SEARCH

SEARCH BY CITATION

Keywords:

  • bone turnover markers;
  • osteoporotic fracture;
  • longitudinal;
  • prospective;
  • TRACP5b

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Osteoporosis is characterized by compromised bone mass and strength, predisposing to an increased risk of fracture. Increased bone metabolism has been suggested to be a risk factor for fracture. The aim of this study was to evaluate whether baseline bone turnover markers are associated with long-term incidence of fracture in a population-based sample of 1040 women who were 75 years old (Malmö OPRA study). Seven bone markers (S-TRACP5b, S-CTX-I, S-OC[1–49], S-TotalOC, S-cOC, S-boneALP, and urinary osteocalcin) were measured at baseline and 1-year follow-up visit. During the mean follow-up of 9.0 years (range 7.4–10.9), 363 women sustained at least one fracture of any type, including 116 hip fractures and 103 clinical vertebral fractures. High S-TRACP5b and S-CTX-I levels were associated with increased risk of any fracture with hazard ratios [HRs (95% confidence interval)] of 1.16 (1.04–1.29) and 1.13 (1.01–1.27) per SD increase, respectively. They also were associated with increased risk of clinical vertebral fracture with HRs of 1.22 (1.01–1.48) and 1.32 (1.05–1.67), respectively. Markers were not associated with risk for hip fracture. Results were similar when we used resorption markers, including urinary osteocalcin, measured at the 1-year visit or an average of the two measurements. The HRs were highest for any fracture in the beginning of the follow-up period, 2.5 years from baseline. For vertebral fractures, the association was more pronounced and lasted for a longer period of time, at least for 5 years. In conclusion, elevated levels of S-TRACP5b, S-CTX-I, and urinary osteocalcin are associated with increased fracture risk for up to a decade in elderly women. © 2010 American Society for Bone and Mineral Research


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Osteoporosis is a skeletal disease characterized by low bone mass and structural deterioration of bone tissue leading to bone fragility. Compromised bone mass and bone strength will increase the susceptibility to fractures.1, 2 Osteoporotic fractures represent a considerable problem in health care owing to the consequences both for the patient and for the health care system, and they increase in number as the population ages.3 Therefore, one of the important challenges in the management of osteoporosis is to identify women who are at high risk of fragility fractures.

Bone mineral density (BMD) measured by dual energy X-ray absorptiometry (DXA) is the single best method available for confirming the diagnosis of osteoporosis, according to World Health Organization (WHO) guidelines,4 as well as for assessing future fracture risk of the individual. In general, the risk of fracture approximately doubles for each standard deviation (SD) reduction in BMD, depending on the site of BMD measurement and the type of fracture evaluated.5, 6 However, there is a great overlap in BMD of individuals with and without fracture.7, 8 Only about half the fractures occur in women who have a BMD below the diagnostic threshold for osteoporosis (T-score ≤ −2.5), but a large number of fractures occur in women with only moderately low BMD, ie, osteopenia (−2.5 < T-score ≤ −1).8, 9 Apparently, BMD is only one of a number of risk factors and can capture only one aspect of the likelihood of fracture.

Owing to the low sensitivity of BMD testing, there is a need for additional measures to identify individuals who are at high risk for fractures and who might need treatment despite not reaching the diagnostic threshold for osteoporosis. Ideally, antiresorptive or anabolic treatments should be targeted on the basis of fracture risk rather than just the T-score for BMD. Over the past few years, the use of potential risk factors that could add information on fracture risk independently of BMD has been evaluated. Kanis and coworkers analyzed a number of epidemiologic studies to identify important clinical risk factors and to determine the effect of adding clinical risk factors to fracture prediction models based on BMD alone.10 They showed that fracture risk assessment can be improved by the integration of clinical risk factors, such as body mass index (BMI), prior fragility fracture, maternal history of hip fracture, or the use of glucocorticoids. Based on the impact of clinical risk factors, an algorithm known as the Fracture Risk Assessment Tool (FRAX) has been developed for the clinical assessment of fracture probability of patients.11 The algorithm integrates the risks associated with selected clinical risk factors as well as BMD measured at the femoral neck.

Bone is continuously subjected to resorption and formation by the coordinated action of osteoclasts and osteoblasts, respectively. Low bone mass and microarchitectural deterioration of bone tissue are both related to abnormalities in bone turnover, and therefore, increased rate of bone turnover is a potential risk factor for fracture.12 In particular, bone resorption markers have been shown to be moderately but consistently associated with increased risk for fractures, and the measurement of bone turnover markers (BTMs) could assist in identifying women at high risk of fracture.13–15 Quantitative changes in bone turnover should occur prior to bone loss and fractures, whereas BMD measurement captures the net result of bone loss that has already occurred before the bone assessment. BTMs thus could add information particularly for individuals with normal or moderately low bone mass (osteopenia) who are asymptomatic but have increased bone metabolism and thus may be at increased risk for developing osteoporosis and osteoporotic fractures in the future.16–18

To assess long-term fracture prediction by BTMs in elderly women well beyond menopause, we measured seven BTMs in a population-based sample of 1040 elderly women followed prospectively for a mean of 9 years (Malmö OPRA study). The purpose of the study was to investigate whether bone markers are associated with long-term incidence of fracture and to evaluate the number of years baseline markers can predict fractures.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Subjects

The Malmö Osteoporosis Prospective Risk Assessment (OPRA) study is a population-based cohort of elderly women, all 75 years of age at inclusion. A letter of invitation was sent out 1 week after her seventy-fifth birthday to 1604 women randomly selected from the population files of the city of Malmö between 1995 and 1999. A total of 1044 women agreed to participate in the baseline evaluation (65%). The OPRA cohort was described in detail previuosly.13 Serum and urine samples were collected at baseline (n = 1040) and at follow-up visit after 1 year (n = 968). Informed consent was obtained from all participants, and the study was in all parts approved by the local ethics committee and in accordance with the Declaration of Helsinki.

Bone mineral density

Areal bone mineral density (aBMD) was measured at baseline at the total body, total hip, and lumbar spine (L2–4) by dual-energy X-ray absorptiometry (DXA, Lunar DPX-L, Madison, WI, USA). The stability of the equipment was checked every morning using a phantom provided by the manufacturer. aBMD was obtained at baseline for 931 women at total body, 926 at total hip, and 974 at lumbar spine. Results were available for at least one measurement site in 995 women and for all three sites in 864 women.

Serum and urine samples

Serum and urine samples were collected at baseline and at follow-up visits after 1 year. Serum and/or urine sample was obtained from 1040 OPRA participants. Serum samples were collected as nonfasting samples between 8:00 and 13:00. Urine samples were obtained as the first morning void, between 02:30 and 10:00. All serum and urine samples were stored at −80°C.

Bone turnover markers

Bone resorption was assessed by serum tartrate–resistant acid phosphatase 5b (S-TRACP5b, BoneTRAP, Immunodiagnostic Systems [IDS], Inc., Bolton, UK) and serum C-terminal cross-linked telopeptides of type I collagen (S-CTX-I, Elecsys β-CrossLaps, Roche Diagnostics, Indianapolis, IN, USA). Bone formation was assessed by serum bone-specific alkaline phosphatase (S-boneALP, Metra BAP Assay, Quidel, Corporation, San Diego, CA, USA) and three assays for different molecular forms of osteocalcin. Serum intact osteocalcin (S-OC[1–49]), serum total osteocalcin (S-TotalOC), and serum γ-carboxylated osteocalcin (S-cOC) were determined by previously described protocols.19 We also analyzed urinary osteocalcin (U-OC) with a two-site assay for osteocalcin midfragment (U-MidOC).20 U-OC results were normalized for urinary creatinine determined in accordance with the alkaline picrate (Jaffe) reaction and expressed as ratios. All analyses were performed blinded and in duplicate. The reported within-assay and between-assay variations for the assays are 1.8% and 2.2% for S-TRACP5b; 5.9% and 5.8% for S-CTX-I; 3.6% and 4.4% for S-boneALP; <5% and <8% for S-OC[1–49], S-TotalOC, and S-cOC19; and 1.7% and <12% for U-MidOC,20 respectively.

Fractures

All prospectively sustained clinical (symptomatic) fractures were identified and verified for all participants of the OPRA study by searching the files of the Department of Radiology at the Malmö University Hospital, which is the only hospital serving the city. The date of first fracture of any type was used as an endpoint for follow-up in the prediction of any fracture. We also analyzed hip fractures (trochanteric and femoral neck fractures) and clinical symptomatic vertebral fractures separately. In the case of death during the follow-up period, this date was registered by the means of the Swedish national population register. Prediction of fractures was analyzed in two ways. First, we used the entire follow-up data until 2006. Second, equally long follow-up time for each woman was used. Women were included in the study during 3.6 years, between 1995 and 1999, and the follow-up time varied from 7.4 to 10.9 years. Since the shortest follow-up time was nearly 7.5 years, the follow-up was finished 7.5 years after the baseline visit (for 18 women, the time was 7.4 years).

An extensive questionnaire employed at the baseline investigation was used to collect information on health, medications, and previously sustained fractures.21 Thirty-three women were taking bisphosphonates at baseline evaluation, and additionally, 19 women started the treatment between baseline and the 1-year visit. The respective numbers for estrogen therapy were 19 and 1.

Statistics

Results for women with and without prospective fracture were compared by t tests or Mann-Whitney tests. The ability of BTMs to predict fracture was estimated by Cox proportional hazard ratios [HRs, with 95% confidence intervals CIs)]. All BTMs were nonnormally distributed (Shapiro-Wilk test < 0.95), and BTM results were used after logarithmic transformation. In Cox model, the follow-up time represents the time from baseline visit to either (1) the occurrence of first fracture or (2) to the end of follow-up period or (3) to death if no fracture had occurred. First fracture was used as the endpoint in the analysis. In the analysis of hip and vertebral fractures, the first hip fracture or the first vertebral fracture was used, respectively. HRs were calculated by comparing women with any type of fracture, hip fracture, or clinical vertebral fracture with the control group of women without any fracture. HRs were calculated either for tertiles of aBMD and BTM (as a categorical variable) or for standard deviation decrease in aBMD and increase in logBTMs (as a continuous variable). In addition, Kaplan-Meier survival analysis with log-rank testing was applied for tertiles (highest versus lowest tertile). In order to estimate how long time baseline markers can predict fractures, Cox regression analysis was done for different time periods: 2.5 years, 5 years, or 7.5 years after baseline visit or from 2.5 to 5 years or from 5 to 7.5 years after baseline visit. For statistical analysis we used SPSS, Version 14.0 (SPSS, Inc., Chicago, IL, USA) (Cox regression) and Statistica software, Release 7.1 (StatSoft, Inc., Tulsa, OK, USA) (basic statistics and Kaplan-Meier analysis). The level of significance was set at p < .05.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Prospective fractures

During the mean follow-up time of 9.0 years (range 7.4–10.9 years), 363 of the 1044 women (35%) sustained at least one fracture of any type in this study of the OPRA cohort: 224 women had one fracture, 94 had two fractures, 33 had three fractures, 9 had four fractures, 2 had five fractures, and one woman had 12 fractures. The average time from baseline to the first fracture (endpoint) was 4.1 years. A total of 116 women sustained a hip fracture and 103 a clinical symptomatic vertebral fracture, 253 women had died by the end of follow-up, and 4 women were lost to the follow-up because they moved away. Women who did not have any fracture (n = 681) were used as the control group in all analyses.

At baseline evaluation, women who had a fracture during follow-up had a lower weight (66.8 kg) and BMI (25.8 kg/m2) than women who did not have a fracture (68.7 kg and 26.7 kg/m2, p = .018 and p = .001, respectively). They also had a significantly lower aBMD at the total body, total hip, and spine (all p < .001) and were more frequently diagnosed with osteoporosis at baseline (39% versus 23%, p < .001). They also were more often current smokers (16% versus 13%, p = .047). At the time of baseline evaluation, 457 women had experienced a fracture earlier in life. This was more frequent in the group of women who later sustained a prospective fracture during the follow-up period (49%) than in those who did not (41%, p = .008). The most recent fracture before baseline evaluation had occurred within the last 2 years (between the ages of 73 and 75 years) in 69 women, and there was no difference between women who prospectively sustained a fracture and those who did not (p = .60) (Table 1).

Table 1. Baseline Characteristics of Participating Women
Fractures during follow-upYes (n = 363)No (n = 681)p
  1. Values are means (SD) or medians (interquartile range) or percentages. p values are for t test or Mann-Whitney test (bone markers) or chi-square test. p < .05 are shown in bold.

Anthropometry   
 Weight (kg)66.8 (11.1)68.7 (11.5)0.018
 Height (m)1.61 (0.06)1.60 (0.06)0.120
 BMI (kg m−2)25.8 (4.0)26.7 (4.2)0.001
Questionnaire   
 Use of bisphosphonates, n (%)16 (4.4%)17 (2.5%)0.093
 Use of HRT, n (%)5 (1.4%)13 (1.9%)0.530
 Use of vitamin D, n (%)65 (29%)69 (14%)<0.001
 Number of smokers, n (%)87 (16%)58 (13%)0.047
Previous fractures   
 Previous fracture, any time, n (%)179 (49%)278 (41%)0.008
 Previous fracture, <2 years, n (%)26 (7.2%)43 (6.3%).599
Baseline aBMD   
 Total body (g/cm2)0.98 (0.09)1.02 (0.10)<.001
 Lumbar spine (g/cm2)0.95 (0.19)1.01 (0.19)<.001
 Total hip (g/cm2)0.81 (0.14)0.87 (0.15)<.001
 Femoral neck T-score−2.08 (1.09)−1.63 (1.15)<.001
 Osteoporosis at hip (%)39%23%<.001
Baseline bone turnover markers   
 S-OC[1–49] (ng/mL)4.8 (3.3–6.1)4.8 (3.5–6.3).672
 S-TotalOC (ng/mL)7.9 (6.0–10.4)8.1 (6.2–10.4).794
 S-cOC (ng/mL)6.8 (5.4–9.1)7.0 (5.3–9.2).630
 S-boneALP (U/L)22 (14–26)21 (17–26).497
 S-TRACP5b (U/L)3.4 (2.7–4.2)3.2 (2.6–3.9).009
 S-CTX-I (nM)281 (189–402)258 (166–393).065
 U-MidOC (µg/mmol crea)1.09 (0.72–1.62)1.02 (0.68–1.55).185

Baseline BTMs and fracture prediction

Baseline bone resorption markers were associated with fracture risk. The hazard ratios per SD for S-TRACP5b and S-CTX-I were 1.16 and 1.13, respectively (Table 2). High S-TRACP5b and S-CTX-I levels also were associated with increased risk for vertebral fractures with HRs of 1.22 and 1.32, respectively (see Table 2). When fracture risk was analyzed for tertiles, women in the highest tertile for S-TRACP5b and U-MidOC had greater fracture risk compared with women in the lowest tertile (Fig. 1C, D; see also Table 2). HR for highest S-CTX-I tertile also was of similar magnitude, but it did not reach statistical significance (see Fig. 1B). The same was observed for vertebral fractures (Fig. 2B–D). Bone formation markers S-OC[1–49], S-TotalOC, S-cOC, and S-boneALP were not associated with fracture risk either when analyzed per SD increase or in tertiles (data not shown), and none of the seven bone markers analyzed was able to predict hip fractures (data not shown).

Table 2. Baseline and 1-Year Bone Markers and the Average Value of Bone Markers and the Hazard Ratio (HR with 95% CI) for Any Fracture and Clinical Vertebral Fracture
HR (95% CI) per SD or TRTBaseline BTM (n = 1044)1-year BTM (n = 915)aAverage 0 + 1 year BTM (n = 963)b
High TRTper SDHigh TRTper SDHigh TRTper SD
  • All women who attended baseline visit (n = 1044) are included. Relative risk is given for the highest tertile (TRT = tertiles based on all 1044 women) and per SD increase in marker levels. The lowest tertile has been used as the reference group (HR = 1.0), and the statistically significant values are shown in bold text. Women with any fracture or vertebral fracture are compared with women without any fracture. For baseline samples, follow-up starts from baseline visit, and the mean follow-up time is 9.0 years (range 7.4–10.9 years). For 1-year samples and the average samples, the follow-up starts from the 1-year visit, and the mean follow up time is 7.9 years (range 6.3–9.9 years).

  • a

    Women who did not attend 1-year visit and/or who sustained a fracture (or died) before 1-year visit are excluded.

  • b

    Women who sustained a fracture (or died) before the 1-year visit are excluded.

Any fracturen = 363 n = 287 n = 300 
 S-CTX1.29 (0.99–1.67)1.13 (1.01–1.27)1.45 (1.08–1.95)1.13 (1.00–1.28)1.40 (1.05–1.87)1.12 (0.99–1.26)
 S-TRACP5b1.40 (1.09–1.81)1.16 (1.04–1.29)1.37 (1.01–1.85)1.15 (1.02–1.29)1.40 (1.05–1.87)1.14 (1.02–1.28)
 U-MidOC1.34 (1.04–1.74)1.08 (0.97–1.20)1.33 (1.00–1.79)1.09 (0.97–1.24)1.33 (1.00–1.76)1.08 (0.96–1.22)
Vertebral fracturen = 103 n = 75 n = 79 
 S-CTX1.42 (0.88–2.28)1.32 (1.05–1.67)1.83 (1.01–3.31)1.26 (0.98–1.61)1.90 (1.09–3.30)1.24 (0.97–1.58)
 S-TRACP5b1.43 (0.90–2.28)1.22 (1.01–1.48)1.70 (0.93–3.10)1.30 (1.04–1.63)1.58 (0.91–2.73)1.22 (0.98–1.53)
 U-MidOC1.43 (0.90–2.26)1.14 (0.93–1.40)1.88 (1.08–3.29)1.39 (1.08–1.78)1.92 (1.12–3.30)1.28 (1.00–1.63)
thumbnail image

Figure 1. Cumulative percentage of women without at least one fracture of any type during the entire study mean follow-up of 9.0 years (range 7.4–10.9 years). Fractures are given for the low tertile (TRT, dotted line), middle tertile (dashed line), and high tertile (solid line) of baseline total body aBMD (A), S-CTX-I (B), S-TRACP5b (C), and U-MidOC (D). Highest and lowest tertile are compared 7.5 years after baseline or for the entire follow-up period of a mean of 9 years (7.4–10.9 years). p values for log-rank tests are given (7.5 years, upper p value; entire follow-up, lower p value).

Download figure to PowerPoint

thumbnail image

Figure 2. Cumulative percentage of women without at least one clinical vertebral fracture during the entire study mean follow-up of 9.0 years (range 7.4–10.9 years). Fractures are given for the low tertile (TRT, dotted line), middle tertile (dashed line), and high tertile (solid line) of baseline total body aBMD (A), S-CTX-I (B), S-TRACP5b (C), and U-MidOC (D). Highest and lowest tertiles are compared 7.5 years after baseline or for the entire follow-up period of a mean of 9 years (7.4–10.9 years). p values for log-rank tests are given (7.5 years, upper p value; entire follow-up, lower p value).

Download figure to PowerPoint

Baseline S-CTX-I and S-TRACP5b also were significantly associated with any fracture after adjustment for kidney function (creatinine clearance), vitamin D status [serum 25(OH)D3], BMI, fracture history, smoking, or cortisol use (data not shown). However, HRs were not statistically significant when adjusted for baseline total body BMD (S-CTX-I HR = 1.08, 95% CI 0.96–1.22; S-TRACP5b HR = 1.07, 95% CI 0.95–1.21), indicating that bone marker levels were not independently associated with fracture risk. Adjustment for age was not considered necessary because all participants were exactly 75 years old.

One-year BTMs or the average of two measurements and fracture prediction

BTMs assessed 1 year after baseline visit were able to predict fractures during the remaining follow-up period of a mean of 7.9 years (range 6.3–9.9 years) similar to BTMs measured at baseline. High S-CTX-I and S-TRACP5b levels were significantly associated with an increased risk for any fracture, and resorption markers also were associated with an increased risk for vertebral fracture (see Table 2). Bone-formation markers (S-OC[1–49], S-TotalOC, S-cOC, and S-boneALP) measured at the 1-year visit were not associated with fracture risk (data not shown). We also analyzed fracture prediction by using the average of baseline and 1-year measurements. When the average of these two independent measurements was used, the HRs were similar and more consistent: HRs for the highest tertiles reached the level of significance for all resorption markers (see Table 2).

Baseline aBMD and fracture prediction

Baseline aBMD of total body, total hip, and lumbar spine were strongly associated with fracture risk for the entire follow-up period (Table 3). A 1 SD decrease in aBMD resulted in a significant increase in risk for any fracture, and HRs for total body aBMD, hip aBMD, and spine aBMD were 1.50, 1.51, and 1.31, respectively. The association of hip aBMD with hip fractures was even more pronounced, with an HR of 2.03, and BMD measured at all three sites also was strongly associated with clinical vertebral fractures (Table 4). Women in the lowest baseline total body aBMD tertile had significantly greater risk for sustaining any fracture (see Fig. 1A) or a vertebral fracture (see Fig. 2A) when compared with women in the highest tertile.

Table 3. Baseline BMD and Bone Markers and the Hazard Ratio (HR with 95% CI) for Any Fracture From Baseline (Time Point 0) to 2.5, 5.0, and 7.5 Years After Baseline
HR per SD (95% CI)Fracture data collected (start to endpoint)
Any fracture0–2.5 years (n = 123)0–5.0 years (n = 231)0–7.5 years (n = 314)2.5–5.0 years (n = 108)5.0–7.5 years (n = 83)0–mean 9 years (n = 363)
  1. Women with any fracture are compared with women without any fracture (n = 921/813/730/773/650). Follow-up starts from baseline visit, and the follow-up times are 2.5, 5.0, 7.5, 5.0, and 7.5 years. For the fourth and fifth columns, only fractures after 2.5 years and after 5.0 years, respectively, are taken into account, and women who had a fracture (or died) before 2.5 and 5.0 years, respectively, have been excluded. In the sixth column, entre follow-up of 9 years is shown as a reference (see Table 1). HR is given per SD increase in BTM or per SD decrease in aBMD. The statistically significant values are shown shown in bold.

aBMD      
 Total body aBMD1.49 (1.34–1.80)1.46 (1.27–1.68)1.51 (1.34–1.71)1.42 (1.15–1.75)1.69 (1.32–2.15)1.50 (1.34–1.67)
 Hip aBMD1.59 (1.29–1.97)1.49 (1.28–1.73)1.57 (1.38–1.78)1.39 (1.12–1.72)1.79 (1.40–2.31)1.51 (1.34–1.70)
 Spine aBMD1.45 (1.18–1.78)1.34 (1.16–1.55)1.35 (1.19–1.52)1.23 (1.00–1.51)1.36 (1.07–1.73)1.31 (1.17–1.47)
BTMs      
 S-OC[1–49]1.29 (1.06–1.57)1.12 (0.97–1.28)1.07 (0.95–1.20)0.96 (0.79–1.16)0.97 (0.78–1,20)1.01 (0.91–1.13)
 S-TotalOC1.22 (1.01–1.47)1.11 (0.97–1.27)1.07 (0.95–1.20)0.99 (0.82–1.21)0.97 (0.78–1.22)1.03 (0.92–1.15)
 S-cOC1.22 (1.01–1.48)1.07 (0.94–1.23)1.05 (0.93–1.18)0.93 (0.77–1.13)0.98 (0.78–1.23)1.01 (0.90–1.12)
 S-boneALP1.13 (0.95–1.36)1.07 (0.94–1.22)1.06 (0.94–1.19)1.01 (0.83–1.22)1.02 (0.82–1.28)1.05 (0.95–1.17)
 S-CTX1.37 (1.10–1.70)1.20 (1.04–1.39)1.22 (1.07–1.38)1.06 (0.86–1.29)1.26 (0.97–1.62)1.13 (1.01–1.27)
 S-TRACP5b1.30 (1.09–1.56)1.21 (1.06–1.38)1.17 (1.04–1.31)1.10 (0.91–1.34)1.06 (0.85–1.32)1.16 (1.04–1.29)
 U-MidOC1.21 (1.00–1.46)1.07 (0.93–1.23)1.08 (0.96–1.22)0.94 (0.77–1.14)1.13 (0.90–1.41)1.08 (0.97–1.20)
Table 4. Baseline BMD and Bone Markers and the Hazard Ratio (HR with 95% CI) for Clinical Vertebral Fracture From Baseline (Time Point 0) to 2.5, 5.0, and 7.5 Years After Baseline
HR per SD (95% CI)Fracture data collected (start to endpoint)
Vertebral fracture0–2.5 years (n = 32)0–5.0 years (n = 60)0–7.5 years (n = 87)2.5–5.0 years (n = 22)5.0–7.5 years (n = 21)0–mean 9 years (n = 103)
  1. Women with vertebral fracture are compared with women without any fracture (n = 921/813/730/773/650). Follow-up starts from baseline visit and the follow-up times are 2.5, 5.0, 7.5, 5.0, and 7.5 years. For the fourth and fifth columns, only fractures after 2.5 and after 5.0 years, respectively, are taken into account, and women who had a fracture (or died) before 2.5 and 5.0 years, respectively, have been excluded. In the sixth column, entire follow up of 9 years is shown as a reference (see Table 1). HR is given per SD increase in BTM or per SD decrease in aBMD. The statistically significant values are shown in bold.

aBMD      
 Total body aBMD1.83 (1.25–2.67)2.19 (1.65–2.89)2.27 (1.80–2.87)2.76 (1.73–4.41)2.42 (1.49–3.93)2.23 (1.80–2.77)
 Hip aBMD1.51 (1.01–2.28)1.97 (1.44–2.69)2.23 (1.71–2.90)2.59 (1.50–4.48)2.68 (1.52–4.73)2.15 (1.70–2.73)
 Spine aBMD2.03 (1.33–3.10)2.11 (1.54–2.87)2.05 (1.58–2.66)2.50 (1.47–4.23)2.00 (1.16–3.44)1.92 (1.52–2.43)
BTMs      
 S-OC[1–49]1.54 (1.03–2.29)1.64 (1.22–2.21)1.36 (1.07–1.74)2.40 (1.44–4.00)0.96 (0.63–1.46)1.18 (0.95–1.46)
 S-TotalOC1.44 (1.00–2.08)1.58 (1.21–2.08)1.31 (1.05–1.64)2.39 (1.48–3.85)0.89 (0.58–1.36)1.17 (0.95–1.43)
 S-cOC1.46 (0.99–2.15)1.52 (1.14–2.03)1.24 (0.98–1.56)2.29 (1.37–3.82)0.87 (0.57–1.33)1.10 (0.89–1.36)
 S-boneALP1.66 (1.19–2.31)1.44 (1.12–1.86)1.36 (1.10–1.67)1.41 (0.91–2.16)1.33 (0.86–2.05)1.19 (0.98–1.45)
 S-CTX1.69 (1.08–2.63)1.61 (1.17–2.23)1.57 (1.20–2.06)1.91 (1.06–3.43)1.54 (0.88–2.68)1.32 (1.05–1.67)
 S-TRACP5b1.46 (1.03–2.07)1.46 (1.13–1.89)1.28 (1.03–1.58)1.70 (1.10–2.63)0.90 (0.59–1.37)1.22 (1.01–1.48)
 U-MidOC1.59 (1.07–2.35)1.43 (1.08–1.89)1.24 (0.99–1.55)1.52 (0.96–2.42)0.99 (0.64–1.52)1.14 (0.93–1.40)

Combining BMD tertile with resorption marker tertile

In an attempt to identify a potential high-risk subgroup in women with low BMD, women belonging to the lowest tertile of total hip aBMD and at the same time to the highest tertile of bone marker (S-TRACP5b, n = 136; S-CTX-I, n = 116; U-MidOC, n = 124) were compared with women belonging to two other tertiles of total hip BMD (n = 618). When hip BMD and bone markers were combined, the HR with S-TRACP5b was 1.94, with S-CTX-I was 1.71, and with U-MidOC was 1.71 (95% CIs 1.46–2.59, 1.25–2.35, and 1.25–2.35, respectively). HRs were not significantly higher than when low hip BMD tertile alone (n = 308) was used, HR = 1.95 (95% CI 1.56–2.43). If only vertebral fractures were analyzed, the HR with S-TRACP5b was 3.68, with S-CTX-I was 2.93, and with U-MidOC was 4.11 (95% CIs 2.23–6.01, 1.67–5.19, and 2.46–6.85, respectively). When low hip BMD tertile alone was used, the HR for vertebral fracture was 3.61 (95% CI 2.39–5.45).

For how long time can markers predict fractures?

HRs for any fracture were greatest when we used the shortest follow-up period of 2.5 years.

Both formation and resorption markers were able to predict fractures 2.5 years from baseline with HRs from 1.13 (S-boneALP) to 1.37 (S-CTX-I). HRs for bone markers were not significant for the fractures sustained only between 2.5 and 5 years and between 5 and 7.5 years (see Table 3). The observed 7.5-year risk of any fracture in the lowest bone marker tertile was 25% to 27%, in the middle tertile 29% to 32%, and in the highest tertile 36% to 38% depending on the resorption marker used. Fracture frequency in each tertile during the entire follow-up of a mean of 9 years is shown in Figure 3.

Both formation and resorption markers also were able to predict vertebral fractures 2.5 years from baseline with HRs from 1.44 (S-TotalOC) to 1.69 (S-CTX-I). HRs for vertebral fracture prediction were of similar magnitude or even greater when the follow-up was extended to 5 or to 7.5 years after baseline (Table 4). Furthermore, baseline bone markers also were able to predict those vertebral fractures sustained only between 2.5 and 5 years. The observed 7.5-year risk of vertebral fracture in the lowest bone marker tertile was 8% to 10%, in the middle tertile also 8% to 10%, and in the highest tertile 14% to 15% depending on the resorption marker used. Vertebral fracture frequency in each tertile during the entire follow-up of a mean of 9 years is shown in Figure 3.

thumbnail image

Figure 3. Fracture frequency in each tertile (TRT) of baseline total body aBMD (A), S-CTX-I (B), S-TRACP5b (C), and U-MidOC (D). Fractures were recorded for the entire mean follow-up of 9 years (range 7.4–10.9 years). Frequencies are shown for any type of fractures and for clinical vertebral fractures.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

In summary, in this study of 1040 elderly women we observed that high levels of the resorption markers S-TRACP5b, S-CTX-I, and U-MidOC were associated with increased fracture risk during the mean follow-up of 9 years. Association between BTMs and fracture risk has been evaluated in several studies, but the data concerning long-term follow-up are more limited. Melton and colleagues followed 224 postmenopausal women for a mean of 14 years with the older bone turnover markers measured in the 1980s, such as total alkaline phosphatase and urinary hydroxyproline. They found no association with long-term fracture risk.22 Sornay-Rendu and colleagues studied 322 postmenopausal osteopenic women of the OFELY study for 9 years and demonstrated that being in the highest quartile for bone-specific alkaline phosphatase was associated with an increased fracture risk.17 In this present study, bone-resorption markers S-TRACP5b, S-CTX-I, and U-MidOC were moderately but significantly associated with fracture risk for the mean follow-up of 9 years, in particular, vertebral fracture risk. This is in line with earlier studies with other resorption markers and shorter follow-up, predominantly S-CTX,12, 14 and with our earlier report on the shorter follow-up in the same cohort.13 Fracture risks observed for two independent samples collected 1 year apart from each other were similar, suggesting that the risk was maintained over the 1-year period. Moreover, the association with fracture risk was similar but more consistent when the precision of measurement was increased by averaging the two measurements. Assays for bone markers typically have high day-to-day variability,23 and measuring turnover several times over a longer period of time may provide more precise information on bone metabolism.24

In this study, bone markers were not significantly associated with fracture risk after adjustment to baseline aBMD. Furthermore, by combining bone marker results with aBMD, fracture prediction was not significantly improved. These findings show that bone marker levels were not independently associated with long-term fracture risk in elderly women and is in contrast to some earlier reports by others14, 25 and by us with shorter follow-up time in this cohort.13 Participants in our study were 75-year-old women well beyond menopause, and the follow-up period was relatively long, on average, 9 years. In this age group of elderly women, low bone mass resulting from earlier postmenopausal bone loss appears to be a more important predictor for long-term fracture risk than the level of bone turnover measured at old age. Other factors affecting fracture risk, such as mobility and balance, also may have deteriorated in the aged population. Therefore, it is not likely that it would be useful to include these bone markers in fracture algorithms, such as the FRAX analysis, for individual patients aiming for long-term fracture prediction in this elderly age group. In another study of 1664 women above the age of 70 years and living in nursing homes, no significant association was found with bone markers (osteocalcin, CTX, osteoprotegerin) and the risk of fracture.26 Bone markers are associated with bone loss more strongly in perimenopausal women than in postmenopausal women,27 and BTMs could be related to future fracture risk more when measured in early postmenopausal women, during the period of accelerated bone turnover and rapid bone loss, compared with elderly women.

The time frame between a bone marker measurement and the fracture assessment should influence the predictive value of markers. Usually the predictive power between measurement and outcome assessment tends to decrease over time; for instance, in the EPIDOS study, the duration of hip fracture prediction was around 5 years for the ultrasound and 10 years for the femoral neck BMD.28 In this present study, we observed that BTMs were most associated for fractures sustained within a few years of BTM measurement. Markers were more associated with clinical vertebral fractures and, importantly, significantly also with vertebral fractures sustained between 2.5 and 5 years of follow-up, suggesting that markers are predictive for vertebral fractures for a longer time period than for fractures of any other type. Vertebral fractures may be more dependent on high bone metabolic activity and bone loss rate compared with fractures at other skeletal sites because of the large proportion of metabolically active trabecular bone in the vertebra and of being less susceptibile to external trauma.

None of the markers was significantly associated with hip fractures. This has also been observed in several previous studies.12 Studies on the injury mechanisms of hip fracture have suggested that a fall onto the hip is a strong risk factor for hip fracture in the elderly, even stronger than the age-related bone loss or osteoporosis.29 Therefore, rate of bone metabolism may play only a relatively minor role in hip fracture prediction, which also may be a reflection of the larger amount of cortical bone and therefore comparably low metabolic activity in the hip.

The strengths of this study include the size of the cohort, random selection, same age of all women, measurement of several BTMs reflecting different aspects of bone metabolism, relatively large number of fracture outcomes, and the long follow-up time of up to 11 years. While there are accumulating data on many conventional markers used in large prospective studies, the data on more recently emerged analytes in fracture prediction are more limited. For recently developed analytes, such as S-TRACP5b and urinary osteocalcin, the data are limited to our previous report on shorter follow-up in the same cohort.13 There are also limitations that should be taken into account when the results of this study are applied. When the study was initiated, information on the effect of feeding on BTMs was not available. Samples were collected without fasting, and the nonfasting status may have affected the results, particularly for S-CTX-I.30 However, since a greater degree of variation is expected in nonfasting samples, sampling after fasting could have enhanced the association between bone markers and fracture. Women were well beyond menopausal age, and the association of BTMs with bone loss has been shown to be stronger in perimenopausal women at the time of rapid bone loss.27 Fractures are, however, more common in the older age groups, and therefore, the study was originally designed for elderly women. We have previously reported in this same cohort that women who are in the highest tertile for bone resorption markers for several years have the greatest bone loss over the same time period.24 It will be interesting to evaluate fracture prediction using long-term bone turnover rate instead of baseline measurements when more long-term fracture data will become available.

We conclude that elevated levels of bone resorption markers are consistently associated with increased fracture risk for up to a decade in elderly women. There was a good agreement between fracture risks obtained for two independent samples collected 1 year apart from each other. BTMs were mostly associated with fractures sustained within a few years of BTM measurement, but the predictive ability remained for a longer time for vertebral fractures.

Disclosures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

HKV has patent licensing arrangements with Immunodiagnostic Systems Limited (IDS). All other authors state that they have no conflicts of interest.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

We thank Dr Patric Garnero (Lyon, France) for valuable comments during manuscript preparation. We also thank Katja Fagerlund for assistance on S-TRACP5b analysis, Anders Isaksson on S-boneALP and S-CTX-I analyses, Jan-Åke Nilsson for statistical advice, and Eva-Lena Forsberg and Åsa Almgren for technical assistance. This work received financial support from the Swedish Medical Research Council and the following foundations: Herman Järnhards Stiftelse (Sweden), Malmö University Hospital Foundation (Sweden), Research Foundation of Orion Corporation (Finland), and Helsingin Sanomain 100-vuotissäätiö (Finland).

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  • 1
    Consensus development conference: diagnosis, prophylaxis, and treatment of osteoporosis. Am J Med. 1993; 94: 646650.
  • 2
    NIH Consensus Development Panel on Osteoporosis Prevention, Diagnosis, and Therapy. JAMA. 2001; 285: 785795.
  • 3
    Cummings SR, Melton LJ. Epidemiology and outcomes of osteoporotic fractures. Lancet. 2002; 359: 17611767.
  • 4
    Assessment of fracture risk and its application to screening for postmenopausal osteoporosis: report of a WHO Study Group. WHO Tech Rep Ser 1994; 843: 1129.
  • 5
    Marshall D, Johnell O, Wedel H. Meta-analysis of how well measures of bone mineral density predict occurrence of osteoporotic fractures. BMJ. 1996; 312: 12541259.
  • 6
    Johnell O, Kanis JA, Oden A, et al. Predictive value of BMD for hip and other fractures. J Bone Miner Res. 2005; 20: 11851194.
  • 7
    Schuit SC, van der Klift M, Weel AE, et al. Fracture incidence and association with bone mineral density in elderly men and women: the Rotterdam Study. Bone. 2004; 34: 195202.
  • 8
    Wainwright SA, Marshall LM, Ensrud KE, et al. Hip fracture in women without osteoporosis. J Clin Endocrinol Metab. 2005; 90: 27872793.
  • 9
    Sanders KM, Nicholson GC, Watts JJ, et al. Half the burden of fragility fractures in the community occur in women without osteoporosis: when is fracture prevention cost-effective? Bone. 2006; 38: 694700.
  • 10
    Kanis JA, Oden A, Johnell O, et al. The use of clinical risk factors enhances the performance of BMD in the prediction of hip and osteoporotic fractures in men and women. Osteoporos Int. 2007; 18: 10331046.
  • 11
    Kanis JA, Johnell O, Oden A, Johansson H, McCloskey E. FRAX and the assessment of fracture probability in men and women from the UK. Osteoporos Int. 2008; 19: 385397.
  • 12
    Garnero P. Markers of bone turnover for the prediction of fracture risk. Osteoporos Int. 2000; 11: S55S65.
  • 13
    Gerdhem P, Ivaska KK, Alatalo SL, et al. Biochemical markers of bone metabolism and prediction of fracture in elderly women. J Bone Miner Res. 2004; 19: 386393.
  • 14
    Garnero P, Hausherr E, Chapuy MC, et al. Markers of bone resorption predict hip fracture in elderly women: the EPIDOS Prospective Study. J Bone Miner Res. 1996; 11: 15311538.
  • 15
    Bruyere O, Collette J, Delmas P, et al. Interest of biochemical markers of bone turnover for long-term prediction of new vertebral fracture in postmenopausal osteoporotic women. Maturitas. 2003; 44: 259265.
  • 16
    Schousboe JT, Bauer DC, Nyman JA, Kane RL, Melton LJ, Ensrud KE. Potential for bone turnover markers to cost-effectively identify and select post-menopausal osteopenic women at high risk of fracture for bisphosphonate therapy. Osteoporos Int. 2007; 18: 201210.
  • 17
    Sornay-Rendu E, Munoz F, Garnero P, Duboeuf F, Delmas PD. Identification of osteopenic women at high risk of fracture: the OFELY study. J Bone Miner Res. 2005; 20: 18131819.
  • 18
    Garnero P. Biomarkers for osteoporosis management: utility in diagnosis, fracture risk prediction and therapy monitoring. Mol Diagn Ther. 2008; 12: 157170.
  • 19
    Kakonen SM, Hellman J, Karp M, et al. Development and evaluation of three immunofluorometric assays that measure different forms of osteocalcin in serum. Clin Chem. 2000; 46: 332337.
  • 20
    Ivaska KK, Kakonen SM, Gerdhem P, Obrant KJ, Pettersson K, Vaananen HK. Urinary osteocalcin as a marker of bone metabolism. Clin Chem. 2005; 51: 618628.
  • 21
    Gerdhem P, Akesson K. Rates of fracture in participants and non-participants in the Osteoporosis Prospective Risk Assessment study. J Bone Joint Surg Br. 2007; 89: 16271631.
  • 22
    Melton LJ III, Crowson CS, O'Fallon WM, Wahner HW, Riggs BL. Relative contributions of bone density, bone turnover, and clinical risk factors to long-term fracture prediction. J Bone Miner Res. 2003; 18: 312318.
  • 23
    Hannon R, Eastell R. Preanalytical variability of biochemical markers of bone turnover. Osteoporos Int. 2000; 11: S30S44.
  • 24
    Ivaska KK, Lenora J, Gerdhem P, Akesson K, Vaananen HK, Obrant KJ. Serial assessment of serum bone metabolism markers identifies women with the highest rate of bone loss and osteoporosis risk. J Clin Endocrinol Metab. 2008; 93: 26222632.
  • 25
    Garnero P, Sornay-Rendu E, Claustrat B, Delmas PD. Biochemical markers of bone turnover, endogenous hormones and the risk of fractures in postmenopausal women: the OFELY study. J Bone Miner Res. 2000; 15: 15261536.
  • 26
    Dobnig H, Piswanger-Solkner JC, Obermayer-Pietsch B, et al. Hip and nonvertebral fracture prediction in nursing home patients: role of bone ultrasound and bone marker measurements. J Clin Endocrinol Metab. 2007; 92: 16781686.
  • 27
    Iki M, Morita A, Ikeda Y, et al. Biochemical markers of bone turnover predict bone loss in perimenopausal women but not in postmenopausal women: the Japanese Population-based Osteoporosis (JPOS) Cohort Study. Osteoporos Int. 2006; 17: 10861095.
  • 28
    Hans D, Schott AM, Duboeuf F, Durosier C, Meunier PJ. Does follow-up duration influence the ultrasound and DXA prediction of hip fracture? The EPIDOS prospective study. Bone. 2004; 35: 357363.
  • 29
    Greenspan SL, Myers ER, Maitland LA, Resnick NM, Hayes WC. Fall severity and bone mineral density as risk factors for hip fracture in ambulatory elderly. JAMA. 1994; 271: 128133.
  • 30
    Clowes JA, Hannon RA, Yap TS, Hoyle NR, Blumsohn A, Eastell R. Effect of feeding on bone turnover markers and its impact on biological variability of measurements. Bone. 2002; 30: 886890.