We used data from the Osteoporotic Fractures in Men (MrOS) study to test the hypothesis that men with higher levels of bone turnover would have accelerated bone loss and an elevated risk of fracture. MrOS enrolled 5995 subjects >65 yr; hip BMD was measured at baseline and after a mean follow-up of 4.6 yr. Nonspine fractures were documented during a mean follow-up of 5.0 yr. Using fasting serum collected at baseline and stored at −190°C, bone turnover measurements (type I collagen N-propeptide [PINP]; β C-terminal cross-linked telopeptide of type I collagen [βCTX]; and TRACP5b) were obtained on 384 men with nonspine fracture (including 72 hip fractures) and 947 men selected at random. Among randomly selected men, total hip bone loss was 0.5%/yr among those in the highest quartile of PINP (>44.3 ng/ml) and 0.3%/yr among those in the lower three quartiles (p = 0.01). Fracture risk was elevated among men in the highest quartile of PINP (hip fracture relative hazard = 2.13; 95% CI: 1.23, 3.68; nonspine relative hazard = 1.57, 95% CI: 1.21, 2.05) or βCTX (hip fracture relative hazard = 1.76, 95 CI: 1.04, 2.98; nonspine relative hazard = 1.29, 95% CI: 0.99, 1.69) but not TRACP5b. Further adjustment for baseline hip BMD eliminated all associations between bone turnover and fracture. We conclude that higher levels of bone turnover are associated with greater hip bone loss in older men, but increased turnover is not independently associated with the risk of hip or nonspine fracture.
Osteoporosis in men is an increasingly important concern,(1) and compared with women, the pathophysiology of bone loss and fragility fractures in men is less well studied. Although sex hormones, cytokines, and other biological determinants likely play an important role,(2–5) it is likely that these factors impact the male skeleton at least in part by their effects on bone turnover. Bone turnover is classically assessed by histomorphometric evaluation of bone biopsy specimens, but biochemical measurements using bone turnover markers (BTMs) are increasingly capable of assessing the linked processes of bone resorption and formation.(6,7) Recent advances in the understanding of BTM accuracy, reproducibility, and clinical indications have been reported.(8,9)
In women, increases in bone turnover, particularly after menopause, are thought to have important adverse skeletal consequences. For example, elevated bone turnover is associated with accelerated bone loss in postmenopausal women, but the association is not sufficiently strong to accurately predict rates of bone loss using measurements of bone turnover.(10) Some but not all studies of bone turnover and fracture in older women have found that elevated BTMs, particularly urine resorption markers, are associated with an increased risk of fracture independent of BMD.(11,12)
There are few prospective studies of the relationship between bone turnover and skeletal outcomes in men.(13) One recently published study found elevated BTMs were associated with modestly increased fracture risk in men,(14) whereas an even more recent publication found no such association.(15) These studies analyzed relatively few fracture events, combined spine and nonspine fractures, and did not include subjects from North America. No published studies have related BTMs to hip fracture risk in men.
To determine the relationship between elevated BTMs and the risk of the hip and other nonspine fractures and to determine whether these relationships were independent of low BMD and other clinical risk factors for osteoporosis, we performed a prospective analysis in a large cohort of community dwelling older men, the Osteoporosis in Men study (MrOS).
MATERIALS AND METHODS
MrOS is a prospective cohort study as designed to assess risk factors for fracture.(16) At the baseline visit, BMD of the hip and lumbar spine and other measurements were obtained, and participants were followed up prospectively to ascertain and validate new fractures.
We recruited 5995 men ≥65 yr of age by local advertisements and mass mailings(17) from several U.S. communities: Birmingham, AL; Minneapolis, MN; Palo Alto, CA; the Monongahela Valley near Pittsburgh, PA; Portland, OR; and San Diego, CA. Additional details about the methods of MrOS have been published elsewhere.(18) Six hundred thirty-three men (10.6%) recruited into MrOS were nonwhite or Hispanic.
We use fasting serum collected at baseline and stored at −190°C to measure biochemical markers of bone turnover at a specialized laboratory (CCBR-Synarc, Lyon, France). As a marker of bone formation, we measured type I collagen N-propeptide (PINP; Roche Diagnostics, Mannheim, Germany),(8) a propeptide of type I collagen reflecting both trimeric and monomeric forms. Intra- and interassay CVs for this assay are <4.4% at this laboratory. For bone resorption, we measured two serum markers, β C-terminal cross-linked telopeptide of type I collagen (βCTX; Roche Diagnostics) with intra- and interassays CVs <4.2%, and TRACP5b (SBA-Sciences, Turku, Finland) with intra- and interassay CVs <7%. βCTX is generated by osteoclastic breakdown of type 1 collagen, whereas TRACP5b is an enzyme thought to primarily reflect osteoclast number.(19)
BMD was measured in the proximal femur and lumbar spine using DXA measured by Hologic QDR 4500 densitometers during the baseline visit in 2000–2002. Repeat BMD was measured using the same devices after an average follow-up of 4.6 yr and strict quality control measures were followed to assure measurement stability.(18)
Follow-up and ascertainment of fractures
We contacted participants every 4 mo through a mailed questionnaire to ask about recent fractures. Follow-up for fractures and vital status was 99% complete during 5.0 yr of follow-up. All reported fractures were validated by physician review of radiology reports or X-rays if no radiology report was available. The physician rated fractures as associated with “severe” trauma if they occurred during a motor vehicle accident or trauma equivalent to falling from more than one stair step above standing height. Main analyses included all fractures, and subanalyses were completed which excluded fractures associated with severe trauma.
Height was measured on a Harpenden stadiometer (DyFed), and weight was measured on balance beam scales (except the MrOS Portland site, which used a digital scale) that were calibrated with standard weights. Body mass index was calculated as weight (kg)/height (m2). Self-reported medical history, physical activity, smoking status, and alcohol use were collected using standard instruments.(18)
To analyze the cross-sectional relationship between bone turnover and BMD at baseline and the prospective relationship between bone turnover and bone loss, we selected a random sample (N = 947) of MrOS participants using a random number generator. After excluding men taking osteoporosis or sex hormone medications and those with nonfasting baseline serum, the analysis of bone loss included 682 men with adequate baseline and follow-up hip DXA measurements.
To analyze the prospective relationship between bone turnover and fracture, we used an efficient case-cohort design. After an average follow-up of 5.0 yr from the baseline visit, we identified all 431 men with confirmed nonspine fracture (including 72 hip fractures). These fracture cases were compared with men in the randomly selected cohort described above after removing those with fracture during follow-up. We censored men who died (N = 103) or withdrew consent (N = 11), and we excluded men who were either taking osteoporosis or sex hormone medications or who did not fast at least 8 h before the blood draw at the baseline visit (N = 102). Thus, after excluding a total of 216 men, the final case-cohort analysis for nonspine fracture included 384 fracture cases and 885 men without nonspine fracture, and the final hip fracture analysis included 72 fracture cases and 933 men without hip fracture.
To examine the baseline characteristics by fracture status, χ2 tests were used for categorical variables, and t-tests were used for continuous variables. The association between BTM and bone loss was analyzed using generalized linear regression models among the group of randomly selected men; BTMs were analyzed a priori by quartiles and as continuous variables. The association between BTM and risk of fracture was analyzed using proportional hazards models that account for the case-cohort sampling.(20,21) Results from these models are reported at relative hazards (RHs) with 95% CIs. In addition to age, race, and clinic, factors associated with bone turnover (p < 0.05), including BMI, diabetes, and grip strength, were included in multivariate models. Additional multivariate models including baseline dietary calcium intake and baseline renal function as estimated by serum creatinine were examined.
Subjects and baseline measurements
Among the 947 randomly selected MrOS participants, the mean age was 73.7 ± 5.9 yr, and the mean total hip BMD was 0.95 ± 0.14 g/cm2. Seventeen percent of men reported previous fractures, and 86% reported good or excellent health. Mean values of PINP (39.0 ± 24.9 ng/ml) and βCTX (0.41 ± 0.21 ng/ml) were slightly lower than values observed in older postmenopausal women in the United States.(22)
The baseline levels of bone turnover among those with and without subsequent fractures are shown in Tables 1 and 2. The correlations between BTMs were moderate to strong (r = 0.37–0.74). Baseline total hip and femoral neck BMD were inversely related to baseline levels of BTMs (all p < 0.001), but the correlations were weak (r = −0.11 to −0.30).
Table Table 1.. Baseline Characteristics Among Men in Bone Marker and Nonspine Fracture Case-Cohort Analysis
Table Table 2.. Baseline Characteristics Among Men in Bone Marker and Hip Fracture Case-Cohort Analysis
BTM and bone loss
After adjustment for age and clinic, higher baseline levels of BTM were associated with greater loss of BMD (Fig. 1; Table 3), and the relationships seemed to be graded. For example, each SD increase in baseline βCTX was associated with an additional 0.17% (95% CI: 0.10, 0.24) annual loss of total hip BMD. Results using models that further adjusted for BMI, race, diabetes, grip strength, and baseline BMD were similar (Table 3). By comparison, in models adjusted for the same covariates, each 5-yr increase in age was associated with an additional 0.18% (95% CI: 0.12, 0.24) annual loss of total hip BMD. However, baseline BTMs accounted for only 1–4% of the observed variability (R2) in change in BMD.
Table Table 3.. Bone Turnover and Bone Loss Among Randomly Selected Men (N = 682)
BTM and fracture risk
After accounting for age and clinic, fracture risk seemed to increase with higher levels of BTMs (Figs. 2 and 3), but with the exception of PINP and nonspine fracture, the relationships between quartile of BTM and the risk of fracture did not reach statistical significance. As previously observed in women, the risk of fracture was most apparent among men in quartile 4 of BTM. When men in the highest quartile of BTM were compared with those in the lower three quartiles, the age- and clinic-adjusted risk of nonspine and hip fracture were increased for PINP and βCTX but not TRACP5b (Table 4). After further adjusting for baseline BMD alone (data not shown), or adjusting for baseline BMD and other potential confounders (Table 4), the relationship between BTM and fracture was markedly attenuated and was no longer statistically significant for any marker.
Table Table 4.. Bone Turnover and Fracture Risk Before and After Adjustment for BMD and Other Factors in Case-Cohort Analysis
We performed a number of additional analyses to determine the robustness of our results. Results were similar when traumatic fractures (N = 55) were excluded and when quintiles of BTM were analyzed. Results were similar in models that adjusted for weight rather than BMI and after adjustment for dietary calcium intake at baseline. We searched for interactions with baseline BMD, but the relationships between bone turnover and nonspine fracture risk were similar among men above and below the median BMD at baseline. Because renal function might differentially affect associations between turnover markers and fracture, we further adjusted our multivariate results for estimated glomerular filtration rate (eGFR) using serum creatinine and the Modification of Diet in Renal Disease (MDRD) formula, but the results were unchanged.
Last, we further analyzed the relationships between fracture and combinations of BTMs, such as the ratio of CTX/TRACP5b, and the “uncoupling index” using βCTX and PINP as described by Eastell et al.,(23) but the results were similar to results using a single BTM (data not shown).
In this large prospective study of men ≥65 yr of age, we found that higher baseline levels of BTMs, as assessed by the formation marker PINP, the resorption marker βCTX, and a marker of osteoclast number (TRACP5b), were independently associated with modestly higher rates of bone loss over nearly 5 yr of follow-up. Comparing the highest age- and clinic-adjusted quartile of marker to the other three quartiles, higher levels of PINP and βCTX but not TRACP5b were associated with an increased risk of hip or any nonspine fracture, but these associations were markedly attenuated and no longer significant in models that further adjusted for baseline hip BMD and other clinical risk factors. Similar to previous studies among women, our results suggest that increased bone turnover is associated with adverse skeletal effects in men, but after accounting for baseline hip BMD, we found no independent effect of elevated bone turnover on fracture risk. Our results suggest that the association between higher PINP or βCTX and fracture risk is caused by lower BMD among men with elevated BTMs. Thus, our results do not support the routine use of BTMs to assess fracture risk in older men if hip BMD is available.
Two previous prospective studies have examined the relationships between elevated BTMs and skeletal outcomes in men. In the Dubbo Osteoporosis Epidemiology study, Meier et al.(14) performed a nested case-control study of three BTMs and fracture risk among the 989 Australian men (mean age, 71 ± 5.2 yr). Bone turnover measurements were obtained on archived serum from 51 men with a mixture of spine and nonspine fracture and 101 men without fracture. Elevated serum carboxyterminal cross-linked telopeptide of type I collagen (ICTP), a marker of bone resorption, was associated with an increased risk of fracture, but neither βCTX nor PINP was associated with fractures. After adjustment for baseline hip BMD and other confounding factors, increased ICTP remained associated with an increased risk of fracture (RR per SD increase = 1.4, 95% CI: 1.1, 1.7). As noted in previous studies of postmenopausal women,(24) the risk of fracture was particularly elevated among men in the highest quartile of BTM (RR = 2.8, 95% CI: 1.4, 5.4 compared with men in the lowest quartile of ICTP, adjusted for age and weight). The relationship between BTMs and rate of change in BMD was not reported in this study.
In the Minos study, Szulc et al.(15) measured serum levels of three serum formation markers (osteocalcin, bone alkaline phosphatase, and PINP) and three resorption markers, (serum βCTX, urine deoxypyridinoline, and urine βCTX) on archived specimens from 723 French men 50–85 yr of age. During an average follow-up of 7.5 yr, 77 incident spine (N = 27) and nonspine fractures were documented. After adjustment for age, BMI, prevalent fractures, and BMD, baseline BTM levels were not associated with the risk of the combined outcome of spine and nonspine fracture. The relative risk of fracture among men in the highest quartile of BTM ranged from 1.0 to 1.4, with p values all >0.37, but the result were not reported with 95% CIs, and the small number of events raises the concern that the study was underpowered to detect significant associations. In the Minos study, elevated BTM levels were associated with accelerated total body bone loss, but surprisingly, BTMs were not associated with hip or spine bone loss.
Our prospective study of BTMs, bone loss, and fracture in men had a number of strengths and several potential weaknesses. Compared with previous prospective studies, our multicenter study was considerably larger and was able to analyze the specific effects of elevated BTMs on nonspine and hip fracture. The MrOS cohort is well characterized, and both risk factors for osteoporosis and fracture outcomes have been carefully measured.(16) We selected representative state-of-the-art bone formation and resorption markers and used automated platforms at a central laboratory that specializes in bone turnover markers. Previous studies have not related skeletal outcomes in men to TRACP5b levels, a resorption marker that reflects osteoclast number, but TRACP5b performed no better than the other BTMs in this analysis, and the overlapping CIs in Table 4 preclude inferences regarding which marker is superior. We did not measure urine resorption markers, which have been more favorably associated with fracture risk in older women, but urine and serum resorption markers are known to be highly correlated.(6) In addition, we only assessed bone turnover at single point, and given the know variability of BTMs, multiple measurements over time may have resulted in stronger associations with bone loss and fracture.(25) These analyses do not include incident vertebral fracture, but those analyses are planned. Last, the results of this study may not generalize to men <65 yr of age or to women of any age.
Additional research relating BTMs and skeletal outcomes in men is needed, and such efforts should focus on newer BTMs (such as isomers of type I collagen C-telopeptide) and use multiple baseline BTM measurements.
In conclusion, in this large prospective study of contemporary BTMs, bone loss, and fracture in older men, we found that elevated serum levels of PINP, βCTX, and TRACP5b are associated with higher rates of hip bone loss, but the associations were of insufficient strength to accurately predict bone loss in any individual subject. Although our data suggest that higher serum levels of PINP and βCTX at baseline are associated with an increased risk of subsequent hip and nonspine fracture in older men, none of the relationships between BTMs and fracture risk were statistically significant after accounting for baseline BMD. Additional prospective studies are warranted, particularly with novel biomarkers, but these results suggest that a single serum measurement of PINP, βCTX, or TRACP5b does not strongly predict future fracture risk in men and should not be incorporated into evolving risk stratification methods.
This project was specifically supported by NIAMS (RO1 AR52862 and K24 AR051895). The Osteoporotic Fracture in Men (MrOS) Study is supported by National Institutes of Health funding. The following institutes provide support the National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS), the National Institute on Aging (NIA), the National Center for Research Resources (NCRR), and the NIH Roadmap for Medical Research under the following grants numbers: UO1 AG18197, UO1 AR45614, UO1 AR45632, UO1 AR45647, UO1 AR45654, UO1 AG45583, UO1 AG18197, UO1 AG27810, and UL1 RR24140. The authors thank the substantial contributions of the MrOS subjects, investigators, and clinic staff.