combined ORs for the incidence of atherosclerotic vascular abnormalities increased as BMD decreased. Of note, after adjusting for age, sex, body mass index (BMI) and other vascular risk factors, decreased BMD remained significantly associated with the incidence of atherosclerotic vascular abnormalities (OR, 2.96, 95% CI [2.25, 3.88], p < 0.00001). Conclusions Based on the results of this study, decreased BMD is an independent predictor for the development of atherosclerosis in elderly individuals. Moreover, the risk of atherosclerotic vascular abnormalities increased as BMD decreased. Future studies focusing on individuals with different severities of atherosclerosis and comorbidities are of interest.


Introduction
Osteoporosis (OP) and atherosclerosis are the two most common diseases in elderly individuals and are associated with significant morbidity, mortality, and disability. OP affects 44 million Americans (55%) over the age of 50 years [1,2]. According to data from the World Health Organization (WHO), elderly individuals will comprise 20% of the population by 2050 [3][4][5], increasing the prevalence of OP and atherosclerosis.
Although mounting epidemiological evidence suggests an association between OP and atherosclerosis, these diseases are traditionally considered unrelated, with their co-existence being caused by independent age-related processes. Recently, a growing number of studies have linked atherosclerosis to OP in elderly individuals [3,4,[6][7][8][9][10][11][12][13]. The association, however, is not fully understood and controversy still exists regarding whether it is gender-related. Moreover, the connection between bone mineral density (BMD) and atherosclerosis is also controversial. Several studies suggested that no relationship exists between decreased BMD and atherosclerosis [9,[14][15][16][17][18]. In another study, osteoporosis and atherosclerosis were independent processes in postmenopausal women, after adjusting for age, gender, BMI, hypertension, and other vascular risk factors [4]. Beer et al. also revealed that low BMD was not associated with coronary atherosclerosis in men [5].
In the absence of any studies that are sufficiently powered to examine the relationship between low BMD and atherosclerosis, a meta-analysis of previously conducted investigations is useful. Thus, we conducted a systematic review and meta-analysis of previous studies for the following purposes: (1) to investigate and compare the incidence of atherosclerotic vascular abnormalities in individuals with normal BMD, osteopenia, and osteoporosis; (2) to explore the relationships between low BMD and atherosclerotic vascular abnormalities among the complete population and in postmenopausal women; and (3) examine the relationships after adjusting for age, sex, body mass index (BMI) and other vascular risk factors.
"CVD" or "coronary artery disease" or "CAD" or "atherosclerotic vascular disease" or "AVD" or "coronary micro-vascular endothelial dysfunction" or "CMED". Details of the search strategy are provided in S1 Table and S1 and S2 Texts. Reference lists were also searched manually for additional studies. In addition, we searched the National Institutes of Health, the Clinical Trial Registry, the Trials Central and the Center Watch, and the Current Controlled Trials for grey literature.

Inclusion criteria
Two reviewers (CY and MX) independently screened manuscript titles and abstracts, and implemented the following inclusion criteria: (1) observational studies that examined the association between low BMD and clinical consequences of atherosclerotic vascular abnormalities, including carotid artery calcification (CAC), cardiovascular disease (CAD), and coronary artery disease (CAD); (2) studies that enrolled individuals aged 18 years and older; and (3) studies that provided data pertaining to BMD in participants with and without atherosclerotic vascular abnormalities.
The following exclusion criteria were used in this study: (1) articles that did not satisfy the inclusion criteria; (2) animal studies, reviews, letters, abstracts, case reports, conference proceedings, and systematic reviews; and (3) studies that did not provide sufficient data on BMD values, including the means, medians, standard deviations (SDs), and/or standard errors for subjects with and without atherosclerotic vascular abnormalities. Any disagreements were evaluated using a kappa test and consensus was achieved by discussion with the corresponding author (RH).

Data extraction and quality assessment
Three independent reviewers (SJ, XC, XZ) obtained relevant data and assessed the accuracy. The following information was extracted from each study: first author's family name, year of publication, country, study design, patient demographics (age, gender, sample size), BMD values, assessment methods, and definition of atherosclerotic vascular abnormalities. Contact with corresponding authors was also attempted to verify the accuracy of the data, as well as to obtain further data for the analysis. The quality of the methodological data included in the studies was assessed using the Newcastle-Ottawa Scale (NOS), as recommended by the Cochrane non-randomized studies methods working group [20]. All articles were scored according the three major categories of NOS (selection, comparability andassessment of outcome). The maximum scores for selection, comparability and exposure were 4, 2 and 3, respectively. Studies were considered of high quality if at least five of the nine criteria were met.

Statistical analysis
Revman software (ver. 5.3; The Nordic Cochrane Centre, Copenhagen, Denmark) was used to pool the data. A p-value 0.05 was considered significant. Summary odds ratio (OR) and its 95% CI were calculated to assess the association between low BMD and the occurrence of atherosclerotic vascular abnormalities across studies. For continuous data, the SD was calculated using the method described by Walter et al. [21]. Fixed-and random-effects models were used to calculate the combined ORs. Q and I 2 statistics were used to assess the heterogeneity of the ORs across multiple studies [22]. A p-value < 0.10 was considered significant. If no statistical heterogeneity was detected between studies (P > 0.10; I 2 < 50%), a fixed-effect test (Mantel-Haenszel test) was used to pool the data. Otherwise, the random-effect (Der Simonian-Laird method) model was applied. A sensitivity analysis was also performed to explore possible explanations for heterogeneity.
Subgroup analyses were conducted according to the severity of low BMD and gender, to evaluate the potential effect on outcomes of modifying these variables. The symmetry of funnel plot was also employed to evaluate publication bias. Furthermore, Egger's test and Begger's test were performed to assess the publication bias.

Study characteristics
The characteristics and quality assessment results of the 25 studies are summarized in Table 1. Of those, two studies were performed in Italy, six in the US, three in China, three in Japan, two in Turkey, two in Korea, one in Morocco, one in Romania, one in Poland, one in Austria, one in Iran, one in Egypt, and one in Finland. Participants ranged in age from 46 to 89 years, and five of the studies included both males and females [5,18,24,26,37]. Nineteen studies included only post-menopausal individuals and one only consisted of male participants [15-17, 23, 25, 28-30, 32-35, 37, 39-42].
The definition of arthrosclerotic vascular abnormalities was determined by the included studies, which referred to: carotid artery calcification (CAC); cardiovascular disease (CVD); and coronary artery disease (CAD). The pulse wave velocity (PWV), breast arterial calcification (BAC), intima-media thickness (IMT) and whole blood viscosity (WBV) were also recorded whenever available. The definition of normal BMD, osteopenia, and osteoporosis was made according to the criteria of the WHO. Osteoporosis was defined as BMD minus 2.5 standard deviations (SD) (or lower than the young adult mean), while osteopenia was defined as BMD between 1 and 2.5 SD below the young adult mean [51]. Low BMD was defined as a combination of osteoporosis and osteopenia. Accordingly, 1,784 individuals were included in the normal BMD group, while 6,959 individuals were included in the low BMD group. There were 3,185 individuals in the osteopenia group, and 3,683 individuals were included in the osteoporosis group.
Seven studies comprised of 5,850 participants were adjusted for age, gender, BMI, hypertension, and other vascular risk factors [23,24,29,36,[39][40][41]. The adjusted ORs with 95% CIs are listed in Table 2. The overall quality of the studies averaged a score of 6 points (range: 5-7 points) on a scale of 0 to 9.
Primary results. The ORs for each study and the combined OR for patients with low and normal BMD are shown in Fig 2. The combined OR for the incidence of atherosclerotic vascular abnormalities in patients with low BMD, compared to patients with normal BMD, was 1.81 (95% CI [1.01, 2.19], p < 0.00001).
Subgroup and sensitivity analysis. The results of the subgroup analyses that were based on the severity of the decreased BMD are shown in Table 3. The combined OR for the incidence of atherosclerotic vascular abnormalities in patients with osteopenia, compared to patients with normal BMD, was 1.84 (95% CI [1.45, 2.32], p < 0.00001). The combined OR for the incidence of atherosclerotic vascular abnormalities in patients with osteoporosis, compared to patients with normal BMD, was 2.05 (95% CI [1.55, 2.71], p < 0.00001). Finally, when comparing patients with and without osteoporosis, the combined OR for the incidence of atherosclerotic vascular abnormalities was 2.45 (95% CI [1.90, 3.17], p < 0.00001) (Fig 3).
No evidence of significant heterogeneity was observed within any subgroup, while a significant association between incidence of atherosclerotic vascular abnormalities and decreased BMD was observed in all subgroups.
To test the robustness of the data, sensitivity analyses were performed by omitting one study at a time. These analyses yielded similar results to those obtained when all studies were analyzed simultaneously. The pooled ORs were all statistically significant.   5).   Thus, the results revealed that under all circumstances, the combined OR for the incidence of atherosclerotic vascular abnormalities increased as BMD decreased.

Adjusted results
After adjusting for age, gender, BMI, hypertension, and other vascular risk factors, the combined OR for the incidence of atherosclerotic vascular abnormalities in patients with low BMD versus patients with normal BMD was 2.96 (95% CI [2.25, 3.88], p < 0.00001; I 2 = 29%) (Fig 6).

Publication bias
No substantial asymmetry was identified using Begg's rank correlation test (z = 1.25, p = 0.211) and Egger's regression test (t = 1.38, p = 0.187). Funnel plots of the combined ORs for patients with low BMDs, compared to patients with normal BMDs also showed no publication bias (Fig 7).  Forest plot shows that the incidence of atherosclerotic vascular abnormalities is significantly higher in individuals (including male and female) with low BMD than those with normal BMD.

Discussion
The systematic review and meta-analysis presented here revealed a strong link between decreased bone mineral density and the risk of atherosclerotic vascular abnormalities. This finding was independent of age, gender, BMI, hypertension, and other vascular risk factors. The combined ORs increased as BMD decreased. Moreover, the link between deceased bone mineral density and atherosclerotic vascular abnormalities was not only seen in postmenopausal women, but also in elderly men (aged above 50 years). These findings suggest that postmenopausal women and elderly men with decreased BMD have a higher risk of developing atherosclerotic vascular abnormalities. Our results not only confirmed a significant relationship between osteoporosis and atherosclerotic vascular abnormalities, as previous studies have suggested [3, 4, 6-13], but also a relationship between osteopenia and atherosclerotic vascular abnormalities. Based on the results of our study, patients who have decreased BMD scores, should be screened for cardiovascular risk, as it is an independent indicator of significant atherosclerotic burden. Besides, an enhanced understanding of the relationship between decreased  BMD and atherosclerosis might lead to the development of therapeutic agents for the prevention and treatment of both osteoporosis and atherosclerotic vascular abnormalities.
In this study, no evidence of significant heterogeneity was observed between studies, except for the comparison between postmenopausal women with and without osteoporosis. The following reasons may account for this finding: the definition of atherosclerotic vascular abnormalities is clear and the severity of bone loss is definite, according to the standard of the WHO. Moreover, we included only high quality studies with a NOS score of five or more (S5 Text). In the subgroup analysis of postmenopausal women with and without osteoporosis, medium heterogeneity (I 2 = 61%) was observed. When excluding the data of Reddy et al. [29], which used the breast arterial calcification (BAC) as a defined index of atherosclerotic vascular abnormalities, the heterogeneity reduced to 41% (p = 0.13).
The key pathology in atherosclerotic vascular abnormalities is the formation of calcified plaque and chronic inflammation on the endothelial surface, which progressively leads to plaque rupture and thrombosis (S3 Text) [52][53][54][55]. Superficial erosion and fibrous cap rupture are the two possible mechanisms of plaque disruption [54][55][56]. Inflammatory processes of the vascular wall should play a particularly important role in regulating the integrity of the fibrous cap. It is reported that inflammatory cytokines produced by leukocytes increase degradation of the structural components of the fibrous cap, which further increase the vulnerability of the plaque Fig 6. Forest plot shows that the incidence of atherosclerotic vascular abnormalities is significantly higher in individuals with low BMD than those with normal BMD, after adjusting for age, gender, BMI, hypertension, and other vascular risk factors.
to rupture [57,58]. Moreover, biomechanical studies showed that the increased circumferential stress also decrease the stability of the atherosclerotic lesion and increase the incidence of plaque rupture and thrombosis (S3 Text) [59].
OPG/RANK/RANKL system is reported to play a key role in calcium deposition and bone formation [68]. RANKL is mainly expressed by osteoblast precursors. The combination between RANKL and RANK promote osteoclast formation and lead to bone resorption [69]. OPG is mainly produced by osteoblasts, and vascular smooth muscle cells (VSMCs), and is a potent inhibitor of osteoclast activation [70]. It is reported that in patients with atherosclerotic vascular abnormalities, OPG is over expressed lining the calcific deposits, while RANKL is expressed in the extracellular matrix around the calcific lesions. The OPG-knockout mice developed both osteoporosis and arterial calcification [71]. These findings suggest that the possible mechanism of OPG/RANK/RANKL on atherosclerosis may due to its differential and site-specific expression [72]. MGP and OC are both Gla-containing proteins that inhibit osteoid formation. It is reported that MGP and OC are both expressed in bone and vascular wall and are both up-regulated in atherosclerotic vessels. Furthermore, as key inhibitors of arterial calcification, MGP and OC inhibit BMP-induced chondrocyte differentiation and osteogenic differentiation of the vascular mesenchyme [73]. In animal experiments, knockout mice models of MGP, displayed both vascular calcification and osteoporosis [60]. In women with osteoporosis and atherosclerosis, serum MGP and OC are also elevated [74]. The possible mechanisms of other bone matrix proteins were shown in (S4 Text), including OPN, ON, BMP etc., which help providing an overview regarding the pathogenesis of concomitant osteoporosis and atherosclerosis.
Besides the potential mechanism introduced above, some characteristics may also influence the risk of atherosclerotic vascular abnormalities in postmenopausal women. Estrogen can prevent the development of atherosclerotic vascular abnormalities. For postmenopausal women, deficiency of estrogen not only impairs normal bone remodeling by breaking the balance of osteoblasts, but also increases the risk of cardiovascular diseases (CAD) [75]. Free iron deposits in tissues, which are responsible for iron overload, also play an important role in inhibiting osteoblast proliferation, which impairs bone remodeling [75]. Postmenopausal women showed a stronger correlation between low BMD and the development of atherosclerotic vascular abnormalities [3,11,76,77]. In a study by Sung et al. [78], the authors showed that both premenopausal and postmenopausal women had a correlation between low BMD and atherosclerotic vascular abnormalities.
Uyl et al. conducted a systematic review in 2011 and indicated that persons with prevalent subclinical CV disease are at increased risk for bone loss, however, the association between different extent of low bone mineral density and increased cardiovascular risk was still not clear [7]. To clarify it, we included 25 studies with a NOS score 5 points and compared the incidence of atherosclerotic vascular abnormalities in people with different extent of decreased BMD. To our knowledge, this is the first systematic review and meta-analysis comparing the incidence of atherosclerotic vascular abnormalities in individuals with normal BMD, osteopenia and osteoporosis. Moreover, it is a relatively comprehensive and up-to-date summary of data on the topic. Male and female participants were included, and gender-specific analyses were performed in the subgroup analysis. Compared to the original studies, the large number of cases and participants increased the detection of significant associations and provided more precise estimates of their effects.
This systematic review and meta-analysis also has several limitations. Because of the paucity of randomized controlled trials, this study only evaluated case-control, cross-sectional and cohort studies, which decreases the robustness of the conclusions. Secondly, most of the included studies were based on healthy adults, which can limit the generalizability of the findings to patients with multiple atherosclerosis risk factors and comorbidities. Although we managed to conduct subgroup analyses regarding different severity of decreased BMD, we had limited ability to assess the relationship between different pathologic stages of atherosclerotic vascular abnormalities and decreased bone mineral density due to the lack of available studies. Future quantitative research focusing on individuals with different severities of atherosclerotic vascular abnormalities, and patients with different comorbidities, will be of interest.

Conclusion
Based on the results of this study, decreased BMD is an independent predictor for the development of atherosclerosis in elderly individuals. Additionally, the risk increases as BMD decreases. Patients who have decreased BMD scores, should be screened for cardiovascular risks. Future studies focusing on individuals with different severities of atherosclerotic vascular abnormalities and patients with different comorbidities would be of interest.