Академический Документы
Профессиональный Документы
Культура Документы
DOI 10.1007/s00198-013-2502-7
ORIGINAL ARTICLE
Received: 6 May 2013 / Accepted: 3 September 2013 / Published online: 1 October 2013
# International Osteoporosis Foundation and National Osteoporosis Foundation 2013
with spine radiographs, with low radiation exposure [16–19]. Clinical assessment includes demographic data: age,
However, not all DXA machines are equipped with VFA height, weight, BMI (body mass index) (in kilogram per
software, and coverage of VFA has not been embraced by square metre), when standing position is possible, and infor-
health insurance in all countries. Thus, there is a need in mation on anti-osteoporotic treatment in the previous year.
clinical setting for a tool able to predict the presence of
vertebral fracture and to indicate appropriately spine imaging BMD and VFA assessment
in selected patients.
The trabecular bone score (TBS) is a bone texture analysis These measurements are performed 4–90 days after the
derived from the DXA image of the lumbar spine [20]. The hospitalization for fracture. aBMD (in gram per square
ex vivo studies conducted on bone specimens (vertebrae, centimetre) is assessed by DXA (Hologic®, QDR 4500A,
radius and femoral neck) show that TBS positively correlates software version 12.6; Bedford, MA) at the lumbar spine
with 3D bone microarchitecture parameters, such as connec- (L2–L4) and hip (total hip and femoral neck). The small
tivity, density and trabecular number, and negatively with detectable difference (SDD) for the device was 0.034, 0.036
trabecular separation [21–24]. In vivo, this texture parameter and 0.027 g/cm2 for the lumbar spine, femoral neck and total
analyses the grey-level variations in DXA images, in the same hip, respectively [30]. The quality control protocol for the
bone region as in lumbar spine aBMD. In clinical practice, DXA device includes daily scanning of a phantom.
lower TBS scores indicate greater fracture susceptibility. A single device was used for the whole current study. The
Cross-sectional studies showed that TBS was lower in post- World Health Organization (WHO) classification was used to
menopausal women with previous fragility fracture compared define osteoporosis as T-score≤−2.5 at the lumbar spine, total
to those without fracture [25] and was lower in women with hip or femoral neck. Vertebral fractures from T4 to L4 were
fractures irrespective of whether their bone mineral density evaluated using VFA software on the DXA device. They were
(BMD) met the criteria for osteoporosis or osteopenia classified with the Genant semiquantitative approach [31].
[26, 27]. Two retrospective historical cohort studies showed Patients with at least one grade 1 fracture were considered as
that lumbar spine TBS and aBMD predicted major fractures fractured. Spinal deformity index (SDI), an assessment tool
similarly; and that the combination of lumbar spine TBS and for vertebral fracture prediction [10, 32] was calculated by
aBMD improves fracture risk prediction in women with summing the grade of each fractured vertebra from T4 to L4;
aBMD in the non-osteoporotic range [27, 28]. We have shown the SDI value can vary between 0 (no fracture) and 39 (all the
previously in patients with rheumatoid arthritis that TBS has a assessed vertebrae are grade 3). The diagnosis was directly
better discrimination than lumbar spine aBMD for prediction assessed on the screen, by one single reader (JF), an expert in
of the presence of vertebral fractures [29]. this field.
Thus, the purpose of our study was to evaluate the value of
TBS, alone or added to aBMD, in the prediction of the TBS
presence and severity of vertebral fractures in a cohort of
patients included in a Fracture Liaison Service. TBS (TBS iNsight® Software version 1.8, Med-Imaps,
Pessac, France) was obtained after reanalysis of DXA lumbar
spine (L2–L4) scans. The study was conducted independently
Patients and methods of the manufacturer. The software uses the AP spine raw
image(s) from the densitometer, including the aBMD region
Patient selection of interest (ROI) and edge detection, so that the TBS calcula-
tion is performed over exactly the same ROI as the aBMD
Patients were selected from the Fracture Liaison Service measurement. For each region of measurement, TBS was
(FLS) of Cochin Hospital. This FLS was established to pro- evaluated based upon grey-level analysis of the DXA images
vide routine assessment for osteoporosis to all men and wom- as the slope at the origin of the log-log representation of the
en over the age of 50 years who had sustained a low-trauma experimental variogram. TBS (L2–L4) was calculated as the
non-vertebral fracture and are hospitalized in the orthopaedic mean value of the individual measurements for vertebrae L2–
surgery department. Low-trauma fractures are defined as those L4. A BMI between 15 and 35 kg/m2 is mandatory for TBS
sustained in falls from standing height or less. Exclusion analysis, according to the manufacturer.
criteria are pathological and traumatic fractures, outpatients
and severe impaired cognitive functions. Statistical analysis
A senior physician, dedicated to the task of identifying
patients with fragility fractures, selects these patients in the Characteristics of patients with hip fractures and non-vertebral
orthopaedic surgery department and prescribes DXA scanning non-hip fractures and with and without vertebral fractures
and VFA. were compared by using chi-square tests or Fisher's exact
Osteoporos Int (2014) 25:243–249 245
tests, or t tests as appropriate. The discriminative values of (n =133); the mean number of vertebral fractures was 1.7±
aBMD at all sites, TBS (L2–L4) and their combination for the 1.2. The characteristics of patients according to the presence
presence of vertebral fracture was assessed by determining the of vertebral fractures were in Table 2.
area under the receiving operator characteristic (ROC) curve Half of our population recruited in the FLS had hip frac-
(area under the curve, AUC). All the analysis was performed ture. Compared to patients with non-hip non-vertebral
on SAS 9.1® statistical software. fractures, patients with hip fracture were older (77.3±10.9 vs
71.3±11.7 years, p ≤0.0001), had frequently more vertebral
fractures (44.5 vs 28.9 %, p =0.002) and osteoporosis
(T≤−2.5 at least one the three sites) (62.6 vs 41.7 %,
Results p ≤0.0001). They had lower femoral neck and hip aBMD than
patients without hip fractures (0.580±0.11 vs 0.639±0.12 g/cm2
Patient characteristics and 0.600±0.15 vs 0.769±0.15 g/cm2, p <0.0001).
with at least one grade 2, and 1.139±0.104 in patients with at TBS in patients with aBMD in the non-osteoporotic range
least one grade 3.
There was a negative correlation between SDI and TBS In patients with aBMD in the non-osteoporotic range (n =173)
(r =−0.31 (p <0.0001)), hip aBMD (−0.32 (p <0.0001)) and (T>−2.5 at all sites), TBS was significantly lower in patients
LS aBMD (r =−0.30 (p <0.0001)). TBS was not correlated with vertebral fractures than in patients without VF (1.187±
with SDI after adjustment for age, hip aBMD and lumbar 0.121 vs 1.253±0.104, p =0.001). AUC of TBS for the dis-
spine aBMD. crimination of VFs was higher than the AUC of LS aBMD
Seventy-six patients had at least one grade 2 VF. The AUC (0.671 vs 0.541, p =0.035) but not of hip aBMD (0.670 vs
for discrimination of these patients as compared to the others 0.585, p =0.264) (Figs. 2 and 3). TBS was negatively corre-
(n =286 having no VF or only grade 1 VFs) were 0.665, lated to SDI (r =−0.25 (p =0.001)). In these patients, LS
0.679, 0.661, 0.712 and 683 for TBS, LS aBMD, hip aBMD and hip aBMD were not correlated with SDI (−0.07
aBMD, TBS + LS aBMD, and TBS + hip aBMD. In this (p =0.360) and −0.04 (p =0.618), respectively).
situation, AUC of TBS for the discrimination of VFs was In patients for whom the non-vertebral fracture was a major
similar to LS aBMD (0.665 vs 0.679, p =0.712) and to hip fracture according to the FRAX® (n =132), TBS was signif-
aBMD (0.665 vs 0.661 p =0.953). Combination of TBS and icantly lower in patients with vertebral fractures than in pa-
LS aBMD or hip aBMD to improve discrimination of patients tients without VF (1.194±0.112 vs 1.264±0.092, p =0.001).
having at least one grade 2 VF was not significant (p =0.095 AUC of TBS for the discrimination of VFs was similar to
and 0.433, respectively) as compared to these parameters LS aBMD (0.697 vs 0.545, p =0.303) and to hip aBMD
alone. Results were similar in patients for whom the non- (0.697 vs 0.588, p =0.264). TBS was negatively correlated to
vertebral fracture was a major fracture according to the SDI (r =−0.29 (p =0.001)). In these patients with aBMD in the
FRAX®. non-osteoporotic range, LS aBMD and hip aBMD were not
0.6
0.5
0.4
0.3
0.2
0.1
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
1 - Specificity
Fig. 2 Discriminative value of ROC curves in the patients with aBMD in the non osteoporotic range
TBS alone and in combination 1
with LS and hip aBMD in patients
with aBMD in the non- 0.9
osteoporotic range 0.8
0.7
Sensitivity
0.6
0.5
0.4
0.3
0.2
0.1
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
1 - Specificity
TB LS Hip TBS + LS TBS + Hip
correlated with SDI (−0.08 (p =0.344) and −0.06 (p =0.491), subjects with hip fractures, explaining the higher prevalence
respectively). of VFs found in our study as compared to other studies [33]. In
these patients with hip fractures, VFs influence functional
outcomes [34], illustrating the relevance of the identification
Discussion of patients with prevalent vertebral fracture.
Our study confirms previous data showing that TBS had an
This study conducted in a population of patients recruited in a additive value for the discrimination of patients with vertebral
Fracture Liaison Service shows that TBS improves the pre- fracture when combined with lumbar spine aBMD [24–26]. In
diction of the presence of VFs as compared to lumbar spine patients with rheumatoid arthritis (RA), TBS measured at the
aBMD alone, but not to hip aBMD. The same result is lumbar spine has a better discrimination value than lumbar
observed in patients with aBMD in the non-osteoporotic spine aBMD for the prediction of the presence of vertebral
range. However, in this population, TBS is correlated to the fractures. In RA patients with osteopenia, the proportion of
severity of the vertebral fractures, whereas lumbar spine and patients with vertebral fracture was higher in the lowest tertile
hip aBMDs are not. of TBS when compared with the highest tertile [29]. In wom-
Almost 40 % of our patients having a recent non-traumatic en older than 50 years of the Manitoba study (n =29 407,
non-vertebral fracture had at least one vertebral fracture, mean follow-up of 4.7 years) (10.7 % of major fractures
confirming the high prevalence of these VFs not previously excluding clinical vertebral fractures), TBS and LS aBMD
diagnosed in this population [33]. Prevalence of VFs differs predicted similarly incident fragility fractures (including clin-
according to the site of non-vertebral fracture; it is higher in ical vertebral fractures), and the combination was superior to
either measurement alone [27]. In the OFELY cohort (17 %
TBS values (mean)
with prevalent fracture), the assessment of TBS in 560 post-
1.24 menopausal women predicted incident fractures (including
1.22 clinical vertebral fractures), as well as LS aBMD, but the
1.2 combination of TBS and LS aBMD added only limited infor-
1.18
mation in the whole population [28]. In our study on patients
*
1.16
with non-vertebral fractures and in the other studies with a
lower prevalence of non-vertebral fractures, TBS is superior to
1.14
LS aBMD for the identification of vertebral fractures.
1.12
However, in clinical practice, measurement of aBMD at
1.1
two sites including the hip is recommended; our data show
1.08
No VF At least one Grade 1 Grade 2 Grade 3 that TBS does not add information when the hip aBMD is
(n=229) VF (n=133) (n=57) (n=47) (n=29) available.
Fig. 3 TBS values according to the vertebral fractures status. The aster- Identification of subjects with aBMD in the non-osteoporotic
isk indicates that TBS value of patients with at least one VF was lower
range at high risk of fractures is a relevant objective. The
than the TBS value of patients without VF (p ≤0.05). TBS was not
statistically significant among patients with only grade 1, patients with OFELY study showed that the combination of normal and
at least one grade 2 and patients with at least one grade 3 osteopenic T-scores with the lowest range of TBS improved
248 Osteoporos Int (2014) 25:243–249
20. Bousson V, Bergot C, Sutter B et al (2012) Trabecular bone score 27. Hans D, Goertzen AL, Krieg MA, Leslie WD (2011) Bone micro-
(TBS): available knowledge, clinical relevance, and future prospects. architecture assessed by TBS predicts osteoporotic fractures indepen-
Osteoporos Int 23:1489–5010 dent of bone density: The Manitoba study. J Bone Miner Res 26:
21. Pothuaud L, Carceller P, Hans D (2008) Correlations between grey-level 2762–2769
variations in 2D projection images (TBS) and 3D microarchitecture: 28. Boutroy S, Hans D, Sornay-Rendu E, Vilayphiou N, Winzenrieth R,
applications in the study of human trabecular bone microarchitecture. Chapurlat R (2013) Trabecular bone score improves fracture risk pre-
Bone 42:775–787 diction in non-osteoporotic women: the OFELY Study. Osteoporos Int
22. Winzenrieth R, Michelet F, Hans D (2012) Three-dimensional (3D) 24:77–85
microarchitecture correlations with 2D projection image grey level 29. Bréban S, Briot K, Kolta S, Paternotte S, Ghazi M, Fechtenbaum J,
variations assessed by TBS using high resolution CT acquisitions: Roux C (2012) Identification of rheumatoid arthritis patients with
effects of resolution and noise. J Clin Densitom 16(3):287–96. doi: vertebral fractures using bone mineral density and trabecular bone
10.1016/j.jocd.2012.05.001 score. J Clin Densitom 15:260–266
23. Hans D, Barthe N, Boutroy S, Winzenrieth R, Pothuaud L, Krieg M-A 30. Kolta S, Ravaud P, Fechtenbaum J, Dougados M, Roux C (2000)
(2011) Correlations between TBS, measured using antero-posterior DXA Follow-up of individual patients on two DXA scanners of the same
acquisition, and 3D parameters of bone microarchitecture: an experimen- manufacturer. Osteoporos Int 11:709–713
tal study on human cadaver vertebrae. J Clin Densitom 14:302–312 31. Genant HK, Wu CY, van Kuijk C, Nevitt MC (1993) Vertebral
24. Pothuaud L, Barthe N, Krieg M-A, Mehsen N, Carceller P, Hans D fracture assessment using a semiquantitative technique. J Bone
(2009) Evaluation of the potential use of trabecular bone score to Miner Res 8:1137–1148
complement bone mineral density in the diagnosis of osteoporosis: a 32. Crans GG, Genant HK, Krege JH (2005) Prognostic utility of a
preliminary spine BMD-matched, case–control study. J Clin semiquantitative spinal deformity index. Bone 37:175–179
Densitom 12:170–176 33. Howat I, Carty D, Harrison J, Fraser M, McLellan AR
25. Rabier B, Héraud A, Grand-Lenoir C, Winzenrieth R, Hans D (2010) (2007) Vertebral fracture assessment in patients presenting
A multicentre, retrospective case–control study assessing the role of with incident nonvertebral fractures. Clin Endocrinol (Oxf) 67:
trabecular bone score (TBS) in menopausal Caucasian women with 923–930
low areal bone mineral density (BMDa): analysing the odds of 34. Gonnelli S, Caffarelli C, Maggi S et al (2013) The assessment of
vertebral fracture. Bone 46:176–181 vertebral fractures in elderly women with recent hip fractures: the
26. Winzenrieth R, Dufour R, Pothuaud L, Hans D (2010) A retrospec- BREAK Study. Osteoporos Int 24:1151–1159
tive case–control study assessing the role of trabecular bone score in 35. van Geel TA, Huntjens KM, van den Bergh JP et al (2010) Timing of
postmenopausal Caucasian women with osteopenia: analyzing the subsequent fractures after an initial fracture. Curr Osteoporos Rep 8:
odds of vertebral fracture. Calcif Tissue Int 86:104–109 118–122