Low bone density and abnormal bone turnover in patients with atherosclerosis of peripheral vessels.
- First Online:
- Cite this article as:
- Pennisi, P., Signorelli, S.S., Riccobene, S. et al. Osteoporos Int (2004) 15: 389. doi:10.1007/s00198-003-1550-9
- 221 Views
Patients with vascular calcifications often have low bone mineral density (BMD), but it is still uncertain if osteoporosis and peripheral vascular disease (VD) are interrelated and linked by a common pathomechanism. Moreover, data on bone turnover in patients with advanced atherosclerosis are lacking. We measured BMD by dual-energy X-ray absorptiometry (DXA) and quantitative bone ultrasound (QUS), as well as the serum levels of osteocalcin (OC), bone-specific alkaline phosphatase (BAP), osteoprotegerin (OPG) and its ligand RANKL, and the urinary concentration of the C-terminal telopeptides of type I collagen (CrossLaps), in 36 patient (20 male and 16 female) with serious atherosclerotic involvement of the carotid and/or femoral artery to investigate the underlying mechanism of vascular and osseous disorders. Thirty age-matched and gender matched healthy individuals served as controls. After adjustment for age, BMD was significantly reduced at the lumbar spine in 23/36 (63%) patients (mean T score −1.71±1.42) and at the proximal femur in 34/36 (93%) patients (neck mean T score −2.5±0.88). Ten patients (27%) had abnormal QUS parameters. Gender and diabetes had no effect on the relationship between vascular calcification and bone density at any site measured. VD subjects had OC and BAP serum levels lower than controls (13.3±3.1 vs 27.7±3.3 ng/ml, P<0.01, and 8.4±2.3 vs 12.5±1.4 μg/l, P<0.01, respectively). Urinary CrossLaps excretion was not significantly different in patients with VD and in controls (257.9±138.9 vs 272.2±79.4 µg/mmol Cr, respectively). Serum OPG and RANKL levels were similar in patients and in controls (3.5±1.07 vs 3.4±1.05 pmol/l, and 0.37±0.07 vs 0.36±0.06 pmol/l, respectively). We proved high occurrence of osteoporosis in VD, with evidence of age and gender independence. Negative bone remodelling balance would be a consequence of reduced bone formation, with no apparent increased activation of the OPG–RANKL system.
KeywordsAtherosclerosisBone mineral densityCrossLapsOsteocalcinOsteoprotegerin
Skeletal changes have been reported in patients with cardiovascular disease, suggesting a relationship between osteoporosis and atherosclerosis. Both cross-sectional and longitudinal studies have, in fact, shown that aortic calcification is a common feature among women with osteoporotic vertebral fractures [1, 2] and that the degree of aortic calcification is inversely related to bone mineral density [3, 4]. Moreover, low bone mineral content at the menopause appears to be a risk factor for increased mortality in later life, especially for cardiovascular disease . Van der Klift et al.  reported an important independent association between peripheral artery disease (PAD) and hip bone mineral density (BMD) in postmenopausal women. A more recent study by Tanko et al.  also showed a significant association between aortic calcification and hip BMD in a population of women aged 60–85 years. Indeed, the nature of this relationship is controversial: today there is literature which reports conflicting results , suggesting an association due to chance, as both osteoporosis and aortic calcification increase with age. The existence of confounding factors in the determination of the association between aortic calcification and low bone mass may be hypothesised by the fact that it has been observed only in women, while no apparent association between the progression of bone loss and the progression of aortic calcification or PAD has been observed in men [3, 6].
We hypothesised that more information could be obtained by the study of a population that presented with atherosclerosis and pathological calcifications of the arterial wall not limited to abdominal or thoracic aorta. Furthermore, a possible linkage between vascular calcification and osteoporosis could be investigated if it were kept in mind that calcification of arterial wall should be regarded as an active process, also regulated by many factors (including oestrogen, vitamin K, vitamin D, and osteoprotegerin) [9, 10, 11, 12, 13] involved in the local regulation of bone turnover.
The aim of this study was to evaluate the relationship between atherosclerotic disease and bone mass assessed with dual-energy X-ray absorptiometry (DXA) in a group of patients with detectable atherosclerotic plaques involving one or more vessels. In addition to densitometry, information regarding the metabolic status of bone was obtained by measurement of the serum levels of osteocalcin (OC) and bone-specific alkaline phosphatase (BAP) (both indicators of osteoblast activity and bone formation), and the urinary excretion of the C-terminal telopeptides of type I collagen (CrossLaps), marker of bone collagen degradation and bone resorption . We also measured serum concentration of osteoprotegerin (OPG) and the receptor activator of nuclear factor, kB ligand (RANKL, or OPG ligand), crucially involved in osteoclast function, bone remodelling , and in vascular biology . We examined whether serum levels of OPG and RANKL are associated with loss of bone and the progression of the vascular disease in our patients.
Material and methods
Characteristics of patients with vascular calcifications and of control subjects
Years since menopause
Body mass index (kg/m2)
Systolic blood pressure (mmHg)
Diastolic blood pressure (mmHg)
Total daily calcium intake (mg)
Physical activity index
Therapy with statins (%)
Common carotid and femoral artery intima–media thickness (IMT) was measured with B-mode ultrasonography used as a non-invasive surrogate indicator of the presence and extent of atherosclerosis [19, 20, 21]. B-mode ultrasound imaging was performed on patients while they were in the supine position by use of a US Apogée CX 800 (ATL-Philips, Ind., USA) with a 7-mHz linear electronic probe. Longitudinal and transversal imaging of the carotid artery was performed, as was anterolateral, lateral and posterolateral imaging of patients’ necks. Moreover, longitudinal and transverse ultrasound (US) imaging of the femoral arteries was carried out. Intima–media thickness of the common carotid artery was measured on frozen US images stored in the PC equipped with a gif card and software.
IMT of the carotid artery was measured from the origin of the common carotid 3 cm above and 1 cm before the first carotid bifurcation, whereas longitudinal imaging in longitudinal axis was used to measure IMT of the femoral artery. IMT was calculated as the mean of three determinations of either the carotid or common femoral artery. Plaques were diagnosed by findings of hyperechogenic images inside the artery and were identified either as faint grey echoes (soft plaques) or bright white echoes (calcified plaques) protruding into the lumen; severity of atherosclerosis was evaluated according to a scoring system which groups plaques into five classes: 1. plaque with low echogenicity; 2. low echogenicity with intraplaque echogenicity <50%; 3. low echogenicity with intraplaque echogenicity >50%; 4. plaque with echogenicity; 4a. hyperechogenicity; 4b. normal echogenicity; 4ab. low echogenicity and hyperechogenic zone; 5. no detectable plaque by calcific posterior barrier. US examination was repeated by two different operators who were not aware of the results of the previous examinations, and results variability did not exceed 2%.
Bone mineral density
Areal BMD (grammes per centimetre squared; bone mineral content relative to projection area) was measured by DXA (XR-36, Norland, Fort Atkinson, Wis., USA) for the total body (TB), at the lumbar spine (L2–L4) and the hip (neck, Ward’s triangle and trochanter). At these measurement sites, the precision of the method (coefficient of variation, CV) at our laboratory was 1.2% for total body; 0.7% for the lumbar spine, and 0.5% for the proximal femur. Results for areal BMD were transformed to T scores (calculated as the difference between the actual measurement and the mean value of healthy gender-matched adult controls, divided by their standard deviation), from the data provided by the densitometer manufacturer.
Bone ultrasound attenuation (BUA, decibels per megahertz) was measured at the left calcaneus, with a gel-coupled scanning calcaneal ultrasonometer (QUS-2; Metra Biosystem, Mountain View, Calif., USA). The procedure includes calcaneal scanning, determination of the region of interest (ROI) and measurement of the ultrasound bone index (UBI), which is used to derive BUA. This method has been described in detail by several authors [22, 23, 24]. The QUS-2 can reproducibly scan and assess the same trabecular-rich ROI within and across individuals; this allows one to measure BUA in individuals with extreme foot size without the aid of a positioning device. In our hands, this procedure has been proven to be a relevant tool for the assessment of bone involvement in patients with multi-systemic disease  and shows a short-term CV of 2.5% and a long-term (with repositioning) CV of 3.2%.
Serum for determination of calcium, phosphate, creatinine, osteocalcin, BAP, OPG and RANKL was obtained between 9.00 and 9.30 a.m., with the patient in the fasting state. After the separation, the serum was stored at −20°C until the measurement was performed. Second-void morning urine samples were collected for determination of degradation products of C-terminal telopeptides of type I collagen. Urine collection was made prior to 10.00 a.m. to obviate any potential influence of diurnal variation. All the measurements were performed in the same session, blind to the subject group. Serum calcium, phosphate and creatinine levels were measured by standard laboratory methods. BAP was measured by an enzyme-linked immunoabsorbent assay (ELISA; Tandem-MP Ostase, Beckman Coulter, USA). The sensitivity of the method was 0.7 µg/l. Within-run precision (per cent CV) was 6.5 for low values, and 4.5 for high values. The mean reference values (microgrammes per litre) were 12.3±4.3 in men and 13.2±4.7 in postmenopausal women. Osteocalcin was measured with a commercially available ELISA kit ( N-MID Osteocalcin One Step) provided by Nordic Bioscience Diagnostics A/S, Herlev, Denmark. The sensitivity of the method was 0.5 ng/ml. Intra-assay CV (per cent) was 3.4 for low values and 2.4 for high values. The mean reference values (nanogrammes per millilitre) were 28.4±9.5 for postmenopausal women and 21.4±9.1 for men.
Serum OPG levels were determined by a sandwich ELISA (Osteoprotegerin, Immun Diagnostik, Bensheim, Germany), which utilises a monoclonal anti-OPG antibody to capture OPG from serum. Captured OPG was detected with biotinylated polyclonal anti-OPG antibody, and tetramethylbenzidine (TMB) substrate. The detection limit of this assay system was 0.14 pmol/l; intra-assay and inter-assay CVs were typically <10% at both low and high concentrations of OPG. Mean reference values in adult control subjects were 3.2±1 pmol/l.
Uncomplexed serum RANKL levels were measured by enzyme immunoassay (sRANKL, Biomedica, Vienna, Austria). In accordance with this method, the sample and the biotinylated polyclonal anti-sRANKL detection antibody were pipetted into the wells. Human sRANKL, if present in the sample, binds to the pre-coated recombinant OPG and forms a sandwich with the detection antibody. TMB was added to the wells as substrate. The detection limit of this assay system was 0.08 pmol/l. Intra-assay and inter-assay CVs (per cent) were 5 and 9, respectively. Mean reference values in adult control subjects were between 0 (undetectable) and 4.6±0.87 pmol/l.
Urinary excretion of C-terminal telopeptides of type I collagen was measured by ELISA, with reagents provided by Osteometer Bio Tech A/S, Herlev, Denmark (CrossLaps). The method utilises polyclonal antibodies raised in rabbits, reactive with the amino acid sequence of EKHD-β-GGR, where the aspartic acid residue (D) is β-isomerised. The detection limit of this assay was 50 μg/l. Intra-assay and inter-assay CVs (per cent) were, respectively, 5.7 and 9.4 for low values and 5.4 and 8.6 for very high values. The mean reference values (microgrammes per millimole creatinine) were 378±184 for postmenopausal women and 244±132 for men.
All data are presented as mean± SD. Sample size estimation and power analysis were performed with the StatMate Package by GraphPad Software, San Diego, Calif., USA. Comparisons between groups for study variables were done with the unpaired Student’s t test for normally distributed parameters. Individual bone-mass measurements were expressed in T score. Differences between groups according to the T-score values were analysed by one-way ANOVA followed by the Tukey-Kramer MC test for the multiple comparison procedures. Individual relationship between bone metabolic markers (BAP, OC and Cross-Laps), biochemical variables (OPG, RANKL) and spine and femur T score were evaluated by linear regression. We performed multivariate regression analysis to adjust risk factors. The dependent variables were spine and femur T scores. The independent variables were age, gender and diabetes. In all analyses, P<0.05 was considered significant. Statistics were done with GraphPad InStat version 3.00 for Windows (GraphPad Software).
Characteristics of men and postmenopausal women in the study sample are shown in Table 1. The mean body mass index (BMI) was 26.7±2.7 in men and 27.03±4.4 in women. All subjects were normocalcaemic. The prevalence of cardiovascular risk factors was 50% in both genders for diabetes, and 45% in men and 50% in women for hyperlipidaemia. Six men and five women were being treated with statins.
On the basis of vascular ultrasound examination, all patients were characterised to have significant atherosclerotic involvement of the carotid and/or femoral artery. Two patients had a score of 3, 32 had a score of 4, and two a score of 5.We found 25 patients with one-vessel lesion (five with femoral only, and 20 with carotid only), and 11 patients with involvement of both carotid and femoral artery.
Systolic and diastolic blood pressure and total daily calcium intake were in the normal range and did not significantly differ between men and women. Physical activity was found to be normally distributed.
Bone density and QUS
Areal BMD and ultrasound parameters in 36 patients with vascular calcification
Subjects with T score between −1 and −2.5
Subjects with T score <−2.5
Mean ± SD
Mean ± SD
Densitometric parameters for BMD (g/cm2)
Triangle of Ward
BUA at the left calcaneus
Serum levels of OPG did not differ statistically in patients and controls (3.5±1.07 and 3.4±1.05 pmol/l, respectively). We found no difference in serum OPG levels when stratifying the results by other variables including gender, age, diabetes, hyperlipidaemia, and the presence of atherosclerotic plaques involving both carotid and femoral artery. Serum RANKL levels were detectable in all patients and controls. There was no significant difference compared with healthy subjects (0.37±0.07 and 0.36±0.06 pmol/l, respectively).
If it is assumed that the smallest difference we think is important between patients and controls is 20% for OPG and RANKL measurements, calculation shows a 80% and a 95% power of our study to find a statistically significant difference in OPG and RANKL serum levels, respectively.
We found that serum OC levels were significantly reduced in patients (16.3±3.1 ng/ml) compared with healthy controls (27.7±3.3 ng/ml; P<0.01). BAP serum levels were significantly lower in patients than in controls (8.4±2.3 and 12.5±1.4 µg/l, respectively; P<0.01). The C-terminal telopeptides of type I collagen (CrossLaps) urinary excretion was not significantly different in patients and in controls (257.9±138.9 µg/mmol Cr, and 272.2±79.4 µg/mmol Cr, respectively).
Serum OPG, RANKL, OC, and BAP, and urinary C-terminal telopeptides of type I collagen (CrossLaps) concentration in patients with vascular calcifications and in controls (NS not significant)
CrossLaps (µg/mmol creatinine)
Our study shows that patients with carotid and/or femoral atherosclerotic plaque have a considerably low total body and regional bone mass. Although the hip appears to be the most affected region in our patients, the loss of bone is not confined to legs, as it is also observed at the lumbar spine, thus suggesting that sites containing different proportions of cancellous and cortical bone are involved to a comparable extent. Our data seem to be in contrast to the earlier findings of Van der Klift et al. , who found no association between lumbar spine BMD and PAD in either men or women. A possible explanation for this difference could be the fact that patients studied by Van der Klift and co-workers were older than those enrolled in our study (72.1±8.09 vs 61.6±6.09 years for women and 71.3±8.4 vs 64.1±7.1 years for men, respectively).
Osteoarthritis, which is associated with an increased BMD, occurs, in fact, quite often in the spine in the elderly, whereas the femoral neck is less often involved . This may then result in an overestimation of the lumbar spine BMD and may have concealed a relationship between BMD and PAD in the population studied by van der Klift et al. Our data are consistent with previous reports showing an association between aortic calcification or carotid atherosclerosis in postmenopausal women and low bone mass [3, 7, 27]. In all these studies, however, multivariate analysis revealed a strong association between low bone mass and age. In the study by Aoyagi et al. , a loss of the significance of the differences between women with and without aortic calcification, after age and other covariates had been adjusted for, was also observed.
The observation of a lack of significant correlation between age and bone loss in our patients, together with the existence of several similarities in the pathophysiology of both vascular calcification and bone mineralisation, would, however, encourage further investigation of a possible unifying aetiology to the finding of an association between bone loss and atherosclerotic vascular calcification. The presence of a generalised osteopenia (osteoporosis) in our patients suggests that the changes leading to systemic atherosclerosis and to calcification of plaques also affect bone metabolism. The underlying mechanism that triggers bone mineral loss in patients with atherosclerosis is unknown.
In a study by Laroche et al.  it was reported that in patients presenting with symptomatic arterial disease of the lower limbs the mean BMC of the leg more severely affected was significantly lower than that of the leg less affected, suggesting that demineralisation was related to a direct local effect of atherosclerosis and not to associated general risk factors. Kiel et al.  suggest that any unifying hypothesis must account for the difference they observed between men and women: according to this author, the lack of an association in men would raise the possibility that hormonal factors unique to women (i.e. oestrogens) may emerge as potential candidates to play a role in the underlying pathophysiology. Oestrogen deficiency contributes to decreased BMD in postmenopause  and is reported to be associated with increased oxidised LDL-cholesterol . Osteoblastic differentiation of pre-osteoblasts from bone is inhibited by minimally oxidised LDL, whereas oxidised lipids enhance differentiation of osteoblast-like cells from the artery wall, ultimately inducing vascular mineralisation and calcification [31, 32]. Oestrogen deficiency is also associated with an increase in parathyroid hormone  (which is in turn associated with low femoral bone density  and cardiovascular disease ), and with increased serum levels of homocysteine, possibly associated with a decrease in BMD at later ages .
Oestrogen plays an important role in skeletal homoeostasis in men as well . Men and women in our study had a comparable degree of severity of vascular lesions as assessed by plaque score. We were not able to find a significant difference between men and women with respect to their BMD for the total body, at the lumbar spine, the femoral neck, and the Ward’s triangle: although we have no information on the oestrogen status in our men, the fact that they appear to lose as much bone as women makes apparently unlikely the hypothesis that osteopenia and osteoporosis in our patients are dependent only on oestrogen deficiency.
Besides oestrogen deficiency, at least two additional factors might be involved in both atherosclerosis and osteopenia. First, there is evidence that excess vitamin D induces both atherosclerosis and osteoporosis in humans and laboratory animals, and that the incidence of osteoporosis and atherosclerosis is increased in countries where vitamin D is added to the food supply . Secondly, population-based studies report radiographic evidence of aortic calcification associated with an impaired vitamin K status and with low bone mass , suggesting that vitamin K might also be involved. Vitamins D and K status was not available in our study; however, patients enrolled in the current study did not receive vitamin D supplementation nor were they treated with oral anticoagulant. Once more, the hypothesis of a vitamin D excess or vitamin K deficiency as a possible cause of the association between atherosclerosis and osteopenia in our patients seems unlikely.
Interpretation of laboratory data is based on the assumption that circulating and urinary levels of markers of bone turnover reflect what happens in the bone microenvironment. The fact that serum levels of OC and BAP are reduced may suggest that in patients with vascular disease osteoblast function is lower than normal. The insignificant difference of RANKL levels between patients with vascular disease and controls contradicts the hypothesis that loss of bone results from increased activation of the RANKL–RANK system. According to our data, there is no evidence that decreased bone density in patients with vascular disease is related to the RANKL–OPG system. It should be stressed, however, that existing data from serum OPG and RANKL in osteoporosis—in which has been demonstrated the involvement of the OPG–RANKL system—are quite discordant. Thus, it is difficult for one to draw conclusions about the role of the OPG–RANKL system in patients with atherosclerosis, based on serum findings.
Patients in this study, despite the involvement of more than one vessel, had no heart disease or coronary artery disease and had shown a normal physical activity index. The possibility that non-specific factors such as reduced activity or loss of weight and muscle strength might play a role in the pathogenesis of osteopenia could, therefore, be ruled out. With respect to the development of bone changes in patients with vascular disease, it would be interesting to mention the animal model of “inflammation-mediated osteopenia” [38, 39]. According to this model, an unspecific inflammation process of a certain extent, induced in rats, is followed by bone loss due to inhibition or impairment of osteoblastic function. It is feasible that patients with vascular disease and/or vascular calcification suffer from a kind of inflammation, as suggested by the increase in serum concentration of the acute-phase reactant C-reactive protein. Bone loss would, therefore, reflect cumulative inflammation stimuli over years or decades. However, data in humans, supporting this hypothesis, are lacking.
The clinical relevance of the decreased bone mass associated with vascular disease is still not clear and needs further longitudinal studies that also include the incidence of osteoporotic fractures. We have, moreover, no data on bone quality in these patients, and there is no study that attempts to investigate whether the fracture-healing process is altered, as is the case of patients with diabetes mellitus .
We suggest that evaluation of bone status should be done in patients with vascular disease in order that preventive or therapeutic intervention may be applied.
This work was supported by funds from the Italian Ministry of Scientific Research (MIUR)