- Email: [email protected]
Spine Trabecular Bone Score Precision, a Comparison Between GE Lunar Standard and High-Resolution Densitometers
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health, vol. -, no. -, 1e7, 2015 Ó Copyright 2015 by The International Society for Clinical Densitometry 1094-6950/-:1e7/$36.00 http://dx.doi.org/10.1016/j.jocd.2014.11.003
Original Article
Spine Trabecular Bone Score Precision, a Comparison Between GE Lunar Standard and High-Resolution Densitometers Diane Krueger,* Jessie Libber, and Neil Binkley Osteoporosis Clinical Research Program, University of Wisconsin, Madison, WI, USA
Abstract Trabecular bone score (TBS) is related to microarchitecture and fracture risk independently of bone mineral density (BMD) and clinical risk factors. Widespread clinical TBS use requires documentation of reproducibility and ideally comparability across scanners. This study evaluated TBS reproducibility and explored differences between Lunar Prodigy and iDXA densitometers. Reproducibility was assessed from replicate scans in 210 men and women participating in various dual-energy X-ray absorptiometry (DXA) precision assessments. iDXA-to-Prodigy comparability was evaluated using 155 participants from 3 study groups. L1eL4 BMD and TBS precision was similar on iDXA and Prodigy (BMD coefficient of variation 5 1.9% and 1.5% and TBS coefficient of variation 5 1.4% and 1.6%, respectively). Precision did not differ between men and women; however, between-technologist differences ( p ! 0.05) of similar magnitude were observed for both BMD and TBS. Prodigy-to-Prodigy TBS values were highly correlated (R2 5 0.85 with bias of 0.010 TBS units). Agreement was less robust comparing Prodigy with iDXA instruments (TBS R2: 0.72e0.81 with biases of 0.012e0.034 TBS units). In conclusion, TBS precision is comparable to that of BMD and does not differ between men and women. Additionally, in these cohorts, slight TBS differences were observed between iDXA and Prodigy scans. These data suggest a potential difference between densitometer models perhaps due to higher iDXA image resolution. Key Words: Correlation; DXA; precision; spine bone mineral density; trabecular bone score.
elevated risk for osteoporosis-related fractures. Currently, bone mineral density (BMD), measured by dual-energy X-ray absorptiometry (DXA), is the clinical gold standard to diagnose osteoporosis and determine fracture risk in the absence of a prior fragility fracture (8). However, as approximately half of low-trauma fractures occur in those with a BMD T-score better than 2.5 (9,10), methods to optimally identify persons likely to fracture despite adequate bone mass by DXA is desirable. As macrogeometry and trabecular bone microarchitecture influence bone strength, and consequently fracture risk, a method to clinically evaluate microarchitecture is an attractive addition to BMD in fracture risk prediction. Trabecular bone score (TBS) is a novel gray-level texture measurement that uses 2-dimensional DXA images to differentiate between 3-dimensional microarchitectures that exhibit the same bone density but different structural characteristics (11,12). A TBS score is determined by measuring the rate of local gray level variation from the DXA image.
Introduction Osteoporosis, a disease of low bone mass and associated microarchitectural deterioration of bone tissue, results in bone fragility and increased fracture risk. Given the aging populations worldwide, this is a major health issue (1). It is estimated that approximately 9 million new osteoporotic fractures occur annually worldwide (2), the ramifications of which are substantial morbidity, mortality, and expenditure of health care resources (3,4). It is projected that the number of osteoporosis-related fractures will double over the next 40e50 years (5). As effective therapies to reduce fractures exist (6,7), there is great interest to improve identification of individuals at Received 09/26/14; Revised 11/14/14; Accepted 11/18/14. *Address correspondence to: Diane Krueger, BS, CBDT, Osteoporosis Clinical Research Program, University of Wisconsin, 2870 University Avenue, Suite 100, Madison, WI 53705. E-mail: dckruege@ wisc.edu
1
2
Krueger, Libber, and Binkley
Consequently, it is not generating a physical measurement but a texture index representative of trabecular bone independent of BMD. This methodology is validated by the high correlation TBS demonstrates with micro-computed tomography (13). Additionally, TBS has documented fracture risk prediction utility both independently and in conjunction with BMD for vertebral and nonvertebral low-trauma fracture (14e20). The literature regarding TBS precision is currently limited (14,18,21), and between densitometer comparisons of TBS values and precision is lacking, particularly in instrument models with differing image resolution. It is plausible, given the use of gray-level assessment in TBS, that higherresolution DXA images may generate different TBS values or have different precision error. This work evaluates TBS measurement and precision comparisons between GE Healthcare Prodigy (standard resolution) and iDXA (high resolution) instruments.
Materials and Methods Subjects Existing data were gathered from 4 distinct cohorts to evaluate precision and TBS measurements from each type of instrument. In the following text are the descriptions of each group, 3 of which have been published; brief descriptions of the original study design, or purpose of the initial data collection, are described here to facilitate clarity. Precision Comparison Study: Designed to compare BMD precision between iDXA and Prodigy, enrolled 30 postmenopausal women having precision scans on each instrument; 1 technologist acquired all scans (22). BMD Instrument Comparison Study: Designed to compare BMD between iDXA and Prodigy, 95 men and women were included from this trial. Each subject was scanned on both instruments, and scans were acquired by multiple technologists (22). Male-Female Precision Comparison Study: Designed to evaluate if precision differs by sex, 180 men and women (90 in each group) were enrolled. Precision scans
acquired on an iDXA; 3 technologists conducted 1 precision assessment for each sex (23). Cross-Calibration Exercise: Data acquired to crosscalibrate 3 clinical instruments, 30 volunteers were scanned on 2 Prodigy and 1 iDXA instruments; 3 technologists acquired scans.
Precision Evaluations of TBS and Spine BMD Precision data from the Precision Comparison and MaleFemale Precision Comparison Studies were evaluated to assess short-term TBS and lumbar spine BMD precision; demographic data for these studies are listed in Table 1. Data from the Precision Comparison Study were used to compare precision values between a Prodigy and an iDXA in 30 women aged 65 years. The male-female precision cohort was used to assess precision variation between DXA technologists and sexes using an iDXA. In this study, 3 technologists performed 2 precision assessments, 1 in 30 men and 1 with 30 women. Each technologist scanned 60 different subjects; consequently, the total sample consisted of 180 subjects.
Interinstrument Comparison of TBS and Spine BMD Data to assess interinstrument agreement of TBS and lumbar spine BMD were evaluated separately in the following 3 cohorts: Precision Comparison Study, BMD Instrument Comparison Study, and Cross-Calibration Exercise. The precision comparison group is described previously. Ninety-five adult men (n 5 37) and women (n 5 58) participated in the BMD instrument comparison study. Finally, 30 adult men (n 5 5) and women (n 5 25) participated in the crosscalibration exercise; all volunteers in this study were scanned on 2 Prodigy instruments and 1 iDXA. Demographic data are listed in Table 1. A strength of this study, but a challenge for data presentation, is that 6 different densitometers (3 iDXA and 3 Prodigy instruments) were used for study of the various cohorts. Table 2 defines which instruments were used in the respective studies. Due to the known variability of individual DXA
Table 1 Demographics N
BMD Instrument Comparison Study Precision Comparison Study Cross-Calibration Exercise Male-Female Precision Comparison Study (male) Male-Female Precision Comparison Study (female)
95 37/58 58.2 16.5 30 0/30 69.5 4.9 30 5/25 57.4 9.6 90 90/0 73.9 6.2 90
M/F
Age, mean SD
Study
0/90 75.8 7.3
BMI, mean SD 27.9 25.9 26.0 26.8
5.7 4.5 4.7 3.4
26.6 5.3
LS BMD, mean SD 1.224 1.111 1.130 1.137
0.214 0.133 0.158 0.257
1.131 0.189
LS T-score, mean SD 0.2 0.7 0.5 0.8
1.7 1.1 1.3 2.1
0.4 1.6
TBS, mean SD 1.390 1.349 1.336 1.340
0.121 0.102 0.112 0.096
1.321 0.088
Abbr: BMD, bone mineral density; BMI, body mass index; F, female; LS, lumbar spine; M, male; SD, standard deviation; TBS, trabecular bone score. Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
TBS Precision and Instrument Correlation
3
Table 2 Study Sample and Acquisition Instrument Study
Evaluation
Participating technologists
Instruments used
Precision Comparison Study
Precision and comparison
1
BMD Instrument Comparison Study
Comparison
4
Male-Female Precision Comparison Study Cross-Calibration Exercise
Precision Comparison
3 3
Research iDXA 1 Research Prodigy 1 Research iDXA 2 Research Prodigy 1 Research iDXA 2 Clinical Prodigy 2 Clinical Prodigy 3 Clinical iDXA 3
Abbr: BMD, bone mineral density.
instruments (24,25), and the clinical need to evaluate a given patient over time, all data were evaluated using individual instrument comparison rather than pooling all Prodigy and all iDXA data.
F-test or factorial analysis of variance (Excel; Microsoft, Redmond, OR).
Precision and Cross-calibration Assessments
Subjects
All precision assessments were conducted per the International Society for Clinical Densitometry (ISCD) official positions; 30 individuals were scanned twice for each evaluation (26). Participants stood from the DXA table between scans, were immediately repositioned, and then rescanned. The first scan performed was analyzed per manufacturer recommendations, with manual adjustment of autoanalysis as needed. The second scans were analyzed using the software copy feature. The cross-calibration volunteers were scanned on 3 instruments (2 Prodigy and 1 iDXA, noted in Table 2). As these instruments were located at various locations within a city, all scans were performed within a 2-week period by 3 technologists.
Demographic ranges of all cohorts combined included an age of 22e92 years, body mass index from 16.4e47.7 kg/ m2, lumbar spine BMD of 0.755e1.945 g/cm2 with associated T-scores of 3.9 to þ6.0 and TBS measurement of 0.937e1.761 TBS units. Specific demographic data are reported by study in Table 1; no between-group comparisons were performed.
Bone Density Scan Acquisition and TBS Measurement As noted previously and in Table 2, 6 densitometers were used in this sampling. For purposes of data presentation here, each model was labeled consecutively on the basis of the instrument installation date. Lumbar spine scans were acquired and analyzed per manufacturer recommendations by ISCDcertified technologists. All scans were reviewed for quality by an ISCD-certified clinician. TBS values were generated by a single ISCD-certified technologist on the lumbar spine BMD files using TBS research software, version 2.1.0 (Medimaps group, Plan-lesOuates, Switzerland). TBS cross-calibration was not performed on any instrument.
Statistical Analyses Prodigy and iDXA TBS values were compared using linear regression and Bland-Altman analysis (Analyse-it, Leeds, United Kingdom). Precision was determined using ISCD Advanced Precision Calculator and compared by
Results
Precision Short-term precision, expressed as root mean square of the standard deviation (RMS-SD) and percent coefficient of variation (%CV), was similar between BMD and TBS and the iDXA and Prodigy instruments. The precision comparison study demonstrated L1eL4 BMD %CV of 1.9% and 1.5% on iDXA and Prodigy, respectively. Similarly, respective TBS %CVs of 1.4% and 1.6% were observed. No betweeninstrument precision differences were demonstrated for either BMD or TBS ( p O 0.10); complete data and values expressed as least significant change calculated at the 95% confidence level are listed in Table 3. The male-female precision study was performed on 1 iDXA and evaluated intertechnologist precision variation as well as sex difference. The %CV for L1eL4 BMD ranged from 1.0% to 1.6% across all technologists and both sexes; TBS %CV values were numerically higher, ranging from 1.5% to 2.5%. As expected, intertechnologist precision differences were observed for both BMD and TBS ( p ! 0.05; Table 3). However, as has previously been reported for BMD in this group (22), there was no difference in TBS reproducibility between sexes. Specific data are presented in Fig. 1 and Table 3.
Between-Instrument Comparison BMD was highly correlated between instruments in all comparison groups; R2 ranged from 0.97 to 0.98 with bias
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
4
Krueger, Libber, and Binkley Table 3 Precision BMD
Prodigya iDXAa iDXA maleb iDXA femaleb iDXA tech 1 male iDXA tech 1 female iDXA tech 2 male iDXA tech 2 female iDXA tech 3 male iDXA tech 3 female
TBS
RMS-SD (g/cm2)
CV (%)
LSC (g/cm2)
RMS-SD (g/cm2)
CV (%)
LSC (g/cm2)
0.016 0.020 0.016 0.016 0.012c 0.014c,d 0.020e,f,g 0.020f 0.014 0.014c
1.51 1.90 1.24 1.35 0.96 1.19 1.48 1.59 1.22 1.23
0.044 0.056 0.044 0.044 0.034 0.037 0.056 0.054 0.040 0.039
0.016 0.018 0.027 0.024 0.025 0.022c 0.033f 0.030e,f 0.020 0.019c
1.57 1.36 2.01 1.82 1.90 1.70 2.51 2.24 1.50 1.45
0.060 0.049 0.074 0.067 0.071 0.062 0.091 0.082 0.055 0.053
Note: LSC calculated at 95% confidence of RMS-SD. Abbr: BMD, bone mineral density; CV, coefficient of variation; LSC, least significant change; RMS-SD, root mean square standard deviation; TBS, trabecular bone score; tech, technologist. a Same female subjects used for each assessment. b Subjects from each assessment were pooled by sex (n 5 90). c Different from tech 2 male. d Different from tech 2 female. e Different from tech 3 female. f Different from tech 1 male. g Different from tech 3 male.
between 0.007 and þ0.007 g/cm2 (Figs. 2 and 3). Similarly, TBS was well correlated between Prodigy instruments (R2 5 0.85, with a bias of 0.010 TBS units [Fig. 2A and B]). No iDXA-to-iDXA data were available to compare. When evaluating TBS iDXA-to-Prodigy performance, the correlation was less robust, demonstrating an R2 range from 0.72 to 0.81 with biases from 0.008 to 0.035 (Fig. 3AeD).
Discussion In this study, we evaluated whether TBS performed similarly on 2 GE Healthcare densitometer models, a standardresolution (Prodigy) and a high-resolution (iDXA) fan-beam densitometer. As TBS uses gray-scale differentiation to calculate values, it is plausible that results and reproducibility
Fig. 1. Similar Precision for BMD and TBS. Precision (RMS-SD) was !3% and did not differ ( p O 0.10) between Prodigy and iDXA. Intertechnologist differences were observed for both BMD and TBS ( p ! 0.05) but not between sexes. BMD, bone mineral density; F, female; M, male; RMS-SD, root mean square-standard deviation; TBS, trabecular bone score. Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
TBS Precision and Instrument Correlation
5
Fig. 2. (AeB) BMD and TBS comparison: Prodigy vs Prodigy. Both BMD and TBS were highly correlated when comparing Prodigy to Prodigy demonstrating respective R2 5 0.98 and 0.85 with biases of 0.007 and 0.010 TBS units. The inset depicts mean (standard deviation) BMD and TBS values, which did not differ between instruments. BMD, bone mineral density; TBS, trabecular bone score. could differ when using technologies of differing image resolution. Consequently, we assessed if TBS precision differed between instruments and also how well correlated TBS measurements were between GE Lunar Prodigy and Lunar iDXA densitometers. These data demonstrate similar precision with a moderate correlation between these Prodigy and iDXA densitometers. These data demonstrate that TBS is precise and comparable to previously reported values (%CV: 1.5%e2.1%) (14,16)
but did demonstrate a slightly larger TBS precision range of 1.2 up to 2.7% CV. However, this could be explained by the fact that the subjects in these precision assessments with the highest %CV were aged at least 65 years, whereas those in the previously reported Manitoba and French cohorts were as young as 45 years (14,16). As a result, it is probable that the subjects reported here have a higher prevalence of spinal degenerative disease and also other age-associated challenges in positioning and analysis for spine BMD
Fig. 3. (AeD) BMD and TBS comparisons: Prodigy vs iDXA. In the sample of 95, BMD was highly correlated (R2 5 0.98) with a bias of þ0.007 g/cm2 (A); TBS was well correlated (R2 5 0.81) with a bias of 0.034 TBS units (B). Similarly, in a sample of 30 from the cross-calibration group, BMD was well correlated (R2 5 0.97) with a bias of 0.007 g/cm2 (C). TBS was less well correlated (R2 5 0.73) with a bias of þ0.012 TBS units (D). The bar graph insets depict mean (standard deviation) BMD and TBS values; mean TBS values were lower ( p ! 0.05) on Prodigy instruments. Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
6 measurement that could provide cause to expect poorer precision values from these data. The numerical difference of only 0.2% CV between densitometer models suggests that TBS reproducibility is not altered by the higher-resolution images generated by the iDXA. Finally, similar to observations for BMD, these data demonstrate comparable TBS precision in men and women, suggesting no gender-related effect and therefore no need to perform gender-specific TBS precision assessments. Similar to BMD, a TBS precision difference was demonstrated between technologists. This is to be expected as precision incorporates 3 components of measurement: the instrument, the subject, and the technologist. Consequently, intertechnologist variation in BMD measurement is widely reported (23,25,26); these same variations could be expected to affect TBS reproducibility in a similar manner. Similar to BMD observations, TBS generates comparable values when measured on various Prodigy instruments. However, the only modest correlation of TBS values between Prodigy and iDXA is noteworthy, particularly because these data consistently demonstrated this finding on several instrument comparisons. It is interesting that these TBS differences exist despite excellent BMD agreement. It seems likely that this difference reflects the higher patterned image (or texture) generated by the iDXA, compared with the lower resolution Prodigy, consequently yielding a different TBS value. It is unknown whether these between-instrument TBS variations reflect differences in fracture association. Regardless of why TBS varies between these different resolution technologies, such variation is not new to the DXA field. It is well accepted that between-densitometer and between-manufacturer differences in BMD measurement exist because of differences in technology; consequently, different BMD values are generated when measured on various instruments. Thus, it is recommended that BMD values obtained from different instruments not be directly compared. Consistent with this approach, these data suggest that it is best to handle the observed difference in TBS results in the same manner, that is, potential difference should be recognized when comparing TBS values generated not only on different manufacturers but extended to differing scanner models. Consequently, TBS change over time should only be evaluated when serial scans have been obtained on the same densitometer or potentially when using TBS cross-calibrated instruments. In this regard, it should be emphasized that these data were generated on instruments that were not TBS cross-calibrated. Specifically, data reported here were generated on historical scans performed using several instruments, consequently preventing crosscalibration of all instruments. Routine TBS software installation includes measurement of a TBS calibration phantom. This allows determination of an adjustment factor that is subsequently incorporated into the software to calibrate an instrument to the original Prodigy densitometer used to develop the TBS technology. Future studies are indicated to evaluate the capability of this approach to harmonize TBS values across densitometers.
Krueger, Libber, and Binkley A strength of this work is that it is the first study to report Prodigy-to-iDXA TBS precision comparisons and also to compare precision in men and women. This adds to the studies noted previously that evaluated TBS precision in humans and a recent publication reporting TBS precision using spine phantoms on a Hologic instrument (Hologic, Bedford, MA) demonstrating TBS comparability between various scan modes on a single densitometer (27). However, limitations to this work are recognized. These data are comprised from relatively small data sets and also lack TBS comparison between iDXA instruments. However, several different iDXA to different Prodigy densitometer comparisons were made with consistent results. Additionally, 13 subjects were included with a body mass index above the optimal range for TBS (15e37 kg/cm2). These data were retained as many were included in the precision assessments, and excluding them would have compromised the statistical approach of the precision calculations. Finally, as these data were acquired using only GE Lunar instruments, these results do not speak to potential intermanufacturer differences in either TBS measurement or TBS precision. Larger studies evaluating the effect of the different technologies used by various manufactures are indicated. In conclusion, TBS precision is similar on iDXA and Prodigy instruments, does not differ between sexes, and intertechnologist variation occurs. Excellent TBS measurement correlation exists between different Prodigy instruments, suggesting that cross-calibration could allow serial TBS comparison on similar model densitometers. However, differences in TBS measurements are observed between iDXA and Prodigy densitometers, suggesting a potential need to evaluate TBS treatment thresholds by model and limiting serial comparisons to scans obtained on the same instrument.
References 1. Kanis JA, Burlet N, Cooper C, et al. 2008 European guidance for the diagnosis and management of osteoporosis in postmenopausal women. Osteoporos Int 19(4):399e428. 2. Johnell O, Kanis JA. 2006 An estimate of the worldwide prevalence and disability associated with osteoporotic fractures. Osteoporos Int 17(12):1726e1733. 3. Browner WS, Pressman AR, Nevitt MC, Cummings SR. 1996 Mortality following fractures in older women. The study of osteoporotic fractures. Arch Intern Med 156(14):1521e1525. 4. Hannan EL, Magaziner J, Wang JJ, et al. 2001 Mortality and locomotion 6 months after hospitalization for hip fracture: risk factors and risk-adjusted hospital outcomes. JAMA 285(21): 2736e2742. 5. Davies KM, Stegman MR, Heaney RP, Recker RR. 1996 Prevalence and severity of vertebral fracture: the Saunders County Bone Quality Study. Osteoporos Int 6(2):160e165. 6. NOF. 2014 Clinician’s Guide to Prevention and Treatment of Osteoporosis. Washington, DC: National Osteoporosis Foundation. 7. Watts NB, Adler RA, Bilezikian JP, et al. 2012 Osteoporosis in men: an Endocrine Society clinical practice guideline. J Clin Endocrinol Metab 97(6):1802e1822. 8. Kanis JA, Melton LJI, Christiansen C, et al. 1994 Perspective: the diagnosis of osteoporosis. J Bone Miner Res 9(8): 1137e1141.
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
TBS Precision and Instrument Correlation 9. Siris ES, Chen YT, Abbott TA, et al. 2004 Bone mineral density thresholds for pharmacological intervention to prevent fractures. Arch Intern Med 164:1108e1112. 10. Schuit SCE, van der Klift M, Weel AEAM, et al. 2004 Fracture incidence and association with bone mineral density in elderly men and women: the Rotterdam study. Bone 34:195e202. 11. Winzenrieth R, Michelet F, Hans D. 2013 Three-dimensional (3D) microarchitecture correlations with 2D projection image gray-level variations assessed by trabecular bone score using high-resolution computed tomographic acquisitions: effects of resolution and noise. J Clin Densitom 16(3):287e296. 12. Hans D, Barthe N, Boutroy S, et al. 2011 Correlations between trabecular bone score, measured using anteroposterior dualenergy X-ray absorptiometry acquisition, and 3-dimensional parameters of bone microarchitecture: an experimental study on human cadaver vertebrae. J Clin Densitom 14(3):302e312. 13. Roux JP, Wegrzyn J, Boutroy D, et al. 2012 Relationship between trabecular bone score (TBS), bone mass and microarchitecture in human vertebrae: an ex vivo study. Osteoporos Int 23(2 Suppl):S327. P597. 14. Hans D, Goertzen AL, Krieg MA, Leslie WD. 2011 Bone microarchitecture assessed by TBS predicts osteoporotic fractures independent of bone density: the Manitoba study. J Bone Miner Res 26(11):2762e2769. 15. Boutroy S, Hans D, Sornay-Rendu E, et al. 2013 Trabecular bone score improves fracture risk prediction in non-osteoporotic women: the OFELY study. Osteoporos Int 24(1):77e85. 16. Winzenrieth R, Dufour R, Pothuaud L, Hans D. 2010 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 86(2):104e109. 17. Del Rio LM, Winzenrieth R, Cormier C, Di Gregorio S. 2013 Is bone microarchitecture status of the lumbar spine assessed by TBS related to femoral neck fracture? A Spanish case-control study. Osteoporos Int 24(3):991e998.
7 18. Popp AW, Meer S, Krieg MA, et al. 2012 Bone mineral density (BMD) combined with micro-architecture parameters (TBS) significantly improves the identification of women at high risk of fracture: the SEMOF Cohort Study. Osteoporos Int 23(2 Suppl):S327. P598. 19. Krueger D, Fidler E, Libber J, et al. 2014 Spine trabecular bone score subsequent to bone mineral density improves fracture discrimination in women. J Clin Densitom 17(1): 60e65. 20. Silva BC, Boutroy S, Zhang C, et al. 2013 Trabecular bone score (TBS)da novel method to evaluate bone microarchitectural texture in patients with primary hyperparathyroidism. J Clin Endocrinol Metab 98(5):1963e1970. 21. Dufour R, Winzenrieth R, Heraud A, et al. 2013 Generation and validation of a normative, age-specifiic reference curve for lumbar spine trabecular bone score (TBS) in French women. Osteoporos Int 24(11):2837e2846. 22. Krueger D, Vallarta-Ast N, Checovich M, et al. 2012 BMD measurement and precision: a comparison of GE Lunar Prodigy and iDXA densitometers. J Clin Densitom 15(1):21e25. 23. Krueger D, Vallarta-Ast N, Libber J, et al. 2013 Does the precision of dual-energy X-ray absorptiometry for bone mineral density differ by sex? J Clin Densitom 17:505e509. 24. Blake GM, Fogelman I. 1997 Technical principles of dual energy X-ray absorptiometry. Semin Nucl Med 27:210e228. 25. Bonnick S, Johnston C, Kleerekoper M, et al. 2001 Importance of precision in bone density measurements. J Clin Densitom 2: 105e110. 26. Baim S, Wilson C, Lewiecki EM, et al. 2006 Precision assessment and radiation safety for dual-energy x-ray absorptiometry: Position paper of the International Society for Clinical Densitometry. J Clin Densitom 8:371e378. 27. Bandirali M, Di Leo G, Messina C, et al. 2014 Reproducibility of trabecular bone score with different scan modes using dualenergy X-ray absorptiometry: a phantom study. Skeletal Radiol.
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
Original Article
Spine Trabecular Bone Score Precision, a Comparison Between GE Lunar Standard and High-Resolution Densitometers Diane Krueger,* Jessie Libber, and Neil Binkley Osteoporosis Clinical Research Program, University of Wisconsin, Madison, WI, USA
Abstract Trabecular bone score (TBS) is related to microarchitecture and fracture risk independently of bone mineral density (BMD) and clinical risk factors. Widespread clinical TBS use requires documentation of reproducibility and ideally comparability across scanners. This study evaluated TBS reproducibility and explored differences between Lunar Prodigy and iDXA densitometers. Reproducibility was assessed from replicate scans in 210 men and women participating in various dual-energy X-ray absorptiometry (DXA) precision assessments. iDXA-to-Prodigy comparability was evaluated using 155 participants from 3 study groups. L1eL4 BMD and TBS precision was similar on iDXA and Prodigy (BMD coefficient of variation 5 1.9% and 1.5% and TBS coefficient of variation 5 1.4% and 1.6%, respectively). Precision did not differ between men and women; however, between-technologist differences ( p ! 0.05) of similar magnitude were observed for both BMD and TBS. Prodigy-to-Prodigy TBS values were highly correlated (R2 5 0.85 with bias of 0.010 TBS units). Agreement was less robust comparing Prodigy with iDXA instruments (TBS R2: 0.72e0.81 with biases of 0.012e0.034 TBS units). In conclusion, TBS precision is comparable to that of BMD and does not differ between men and women. Additionally, in these cohorts, slight TBS differences were observed between iDXA and Prodigy scans. These data suggest a potential difference between densitometer models perhaps due to higher iDXA image resolution. Key Words: Correlation; DXA; precision; spine bone mineral density; trabecular bone score.
elevated risk for osteoporosis-related fractures. Currently, bone mineral density (BMD), measured by dual-energy X-ray absorptiometry (DXA), is the clinical gold standard to diagnose osteoporosis and determine fracture risk in the absence of a prior fragility fracture (8). However, as approximately half of low-trauma fractures occur in those with a BMD T-score better than 2.5 (9,10), methods to optimally identify persons likely to fracture despite adequate bone mass by DXA is desirable. As macrogeometry and trabecular bone microarchitecture influence bone strength, and consequently fracture risk, a method to clinically evaluate microarchitecture is an attractive addition to BMD in fracture risk prediction. Trabecular bone score (TBS) is a novel gray-level texture measurement that uses 2-dimensional DXA images to differentiate between 3-dimensional microarchitectures that exhibit the same bone density but different structural characteristics (11,12). A TBS score is determined by measuring the rate of local gray level variation from the DXA image.
Introduction Osteoporosis, a disease of low bone mass and associated microarchitectural deterioration of bone tissue, results in bone fragility and increased fracture risk. Given the aging populations worldwide, this is a major health issue (1). It is estimated that approximately 9 million new osteoporotic fractures occur annually worldwide (2), the ramifications of which are substantial morbidity, mortality, and expenditure of health care resources (3,4). It is projected that the number of osteoporosis-related fractures will double over the next 40e50 years (5). As effective therapies to reduce fractures exist (6,7), there is great interest to improve identification of individuals at Received 09/26/14; Revised 11/14/14; Accepted 11/18/14. *Address correspondence to: Diane Krueger, BS, CBDT, Osteoporosis Clinical Research Program, University of Wisconsin, 2870 University Avenue, Suite 100, Madison, WI 53705. E-mail: dckruege@ wisc.edu
1
2
Krueger, Libber, and Binkley
Consequently, it is not generating a physical measurement but a texture index representative of trabecular bone independent of BMD. This methodology is validated by the high correlation TBS demonstrates with micro-computed tomography (13). Additionally, TBS has documented fracture risk prediction utility both independently and in conjunction with BMD for vertebral and nonvertebral low-trauma fracture (14e20). The literature regarding TBS precision is currently limited (14,18,21), and between densitometer comparisons of TBS values and precision is lacking, particularly in instrument models with differing image resolution. It is plausible, given the use of gray-level assessment in TBS, that higherresolution DXA images may generate different TBS values or have different precision error. This work evaluates TBS measurement and precision comparisons between GE Healthcare Prodigy (standard resolution) and iDXA (high resolution) instruments.
Materials and Methods Subjects Existing data were gathered from 4 distinct cohorts to evaluate precision and TBS measurements from each type of instrument. In the following text are the descriptions of each group, 3 of which have been published; brief descriptions of the original study design, or purpose of the initial data collection, are described here to facilitate clarity. Precision Comparison Study: Designed to compare BMD precision between iDXA and Prodigy, enrolled 30 postmenopausal women having precision scans on each instrument; 1 technologist acquired all scans (22). BMD Instrument Comparison Study: Designed to compare BMD between iDXA and Prodigy, 95 men and women were included from this trial. Each subject was scanned on both instruments, and scans were acquired by multiple technologists (22). Male-Female Precision Comparison Study: Designed to evaluate if precision differs by sex, 180 men and women (90 in each group) were enrolled. Precision scans
acquired on an iDXA; 3 technologists conducted 1 precision assessment for each sex (23). Cross-Calibration Exercise: Data acquired to crosscalibrate 3 clinical instruments, 30 volunteers were scanned on 2 Prodigy and 1 iDXA instruments; 3 technologists acquired scans.
Precision Evaluations of TBS and Spine BMD Precision data from the Precision Comparison and MaleFemale Precision Comparison Studies were evaluated to assess short-term TBS and lumbar spine BMD precision; demographic data for these studies are listed in Table 1. Data from the Precision Comparison Study were used to compare precision values between a Prodigy and an iDXA in 30 women aged 65 years. The male-female precision cohort was used to assess precision variation between DXA technologists and sexes using an iDXA. In this study, 3 technologists performed 2 precision assessments, 1 in 30 men and 1 with 30 women. Each technologist scanned 60 different subjects; consequently, the total sample consisted of 180 subjects.
Interinstrument Comparison of TBS and Spine BMD Data to assess interinstrument agreement of TBS and lumbar spine BMD were evaluated separately in the following 3 cohorts: Precision Comparison Study, BMD Instrument Comparison Study, and Cross-Calibration Exercise. The precision comparison group is described previously. Ninety-five adult men (n 5 37) and women (n 5 58) participated in the BMD instrument comparison study. Finally, 30 adult men (n 5 5) and women (n 5 25) participated in the crosscalibration exercise; all volunteers in this study were scanned on 2 Prodigy instruments and 1 iDXA. Demographic data are listed in Table 1. A strength of this study, but a challenge for data presentation, is that 6 different densitometers (3 iDXA and 3 Prodigy instruments) were used for study of the various cohorts. Table 2 defines which instruments were used in the respective studies. Due to the known variability of individual DXA
Table 1 Demographics N
BMD Instrument Comparison Study Precision Comparison Study Cross-Calibration Exercise Male-Female Precision Comparison Study (male) Male-Female Precision Comparison Study (female)
95 37/58 58.2 16.5 30 0/30 69.5 4.9 30 5/25 57.4 9.6 90 90/0 73.9 6.2 90
M/F
Age, mean SD
Study
0/90 75.8 7.3
BMI, mean SD 27.9 25.9 26.0 26.8
5.7 4.5 4.7 3.4
26.6 5.3
LS BMD, mean SD 1.224 1.111 1.130 1.137
0.214 0.133 0.158 0.257
1.131 0.189
LS T-score, mean SD 0.2 0.7 0.5 0.8
1.7 1.1 1.3 2.1
0.4 1.6
TBS, mean SD 1.390 1.349 1.336 1.340
0.121 0.102 0.112 0.096
1.321 0.088
Abbr: BMD, bone mineral density; BMI, body mass index; F, female; LS, lumbar spine; M, male; SD, standard deviation; TBS, trabecular bone score. Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
TBS Precision and Instrument Correlation
3
Table 2 Study Sample and Acquisition Instrument Study
Evaluation
Participating technologists
Instruments used
Precision Comparison Study
Precision and comparison
1
BMD Instrument Comparison Study
Comparison
4
Male-Female Precision Comparison Study Cross-Calibration Exercise
Precision Comparison
3 3
Research iDXA 1 Research Prodigy 1 Research iDXA 2 Research Prodigy 1 Research iDXA 2 Clinical Prodigy 2 Clinical Prodigy 3 Clinical iDXA 3
Abbr: BMD, bone mineral density.
instruments (24,25), and the clinical need to evaluate a given patient over time, all data were evaluated using individual instrument comparison rather than pooling all Prodigy and all iDXA data.
F-test or factorial analysis of variance (Excel; Microsoft, Redmond, OR).
Precision and Cross-calibration Assessments
Subjects
All precision assessments were conducted per the International Society for Clinical Densitometry (ISCD) official positions; 30 individuals were scanned twice for each evaluation (26). Participants stood from the DXA table between scans, were immediately repositioned, and then rescanned. The first scan performed was analyzed per manufacturer recommendations, with manual adjustment of autoanalysis as needed. The second scans were analyzed using the software copy feature. The cross-calibration volunteers were scanned on 3 instruments (2 Prodigy and 1 iDXA, noted in Table 2). As these instruments were located at various locations within a city, all scans were performed within a 2-week period by 3 technologists.
Demographic ranges of all cohorts combined included an age of 22e92 years, body mass index from 16.4e47.7 kg/ m2, lumbar spine BMD of 0.755e1.945 g/cm2 with associated T-scores of 3.9 to þ6.0 and TBS measurement of 0.937e1.761 TBS units. Specific demographic data are reported by study in Table 1; no between-group comparisons were performed.
Bone Density Scan Acquisition and TBS Measurement As noted previously and in Table 2, 6 densitometers were used in this sampling. For purposes of data presentation here, each model was labeled consecutively on the basis of the instrument installation date. Lumbar spine scans were acquired and analyzed per manufacturer recommendations by ISCDcertified technologists. All scans were reviewed for quality by an ISCD-certified clinician. TBS values were generated by a single ISCD-certified technologist on the lumbar spine BMD files using TBS research software, version 2.1.0 (Medimaps group, Plan-lesOuates, Switzerland). TBS cross-calibration was not performed on any instrument.
Statistical Analyses Prodigy and iDXA TBS values were compared using linear regression and Bland-Altman analysis (Analyse-it, Leeds, United Kingdom). Precision was determined using ISCD Advanced Precision Calculator and compared by
Results
Precision Short-term precision, expressed as root mean square of the standard deviation (RMS-SD) and percent coefficient of variation (%CV), was similar between BMD and TBS and the iDXA and Prodigy instruments. The precision comparison study demonstrated L1eL4 BMD %CV of 1.9% and 1.5% on iDXA and Prodigy, respectively. Similarly, respective TBS %CVs of 1.4% and 1.6% were observed. No betweeninstrument precision differences were demonstrated for either BMD or TBS ( p O 0.10); complete data and values expressed as least significant change calculated at the 95% confidence level are listed in Table 3. The male-female precision study was performed on 1 iDXA and evaluated intertechnologist precision variation as well as sex difference. The %CV for L1eL4 BMD ranged from 1.0% to 1.6% across all technologists and both sexes; TBS %CV values were numerically higher, ranging from 1.5% to 2.5%. As expected, intertechnologist precision differences were observed for both BMD and TBS ( p ! 0.05; Table 3). However, as has previously been reported for BMD in this group (22), there was no difference in TBS reproducibility between sexes. Specific data are presented in Fig. 1 and Table 3.
Between-Instrument Comparison BMD was highly correlated between instruments in all comparison groups; R2 ranged from 0.97 to 0.98 with bias
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
4
Krueger, Libber, and Binkley Table 3 Precision BMD
Prodigya iDXAa iDXA maleb iDXA femaleb iDXA tech 1 male iDXA tech 1 female iDXA tech 2 male iDXA tech 2 female iDXA tech 3 male iDXA tech 3 female
TBS
RMS-SD (g/cm2)
CV (%)
LSC (g/cm2)
RMS-SD (g/cm2)
CV (%)
LSC (g/cm2)
0.016 0.020 0.016 0.016 0.012c 0.014c,d 0.020e,f,g 0.020f 0.014 0.014c
1.51 1.90 1.24 1.35 0.96 1.19 1.48 1.59 1.22 1.23
0.044 0.056 0.044 0.044 0.034 0.037 0.056 0.054 0.040 0.039
0.016 0.018 0.027 0.024 0.025 0.022c 0.033f 0.030e,f 0.020 0.019c
1.57 1.36 2.01 1.82 1.90 1.70 2.51 2.24 1.50 1.45
0.060 0.049 0.074 0.067 0.071 0.062 0.091 0.082 0.055 0.053
Note: LSC calculated at 95% confidence of RMS-SD. Abbr: BMD, bone mineral density; CV, coefficient of variation; LSC, least significant change; RMS-SD, root mean square standard deviation; TBS, trabecular bone score; tech, technologist. a Same female subjects used for each assessment. b Subjects from each assessment were pooled by sex (n 5 90). c Different from tech 2 male. d Different from tech 2 female. e Different from tech 3 female. f Different from tech 1 male. g Different from tech 3 male.
between 0.007 and þ0.007 g/cm2 (Figs. 2 and 3). Similarly, TBS was well correlated between Prodigy instruments (R2 5 0.85, with a bias of 0.010 TBS units [Fig. 2A and B]). No iDXA-to-iDXA data were available to compare. When evaluating TBS iDXA-to-Prodigy performance, the correlation was less robust, demonstrating an R2 range from 0.72 to 0.81 with biases from 0.008 to 0.035 (Fig. 3AeD).
Discussion In this study, we evaluated whether TBS performed similarly on 2 GE Healthcare densitometer models, a standardresolution (Prodigy) and a high-resolution (iDXA) fan-beam densitometer. As TBS uses gray-scale differentiation to calculate values, it is plausible that results and reproducibility
Fig. 1. Similar Precision for BMD and TBS. Precision (RMS-SD) was !3% and did not differ ( p O 0.10) between Prodigy and iDXA. Intertechnologist differences were observed for both BMD and TBS ( p ! 0.05) but not between sexes. BMD, bone mineral density; F, female; M, male; RMS-SD, root mean square-standard deviation; TBS, trabecular bone score. Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
TBS Precision and Instrument Correlation
5
Fig. 2. (AeB) BMD and TBS comparison: Prodigy vs Prodigy. Both BMD and TBS were highly correlated when comparing Prodigy to Prodigy demonstrating respective R2 5 0.98 and 0.85 with biases of 0.007 and 0.010 TBS units. The inset depicts mean (standard deviation) BMD and TBS values, which did not differ between instruments. BMD, bone mineral density; TBS, trabecular bone score. could differ when using technologies of differing image resolution. Consequently, we assessed if TBS precision differed between instruments and also how well correlated TBS measurements were between GE Lunar Prodigy and Lunar iDXA densitometers. These data demonstrate similar precision with a moderate correlation between these Prodigy and iDXA densitometers. These data demonstrate that TBS is precise and comparable to previously reported values (%CV: 1.5%e2.1%) (14,16)
but did demonstrate a slightly larger TBS precision range of 1.2 up to 2.7% CV. However, this could be explained by the fact that the subjects in these precision assessments with the highest %CV were aged at least 65 years, whereas those in the previously reported Manitoba and French cohorts were as young as 45 years (14,16). As a result, it is probable that the subjects reported here have a higher prevalence of spinal degenerative disease and also other age-associated challenges in positioning and analysis for spine BMD
Fig. 3. (AeD) BMD and TBS comparisons: Prodigy vs iDXA. In the sample of 95, BMD was highly correlated (R2 5 0.98) with a bias of þ0.007 g/cm2 (A); TBS was well correlated (R2 5 0.81) with a bias of 0.034 TBS units (B). Similarly, in a sample of 30 from the cross-calibration group, BMD was well correlated (R2 5 0.97) with a bias of 0.007 g/cm2 (C). TBS was less well correlated (R2 5 0.73) with a bias of þ0.012 TBS units (D). The bar graph insets depict mean (standard deviation) BMD and TBS values; mean TBS values were lower ( p ! 0.05) on Prodigy instruments. Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
6 measurement that could provide cause to expect poorer precision values from these data. The numerical difference of only 0.2% CV between densitometer models suggests that TBS reproducibility is not altered by the higher-resolution images generated by the iDXA. Finally, similar to observations for BMD, these data demonstrate comparable TBS precision in men and women, suggesting no gender-related effect and therefore no need to perform gender-specific TBS precision assessments. Similar to BMD, a TBS precision difference was demonstrated between technologists. This is to be expected as precision incorporates 3 components of measurement: the instrument, the subject, and the technologist. Consequently, intertechnologist variation in BMD measurement is widely reported (23,25,26); these same variations could be expected to affect TBS reproducibility in a similar manner. Similar to BMD observations, TBS generates comparable values when measured on various Prodigy instruments. However, the only modest correlation of TBS values between Prodigy and iDXA is noteworthy, particularly because these data consistently demonstrated this finding on several instrument comparisons. It is interesting that these TBS differences exist despite excellent BMD agreement. It seems likely that this difference reflects the higher patterned image (or texture) generated by the iDXA, compared with the lower resolution Prodigy, consequently yielding a different TBS value. It is unknown whether these between-instrument TBS variations reflect differences in fracture association. Regardless of why TBS varies between these different resolution technologies, such variation is not new to the DXA field. It is well accepted that between-densitometer and between-manufacturer differences in BMD measurement exist because of differences in technology; consequently, different BMD values are generated when measured on various instruments. Thus, it is recommended that BMD values obtained from different instruments not be directly compared. Consistent with this approach, these data suggest that it is best to handle the observed difference in TBS results in the same manner, that is, potential difference should be recognized when comparing TBS values generated not only on different manufacturers but extended to differing scanner models. Consequently, TBS change over time should only be evaluated when serial scans have been obtained on the same densitometer or potentially when using TBS cross-calibrated instruments. In this regard, it should be emphasized that these data were generated on instruments that were not TBS cross-calibrated. Specifically, data reported here were generated on historical scans performed using several instruments, consequently preventing crosscalibration of all instruments. Routine TBS software installation includes measurement of a TBS calibration phantom. This allows determination of an adjustment factor that is subsequently incorporated into the software to calibrate an instrument to the original Prodigy densitometer used to develop the TBS technology. Future studies are indicated to evaluate the capability of this approach to harmonize TBS values across densitometers.
Krueger, Libber, and Binkley A strength of this work is that it is the first study to report Prodigy-to-iDXA TBS precision comparisons and also to compare precision in men and women. This adds to the studies noted previously that evaluated TBS precision in humans and a recent publication reporting TBS precision using spine phantoms on a Hologic instrument (Hologic, Bedford, MA) demonstrating TBS comparability between various scan modes on a single densitometer (27). However, limitations to this work are recognized. These data are comprised from relatively small data sets and also lack TBS comparison between iDXA instruments. However, several different iDXA to different Prodigy densitometer comparisons were made with consistent results. Additionally, 13 subjects were included with a body mass index above the optimal range for TBS (15e37 kg/cm2). These data were retained as many were included in the precision assessments, and excluding them would have compromised the statistical approach of the precision calculations. Finally, as these data were acquired using only GE Lunar instruments, these results do not speak to potential intermanufacturer differences in either TBS measurement or TBS precision. Larger studies evaluating the effect of the different technologies used by various manufactures are indicated. In conclusion, TBS precision is similar on iDXA and Prodigy instruments, does not differ between sexes, and intertechnologist variation occurs. Excellent TBS measurement correlation exists between different Prodigy instruments, suggesting that cross-calibration could allow serial TBS comparison on similar model densitometers. However, differences in TBS measurements are observed between iDXA and Prodigy densitometers, suggesting a potential need to evaluate TBS treatment thresholds by model and limiting serial comparisons to scans obtained on the same instrument.
References 1. Kanis JA, Burlet N, Cooper C, et al. 2008 European guidance for the diagnosis and management of osteoporosis in postmenopausal women. Osteoporos Int 19(4):399e428. 2. Johnell O, Kanis JA. 2006 An estimate of the worldwide prevalence and disability associated with osteoporotic fractures. Osteoporos Int 17(12):1726e1733. 3. Browner WS, Pressman AR, Nevitt MC, Cummings SR. 1996 Mortality following fractures in older women. The study of osteoporotic fractures. Arch Intern Med 156(14):1521e1525. 4. Hannan EL, Magaziner J, Wang JJ, et al. 2001 Mortality and locomotion 6 months after hospitalization for hip fracture: risk factors and risk-adjusted hospital outcomes. JAMA 285(21): 2736e2742. 5. Davies KM, Stegman MR, Heaney RP, Recker RR. 1996 Prevalence and severity of vertebral fracture: the Saunders County Bone Quality Study. Osteoporos Int 6(2):160e165. 6. NOF. 2014 Clinician’s Guide to Prevention and Treatment of Osteoporosis. Washington, DC: National Osteoporosis Foundation. 7. Watts NB, Adler RA, Bilezikian JP, et al. 2012 Osteoporosis in men: an Endocrine Society clinical practice guideline. J Clin Endocrinol Metab 97(6):1802e1822. 8. Kanis JA, Melton LJI, Christiansen C, et al. 1994 Perspective: the diagnosis of osteoporosis. J Bone Miner Res 9(8): 1137e1141.
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
TBS Precision and Instrument Correlation 9. Siris ES, Chen YT, Abbott TA, et al. 2004 Bone mineral density thresholds for pharmacological intervention to prevent fractures. Arch Intern Med 164:1108e1112. 10. Schuit SCE, van der Klift M, Weel AEAM, et al. 2004 Fracture incidence and association with bone mineral density in elderly men and women: the Rotterdam study. Bone 34:195e202. 11. Winzenrieth R, Michelet F, Hans D. 2013 Three-dimensional (3D) microarchitecture correlations with 2D projection image gray-level variations assessed by trabecular bone score using high-resolution computed tomographic acquisitions: effects of resolution and noise. J Clin Densitom 16(3):287e296. 12. Hans D, Barthe N, Boutroy S, et al. 2011 Correlations between trabecular bone score, measured using anteroposterior dualenergy X-ray absorptiometry acquisition, and 3-dimensional parameters of bone microarchitecture: an experimental study on human cadaver vertebrae. J Clin Densitom 14(3):302e312. 13. Roux JP, Wegrzyn J, Boutroy D, et al. 2012 Relationship between trabecular bone score (TBS), bone mass and microarchitecture in human vertebrae: an ex vivo study. Osteoporos Int 23(2 Suppl):S327. P597. 14. Hans D, Goertzen AL, Krieg MA, Leslie WD. 2011 Bone microarchitecture assessed by TBS predicts osteoporotic fractures independent of bone density: the Manitoba study. J Bone Miner Res 26(11):2762e2769. 15. Boutroy S, Hans D, Sornay-Rendu E, et al. 2013 Trabecular bone score improves fracture risk prediction in non-osteoporotic women: the OFELY study. Osteoporos Int 24(1):77e85. 16. Winzenrieth R, Dufour R, Pothuaud L, Hans D. 2010 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 86(2):104e109. 17. Del Rio LM, Winzenrieth R, Cormier C, Di Gregorio S. 2013 Is bone microarchitecture status of the lumbar spine assessed by TBS related to femoral neck fracture? A Spanish case-control study. Osteoporos Int 24(3):991e998.
7 18. Popp AW, Meer S, Krieg MA, et al. 2012 Bone mineral density (BMD) combined with micro-architecture parameters (TBS) significantly improves the identification of women at high risk of fracture: the SEMOF Cohort Study. Osteoporos Int 23(2 Suppl):S327. P598. 19. Krueger D, Fidler E, Libber J, et al. 2014 Spine trabecular bone score subsequent to bone mineral density improves fracture discrimination in women. J Clin Densitom 17(1): 60e65. 20. Silva BC, Boutroy S, Zhang C, et al. 2013 Trabecular bone score (TBS)da novel method to evaluate bone microarchitectural texture in patients with primary hyperparathyroidism. J Clin Endocrinol Metab 98(5):1963e1970. 21. Dufour R, Winzenrieth R, Heraud A, et al. 2013 Generation and validation of a normative, age-specifiic reference curve for lumbar spine trabecular bone score (TBS) in French women. Osteoporos Int 24(11):2837e2846. 22. Krueger D, Vallarta-Ast N, Checovich M, et al. 2012 BMD measurement and precision: a comparison of GE Lunar Prodigy and iDXA densitometers. J Clin Densitom 15(1):21e25. 23. Krueger D, Vallarta-Ast N, Libber J, et al. 2013 Does the precision of dual-energy X-ray absorptiometry for bone mineral density differ by sex? J Clin Densitom 17:505e509. 24. Blake GM, Fogelman I. 1997 Technical principles of dual energy X-ray absorptiometry. Semin Nucl Med 27:210e228. 25. Bonnick S, Johnston C, Kleerekoper M, et al. 2001 Importance of precision in bone density measurements. J Clin Densitom 2: 105e110. 26. Baim S, Wilson C, Lewiecki EM, et al. 2006 Precision assessment and radiation safety for dual-energy x-ray absorptiometry: Position paper of the International Society for Clinical Densitometry. J Clin Densitom 8:371e378. 27. Bandirali M, Di Leo G, Messina C, et al. 2014 Reproducibility of trabecular bone score with different scan modes using dualenergy X-ray absorptiometry: a phantom study. Skeletal Radiol.
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015