SEARCH

SEARCH BY CITATION

Keywords:

  • Osteoporosis;
  • Fracture risk;
  • Bone mineral density;
  • Trabecular bone score;
  • Postmenopausal women

Abstract

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

The measurement of BMD by dual-energy X-ray absorptiometry (DXA) is the “gold standard” for diagnosing osteoporosis but does not directly reflect deterioration in bone microarchitecture. The trabecular bone score (TBS), a novel gray-level texture measurement that can be extracted from DXA images, correlates with 3D parameters of bone microarchitecture. Our aim was to evaluate the ability of lumbar spine TBS to predict future clinical osteoporotic fractures. A total of 29,407 women 50 years of age or older at the time of baseline hip and spine DXA were identified from a database containing all clinical results for the Province of Manitoba, Canada. Health service records were assessed for the incidence of nontraumatic osteoporotic fracture codes subsequent to BMD testing (mean follow-up 4.7 years). Lumbar spine TBS was derived for each spine DXA examination blinded to clinical parameters and outcomes. Osteoporotic fractures were identified in 1668 (5.7%) women, including 439 (1.5%) spine and 293 (1.0%) hip fractures. Significantly lower spine TBS and BMD were identified in women with major osteoporotic, spine, and hip fractures (all p < 0.0001). Spine TBS and BMD predicted fractures equally well, and the combination was superior to either measurement alone (p < 0.001). Spine TBS predicts osteoporotic fractures and provides information that is independent of spine and hip BMD. Combining the TBS trabecular texture index with BMD incrementally improves fracture prediction in postmenopausal women. © 2011 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
  10. Supporting Information

Osteoporosis has emerged as a major health concern in almost all industrialized countries,1–3 with up to 9 million new osteoporotic fractures expected each year.4 The mortality rate associated with hip and spine fractures can exceed 20%.5, 6 In the United States, osteoporosis affects as many as 4 to 6 million postmenopausal women,7 with 2 million fractures occurring annually.8 Up to 10% of women in their fifties have already experienced an osteoporotic fracture.9 In Canada, osteoporosis affects more than one in four women older than age 50 years.10 Other investigators have identified significant osteoporosis risks in men as well.10 Moreover, given steady increases in the lifespans of both men and women, these numbers have been projected to double over the next 40 to 50 years.11

BMD, measured by dual-energy X-ray absorptiometry (DXA), has been the reference standard for osteoporosis diagnosis in the absence of established fragility fractures.12 BMD is one of the major determinants of bone strength and fracture risk,13 but there is considerable overlap in BMD values between individuals who develop fractures and those who do not.14 Other factors influence bone strength and fracture risk, including the macrogeometry of cortical bone, the microarchitecture of trabecular bone, bone microdamage, mineralization, and turnover.15, 16 In recent years, a number of techniques have been developed for bone microarchitecture assessment.15–20 Among the noninvasive techniques, (peripheral) quantitative computed tomography (pQCT, QCT) and magnetic resonance imaging (MRI) allow for the direct measurement of bone microarchitecture, and both have benefited from significant enhancements in either acquisition technology or image characterization. However, these two techniques remain impractical for routine screening and clinical management owing to high costs and the inconvenience of having patients return to undergo another time-consuming assessment after DXA has been performed. Two-dimensional (2D) X-ray-based images, such as plain radiographs, have been investigated as a more practical alternative for the noninvasive assessment of bone microarchitecture. Different gray-level features have been explored, including fractal dimension and Fourier analysis,15–20 but none has clearly demonstrated added value in routine clinical practice.

The trabecular bone score (TBS, unitless) is a new texture measurement that can be applied to any X-ray images including DXA images by quantifying local variations in gray level.21, 22 TBS uses experimental variograms of 2D projection images to differentiate between 3D microarchitectures that exhibit the same BMD but different trabecular characteristics.21, 22 In human cadavers,22 significant correlations have been identified between TBS and 3D parameters of bone microarchitecture, independent of any correlation between TBS and BMD. The greatest correlation was between TBS and the density of connectivity (Conn.D), with TBS explaining about 67.2% of the variance in Conn.D. Based on multivariate linear regression modeling, a model has been established to allow for interpretation of the relationship between TBS and 3D bone microarchitecture parameters. Higher scores reflect stronger and more fracture-resistant microarchitecture, whereas lower scores indicate bone that is weaker and more susceptible to fracture.21, 22 Since it is constrained by neither the size nor shape of the region measured, TBS can be applied to small and/or irregular surfaces, such as the standard regions of measurement used in DXA images. TBS can be applied retrospectively to an existing DXA exam without the need for any further imaging and can be compared directly with BMD because both evaluate the same region of bone. The added value of TBS in bone mineral densitometry for fracture risk assessment has been documented in cross-sectional studies.23–25 Indeed, TBS has been found (1) to be lower in postmenopausal women with a past osteoporotic fracture compared with age- and BMD-matched women without fracture,23 (2) to give an incremental increase in the odds ratio for spine fracture when combined with spine BMD.24, 25 and (3) to be lower in women with (versus without) fractures irrespective of whether their BMD met the criteria for osteoporosis or osteopenia.24, 25

The objective of this study was to determine whether TBS can predict osteoporosis-related fractures independent of BMD in a large cohort of postmenopausal women.

Methods

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

Patient population

In this retrospective historical cohort study, 2D gray-scale DXA images of the lumbar spine, collected from a large cohort of postmenopausal women from the Canadian Province of Manitoba, were sent to the University of Lausanne, Switzerland, for the calculation of spine TBS. The Manitoba Bone Density Program is a targeted case-finding clinical program. The associated database has been validated and described elsewhere26–29 and shown to exceed 99% in terms of completeness and accuracy.30 All women 50 years of age or older who had undergone BMD measurement of the spine and hip by DXA using a single, narrow, fan-beam scanner configuration (Prodigy, GE Healthcare, Madison, WI, USA) were eligible for inclusion, provided that they had medical coverage during the observation period ending March 31, 2007. For women with more than one eligible set of measurements, only the first record was included in the analysis. The final sample consisted of 29,407 women. The study was approved by the Research Ethics Board for the University of Manitoba and the Health Information Privacy Committee of Manitoba Health.

Data sources

In the Province of Manitoba, Canada, health services are provided to virtually all residents through a single public health care system. Manitoba Health maintains computerized databases of physician billing claims and hospital separations for all residents of the province eligible to receive health services. Each health system contact includes information on a patient's demographics, date and type of service, and diagnoses from (1) physician billing claims (inpatient, outpatient, and private office) coded using the International Classification of Disease, 9th edition, Clinical Modification (ICD-9-CM) system and (2) hospital discharge abstracts, for which the diagnoses and procedures have been coded using the ICD-9-CM system prior to 2005 and the ICD-10-CA system thereafter.29 Anonymous linkage of these databases to the BMD database was possible via a unique scrambled health identification number, thereby allowing for the creation of a longitudinal record of health services and outcomes. Longitudinal health service records were assessed for the presence of fracture codes before and after BMD testing that were not associated with trauma codes.30 Hip fractures and major osteoporotic fractures (ie, hip, clinical spine, forearm, and humerus fractures) were studied because these are the basis for the 10-year absolute fracture risk estimates published by Kanis and colleagues.31, 32 We required that hip and forearm fractures be accompanied by a site-specific fracture reduction, fixation, or casting code, which enhances the diagnostic and temporal specificity of an acute fracture. These same fracture definitions have been used in previous analyses to show that BMD measurements predict fractures in our clinical cohort as well as those reported in large meta-analyses.33 In addition to incident and prior osteoporotic fractures, we identified prior diagnoses of rheumatoid arthritis, diabetes, chronic obstructive pulmonary disease (COPD, a proxy for smoking), substance abuse (a proxy for high alcohol intake), prolonged (>3 months) systemic corticosteroid use in the last year, and pharmacologic treatment for osteoporosis dispensed in the last year. A global comorbidity measure was constructed using the Johns Hopkins Ambulatory Care Group (ACG) Case-Mix System (Version 8).34 This was based on the number of ambulatory diagnostic groups (ADGs), which represent 32 comorbidity clusters of every ICD-9-CM diagnostic code. The number of ADGs (categorized as none, 1 to 2, 3 to 5 and 6 or more) is strongly linked with fracture risk.35

Measurement of BMD

All DXA scans were performed using Prodigy scanners (GE Healthcare) and analyzed (Encore Software 12.4) in accordance with the manufacturer's recommendations. BMD measurements were recorded for the lumbar spine BMD for L1 through L4 (L1–L4), the femoral neck, and the total hip. Hip T- and Z-scores were calculated using the revised National Health and Nutrition Examination Survey (NHANES) III white female reference values.36, 37 For the lumbar spine, manufacturer reference data for white US women were used. TBS and BMD values that fell below the 0.1 percentile or above the 99.9 percentile were treated as outliers and excluded from further analysis. The resulting data approximated a normal distribution. Instruments were cross-calibrated using anthropomorphic phantoms. No clinically significant differences were identified; therefore, all analyses are based on unadjusted numerical results generated by the instrument. All three instruments used for this study exhibited stable long-term performance (coefficient of variation [CV] < 0.5%) and satisfactory in vivo precision. Anthropomorphic data (ie, height and weight) were measured at the time of DXA, and BMI was calculated.

Measurement of trabecular bone score (TBS)

All TBS measurements were performed in the Bone Disease Unit at the University of Lausanne, Lausanne, Switzerland (TBS iNsight Software, Version 1.8, Med-Imaps, Pessac, France), using anonymized spine DXA files from the Manitoba database to ensure blinding of the Swiss investigators to all clinical parameters and outcomes. For each region of measurement, TBS was evaluated based on gray-level analysis of the DXA images as the slope at the origin of the log-log representation of the experimental variogram.22 The TBS iNsight software can be installed either on the DXA devices directly (Hologic and GE Healthcare Lunar) for operator-independent automated analysis or on a stand-alone workstation. In both cases, the software uses the anteroposterior spine raw image(s) from the densitometer, including the BMD region of interest (ROI) and edge detection, so that the TBS calculation is performed over exactly the same ROI as the BMD measurement. The average automated analysis time is about 20 seconds. In the current analysis, we use a research version of the commercialized TBS iNsight software that allows for large-batched analyses from a workstation. No significant differences in mean TBS measurements were seen for the three DXA scanners used. Short-term reproducibility (CV) for TBS calculated from all three instruments used for this study was 2.1% and for spine BMD was 1.7% in 92 individuals with repeat spine DXA scans performed within 28 days (51 same day, 41 different day).

Statistical analysis

All statistical analyses were performed using Statistica (Version 8.0, StatSoft, Inc., Tulsa, OK, USA). A p value of 0.001 was set as the threshold for statistical significance in all intergroup and intervariable comparisons to adjust for multiple comparisons, and all inferential tests were two-tailed. Continuous variables were reported as means with standard deviations and all counts as a percentage of the total sample. Nonpaired group comparisons in the means for subject age, morphometric variables, spine BMD, and spine TBS were assessed in women with and without fractures using the Student's t test. Pearson correlations were used to identify the linear relationship between the various bone measurements. The Cochran-Armitage test was performed to identify linear trends in incident fractures according to TBS tertile, stratified by BMD T-score (categorized as normal, osteopenia, or osteoporosis) for the total hip, lumbar spine, and minimum site (lowest BMD T-score of spine, total hip, and femoral neck). Odds ratios (ORs) for fracture were computed between TBS tertiles, also stratified by BMD T-score. Tests for trend were conducted both including and excluding subjects who had had a major osteoporotic fracture prior to their initial BMD test. Finally, univariate and multivariate hazard ratios (HRs) were calculated to identify predictors of post-BMD osteoporotic fracture, determined from Cox proportional-hazards models. The likelihood-ratio test was used to assess the incremental value of combining BMD and TBS measurements.38 The likelihood-ratio chi-square statistic from the Cox proportional-hazards model provides a global measure of model fit, and the difference between chi-square values is used to test the improvement in model fit. Finally, we performed receiver operator curve (ROC) analysis for the fracture-prediction models. ROC areas under the curve (AUC) were compared using the nonparametric method of DeLong and colleagues,39 which allows for efficient comparison of the highly correlated curves originating from a common population (AccuROC 2.5; Accumetric Corp, Montreal, Quebec, Canada).

Results

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

Demographics and baseline clinical, BMD, and TBS measurements

A total of 29,407 women were included in the analyses. Baseline data on the study population are given in Table 1. Mean age for the women was 65.4 years, ranging from 50 to over 95 years. Eight percent had diabetes, 7.6% COPD, 3.4% rheumatoid arthritis, and 2.3% a substance-abuse diagnosis. Three-thousand nine-hundred and eighty-six women (13.6%) had a history of major osteoporotic fracture diagnosed before BMD testing, including 826 (2.8%) clinical spine fractures and 361 (1.2%) hip fractures.

Table 1. Demographics and Baseline Characteristics of the Population (n = 29,407)
 Mean ± SD
  • Data are mean ± SD or percent.

  • a

    Clinical spine, hip, forearm, humerus.

  • b

    Bisphosphonates, raloxifene, sCT, systemic estrogen.

  • c

    Number of Johns Hopkins Adjusted Care Groups.

Age (years)65.4 ± 9.5
Height (cm)160.2 ± 6.4
Weight (kg)68.5 ± 13.8
Body mass index (BMI)26.7 ± 5.1
Lumbar spine T-score−1.19 ± 1.50
Lumbar spine Z-score−0.04 ± 1.41
Lumbar spine TBS1.24 ± 0.12
Femoral neck T-score−1.47 ± 0.94
Total-hip T-score−1.03 ± 1.16
Major osteoporotic fractures pre-BMDa3986 (13.6%)
Hip fractures pre-BMD361 (1.2%)
Clinical spine fractures pre-BMD826 (2.8%)
Osteoporotic fractures post-BMDa1668 (5.7%)
Hip fractures post-BMD293 (1.0%)
Clinical spine fractures post-BMD439 (1.5%)
Years of coverage after BMD4.7 ± 2.2
Osteoporosis therapy pre-BMD (in past year)b30.3%
Systemic corticosteroids pre-BMD (≥3 months in past year)4.1%
Rheumatoid arthritis3.4%
Diabetes8.0%
COPD7.6%
Substance abuse2.3%
ADG comorbidity indexc4.4 ± 2.6

The mean lumbar spine Z-score of −0.04 (SD 1.41) was close to zero, demonstrating the lack of detectable referral bias among the women tested. The mean lumbar spine T-score was −1.19 (SD 1.50), with 7157 (24.3% of the population) meeting the WHO criteria for osteoporotic spine BMD; for the femoral neck and total hip, 13.0% and 9.6% met the WHO criteria for osteoporosis. Considering all three BMD measurements together, 31.2% of the women had at least one measurement within the osteoporotic range. Lumbar spine TBS was weakly correlated with BMD of the lumbar spine (r = 0.33), femoral neck (r = 0.27), and total hip (r = 0.26), indicating that only 6.7% to 10.7% of the variance in BMD could be explained by lumbar spine TBS. There was a stronger correlation between spine BMD and hip BMD (r = 0.72).

Sixteen-hundred and sixty-eight women (5.67%) experienced a major osteoporotic fracture after BMD testing, including 439 (1.5%) with clinical spine fractures and 293 (1.0%) with hip fractures, during a mean 4.7 years (SD 2.2 years) of follow-up. Subjects with major osteoporotic fractures after initial BMD testing averaged 4.7 years older than those without, were slightly shorter and lighter, and had lower BMI values (Table 2). BMD values at the lumbar spine, the femoral neck, and the total hip were lower in those with osteoporotic fractures, as was mean lumbar spine TBS (all p < 0.001; Table 2). Similarly, subjects with spine fractures (n = 439) averaged more than 6 years older than those without fractures (p < 0.001), were slightly shorter (p < 0.001), lighter (p < 0.001), and had slightly lower BMI values. As with all major osteoporotic fractures, T-scores at the lumbar spine, femoral neck, and total hip were lower among those with spine fractures (all p < 0.001), as was the lumbar spine TBS (p < 0.001). Those with hip fractures were almost a full decade older than their counterparts without hip fracture, were more than 7 kg lighter, were slightly shorter, had lower BMI values, and lower BMD scores and TBS values (all p < 0.001).

Table 2. Comparisons of Women With and Without Incident Fractures
 No major osteoporotic fracture (mean ± SD)Any major osteoporotic fracturea (mean ± SD)p Value
Age (years)65.1 ± 9.469.8 ± 9.7<0.001
Height160.2 ± 6.4159.5 ± 6.7<0.001
Weight68.6 ± 13.866.1 ± 13.3<0.001
BMI26.7 ± 5.126.0 ± 4.9<0.001
L1–L4T-score−1.15 ± 1.50−1.89 ± 1.39<0.001
Femoral neck T-score−1.44 ± 0.93−2.02 ± 0.89<0.001
Total-hip T-score−0.99 ± 1.15−1.72 ± 1.13<0.001
L1–L4 TBS1.24 ± 0.121.19 ± 0.12<0.001
 No spine fractureSpine fracture 
Age (years)65.3 ± 9.471.8 ± 9.1<0.001
Height160.2 ± 6.4159.0 ± 6.7<0.001
Weight68.5 ± 13.866.0 ± 13.4<0.001
BMI26.7 ± 5.126.1 ± 5.00.014
L1–L4T-score−1.18 ± 1.50−2.18 ± 1.38<0.001
Femoral neck T-score−1.46 ± 0.93−2.14 ± 0.88<0.001
Total-hip T-score−1.01 ± 1.15−1.87 ± 1.14<0.001
L1–L4 TBS1.24 ± 0.121.17 ± 0.12<0.001
 No hip fractureHip fracture 
  • a

    Clinical spine hip, forearm, humerus.

Age (years)65.3 ± 9.475.0 ± 8.8<0.001
Height160.2 ± 6.4158.6 ± 7.2<0.001
Weight68.6 ± 13.861.0 ± 12.2<0.001
BMI26.7 ± 5.124.2 ± 4.3<0.001
L1–L4T-score−1.18 ± 1.50−1.99 ± 1.44<0.001
Femoral neck T-score−1.46 ± 0.93−2.48 ± 0.78<0.001
Total-hip T-score−1.01 ± 1.15−2.34 ± 1.00<0.001
L1–L4 TBS1.24 ± 0.121.16 ± 0.13<0.001

Women were stratified according to lumbar spine TBS (tertiles) and lumbar spine BMD category (ie, normal, osteopenia, or osteoporosis). A consistent trend of lower fracture rates with higher TBS scores, overall and for specific levels of BMD, was seen for the lumbar spine, total hip, femoral neck, and minimum T-score (all p < 0.05; Table 3). Similar results were seen when women with major fractures prior to BMD measurement were excluded (data not shown). For all women combined, the OR for fracture for the lowest TBS tertile compared with the middle tertile was 1.57 (95% CI 1.46–1.68) and for the lowest TBS tertile compared with the highest tertile was 2.88 (95% CI 2.74–3.01; Table 3). When stratified by WHO BMD category, there was some attenuation in the ORs, but for most, the 95% CI still excluded unity. Within the osteopenia subgroup (whether defined from minimum T-score or individual BMD sites), the OR for fracture for the lowest TBS tertile compared with the highest tertile was consistently greater than 2. Major osteoporotic fracture rates per 1000 person-years were calculated and showed independent effects of TBS and BMD (Fig. 1).

Table 3. Proportion of Women Sustaining Incident Major Osteoporotic Fractures During Follow-up According to TBS Tertile and WHO BMD Classification
BMD siteBMD categoryFracture risk by lumbar spine TBS tertileTBS odds ratios (95% CI)
LowestMiddleHighestp for trendaLowest versus middleLowest versus highest
  • a

    p Value is from the Cochran-Armitage test.

  • b

    Minimum BMD T-score of total hip, femoral neck, and AP spine.

SpineNormal4.0%3.3%2.2%<0.0011.21 (0.93–1.49)1.86 (1.59–2.14)
 Osteopenia7.3%5.7%3.8%<0.0011.31 (1.13–1.49)2.01 (1.80–2.22)
 Osteoporosis12.1%8.1%5.7%<0.0011.56 (1.39–1.73)2.30 (2.00–2.60)
 All8.4%5.5%3.1%<0.0011.57 (1.46–1.68)2.88 (2.74–3.01)
Total hipNormal4.5%3.3%1.9%<0.0011.40 (1.17–1.63)2.38 (2.14–2.63)
 Osteopenia8.7%6.3%4.5%<0.0011.43 (1.27–1.58)2.04 (1.85–2.23)
 Osteoporosis16.3%12.8%10.6%<0.0011.33 (1.10–1.57)1.65 (1.26–2.04)
 All8.4%5.5%3.1%<0.0011.57 (1.46–1.68)2.88 (2.74–3.01)
Femoral neckNormal3.5%3.0%1.7%<0.0011.18 (0.85–1.51)2.05 (1.71–2.39)
 Osteopenia7.7%5.1%3.5%<0.0011.55 (1.40–1.70)2.33 (2.15–2.50)
 Osteoporosis15.2%12.9%10.0%<0.0011.21 (1.01–1.42)1.61 (1.30–1.93)
 All8.4%5.5%3.1%<0.0011.57 (1.46–1.68)2.88 (2.74–3.01)
MinimumbNormal2.6%2.3%1.6%<0.0301.13 (0.64–1.63)1.62 (1.14–2.09)
 Osteopenia6.3%4.5%3.0%<0.0011.43 (1.24–1.61)2.14 (1.94–2.34)
 Osteoporosis11.9%8.8%6.2%<0.0011.40 (1.24–1.55)2.02 (1.79–2.25)
 All8.4%5.5%3.1%<0.0011.57 (1.46–1.68)2.88 (2.74–3.01)
thumbnail image

Figure 1. Fracture incidence rates in the highest, middle, and lowest TBS tertiles according to lumbar spine BMD category.

Download figure to PowerPoint

For each SD decline in total-hip BMD, there was a 75% increase in the age-adjusted hazard of spine fracture versus 72% with the lumbar spine BMD and 45% with lumbar spine TBS (Table 4). The combined model for total-hip BMD and lumbar spine TBS showed a significant improvement in fracture prediction over models based on either BMD or TBS alone (p < 0.0001). Spine fracture risk increased by 90% for each SD decline in the combination. The same pattern was seen for models combining lumbar spine BMD and lumbar spine TBS, with the combined model again superior to either BMD or TBS alone (p < 0.0001). Compared with total-hip BMD alone, the addition of lumbar spine TBS significantly improved prediction for hip fractures (p = 0.0002) and for any major osteoporotic fracture (p < 0.0001), as did the combination of lumbar spine BMD and lumbar spine TBS (p < 0.0001). Results were only slightly attenuated after further adjustments were made for ADG comorbidity score, rheumatoid arthritis, COPD, diabetes, substance abuse, BMI, prior osteoporotic fracture, systemic corticosteroid use in the last year, and osteoporosis treatment in the last year. Similar results were found from ROC analysis (Table 5).

Table 4. Univariate and Multivariate Hazard Ratios (HRs) for Fracture Prediction
 Clinical spine fractureHip fractureAny major osteoporotic fracture
HR/SD (95% CI)pHR/SD (95% CI)pHR/SD (95% CI)p
  • Note: All models are age-adjusted. p Value is for improvement in model fit using the likelihood-ratio test when TBS is added to BMD.

  • a

    Adjusted for age and the following clinical risk factors (CRF): ADG comorbidity score, rheumatoid arthritis, COPD, diabetes, substance abuse, BMI, prior osteoporotic fracture, systemic corticosteroid use in the last year, and osteoporosis treatment in the last year.

UnivariateBMD total hip1.75 (1.58–1.96)N/A2.55 (2.22–2.93)N/A1.67 (1.58–1.76)N/A
 BMD femoral neck1.76 (1.57–1.98) 2.60 (2.23–3.03) 1.68 (1.58–1.78) 
 BMD spine1.72 (1.55–1.91) 1.31 (1.16–1.48) 1.47 (1.39–1.55) 
 TBS spine1.45 (1.32–1.58) 1.46 (1.30–1.63) 1.35 (1.29–1.42) 
TBS with total-hip BMDBMD total hip1.64 (1.47–1.83)<0.00012.40 (2.09–2.76)0.00021.59 (1.50–1.68)<0.0001
 TBS spine1.34 (1.22–1.48) 1.27 (1.12–1.43) 1.26 (1.20–1.33) 
 Combination1.90 (1.71–2.11) 2.63 (2.29–3.01) 1.76 (1.67–1.86) 
CRF-adjusted combinationaBMD total hip1.60 (1.41–1.81)0.00042.06 (1.76–2.42)0.00081.54 (1.45–1.64)<0.0001
 TBS spine1.20 (1.09–1.33) 1.25 (1.10–1.41) 1.18 (1.12–1.25) 
TBS with femoral neck BMDBMD femoral neck1.65 (1.47–1.86)<0.00012.45 (2.09–2.86)<0.00011.60 (1.51–1.70)<0.0001
 TBS spine1.35 (1.23–1.48) 1.29 (1.15–1.46) 1.26 (1.20–1.33) 
 Combination1.92 (1.71–2.14) 2.70 (2.32–3.14) 1.78 (1.68–1.89) 
CRF-adjusted combinationaBMD femoral neck1.55 (1.36–1.75)0.00012.03 (1.72–2.39)0.00011.51 (1.42–1.61)<0.0001
 TBS spine1.22 (1.10–1.34) 1.28 (1.13–1.46) 1.20 (1.14–1.26) 
TBS with spine BMDBMD spine1.58 (1.42–1.77)<0.00011.19 (1.05–1.35)<0.00011.37 (1.30–1.45)<0.0001
 TBS spine1.29 (1.17–1.42) 1.40 (1.24–1.57) 1.25 (1.19–1.31) 
 Combination1.81 (1.63–2.01) 1.53 (1.36–1.72) 1.55 (1.48–1.64) 
CRF-adjusted combinationaBMD spine1.62 (1.44–1.83)0.01300.98 (0.86–1.13)<0.00011.36 (1.26–1.47)<0.0001
 TBS spine1.14 (1.03–1.26) 1.47 (1.30–1.67) 1.17 (1.09–1.25) 
Table 5. AUCs (95% CI) for Prediction of Fracture
 Clinical spine fractureHip fractureAny major osteoporotic fracture
  • a

    Significant improvement in AUC when TBS added to corresponding model for BMD alone.

UnivariateBMD total hip0.71 (0.68–0.73)0.81 (0.79–0.83)0.68 (0.66–0.69)
 BMD femoral neck0.71 (0.68–0.73)0.80 (0.77–0.82)0.68 (0.66–0.69)
 BMD spine0.69 (0.67–0.72)0.65 (0.62–0.69)0.64 (0.63–0.66)
 TBS spine0.66 (0.64–0.69)0.68 (0.65–0.71)0.63 (0.61–0.64)
CombinedBMD total hip0.73 (0.71–0.75)a0.82 (0.79–0.84)a0.69 (0.68–0.71)a
 TBS spine   
CombinedBMD femoral neck0.73 (0.71–0.75)a0.81 (0.79–0.83)a0.69 (0.68–0.71)a
 TBS spine   
CombinedBMD spine0.71 (0.69–0.74)a0.69 (0.66–0.72)a0.66 (0.65–0.68)a
 TBS spine   

Discussion

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

We found significantly lower lumbar spine TBS and BMD scores in women with major osteoporotic fractures, spine fractures, and hip fractures. The correlation between spine BMD and spine TBS was modest (r = 0.32) and less than that between spine and hip BMD (r = 0.72). Spine TBS predicted fractures almost as well as lumbar spine BMD, and the combination was superior to either measurement alone (p < 0.001). Incremental improvement in the performance of the combination of BMD and TBS remained significant even after adjustment for multiple clinical risk factors.

Although BMD testing by DXA remains the “gold standard” for the diagnosis of osteoporosis and the prediction of subsequent fracture risk, the majority of those who meet the WHO criteria for osteoporosis do not develop an osteoporotic fracture, whereas the majority of osteoporotic fractures occur in individuals who have BMD values in the nonosteoporotic range.13, 14 An attractive potential use for TBS is when BMD alone is not sufficient for risk stratification. For example, this may be the case when BMD is in the osteopenic range. Given the need to balance the costs of treatment with the direct and indirect costs of osteoporotic fractures,4, 7, 11 it is important to optimize the prediction of fracture risk in order to identify those most likely to benefit from treatment. Trabecular bone microarchitecture is an important contributor to bone strength independent of BMD, but it has not been used to assess fracture risk in clinical practice. Imaging techniques other than DXA have been proposed as potential candidates for the clinical evaluation of trabecular bone microarchitecture.18, 20 Meanwhile, DXA technology has developed to the point where newer DXA systems provide considerably more accurate and reproducible measurements of BMD.40 High-quality DXA scans now can be used for other purposes, such as confirming and characterizing spine fractures from vertebral fracture assessment (VFA)41 and evaluating macroscopic geometry.40–42 There is also some investigation regarding extracting structure parameters from DXA images of the calcaneus.43, 44

Unlike the three previous case-control studies,23–25 this study looked at prediction of incident fracture events. Our results do not support replacing BMD in favor of TBS. Rather, there may be a role for using these two measurements in combination, especially in those at intermediate risk, such as individuals with BMD values in the osteopenic range. Among osteopenic patients with TBS in the lowest tertile, ORs for fracture were significantly higher than among those in the middle tertile (ORs ranged from 1.31 to 1.55) or highest tertile (ORs all exceeded 2). In principle, a protocol could be established to perform TBS only on scans with BMD values or risk scores within a specified range. This has the additional advantage over some other techniques of being potentially applicable to almost any bone site, including spine, femoral neck, hip, and forearm. Alternatively, if the information is easily extracted from DXA and is incremental to BMD, then it might be appropriate to use it in all cases. Such an approach could help to define the fracture risk profile by taking into account both the density and the microstructure of the bone. One advantage of TBS over other proposed methods for assessment of bone microarchitecture is that the measurement can be extracted from previously obtained DXA images, unlike pQCT and MRI, which require a patient to return for further costly and time-consuming measurements. Other methods have been tested using textural analysis from X-ray or DXA of the calcaneus.15, 19, 43, 44 For example, Vokes and colleagues have developed radiographic texture analysis (RTA) for calcaneal images as assessed by peripheral DXA.44 They also have found in 900 subjects that RTA has a potential to enhance identification of patients at increased risk for osteoporotic fractures. In another study, Vokes and colleagues suggested that RTA of densitometer-generated calcaneus images provides an estimate of bone fragility independent of and complementary to BMD measurement and age.44 However, in contrast to TBS, RTA requires an additional measurement at another skeletal site that is not part of clinical routine and current recommendations.

A major limitation of our study was in assessing clinical spine fractures from administrative data. Vertebral fracture assessment (VFA) and bone turnover markers (BTMs) were not available for this large cohort. Whether VFA or BTMs would reduce the added value of TBS or whether there would be additional additive value in incorporating these parameters cannot be determined. The relatively small incremental change in AUCs from using TBS in combination with BMD is not unexpected given the already existing high correlation between BMD and fracture, and there has been criticism of overreliance on ROC analysis for assessing additional risk factors.45, 46 When we examined risk stratification within TBS tertiles stratified by BMD category, the incremental value of using TBS was clearly evident.

Confirmatory studies are required before developing recommendations for the clinical use of TBS. This includes optimizing the BMD and TBS thresholds for clinical decision making, and validating that a combined approach can achieve a favorable balance in enhancing sensitivity and specificity is an important next step. Furthermore, the value of TBS in predicting fracture risk also needs to be evaluated in premenopausal women, men, and patients on steroids or with other risk factors for osteoporosis. Nevertheless, this study shows that TBS holds promise as a low-cost and easily applied adjunct to BMD testing in the assessment of fracture risk.

Disclosures

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

TBS iNsight Software is a product of Med-Imaps. Didier Hans is co-owner of the TBS patent and has corresponding ownership shares. All the 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
  10. Supporting Information

We are indebted to Manitoba Health for the provision of data (HIPC No. 2008/2009-33). The results and conclusions are those of the authors, and no official endorsement by Manitoba Health is intended or should be inferred. This article has been reviewed and approved by the members of the Manitoba Bone Density Program Committee.

Authors' roles: WD Leslie is the principal Investigator of the Manitoba study. He participated in the set up of the study, provided the clinical data, performed all the statistical analysis, and contributed to the writing of the manuscript. MA Krieg participated in the set up of the study, interpreted the clinical relevance of the outcomes, and contributed to the writing of the manuscript. AL Goertzen provided all the blinded DXA scans and was in charge of the QC of the study. He also participated in the set up of the study, helped in the statistical analysis, and contributed to the writing of the manuscript. D Hans was the initiator of the study idea. He participated in the set up of the study, performed the Blind TBS calculation, and wrote the manuscript. D Hans and MA Krieg were completely blind to the clinical data until the analysis were finalized by WD Leslie.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information
  • 1
    Czerwinski E, Badurski JE, Marcinowska-Suchowierska E, Osieleniac J. Current understanding of osteoporosis according to the position of the World Health Organization (WHO) and International Osteoporosis Foundation. Ortop Traumatol Rehabil. 2007; 9: 33756.
  • 2
    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. 2009; 19(4): 38597.
  • 3
    Meunier PJ, Delmas PD, Eastell R, McClung MR, Papapoulos S, Rizzoli R, et al. Diagnosis and management of osteoporosis in postmenopausal women: clinical guidelines. International Committee for Osteoporosis Clinical Guidelines. Clin Ther. 1999; 21(6): 102544.
  • 4
    Johnnell O, Kanis JA. An estimate of the worldwide prevalence and disability associated with osteoporotic fractures. Osteoporos Int. 2006; 17: 172633.
  • 5
    Browner WS, Pressman AR, Nevitt MC, et al. Mortality following fractures in older women. The study of osteoporotic fractures. Arch Intern Med. 1996; 156: 15215.
  • 6
    Hannan EL, Magaziner J, Wang JJ, et al. Mortality and locomotion 6 months after hospitalization for hip fracture: risk factors and risk-adjusted hospital outcomes. JAMA. 2001; 285: 273642.
  • 7
    Looker AC, Orwoll ES, Johnston CCJr, Lindsay RL, Wahner HW, Dunn WL, et al. Prevalence of low femoral bone density in older U.S. adults from NHANES III. J Bone Miner Res. 2009; 12: 17618.
  • 8
    Burge R, Dawson-Hughes B, Solomon DH, Wong JB, King A, Tosteson A. Incidence and economic burden of osteoporosis-related fractures in the United States, 2005–2025. J Bone Miner Res. 2007; 22: 46575.
  • 9
    Davies KM, Stegman MR, Heaney RP, Recker RR. Prevalence and severity of vertebral fracture: The Saunders County Bone Quality Study. Osteoporos Int. 2005; 6(2): 1605.
  • 10
    Leslie WD, Lix LM, Langsetmo L, Berger C, Goltzman D, Hanley DA, Adachi JD, Johansson H, Oden A, McCloskey E, Kanis JA. Construction of a FRAX® model for the assessment of fracture probability in Canada and implications for treatment. Osteoporos Int. 2011 Mar; 22(3): 81727.
  • 11
    Johnell O, Kanis JA. An estimate of the worldwide prevalence and disability associated with osteoporotic fractures. Osteoporos Int. 2006; 17(12): 172633.
  • 12
    Assessment of fracture risk and its application to screening for postmenopausal osteoporosis. Report of a WHO Study Group. World Health Organ Tech Rep Ser. 1994; 843: 1129.
  • 13
    Johnell O, Kanis JA, Oden E, Johansson H, De Laet C, Delmas P, et al. Predictive value of BMD for hip and other fractures. J Bone Miner Res. 2005; 20(7): 118594.
  • 14
    Hordon LD, Raisi M, Paxton S, Beneton MM, Kanis JA, Aaron JE. Trabecular architecture in women and men of similar bone mass with and without vertebral fracture: Part I. 2-D histology. Bone. 2000; 27(2): 2716.
  • 15
    Link TM, Majumdar S. Current diagnostic techniques in the evaluation of bone architecture. Curr Osteoporos Rep. 2004; 2(2): 4752.
  • 16
    Rubin CD. Emerging concepts in osteoporosis and bone strength. Curr Med Res Opin. 2005; 21(7): 104956.
  • 17
    Boutroy S, Bouxsein ML, Munoz F, Delmas PD. In vivo assessment of trabecular bone microarchitecture by high-resolution peripheral quantitative computed tomography. J Clin Endocrinol Metab. 2005; 90(12): 650815.
  • 18
    Valentinitsch A, Patsch J, Mueller D, Kainberger F, Langs G. Texture analysis in quantitative osteoporosis assessment: Characterizing microarchitecture. In Biomedical Imaging: From Nano to Macro, 2010 IEEE International Symposium on, pages 1361{1364. IEEE, 2010.
  • 19
    Griffith JF, Engelke K, Genant HK. Looking Beyond bone mineral density: imaging assessment of bone quality. Ann N Y Acad Sci. 2010 Mar; 1192: 4556.
  • 20
    Pothuaud L, Carceller P, Hans D. Correlations between grey-level variations in 2D projection images (TBS) and 3D microarchitecture: applications in the study of human trabecular bone microarchitecture. Bone. 2008; 42(4): 77587.
  • 21
    Hans D, Barthe N, Boutroy S, Winzenrieth R, Pothuaud L, Krieg M-A. Correlations between TBS, measured using antero-posterior DXA acquisition, and 3D parameters of bone micro-architecture: an experimental study on human cadavre vertebrae. J Clin Densitom. 2011 Jul-Sep; 14(3): 30212.
  • 22
    Pothuaud L, Barthe N, Krieg M-A, Mehsen N, Carceller P, Hans D. Evaluation of the potential use of trabecular bone score to complement bone mineral density in the diagnosis of osteoporosis: a preliminary spine BMD-matched, case-control study. J Clin Densitom. 2009 Apr–Jun; 12(2): 1706.
  • 23
    Rabier B, Héraud A, Grand-Lenoir C, Winzenrieth R, Hans D. A multicentre, retrospective case-control study assessing the role of trabecular bone score (TBS) in menopausal Caucasian women with low areal bone mineral density (BMDa): Analysing the odds of vertebral fracture. Bone. 2010 Jan; 46(1): 17681.
  • 24
    Winzenrieth R, Dufour R, Pothuaud L, Hans D. A retrospective case-control study assessing the role of trabecular bone score in postmenopausal Caucasian women with osteopenia: analyzing the odds of vertebral fracture. Calcif Tissue Int. 2010 Feb; 86(2): 1049.
  • 25
    Leslie WD, Pahlavan PS, Tsang JF, Lix LM. Prediction of hip and other osteoporotic fractures from hip geometry in a large clinical cohort. Osteoporos Int. 2009; 10: 176774.
  • 26
    Leslie WD, Metge C. Establishing a regional bone density program: lessons from the Manitoba experience. J Clin Densitom. 2003; 6: 27582.
  • 27
    Leslie WD, Caetano PA, MacWilliam LR, et al. Construction and validation of a population-based bone densitometry database. J Clin Densitom. 2005; 8: 2530.
  • 28
    Roos NP, Shapiro E. Revisiting the Manitoba Centre for Health Policy and Evaluation and its population-based health information system. Med Care. 1999; 37: JS104.
  • 29
    Leslie WD, Tsang JF, Caetano PA, Lix LM. Effectiveness of bone density measurement for predicting osteoporotic fractures in clinical practice. J Clin Endocrinol Metab. 2007 January; 92(1): 7781.
  • 30
    Kanis JA, Johnell O, Oden A, Dawson A, De Laet C, Jonsson B. Ten year probabilities of osteoporotic fractures according to BMD and diagnostic thresholds. Osteoporos Int. 2001 December; 12(12): 98995.
  • 31
    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 April; 19(4): 38597.
  • 32
    Leslie WD, Tsang JF, Caetano PA, Lix LM. Effectiveness of bone density measurement for predicting osteoporotic fractures in clinical practice. J Clin Endocrinol Metab. 2007 Jan; 92(1): 7781.
  • 33
    Smith NS, Weiner JP. Applying population-based case mix adjustment in managed care: the Johns Hopkins Ambulatory Care Group system. Manag Care Q. 1994; 2(3): 2134.
  • 34
    Leslie WD, Derksen S, Prior HJ, Lix LM, Metge C, O'neil J. The interaction of ethnicity and chronic disease as risk factors for osteoporotic fractures: a comparison in Canadian Aboriginals and non-Aboriginals. Osteoporos Int. 2006 September; 17(9): 135868.
  • 35
    Kanis JA, McCloskey E, Johansson H, et al. A reference standard for the description of osteoporosis. Bone. 2008; 42: 46775.
  • 36
    Binkley N, Kiebzak GM, Lewiecki EM, et al. Recalculation of the NHANES database SD improves T-score agreement and reduces osteoporosis prevalence. J Bone Miner Res. 2005; 20: 195201.
  • 37
    Singer JD, Willet JB. Applied longitudinal data analysis: Modeling change and event occurrence. New York: Oxford University Press; 2003.
  • 38
    DeLong ER, DeLong DM, Clarke-Pearson DL., Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988; 44: 83745.
  • 39
    Janes H, Pepe MS, Gu W. Assessing the value of risk predictions by using risk stratification tables. Ann Intern Med. 2008; 18; 149(10): 75160.
  • 40
    Bonnick SL. Bone densitometry in clinical practice - Application and interpretation (second edition). Totawa, New Jersey: Humana Press Inc; 2004.
  • 41
    Ahmad O, Ramamurthi K, Wilson KE, Engelke K, Prince RL, Taylor RH. Volumetric DXA (VXA): A new method to extract 3D information from multiple in vivo DXA images. J Bone Miner Res. 2010 Dec; 25(12): 274451.
  • 42
    Kolta S, Quiligotti S, Ruyssen-Witrand A, Amido A, Mitton D, Bras AL, Skalli W, Roux C. In vivo 3D reconstruction of human vertebrae with the three-dimensional X-ray absorptiometry (3D-XA) method. Osteoporos Int. 2008 Feb; 19(2): 18592.
  • 43
    Vokes TJ, Giger ML, Chinander MR, Karrison TG, Favus MJ, Dixon LB. Radiographic texture analysis of densitometer-generated calcaneus images differentiates postmenopausal women with and without fractures. Osteoporos Int. 2006; 17: 1472148.
  • 44
    Vokes TJ, Lauderdale D, Ma S-L, Chinander MR, Childs K, Giger ML. Radiographic texture analysis of densitometric calcaneal images: Relationship to clinical characteristics and to bone fragility. J Bone Miner Res. 2010; 25: 5663.
  • 45
    Cook NR. Use and misuse of the receiver operating characteristic curve in risk prediction. Circulation. 2007; 115: 92835.
  • 46
    Pencina MJ, D'Agostino RB Sr, D'Agostino RB Jr, Vasan RS. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med. 2008; 27: 15772.

Supporting Information

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

Additional supporting information may be found in the online version of this article.

FilenameFormatSizeDescription
jbmr_499_sm_chinese_translation.pdf815KChinese Translation

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.