Calcifications in the Abdominal Aorta Predict Fractures in Men: MINOS Study

Authors


  • The authors state that they have no conflicts of interest.

Abstract

In a cohort of 781 men ≥50 yr of age followed up for 10 yr, extended calcifications in the abdominal aorta were associated with a 2- to 3-fold increase in the risk of osteoporotic fractures regardless of BMD and falls.

Introduction: Cardiovascular disease and osteoporotic fractures are public health problems that frequently coexist.

Materials and Methods: We assessed the relation of the severity of aortic calcifications with BMD and the risk of fracture in 781 men ≥50 yr of age. During a 10-year follow-up, 66 men sustained incident clinical fractures. Calcifications in the abdominal aorta expressed as an aortic calcification score (ACS) were assessed by a semiquantitative method. BMD was measured at the lumbar spine, hip, whole body, and distal forearm.

Results: ACS > 2 was associated with a 2-fold increase in the mortality risk after adjustment for age, weight, smoking, comorbidity, and medications. After adjustment for age, body mass index (BMI), smoking, and comorbidity, men in the highest quartile of ACS (>6) had lower BMD of distal forearm, ultradistal radius, and whole body than men in the lower quartiles. Log-transformed ACS predicted fractures when adjusted for age, BMI, age by BMI interaction, prevalent fractures, BMD, and history of two or more falls (e.g., hip BMD; OR = 1.44; p < 0.02). ACS, BMD at all the skeletal sites, and history of two or more falls were independent predictors of fracture. Men with ACS > 6 had a 2- to 3-fold increased risk of fracture after adjustment for confounding variables (OR = 2.54-3.04; p < 0.005-0.001 according to the site).

Conclusions: This long-term prospective study showed that elevated ACS (>6) is a robust and independent risk factor for incident fracture in older men regardless of age, BMI, BMD, prevalent fractures, history of two or more falls, comorbidities, and medications.

INTRODUCTION

Cardiovascular disease and osteoporotic fractures are two major public health problems. Cardiovascular disease and osteoporosis coexist in women.(1-3) Progression of aortic calcifications has been associated with faster bone loss.(3,4) Low BMD has been shown to predict cardiovascular events and cardiovascular mortality,(5,6) whereas the association between the extension of aortic calcifications and hip fracture risk is controversial.(7,8)

In contrast to these findings in women, few studies concern the relationship between osteoporosis and cardiovascular disease in men. Vascular calcifications have been associated with lower BMD in men,(9-11) yet data on the association between cardiovascular disease and low BMD in men are discordant,(2,10,12) despite the fact that low BMD in men has in some studies been shown to be associated with higher cardiovascular mortality.(13-15)

Furthermore, the relationship between cardiovascular pathology and BMD is weaker in men. Low femoral neck BMD has been associated with a higher prevalence of peripheral arterial disease and stroke in women but not in men.(16,17) Decline in metacarpal cortical area was shown to be correlated with the progression of aortic calcifications only in women.(18) Low metacarpal cortical bone mass has been associated with a higher incidence of coronary heart disease in women but not in men.(19) Extended aortic calcifications were not associated with the hip fracture risk in a long-term follow-up of middle-age men.(8)

Thus, the association of cardiovascular pathology with osteoporosis in men, especially with fracture risk, is poorly understood. Aortic calcifications are frequent in elderly men.(20) The extent of aortic calcification is correlated with cardiovascular morbidity and mortality.(21,22) Thus, a study of the association of aortic calcifications with BMD and fracture risk in men can improve our understanding of the relationship between cardiovascular morbidity and osteoporosis. Moreover, if the extended aortic calcifications are associated with an increased susceptibility to fracture, they could be a useful clinical indicator of the risk of fracture easily assessed on lateral radiography of lumbar spine.

There were two aims of this study: to assess prospectively the predictive value of aortic calcifications for the risk of incident fracture in a cohort of men ≥50 yr of age followed prospectively for 10 yr and to assess the relation between the severity of aortic calcifications and BMD.

MATERIALS AND METHODS

Cohort

MINOS is a prospective study of male osteoporosis initiated in 1995 as a collaboration between the INSERM and Société de Secours Minière de Bourgogne (SSMB) in Montceau les Mines.(23) Letters inviting to participate in the study were sent to a randomly selected sample of 3400 men 50-85 yr of age insured by SSMB. Among 841 volunteers who responded to the invitation and signed an informed consent, 781 had a BMD measurement by DXA and lateral radiographs of spine. Their mean age was 65 ± 7 (SD) yr. Sixty men did not have DXA or had radiographs of poor quality. The study obtained authorization of the local ethics committee and was performed in accord with the Helsinki Declaration of 1975 and 1983. We obtained dates of death from the administration of SSBM.

At baseline, the men responded to a written epidemiological questionnaire. Tobacco smoking was assessed according to two groups: current smoker versus currently nonsmoker. Current leisure physical activity was assessed as described previously.(24) Number of medications taken at the recruitment was also registered. Prevalent fragility fractures were observed in 108 men (Table 1). They were comprised of 74 prevalent vertebral fractures of grades 2 and 3 (according to semiquantitative criteria described previously) and 69 self-reported low-trauma non-spine fractures that occurred after the age of 40 yr.(25,26) This threshold was chosen arbitrarily; however, most fractures before this age were high-trauma fractures (accidents), and the elderly participants could not recall circumstances of fractures that occurred many years ago. We excluded high-trauma fractures (motor vehicle accidents, professional traumas), as well as those of fingers, toes, and skull. We considered as prevalent falls all self-reported nonaccidental falls that occurred during the year preceding the recruitment. Men who sustained at least two falls were defined as frequent fallers.

Table Table 1.. Distribution of Prevalent and Incident Fractures According to Skeletal Site
original image

Data on incident fractures were recorded by epidemiological questionnaire every 18 mo during 7.5 yr and then during the last 2.5 yr by telephone. For men who died before the end of follow-up, information on fractures was obtained from the proxies. Seventy-one clinical incident fractures were reported in 66 men (Table 1). This analysis concerns the 66 first incident fractures confirmed by the physician on the basis of medical records or radiography. Average duration of follow-up until the first fracture, death, loss to follow-up, or 10 yr since the baseline examination was 8.83 yr (median follow-up: 10 yr). Vertebral fractures were diagnosed on the basis of 20% decrease of any height of the vertebral body in comparison with baseline. We excluded from the analysis two fractures caused by a motor vehicle accident, two fractures of fingers and toes, two forearm fractures after a fall from the height of >4 m, and one pathological fracture (bone metastasis).

Bone mass measurement

Areal BMD (aBMD) was measured at the lumbar spine, right hip, and whole body by DXA (QDR-1500; Hologic, Waltham, MA, USA) and at the distal forearm using single energy X-ray absorptiometry (DTX100; Osteometer, Rodoure, Denmark) as described previously.(23) CV of aBMD was 0.33% for the spine phantom, 0.94% for femoral neck of hip phantom, and 0.62% for human lumbar spine phantom. At the forearm, two regions of interest (ROI) were evaluated. The distal ROI includes 24 mm of ulna and radius situated proximally to the site where the spacing between the two bones is 8 mm. Ultradistal ROI is situated distally to the previous one. All scans were analyzed manually. Long-term CV for aBMD of a calibration standard was 0.47%. Three hip scans and three forearm scans with evident positioning error were excluded. Width of tubular bones (femoral neck, distal radius, distal ulna) was calculated by dividing their projected area by their length (15 and 20 mm, respectively).(27)

Assessment of aortic calcifications

Aortic calcifications were assessed on the baseline lateral radiographs of lumbar spine by the semiquantitative method.(28) Calcific deposits in the abdominal aorta adjacent to the first four lumbar vertebrae were assessed for the posterior and anterior wall of the aorta using the midpoint of the intervertebral space above and below the vertebrae as boundaries. Individual level-specific severity scores for both the posterior and anterior walls (0-3) were added to yield abdominal aortic calcification score (ACS) ranging from 0 to 24. Intraobserver reproducibility assessed by intraclass correlation coefficient was 0.92-0.95. Interobserver reproducibility was 0.91. Lumbar arthritis was assessed on the lateral radiographies of spine as described previously.(23)

Statistical analysis

All calculations were performed by the SAS software (SAS Institute, Cary, NC, USA). The ACS disclosed a highly skewed distribution and were thus log-transformed. We used logistic regression to model the risk of mortality according to ACS. Data are presented as OR and 95% CI. For the continuous analysis, OR is presented per 1 unit of log-transformed ACS. Because the mortality rate was very similar in the two lowest quartiles, they were pooled and analyzed jointly. Analyses were adjusted for age, weight, smoking (smoker versus nonsmoker), self-reported comorbidities (arterial hypertension, coronary heart disease, type 2 diabetes mellitus, cerebrovascular disease, hemiplegia, parkinsonism), and medications. Comorbidities were dichotomized and included separately or as a score (number of diseases). Other than diabetes, none of the other diseases were significant (p > 0.15) and were thus not included in the model. Number of drugs followed a skewed distribution (0-15; median and IQ range: 2 [1, 5]) and was dichotomized (0-5 versus >5). BMD was compared between the highest quartile of ACS and the three lowest quartiles using analysis of covariance adjusted for age, body weight, smoking, and comorbidities. Analyses of the lumbar spine BMD were also adjusted for the arthritis score (0-3 versus >3).

Fracture risk was assessed by logistic regression. Increase in fracture risk was assessed per 1 unit increase in the log-transformed ACS or using the threshold of 6: ACS > 6 (fourth quartile) versus ACS 0-6 (three lowest quartiles). All the analyses were adjusted for age, BMI, age and BMI interaction,(25) smoking (smoker versus nonsmoker), prevalent fractures (yes versus no), history of falls in the past year (≥2 versus <2 falls), and BMD. Models for femoral neck, distal radius, or distal ulna were also adjusted for bone width because we have shown that low bone width predicted fractures regardless of BMD.(29) Preliminary analyses were adjusted for medications and self-reported comorbidities that could be associated with ACS and that might increase the risk of falls such as arterial hypertension, coronary heart disease, diabetes mellitus, cerebrovascular disease, hemiplegia, and parkinsonism. They were dichotomized (yes = 1 or no = 0) and included separately in the models or as a score equal to the number of diseases. The comorbidities and medications were not significant and were not included in the final models.

RESULTS

General analysis

The distribution of ACS was skewed: 220 men (28.2%) had no calcifications, 181 men (23.2%) had ACSs of 1 and 2, 188 men (24.1%) had scores between 3 and 6, and 192 men (24.6%) had a score of >6.(7-21) As these thresholds correspond best to the quartiles of the distribution, all the analyses were performed according to this classification.

ACS increased with age (Fig. 1). Log-transformed ACS was a predictor of death in the univariate analysis and in the model adjusted for age, weight, smoking, comorbidities, and medications (Table 2). The crude mortality rate per quartile of ACS was: 11.4%, 13.4%, 26.7%, and 41.9%. In comparison with the two lowest quartiles of ACS (ACS = 0-2), the risk of death was higher both in the third (ACS = 3-6; OR = 1.90; 95% CI, 1.18-3.05; p < 0.01) and in the fourth (ACS > 6; OR = 2.91; 95% CI, 1.83-4.63; p < 0.0001) quartiles (Fig. 2; Table 2). After adjustment for age, weight, smoking, comorbidities, and medications, increased ACS (>2) was still associated with a higher mortality.

Table Table 2.. Prediction of the Risk of Death According to ACS at Baseline
original image
Figure Fig. 1..

Age-related increase in the ACS. ACS presented as median and interquartile range per 5-yr age group.

Figure Fig. 2..

Survival of men according to the baseline ACS. Survival of men from the MINOS cohort during the 10 yr of the follow-up according to the ACS at baseline: ACS in two first quartiles (0-2), third quartile (3-6), and fourth quartile (>6).

Cross-sectional analysis

For all the skeletal sites, BMD did not differ between the three lower quartiles of ACS (Table 3). For whole body, distal forearm, and ultradistal radius, BMD decreased across quartiles of ACS. After adjustment for age, body weight, smoking, and comorbidity, men in the highest quartile of ACS had lower BMD of distal forearm and ultradistal radius as well as lower BMC and BMD of the whole body compared with men in the three lower quartiles. At the distal forearm, difference in BMD was caused by the lower BMC in the highest quartile of ACS (e.g., radius: 2.64 [0.46] versus 2.75 [0.38] g, p < 0.002). Bone width was similar in both groups.

Table Table 3.. aBMD Across Quartiles of ACS
original image

Prediction of fractures

In the univariate analysis, higher ACS was associated with an increased risk of having at least one fracture during the follow-up (OR = 1.32 per 1 log-transformed unit; 95% CI, 1.01-1.74; p < 0.05), and this was unchanged after adjustment for age (OR = 1.37; p < 0.04) and after additional adjustment for BMI, smoking, and prevalent fractures (OR = 1.38; p < 0.03). Furthermore, ACS remained predictive after adjustment for BMD and history of two or more falls (e.g., for spine BMD [OR = 1.45, p < 0.03], total hip BMD [OR = 1.44, p < 0.02], or distal forearm BMD [OR = 1.39, p < 0.04]). For the femoral neck, ACS remained predictive of fractures after adjustment for BMD and bone width (OR = 1.41; 95% CI, 1.04-1.91; p < 0.03).

Percentage of men with incident fractures across ACS quartiles was 6.4% (14 men), 9.0% (16 men), 4.8% (9 men), and 14.4% (27 men). The fracture incidence in the fourth quartile of ACS was higher (OR = 2.33; 95% CI, 1.39-3.93; p < 0.002) compared with men in the three lowest quartiles. Adjustment for age, BMI, age and BMI interaction, smoking, and prevalent fractures did not decrease OR for the fourth quartile of ACS (OR = 2.89, p < 0.001). Additional adjustment for history of two or more falls in the past year did not change the predictive value of ACS (OR = 2.90, p < 0.001). History of two or more falls was predictive of fractures (OR = 2.53; 95% CI, 1.29-4.99; p < 0.01).

Additional adjustment for BMD did not influence the association of ACS with incident fracture. Regardless of the skeletal site, OR for BMD, fourth quartile of ACS, and history of two or more falls remained significant (Table 4). Multivariate models for the femoral neck, distal radius, and distal ulna were also adjusted for the bone width. For the three sites, BMD, bone width, fourth quartile of ACS, and history of two or more falls were all significant predictors of fracture (Table 5).

Table Table 4.. Adjusted ORs for the Risk of Fracture for BMD of Different Skeletal Sites, Fourth Quartile of ACS (ACS > 6), and History of Two or More Falls
original image
Table Table 5.. Adjusted ORs for the Risk of Fracture for BMD and Width of Femoral Neck, Distal Radius, and Distal Ulna, Fourth Quartile of ACS (ACS > 6), and History of Two or More Falls
original image

Low BMD (T-score < −2) and high ACS (>6) were independent predictors of fracture (Table 6). Low BMD predicted fractures in men with normal ACS (0-6) and high ACS predicted fractures in men with normal BMD (T-score ≥ −2). In men with low BMD and high ACS, fracture risk was markedly higher than in men with normal BMD and normal ACS.

Table Table 6.. Fracture Incidence (per 1000 Person-Years) According to BMD and ACS
original image

DISCUSSION

In elderly men, extended aortic calcifications are indicators of the increased risk for incident fracture regardless of age, BMI, BMD, prevalent fractures, smoking, history of two or more falls, comorbidities, and medications. As an indicator of the risk of fracture, ACS > 6 is both independent (significant after adjustment for other confounding variables including BMD) and robust (adjustment for other variables has a limited effect on the OR of ACS).

Men in the highest quartile of ACS had lower BMD at the distal forearm and whole body, which is consistent with previous reports of lower BMD in men with cardiovascular disease.(2,8,10) BMD of spine and hip did not differ by ACS. At the spine, the extent of aortic calcifications is more correlated with BMD of trabecular bone because the difference in BMD may be obscured by arthritis.(1,3) BMD of femoral neck and hip was decreased in survivors of myocardial infarction,(12) whereas in a recent study, cardiovascular disease was not associated with hip BMD.(2) The association between the cardiovascular pathology and whole body BMD and BMC has not been studied until now.

Aortic calcification is a highly regulated process; however, mechanisms linking them to the risk of fracture are not clear. ACS and fracture incidence increase with age and share common risk factors such as lifestyle (smoking, sedentary lifestyle), hormonal (hypogonadism, PTH), or nutritional (vitamin K deficiency, fat intake).(30,31) Aortic calcifications and skeletal changes share biochemical pathways including proteins expressed in bone and atherosclerotic plaques (matrix GLA protein, osteopontin), as well as cytokines regulating bone and vascular metabolism (e.g., osteoprotegerin, bone morphogenetic proteins [mainly BMP-2], interleukin 6 or TNF-α).(20,30,32-34)

A possible link between aortic calcifications and osteoporosis is the metabolic syndrome characterized by abdominal obesity, hypertension, glucose intolerance, hypercholesterolemia, and hyperglyceridemia.(35) Metabolic syndrome is associated with an increased prevalence of cardiovascular calcifications, lower BMD, and higher fracture incidence.(36-38) In older men, low testosterone level is associated with a higher prevalence of the metabolic syndrome, aortic calcifications, low BMD, faster bone loss, and higher risk of fracture.(31,39-43) Some but not all studies showed that higher blood pressure, part of the metabolic syndrome, is correlated with lower BMD and faster bone loss.(44-46) In preclinical studies, a high-fat high-cholesterol diet (Western diet) induced aortic calcifications and reduced BMC.(47,48) Also in clinical studies, high-fat high-energy intake was associated with lower BMD in men.(49,50)

PTH plays a role mainly in end-stage renal disease (ESRD), where vascular calcification, bone disease, and cardiovascular disease coexist and worsen in parallel.(51) Arterial calcifications were most extensive in ESRD patients with adynamic bone disease characterized by low PTH level, low bone formation, and low bone resorption.(52) Inhibition of PTH secretion reduced the risk of fracture and cardiovascular event.(53) Interestingly, in male rats, PTH(1-34) increases bone mass but suppresses cardiovascular mineral deposition.(54) Thus, contribution of endogenous versus exogenous PTH to cardiovascular mineral deposition is not straightforward.

A putative link between aortic calcifications and bone fragility is homocysteine (Hcy). Higher Hcy level has been found to be a predictor of fracture (after adjustment for BMD), especially in frail elderly persons.(55-58) Increased Hcy level is a strong indicator of the cardiovascular morbidity and mortality(59,60) and has been correlated with the extent of aortic calcifications.(61) In vitro Hcy may impair the synthesis of covalent cross-links of the type I collagen(62,63) and potentiate the calcification of vascular smooth muscle cells.(64) However, the role of Hcy as a link between fracture risk and cardiovascular pathology is purely speculative.

Because ACS is higher in men with coronary heart disease, arterial hypertension, or diabetes, one explanation of the relation between ACS and fracture might be a higher risk of fall caused by complications or treatment.(65-68) We adjusted all analyses for history of two or more falls, which itself increased the risk of fracture, similarly to previous data.(69) Higher ACS was an independent risk factor for fracture even after adjusting for falls and BMD. Our findings extend the previous data in being able to exclude this potential confounder, even though we did not adjust for falls during the follow-up nor for the severity of falls. It is possible that persons with cardiovascular diseases sustain “worse” falls. Persons with diabetic neuropathy or neurological deficits may have weaker protective reflexes such that they cannot effectively use protective reflexes at the time of hypotension, hypoglycemia or arrhythmia.

The strengths of our study are the number of men, measurements of BMD at four skeletal sites, and prospective 10-yr follow-up. There are also limitations. The major limitation is the lack of data concerning the metabolic syndrome. Population of Montceau les Mines may be not representative of the French population. We recruited volunteers who are mainly healthier elderly men. Reliability of the self-reported fractures or falls was not verified by formal adjudication, but rather relied on data obtained from the medical records. Our analysis is limited to the first incident fractures because there were few second fractures, and their assessment was not exhaustive because we could not contact three men who had first incident fractures. Our study was not planned to assess cardiovascular diseases. Evaluation of cardiovascular diseases was limited to self-report (yes/no, no data on the duration or severity of diseases). Certain data such as blood pressure, concentration of cholesterol, or glucose were not collected. Nutritional assessment was limited to the estimation of calcium intake.

In conclusion, in this prospective study, elevated ACS (>6) was an independent and robust indicator of the increased risk for incident fracture in men regardless of age, BMI, BMD, prevalent fractures, tendency to fall, smoking, comorbidities, and medications. Our data are consistent with the association between osteoporosis and cardiovascular disease. Extended aortic calcifications might be a clinical indicator of fracture risk easily assessed on lateral radiography of the spine. However, the mechanism of the association between aortic calcifications and fracture risk, regardless of BMD and falls, remains to be elucidated.

Acknowledgements

The authors thank Dr Elizabeth J Samelson for the careful reading of the manuscript and helpful comments.

Ancillary