QCT Volumetric Bone Mineral Density and Vascular and Valvular Calcification: The Framingham Study

Abstract

There is increasing evidence that bone and vascular calcification share common pathogenesis. Little is known about potential links between bone and valvular calcification. The purpose of this study was to determine the association between spine bone mineral density (BMD) and vascular and valvular calcification. Participants included 1317 participants (689 women, 628 men) in the Framingham Offspring Study (mean age 60 years). Integral, trabecular, and cortical volumetric bone density (vBMD) and arterial and valvular calcification were measured from computed tomography (CT) scans and categorized by sex-specific quartiles (Q4 = high vBMD). Calcification of the coronary arteries (CAC), abdominal aorta (AAC), aortic valve (AVC), and mitral valve (MVC) were quantified using the Agatston Score (AS). Prevalence of any calcium (AS >0) was 69% for CAC, 81% for AAC, 39% for AVC, and 20% for MVC. In women, CAC increased with decreasing quartile of trabecular vBMD: adjusted mean CAC = 2.1 (Q4), 2.2 (Q3), 2.5 (Q2), 2.6 (Q1); trend p = 0.04. However, there was no inverse trend between CAC and trabecular vBMD in men: CAC = 4.3 (Q4), 4.3 (Q3), 4.2 (Q2), 4.3 (Q1); trend p = 0.92. AAC increased with decreasing quartile of trabecular vBMD in both women (AAC = 4.5 [Q4], 4.8 [Q3], 5.4 [Q2], 5.1 [Q1]; trend p = 0.01) and men (AAC = 5.5 [Q4], 5.8 [Q3], 5.9 [Q2], 6.2 [Q1]; trend p = 0.01). We observed no association between trabecular vBMD and AVC or MVC in women or men. Finally, cortical vBMD was unrelated to vascular calcification and valvular calcification in women and men. Women and men with low spine vBMD have greater severity of vascular calcification, particularly at the abdominal aorta. The inverse relation between AAC and spine vBMD in women and men may be attributable to shared etiology and may be an important link on which to focus treatment efforts that can target individuals at high risk of both fracture and cardiovascular events.

Introduction

Atherosclerosis, heart valve sclerosis, and osteoporosis are widespread disorders that contribute to serious morbidity and mortality in our aging population. Osteoporosis is a skeletal disease characterized by reduced bone mineral density (BMD), bone structure deterioration, and increased frequency of fractures. Atherosclerosis is characterized by the deposition of calcified plaques in the vascular bed, whereas heart valve sclerosis is preceded by valvular calcification and fibrosis.⁽¹⁾ Growing evidence suggests cardiovascular calcification is inversely associated with BMD independent of age and other clinical risk factors shared by both disorders.^(2,3) Further, there is some evidence to suggest a similar relationship between valvular calcification and BMD.^(4–6)

Several molecular mechanisms have been suggested for a potential link between cardiovascular calcification and bone metabolism.^(7,8) Vascular calcification has been shown to be an active process that resembles bone mineralization.^(7–9) Minerals in calcified plaques in blood vessel walls and valve leaflets have been identified as calcium hydroxyapatite. In addition, bone marrow cells such as osteoblasts, osteoclasts, chondrocytes, and hematopoietic cells have been identified in calcified plaques.⁽¹⁰⁾ Moreover, alkaline phosphatase (APL)-containing matrix vesicles and several bone matrix proteins, such as osteopontin and osteocalcin, are also found at the site of cardiovascular mineralization.^(11–13) In knockout mouse models, deletion of matrix Gla protein and osteoprotegerin causes both cardiovascular calcification and osteoporosis.^(14,15)

Prior studies of bone density and vascular calcification often include women only,^(16–20) use plain radiographs to evaluate calcification of a single vascular bed, the abdominal aorta,^{(2,16,18–21)} and rely on dual-energy X-ray absorptiometry (DXA) measurements of areal BMD (aBMD).^(16,19–24) DXA poses limitations for studies of bone and vascular calcification, particularly aBMD measures at the lumbar spine. As a 2D measure, aBMD is confounded by bone size.⁽²⁵⁾ In addition, DXA is not able to distinguish between trabecular and cortical compartments of bone. These issues are important limitations because the individual compartments differ with respect to the influence of several factors, including hormones, loading, and aging. At the lumbar spine, the skeletal site frequently evaluated in the clinical setting, DXA aBMD is overestimated in the presence of aortic calcification⁽²⁶⁾ and features of spinal degeneration, such as osteophytes. In contrast to DXA, computed tomography (CT) allows quantification of volumetric bone density, separately for the trabecular and cortical compartments, not confounded by aortic calcification and degenerative disease.

To our knowledge, no prior investigations have used multidetector computed tomography (MDCT) to assess both bone and vascular calcification, as well as valvular calcification, in a population-based study of women and men. Therefore, the purpose of our study was to determine the association between CT-derived volumetric BMD (vBMD) at the lumbar spine, including both trabecular vBMD and cortical vBMD, and vascular and valvular calcification, in women and men.

Materials and Methods

Participants

The current study included 1317 participants of the Framingham Offspring Study (689 women, 628 men, mean age 60 years, range 36 to 83 years), who underwent CT examinations in 2002 to 2005, and had complete information for calcium scores, volumetric bone mineral density, and covariates. One hundred five individuals were excluded because of uninterpretable CT images (n = 54) or missing covariate data (n = 51). Boston University Institutional Review Board approved the study, and participants provided written informed consent.

CT imaging

Participants underwent cardiac imaging using an 8-slice multidetector CT scanner (Lightspeed Ultra; General Electric Medical Systems, Milwaukee, WI, USA) in 2002 to 2005. Two scans were obtained for each individual using a sequential scan protocol with a slice collimation of 8 × 2.5 mm (120 KVp, 320/400 mA for <220 and >220 lb body weight, respectively) during a single end-inspiratory breath hold (typical duration, 18 seconds). Image acquisition (330 ms) was prospectively initiated at 50% of the cardiac cycle. For the chest, 2.5-mm slices were acquired from the carina to the diaphragm. For the abdomen, 2.5-mm slices were obtained of a 125-mm abdominal segment above S1.⁽²⁷⁾

Vascular and valvular calcium scoring

Vascular and valvular calcium measurements were performed using an offline workstation (Acquarius, Terarecon, San Mateo, CA, USA) by four experienced readers, who independently analyzed the axial images. A calcified lesion was defined as an area of ≥3 connected pixels with a CT attenuation of ≥130 Hounsfield units, with the use of 3-dimensional connectivity criteria (6 points). Scans for each individual were evaluated for coronary artery calcification (CAC), aortic artery calcification (AAC), aortic valve calcification (AVC), and mitral valve calcification (MVC).

The Agatston Score was used to quantify the level of calcification. For each individual, the Agatston Score is calculated by multiplying the area of each calcified lesion by a density factor, dependent on the maximal attenuation (HU) within the lesion, and summing each of these values for a total calcification score.^(28,29) The density factor, ranging from zero to four, is determined as follows: 1 = 130 to 199 HU, 2 = 200 to 299 HU, 3 = 300 to 399 HU, and 4 > 400 HU. Aortic valve calcium was defined as calcium deposits of the aortic cusps or nodular deposits at the coaptation points of the aortic cusps. Calcium deposits restricted to the aortic wall were excluded in scoring aortic valve calcium. Mitral valve calcium was defined as calcium deposits in the region of the annulus and/or the mitral valve leaflets. Interobserver reliability of calcium measurements was high, with intraclass correlation coefficients (ICCs) greater than 0.96 for both vascular and valvular calcium scores.^(1,27) Prevalence of any calcification was defined as Agatston Score greater than zero (Agatston Score >0).

Volumetric bone mineral density

Integral, trabecular, and cortical volumetric bone density (vBMD; g/cm³) of L₃ was measured from the CT scans using previously published algorithms.^(30,31) The volume of interest for integral vBMD included the entire vertebral body (both cortical and trabecular compartments) but excluded the transverse and posterior processes. The volume of interest for trabecular vBMD measurements was an elliptical region encompassing the anterior vertebral body, centered at the midvertebral level and encompassing 70% of the volume between vertebral endplates. The volume of interest for cortical vBMD measurements encompassed approximately the outer 1 mm of the peripheral bone of the vertebral body.⁽³⁰⁾

Covariates

Information on clinical risk factors was obtained from study visits, conducted in 1998 to 2001, that included physical examinations, laboratory testing, and questionnaires. Height, to the nearest one-quarter inch, and weight, to the nearest half pound, were measured using a stadiometer and balance-beam scale, respectively. Body mass index (BMI) was calculated as weight in kilograms divided by the square of height in meters (kg/m²). Information on cigarette smoking, physical activity (Framingham Physical Activity Index), ⁽³²⁾ and, in women, menopause status (cessation of menses for at least 1 year) and use of hormone-replacement therapy were determined by questionnaire.

Statistical analysis

Multivariable linear regression was used to model mean Agatston Score as a function of sex-specific quartile of vBMD (Q1 = low vBMD). Because of a large proportion of participants with Agatston Score = 0, we performed a log transformation, calculated as: log (Agatston Score + 1). Separate models were used for integral, trabecular, and cortical vBMD. We performed the analysis separately for women and men. Models included the following covariates: age (years), height (inches), BMI (kg/m²), Framingham physical activity index, current smoking (yes/no), current hormone-replacement therapy (yes/no), and menopause status (post-/premenopausal). We repeated our analyses considering vBMD as a continuous variable and found no differences in the interpretation of results based on vBMD quartiles, presented herein.

Results

The study included 1317 participants. Age ranged from 36 to 83 years, and mean age was 60 years (Table 1). Prevalence of any calcification (Agatston Score > 0) was higher for vascular calcification than valve calcification in both women and men (Table 2). Women had lower prevalence of any calcification than men for CAC, AAC, and AVC, but prevalence was similar in women and men for MVC.

Table 1.

Characteristics of Participants in CT Study of Volumetric Bone Mineral Density of the Lumbar Spine and Vascular and Valvular Calcification, Framingham Offspring Cohort, 1998 to 2001

	Women (n=689)		Men (n=628)
Characteristics	Mean or %	SD or n	Mean or %	SD or n
Age (years)	60	9	60	9
Height (inches)	64	2	69	3
Weight (lb)	159	33	195	33
Body mass index (kg/m2)	28	6	29	5
Physical activity index	37	5	39	7
Current smoker (%)	10	68	11	67
Postmenopausal (%)	84	578	–	–
Current hormone-replacement  therapy use (%)	35	241	–	–

Table 2.

Distribution of Vascular and Valvular Calcification (Agatston Scores)^a

	Coronary artery calcification (CAC)		Aortic artery calcification (AAC)		Aortic valve calcification (AVC)		Mitral valve calcification (MVC)
	Agatston Score >0	Mean (SD)	Agatston Score >0	Mean (SD)	Agatston Score >0	Mean (SD)	Agatston Score >0	Mean (SD)
Women  (n=689)	57%	2.3 (2.4)	77%	5.0 (3.2)	31%	1.1 (1.9)	20%	0.9 (2.0)
Men  (n=628)	83%	4.3 (2.6)	86%	5.8 (3.0)	47%	1.9 (2.3)	21%	0.9 (1.9)

^aLog (Agatston Score+1).

Integral vBMD and vascular and valvular calcification

CAC increased in women from 2.0 to 2.6 with decreasing quartile of integral vBMD (trend, p = 0.02; Fig. 1). AAC similarly increased in women from 4.4 to 5.0 with decreasing quartile of integral vBMD (trend, p = 0.01). In contrast, no inverse trend between CAC (p = 0.69) or AAC (p = 0.14) and integral vBMD was observed in men.

Fig. 1.

Multivariable-adjusted mean Agatston Score for coronary artery calcification (CAC) and aortic artery calcification (AAC) by quartile of integral vBMD (Q1=low) in women (A) and men (B).

There was no association between integral vBMD and valvular calcification (AVC, MVC) in women or men (Fig. 2). For example, mean AVC was 1.2 for women in the lowest quartile of integral vBMD as well as women in the highest quartile of integral vBMD (trend, p = 0.77), and mean AVC was 1.8 for men in the lowest quartile of integral vBMD and 2.0 for the highest quartile of integral vBMD (trend, p = 0.63).

Fig. 2.

Multivariable-adjusted mean Agatston Score for aortic valve calcification (AVC) and mitral valve calcification (MVC) by quartile of integral vBMD (Q1 = low) in women (A) and men (B).

Trabecular vBMD and vascular and valvular calcification

Similar to results for integral vBMD, women with the lowest trabecular vBMD had the greatest mean CAC (trend, p = 0.04) and AAC (trend, p = 0.01; Fig. 3). In men, CAC was unrelated to trabecular vBMD (trend, p = 0.92), whereas mean AAC increased from 5.5 for the highest quartile to 6.2 for the lowest quartile of trabecular vBMD (trend, p = 0.01).

Fig. 3.

Multivariable-adjusted mean Agatston Score for coronary artery calcification (CAC) and aortic artery calcification (AAC) by quartile of trabecular vBMD (Q1 =low) in women (A) and men (B).

There was no association between trabecular vBMD and valvular calcification in women or men (Fig. 4).

Fig. 4.

Multivariable-adjusted mean Agatston Score for aortic valve calcification (AVC) and mitral valve calcification (MVC) by quartile of trabecular vBMD (Q1 = low) in women (A) and men (B).

Cortical vBMD and vascular and valvular calcification

Finally, we found no association between cortical vBMD and vascular calcification (CAC, AAC; Fig. 5) or valvular calcification (AVC, MVC; Fig. 6) in women or men.

Fig. 5.

Multivariable-adjusted mean Agatston Score for coronary artery calcification (CAC) and aortic artery calcification (AAC) by quartile of cortical vBMD (Q1 = low) in women (A) and men (B).

Fig. 6.

Multivariable-adjusted mean Agatston Score for aortic valve calcification (AVC) and mitral valve calcification (MVC) by quartile of cortical vBMD (Q1 = low) in women (A) and men (B).

Discussion

In this large, population-based cohort of women and men, we used CT scans to evaluate trabecular and cortical vBMD and severity of vascular and valvular calcium of the coronary arteries, abdominal aorta, and mitral and aortic valves. We found: 1) CAC increased with decreasing trabecular vBMD in women but not in men; 2) AAC increased with decreasing trabecular vBMD in both women and men; 3) calcification of the mitral and aortic valves was not related to vBMD.

Our findings are consistent with prior work on bone and vascular calcification; however, only one investigation including both women and men employed methods similar to ours, using CT to evaluate CAC, AAC, and volumetric bone density.⁽²⁾ Identical to our results, the Multi-Ethnic Study of Atherosclerosis (MESA) observed an inverse association between trabecular vBMD and CAC in women but not men, and an inverse association between trabecular vBMD and AAC in both women and men.⁽²⁾ Studies limited to women, including SWAN (Study of Women’s Health Across the Nation)⁽¹⁷⁾ and a clinically based study,⁽¹⁸⁾ also found that women with the lowest trabecular vBMD had the highest AAC. In contrast, the Rochester Epidemiology Project⁽³³⁾ and a study in the Amish⁽²⁴⁾ reported no association between trabecular vBMD and AAC in women or men. Reasons for discrepant findings between studies are not readily apparent. Differences between studies in characteristics of participants, imaging modalities, and other factors may explain, at least in part, discordant results. For example, the Amish study used electron beam computed tomography to evaluate calcium scores and DXA to evaluate areal BMD. In the Rochester study, prevalence of any AAC was higher in women but lower in men compared with those our study.

This work extends results of prior studies in which measurements of vascular calcification were limited to radiographic assessments of the abdominal aorta, and measurements of bone density were based on DXA or radiographic assessments. A 25-year longitudinal study of the original cohort of the Framingham Heart Study found that women, but not men, with the greatest loss of metacarpal cortical area, evaluated from hand radiographs, had the greatest increase in abdominal aortic calcification, also evaluated from plain radiographs.⁽³⁾ Reasons for sex differences in results are not readily apparent but may indicate potential differences between women and men in the underlying pathophysiology of bone mineralization, vascular calcification, or both. Differences in methodologies across studies may also account for differences in results for women and men.

There are several potential mechanisms that link vascular calcification with osteoporosis. One plausible explanation is inflammation and lipids.^(34–37) In atherosclerosis, deposition and oxidation of lipids in the subendothelial space leads to inflammation, which results in further progression of disease.⁽³⁸⁾ Osteolysis has also been shown to accompany inflammation, and lipid deposits have also been found in the subendothelial spaces of osteoporotic bone tissues.⁽³⁹⁾ Through Msx2 and Wnt signaling pathways, inflammatory cytokines, such as TNF-alpha, are strong regulators of osteochondrogenic differentiation in osteoblastic and vascular cells, which subsequently exert effects on biomineralization.^(40–42) The seemingly paradoxical effects of cytokines on reducing bone formation in the skeleton, yet upregulating bone formation in the vascular wall, are not well understood. Towler and colleagues⁽⁴⁰⁾ have demonstrated, for example, that TNF-alpha upregulates the MSX2-WNT signaling pathway to promote vascular calcification in diabetic mice. Yet, in the skeleton, TNF-alpha has been shown to potently suppress Smad signaling, induced by TGF-beta and BMP-2, downregulate BMP-2-mediated Runx2 expression, and inhibit mineralization of osteoblasts.⁽⁴³⁾ Thus, the same cytokine may signal differently in the two tissue microenvironments, although the mechanisms are not clear. Additional potential mechanisms linking vascular calcification with osteoporosis include fat intake, as an association between lipid level or saturated fat intake and bone density is supported by some but not all studies.^(44–46) Shared endogenous regulatory factors, such as transglutaminase 2, fetuin, and osteoprotegerin,^(34,47,48) may play a role in the link between bone and vascular calcification.

We observed no association between vBMD and aortic or mitral valve calcification. In contrast to numerous investigations of bone density and vascular calcification, few studies have investigated associations between bone density and valvular calcification. Small studies, based on referral samples, found an association between lower T-score and aortic valve calcification, evaluated by echocardiography examination.^(4–6) These findings are concordant with our results for bone and vascular calcification, however, not with our null findings for bone and valvular calcification. Of note, prevalence of valvular calcification was low relative to prevalence of vascular calcification. For example, prevalence of any calcification was 20% for mitral valves compared with 82% for the abdominal aorta. Thus, it is possible that the power of our study was not adequate to detect a potential association between valve calcium and vBMD, particularly an association of small magnitude. Vascular and valvular calcification share similar underlying pathophysiology. Hemodynamic stress and reactive oxygen species trigger both processes.^(38,49,50) Both valvular myofibroblast-like cells and vascular smooth muscle cells respond similarly to macrophage-derived proteinases,⁽⁵¹⁾ and both cell types could be converted into osteoblastic cells.^(52,53)

We found that cortical vBMD was unrelated to vascular calcification in women and men. Griffith and colleagues illustrated that decreasing marrow perfusion leads to increasing fat content with trabecular bone thinning.⁽⁵⁴⁾ In women, the rapid loss of trabecular bone in early menopause corresponds with the period of increased development of atherosclerosis of the aorta among women.⁽⁵⁵⁾ It is thought that the loss of trabecular elements is more common in women because of a menopause-associated increase in osteoclastic activity that leads to a more discontinuous trabecular network.⁽⁵⁶⁾

There are several strengths of our study. First, we used CT imaging to measure both calcification and vBMD. Second, Agatston Scores measured from CT scans are sensitive markers of vascular calcification and have been shown to successfully track progression of coronary artery calcification in patients with coronary artery disease.^(57,58) However, the use of density factors, truncated to single digits (1 to 4), to calculate Agatston Scores may reduce both accuracy and precision. We would expect such misclassification to be random, thereby reducing our ability to detect associations. Third, we examined both vascular and valvular calcification, providing a comprehensive evaluation of calcification. Fourth, we controlled for important confounders, including risk factors for both osteoporosis and cardiovascular disease, collected using highly standardized, rigorous data collection methods. Finally, our study included a large number of both women and men from a community-based cohort, allowing us to generalize our results to similar populations.

A limitation of this study is the observational design. Results may have been affected by uncontrolled confounding and causation cannot be inferred. In addition, members of the Framingham Study largely represent the predominantly white sociodemographic makeup of the town of Framingham, MA, at the time of enrollment of the original cohort in 1949.⁽⁵⁹⁾ Therefore, the results of our study may not be directly applicable to other race groups. However, the similarity of our results to those found in the Multi-Ethnic Study of Atherosclerosis (MESA)⁽²⁾ suggests that our findings may be applicable to individuals with race and ethnic backgrounds that differ from our study cohort.

In conclusion, women and men with low spine vBMD have greater severity of vascular calcification, particularly at the abdominal aorta. The inverse relation between abdominal aortic calcification and spine vBMD in women and men may be an important link on which to focus prevention efforts that can target individuals at high risk of both fracture and cardiovascular events.