Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health, vol. 21, no.

2, 185–192, 2018
© 2017 The International Society for Clinical Densitometry.
1094-6950/21:185–192/$36.00
https://doi.org/10.1016/j.jocd.2017.09.003

Original Article

Lumbar Spine Trabecular Bone Score (TBS) Reflects Diminished

Bone Quality in Patients With Diabetes Mellitus and Oral
Glucocorticoid Therapy
Yunfeng Xue,1 Andrea L. Baker,2 Shahla Nader,1 Philip Orlander,1
Anthony J. Sanchez,1 James Kellam,2 Nahid J. Rianon,1 and
Catherine G. Ambrose2,*
1
Department of Internal Medicine, The University of Texas Health Science Center at Houston McGovern Medical School,
Houston, TX, USA; and 2Department of Orthopaedic Surgery, The University of Texas Health Science Center at Houston
McGovern Medical School, Houston, TX, USA

Abstract
Trabecular bone score (TBS) is a texture parameter that measures the grayscale variation within dual-
energy X-ray absorptiometry (DXA) images, and has been shown to significantly correlate with the 3-dimensional
bone microarchitecture. The objective of this study was to determine whether TBS is a better clinical tool
than traditionally used bone mineral density (BMD) to detect the skeletal deterioration seen in patients with
diabetes (DM), patients undergoing oral glucocorticoid (GC) therapy, and patients who are both diabetic and
taking steroids (GC + DM). We performed retrospective, cross-sectional study using DXA images of pa-
tients who visited UTHealth Department of Internal Medicine DXA clinic in Houston, TX, from May 30,
2014 to May 30, 2016. A total of 477 men and women, who were 55 years or older, were included in the study.
Lumbar spine (LS) BMD and TBS were collected. Electronic medical records were reviewed to collect clini-
cal information for each patient. When both men and women were analyzed as a single group, LS-BMD was
significantly higher in the diabetic group than in the control group (1.14 vs 1.10, p = 0.038), whereas mean
TBS of L1–L4 was significantly lower in the diabetic group (1.21 vs 1.26, p = 0.004). LS-TBS was also signifi-
cantly lower in diabetic women than in nondiabetic women (1.20 vs 1.26, p = 0.002). Receiver operating char-
acteristic curves and areas under the curve indicated that LS-TBS provided better ability than LS-BMD to
discriminate between control subjects and those in the DM, GC, or GC + DM groups (areas under the curve
between 0.645 and 0.697, p < 0.010 for all). LS-TBS is a BMD-independent parameter that is capable of cap-
turing a larger portion of bone quality deterioration undetected by BMD alone in patients with DM and un-
dergoing oral GC therapy.
Key Words: Bone mineral density; diabetes mellitus; glucocorticoid; trabecular bone score.

Introduction disease characterized by loss of bone mass and deteriora-

According to the Consensus Development Confer-
ence in 1993, osteoporosis is defined as a systemic skeletal
Received 08/4/17; Accepted 09/14/17.
*Address correspondence to: Catherine G. Ambrose, PhD,
UTHSC-Houston, McGovern Medical School, 6431 Fannin, Rm
6.154, Houston, TX 77030. E-mail: catherine.g.ambrose@
uth.tmc.edu
tion of the microarchitectural integrity of bone tissue,
which increases bone fragility and fracture risk (1). In
patients with osteoporosis, fractures occur following minimal
trauma, or in some cases, without any trauma. Annually, 2
million fractures can be attributed to osteoporosis, leading
to more than 432,000 hospital admissions, 2.5 million medical
office visits, and about 180,000 nursing home admissions
in the United States alone (2). In part because of an
aging population, the cost of care is expected to rise to

185

Downloaded for Budi Lakukua (googledotbudi@yahoo.co.id) at Universitas Tarumanagara from ClinicalKey.com by Elsevier on October 01, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
186 Xue et al.

25.3 billion by 2025 (3,4). As the burden of osteoporosis estimate 3-dimensional parameters of bone microarchi-
on our society continues to rise, there is a pressing need tecture (17).
for more effective methods to identify and treat patients Previous studies have found TBS to be a useful tool in
who are at risk of fracture. Early identification and inter- assessing fracture risk in patients with type 2 diabetes
vention have been shown to significantly improve outcomes (11,12). In most previously published studies of subjects with
in patients with osteoporosis (5). Currently, areal bone diabetes, steroid use was not investigated, although it has
mineral density (BMD) as measured by dual-energy X-ray also been shown to have a deleterious effect on both BMD
absorptiometry (DXA) is the gold standard in both the and TBS (23). In this study, we retrospectively assessed TBS
diagnosis and the treatment monitoring of osteoporosis for both men and women, and accounted for the use of
(6). BMD is integral to the World Health Organization steroid medications to investigate the possible utility of TBS
disease classification system, which defines osteoporosis in populations with patients who have DM, are taking glu-
as a BMD that lies 2.5 standard deviations (SD) below cocorticoids (GCs), or whose bone may be deteriorated by
the average of healthy young men or women (a T-score both (DM + GC).
of <−2.5 SD) (7). BMD is also a major determinant in the
Fracture Risk Assessment Tool, which predicts fracture
risk using a number of established risk factors for osteo-
Materials and Methods
porosis (8). Patient Selection
It has been well documented in literature that both type
In this retrospective cross-sectional study (reviewed and
1 and type 2 diabetes mellitus (DM) are independent risk
approved by the UTHealth IRB), DXA images were re-
factors for osteoporotic fractures (9–11). Paradoxically, pa-
analyzed for patients who presented to UTHealth Depart-
tients with diabetes have normal or higher BMD than do
ment of Internal Medicine DXA clinic in Houston, Texas,
nondiabetic controls (12,13). The commonly accepted hy-
from May 30, 2014 to May 30, 2016. Both men and women
pothesis for such observation is that BMD does not reflect
who were 55 years old and older and who had undergone
the microarchitectural integrity of the bone, which is a major
LS-BMD measurement of the lumbar spine by Prodigy
contributor to bone strength (14). Because of BMD’s central
DXA system (GE Medical-Lunar, Madison, WI) were eli-
role in most fracture prediction models used in clinical prac-
gible for inclusion. The DXA images were assessed using
tice, this paradox results in a lack of appropriate tools to
TBS iNsight v2.1 (Medimaps, Merignac, France) in accor-
identify patients with diabetes who are at increased frac-
dance with the manufacturer recommendations to gener-
ture risk (15,16).
ate the TBS parameters of the lumbar spine (L1–L4). In
Assessment of skeletal microstructure can be made via
2 patients, LS-BMD information was available only for
histomorphometric analysis of an iliac bone biopsy. Al-
L2–L4, and in these subjects, TBS was assessed for 3 ver-
though it is the gold standard for assessing bone
tebrae only. Patients with missing height or weight input
microarchitecture, the procedure is invasive, costly, and some
were excluded from the study. In patients with more than
argue that iliac bone is not representative of sites that are
1 eligible set of LS-BMD measurements, only the most
truly at risk of fracture (17). For these reasons, iliac bone
recent one was included. Finally, the patients’ electronic
biopsy remains primarily a research tool. Other measures
medical records were reviewed to collect additional infor-
have been demonstrated to either directly measure archi-
mation including age, gender, body mass index (BMI), eth-
tecture, such as high-resolution peripheral quantitative com-
nicity, diagnosis of DM, hemoglobin A1c (HbA1c), fracture
puted tomography (18), or be significantly correlated to
history, history of rheumatoid disease, use of corticoste-
architecture (flat-panel volume computed tomography, mag-
roid, serum level of calcium, and serum level of 25-
netic resonance imaging), but these techniques can be ex-
hydroxyvitamin D. We divided subjects into 4 groups
pensive and some can result in increased radiation dose
including controls: subjects with a diagnosis of diabetes,
for the patient (19,20). Trabecular bone score (TBS) is a
whether type 1 or type 2, but not taking steroids were in-
noninvasive imaging technology developed to provide an
cluded in the DM group. Subjects who had taken cortico-
indirect index of the trabecular microarchitecture using
steroids for >3 months but who did not have diabetes were
pixel gray-level variations in the lumbar spine DXA images
included in the GC group. Subjects with both diabetes and
(21). The underlying principle of TBS is that a healthy,
documented corticosteroid use were included in the
dense trabecular microstructure projects a DXA image
GC + DM group.
containing a large number of pixel value variations of small
amplitude, whereas a porous trabecular network pro-
duces a 2-dimensional (2D) image with low number of Data Analysis
variations of high amplitude (17,22). The numerical score Descriptive data for subject characteristics were ex-
is derived from experimental variogram of the 2D projec- pressed in mean ± SD for continuous variables or count
tion image, with a higher TBS indicating better bone struc- (percentage) for categorical variables. Group comparison
ture (22). Studies in human cadavers have shown that TBS was made with 1-way analysis of variance (least signifi-
can differentiate skeletal samples with similar BMD but cant difference post hoc tests) and Pearson chi-square
different microstructure, and TBS can be used to reliably test, respectively. Receiver operating characteristic (ROC)

Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health Volume 21, 2018

Downloaded for Budi Lakukua (googledotbudi@yahoo.co.id) at Universitas Tarumanagara from ClinicalKey.com by Elsevier on October 01, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
Lumbar Spine TBS Reflects Diminished Bone Quality 187

curves and areas under the curve (AUC) were calculated
to represent the ability of the LS-BMD and TBS values
to discriminate between control subjects and those in the
DM, GC, or GC + DM groups. We used the following
ranges to categorize the lumbar TBS values as described
by other investigators (21): TBS ≥1.350 was considered
category 1 or normal; TBS between 1.200 and 1.350 was
defined as category 2 or partially degraded
microarchitecture; and TBS ≤1.200 was category 3 or de-
graded microarchitecture. We also categorized the LS-
BMD results according to the World Health Organization
classification for osteoporosis: category 1—normal (t-
score ≥−1.0), category 2—low bone mass (t-score between
−2.5 and −1.0), and category 3—osteoporotic (t-score ≤−2.5).
The distribution of TBS and LS-BMD categories in each
subgroup was expressed in n (percentage). For each of
the 4 subgroups, Kendall tau-b was calculated to deter-
mine if the TBS and LS-BMD categorical classifications
were correlated. All statistical analyses were performed
in IBM SPSS Statistics for Windows, version 24 (IBM
Corp, Armonk, NY).
Results
Of the 477 subjects included in the study, 404 (84.7%)
were women and 73 (15.3%) were men. There were 231
(48.4%) Caucasian subjects, 117 (24.55) African Ameri-
can subjects, 50 (10.5%) Hispanic subjects, and 1 Asian
subject. For 78 subjects (16.4%), no data were available to
classify ethnicity. A total of 106 (22.2%) had a diagnosis
of DM and 95 (19.9%) had a documented history of oral
GC therapy. There were 291 subjects in the control group
(257 women and 34 men), 80 subjects in the GC group (62
women, 18 men), 91 subjects in the DM group (74 women,
17 men), and 15 subjects in the GC + DM group (11 women
and 4 men).
Table 1 summarizes the baseline characteristics of the
study cohort, divided by gender and group. Our data showed

Table 1
(A) Baseline Characteristics of Male Study Subjects. (B) Baseline Characteristics of Female Study Subjects

A
With diabetes and
Control group Taking steroids With diabetes taking steroids
(N = 34) (GC) (N = 18) (DM) (N = 17) (GC + DM) (N = 4) p

Age, yr 68.79 ± 9.09 67.61 ± 9.00 72.24 ± 7.22 60.75 ± 5.44 0.094
Body mass index, kg/m2 25.59 ± 4.33 27.68 ± 4.43 27.64 ± 4.21 34.70 ± 4.07 0.001
Fasting plasma glucose, mg/dL 98.30 ± 22.87 104.64 ± 25.70 160.31 ± 47.70 167.33 ± 52.16 <0.001
HbA1c, % 5.61 ± 0.51 6.00 ± 0.64 6.99 ± 1.00 6.50 ± 1.70 0.001
Est. glomerular filtration rate, 76.12 ± 16.61 68.71 ± 23.6 56.71 ± 15.98 32.00 ± 26.64 0.001
mL/min/1.73 m2
Vitamin D, ng/mL 39.49 ± 18.07 32.93 ± 9.49 28.52 ± 6.55 20.65 ± 10.82 0.114
Calcium, mg/dL 9.24 ± 0.57 9.29 ± 0.50 9.25 ± 0.49 7.58 ± 2.6 0.004
LS-BMD (L1–L4), g/cm2 1.18 ± 0.25 1.23 ± 0.19 1.33 ± 0.22 1.21 ± 0.13 0.190
LS-TBS (L1–L4), unitless 1.30 ± 0.13 1.22 ± 0.12 1.28 ± 0.12 1.18 ± 0.06 0.068

With diabetes and

Control group Taking steroids With diabetes taking steroids
(N = 257) (GC) (N = 62) (DM) (N = 74) (GC + DM) (N = 11) p

Age, yr 68.00 ± 7.09 68.29 ± 9.26 69.72 ± 8.09 70.27 ± 9.88 0.321
Body mass index, kg/m2 25.84 ± 5.17 27.38 ± 5.25 28.31 ± 4.92 25.14 ± 5.19 0.001
Fasting plasma glucose, mg/dL 94.17 ± 12.68 102.97 ± 23.69 142.21 ± 61.10 175.73 ± 107.32 <0.001
HbA1c, % 5.70 ± 0.38 6.31 ± 0.75 6.71 ± 1.20 6.52 ± 0.84 <0.001
Est. glomerular filtration rate, 75.32 ± 18.45 71.41 ± 24.53 66.70 ± 23.01 66.91 ± 19.23 0.015
mL/min/1.73 m2
Vitamin D, ng/mL 35.85 ± 16.74 35.97 ± 13.43 33.64 ± 14.46 31.83 ± 13.94 0.707
Calcium, mg/dL 9.37 ± 0.69 9.42 ± 0.52 9.42 ± 0.50 9.23 ± 0.79 0.769
LS-BMD (L1–L4), g/cm2 1.09 ± 0.19 1.05 ± 0.18 1.12 ± 0.16 0.99 ± 0.18 0.028
LS-TBS (L1–L4), unitless 1.28 ± 0.11 1.22 ± 0.13 1.20 ± 0.11 1.23 ± 0.22 <0.001

Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health Volume 21, 2018

Downloaded for Budi Lakukua (googledotbudi@yahoo.co.id) at Universitas Tarumanagara from ClinicalKey.com by Elsevier on October 01, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
188 Xue et al.

Fig. 1. Lumbar spine BMD mean ± SD for men and women in the control group, GC group, DM group, and GC + DM
group.

that, in either men or women, the diagnosis of diabetes was
associated with a higher BMI, higher fasting plasma glucose
level, higher HbA1c, and lower estimated glomerular fil-
tration rate. For women, subjects in the DM group had the
highest LS-BMD but the lowest TBS. In men, subjects in
the DM group had the highest LS-BMD, and subjects in
the GC + DM group had the lowest TBS. For all subjects,
TBS was negatively correlated to HbA1c (Pearson corre-
lation coefficient −0.121, p = 0.047) and serum creatinine
(Pearson correlation coefficient −0.130, p = 0.007).
One hundred thirty-two patients (27.7%) had at least
1 diagnosed rheumatoid disease, which included the fol-
lowing: rheumatoid arthritis, systemic lupus erythemato-
sus, temporal arteritis, Graves disease, psoriatic arthritis,
Sjogren syndrome, Raynaud disease, systemic sclerosis,
CREST syndrome, Churg-Strauss vasculitis, ankylosing
spondylitis, or fibromyalgia rheumatica. Ninety-five sub-
jects (19.9%) had a history of systemic corticosteroid use
with a minimum duration of at least 3 months. The mean
duration of corticosteroid use was 33.78 ± 23.15 months. Sig-
nificantly more men (30.1%) than women (18.1%) had a
history of corticosteroid use (p = 0.018), but there was no
difference in the percentage of subjects with a history of
corticosteroid use when compared by diabetes status
(p = 0.092).
The average LS-BMD and TBS scores by group and
separated by gender are shown in Figs. 1 and 2. For either
men or women, LS-BMD was higher in the DM group than
in any other group, but this difference was significant only
in women when the DM group was compared with the
GC + DM group (p = 0.028). TBS was not significantly dif-
ferent among the 4 groups for men, but the women in the
DM group were significantly lower than the female control
group (p < 0.001).
The study subjects were stratified with aforemen-
tioned criteria for both LS-BMD and TBS, and the distri-
bution for each of the 4 groups is depicted in Fig. 3. For
both men and women, there were very few subjects clas-
sified as osteoporotic by LS-BMD—most subjects were clas-
sified as having normal LS-BMD. For all groups, there were
more subjects stratified into the degraded microarchitecture
(lowest category) group by TBS than were classified as

Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health Volume 21, 2018

Downloaded for Budi Lakukua (googledotbudi@yahoo.co.id) at Universitas Tarumanagara from ClinicalKey.com by Elsevier on October 01, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
Lumbar Spine TBS Reflects Diminished Bone Quality 189

Fig. 2. Lumbar spine TBS mean ± SD for men and women in the control group, GC group, DM group, and GC + DM
group.

osteoporotic (lowest category) by LS-BMD. These differ- Discussion

ences were smallest in the control and most apparent for
subjects with diabetes (those in the DM and GC + DM
groups). Kendall tau-b statistic showed that the distribu-
tion into LS-BMD and TBS categories were weakly but sig-
nificantly correlated to the control group (correlation 0.144,
p = 0.005) and the GC group (correlation 0.314, p = 0.001).
LS-BMD and TBS categorization were not correlated for
the DM (p = 0.384) or the GC + DM (p = 0.733) groups,
which may indicate that the 2 measures are independent
of each other for subjects with diabetes.
The discriminatory ability of LS-BMD or TBS to dis-
tinguish subjects taking GC or who have diabetes is shown
in Table 2 as area under the ROC curves. Both LS-BMD
and TBS were able to discriminate subjects in the GC group
compared with controls, but LS-BMD would be consid-
ered a failing discriminatory test (AUC < 0.6). For all 3
groups of subjects, TBS provided better ability to discrimi-
nate between the study groups and the control group than
LS-BMD, with AUCs significantly higher than 0.5 (between
0.6 and 0.7).
In our study population consisting of men and women
age 55 years or older, diabetic patients on average had a
higher BMD and lower TBS compared with nondiabetic
controls. The percentage of subjects with normal BMD was
increased in diabetic groups compared with the control
group, whereas when subjects were categorized accord-
ing to their TBS score, the diabetic group demonstrated a
higher percentage of subjects with degraded microarchi-
tecture compared with the control group. The opposing
effect of diabetes on BMD and TBS demonstrated in this
study is consistent with other population studies in litera-
ture (9,11,12). Some studies have demonstrated that BMI
is positively associated with BMD score (10,24), and as the
subjects with diabetes had a higher BMI, this may par-
tially account for the increased BMD seen in subjects with
diabetes. We also found significant, positive correlations
between BMI and LS-BMD in all groups except in the GC
group. In our dataset, TBS was either negatively corre-
lated to BMI (control group) or not correlated (GC, DM,

Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health Volume 21, 2018

Downloaded for Budi Lakukua (googledotbudi@yahoo.co.id) at Universitas Tarumanagara from ClinicalKey.com by Elsevier on October 01, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
190 Xue et al.

Normal Osteopenic or Paritally Degraded Osteoporotic or Degraded Architecture

0% 0% 0%
11.8%
20.6% 17.6%
25.0%
35.3% 33.3% 33.3%
50.0%
32.4%

47.1%
29.4%
82.4% 50.0%
75.0%
66.7%
55.9%
50.0%
32.4% 35.3%
16.7%
0%
TBS LS-BMD TBS LS-BMD TBS LS-BMD TBS LS-BMD
CONTROL DM GC DM+GC
MALES

Normal Osteopenic or Paritally Degraded Osteoporotic or Degraded Architecture

2.7%
11.3% 14.5%
21.0%
27.3%
33.8% 38.7%
51.4% 54.5%
38.1%
48.4%
53.7% 36.4%

46.8%
36.5% 63.5% 27.3%
50.6%
37.1% 36.4%
25.3%
14.5% 18.2%
12.2%

TBS LS-BMD TBS LS-BMD TBS LS-BMD TBS LS-BMD

CONTROL DM GC DM+GC
FEMALES

Fig. 3. Distributions of TBS and DXA categories in men (top) and women (bottom).

Table 2
The Ability of LS-BMD or TBS to Discriminate Between Healthy Controls and Subjects With Diabetes, Subjects
Taking Corticosteroids, or Subjects With Diabetes and Taking Corticosteroids

LS-BMD TBS
Study group AUC p value 95% CI AUC p value 95% CI
Control vs GC 0.519 0.005 0.446–0.593 0.645 0.005 0.575–0.716
Control vs DM 0.597 0.595 0.533–0.661 0.668 <0.001 0.602–0.734
Control vs GC + DM 0.557 0.457 0.405–0.709 0.697 0.010 0.532–0.863
The data shown are receiver operating characteristic (ROC) analysis area under the curve (AUC) with 95% confidence intervals.

Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health Volume 21, 2018

Downloaded for Budi Lakukua (googledotbudi@yahoo.co.id) at Universitas Tarumanagara from ClinicalKey.com by Elsevier on October 01, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
Lumbar Spine TBS Reflects Diminished Bone Quality
GC + DM groups), which may indicate that TBS is inde-
pendent of BMI.
Despite having normal or higher BMD, patients with dia-
betes have increased risk of fracture (9,11). The mecha-
nism through which diabetes influences fracture risk is likely
multifactorial. Extraskeletal factors such as vision distur-
bance and peripheral neuropathy may play a role through
increasing the fall risk (25). However, even after adjust-
ing for fall frequency, the fracture risk in patients with dia-
betes remains higher than in the control group (13,26). Our
result shows that lumbar spine TBS is significantly lower
in patients with diabetes, which suggests that diabetes may
have a direct adverse effect on bone integrity. TBS, an
overall score computed from gray textures of 2D DXA
images, provides an estimate of the 3-dimensional trabecu-
lar microstructure and global bone quality (21). Low TBS
can identify poor bone strength that is not otherwise cap-
tured by DXA, and may explain the high fracture risk seen
in patients with diabetes (17).
Leslie et al (11) demonstrated in their Canadian cohort
study that lumbar spine TBS was a BMD-independent pre-
dictor of fracture risk in older diabetic women. Although
we did not assess fracture risk directly, our current study
shows TBS is related to metabolic risk factors for frac-
ture in both men and women. In our subjects, low TBS was
associated with high HbA1c and serum creatinine level. Kim
et al (12) had shown similar results with their cross-
sectional study of Korean men and women. They found that
a high blood glucose level and HbA1c were associated with
low TBS.
The deposition of advanced glycosylation end-products
in bone collagen may directly alter its physical properties
(26) including increased brittleness (27). The deteriora-
tion of bone strength in subjects with diabetes has been
shown to be associated with accumulation of advanced
glycosylation end-products in bone collagen (28). In addi-
tion, other possible mechanisms may contribute to the loss
of bone strength in diabetes. These include hyperglyce-
mia, weight loss, insulin deficiency or resistance,
hypercalciuria, impaired renal function, and microvascu-
lar complications (26). Further work is needed to clarify
exactly which properties in diabetes affect lumbar spine
TBS.
The use of GCs is known to exert significant deleteri-
ous effects on bone quality, and GC-induced osteoporosis
is the most common cause of secondary osteoporosis (29).
GCs have been shown to inhibit bone formation through
increased apoptosis of osteoblasts and osteocytes, and
promote bone resorption by prolonging the lifespan of os-
teoclasts (30). The adverse effects are dose dependent and
rapid in onset (often can be observed 3–6 months after the
start of GC therapy) (31). Previous studies have found that
BMD alone is not effective in identifying bone deteriora-
tion in patients undergoing GC therapy. TBS or TBS com-
bined with BMD has been shown to be more useful tools
in detecting microstructural skeletal changes in such patient
group (23,32,33). This is consistent with the results of our
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Downloaded for Budi Lakukua (googledotbudi@yahoo.co.id) at Universitas Tarumanagara from ClinicalKey.com by Elsevier on October 01, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
191
study. Using ROC analysis, TBS was more effective in dis-
tinguishing between patients under GC therapy and con-
trols than BMD (AUC 0.645 vs 0.519 respectively, p = 0.005
for both). TBS retains its discriminatory ability when com-
paring the group of patients with diabetes with GC therapy
with the control group, whereas BMD is not able to dis-
tinguish subjects in these 2 groups.
There are limitations to our study. We were unable to
distinguish patients with type 1 from those with type 2
DM, and therefore were not able to account for the dif-
ferences between them. Given the age of our cohort, it is
likely that the vast majority of subjects in the diabetic group
have type 2 diabetes. We did not have information con-
cerning the onset, duration and severity of diabetes, or
the use of diabetic medications. Particularly, the use of
thiazolidinediones have recently been associated with higher
fracture risk in women (34). A randomized trial enrolling
healthy postmenopausal women has linked rosiglitazone
use to more rapid bone loss (35). In this study, bone turn-
over markers were available only for a few subjects; we
suggest that future studies should include these markers
to further assess the effects of diabetes on bone metabo-
lism. Finally, as a cross-sectional study, we did not link our
study results to future fracture outcomes.
Conclusion
Lumbar spine TBS is a BMD-independent parameter
that is capable of detecting microstructural skeletal dete-
riorations in patients with type 1 and type 2 DM. In our
analysis, TBS appears to better discriminate patients with
diabetes who are concomitantly undergoing GC therapy
from control population than BMD. Its ability to detect
changes in bone microarchitecture may be a valuable tool
for the clinical management of osteoporosis, and serve par-
ticular importance in treatment monitoring for patients with
diabetes and GC use.
References
1. 1993 Consensus development conference: diagnosis, prophy-
laxis, and treatment of osteoporosis. Am J Med 94(6):646–
650.
2. Office of the Surgeon General. 2004 Reports of the Surgeon
General. In: Bone Health and Osteoporosis: A Report of the
Surgeon General. Rockville (MD): Office of the Surgeon
General (US).
3. Keen RW. 2003 Burden of osteoporosis and fractures. Curr
Osteoporos Rep 1(2):66–70.
4. Burge R, Dawson-Hughes B, Solomon DH, et al. 2007 Inci-
dence and economic burden of osteoporosis-related frac-
tures in the United States, 2005–2025. J Bone Miner Res
22(3):465–475.
5. Vaile J, Sullivan L, Bennett C, Bleasel J. 2007 First Fracture
Project: addressing the osteoporosis care gap. Intern Med J
37(10):717–720.
6. Vasikaran S, Eastell R, Bruyère O, et al. 2011 Markers of bone
turnover for the prediction of fracture risk and monitoring
Volume 21, 2018
192
of osteoporosis treatment: a need for international refer-
ence standards. Osteoporos Int 22(2):391–420.
7. Tosteson AN, Melton LJ, Dawson-Hughes B, et al. 2008 Cost-
effective osteoporosis treatment thresholds: the United States
perspective. Osteoporos Int 19(4):437–447.
8. Kanis JA, Borgstrom F, De Laet C, et al. 2005 Assessment
of fracture risk. Osteoporos Int 16(6):581–589.
9. Melton LJ, Leibson CL, Achenbach SJ, et al. 2008 Fracture
risk in type 2 diabetes: update of a population-based study.
J Bone Miner Res 23(8):1334–1342.
10. Schwartz AV, Sellmeyer DE, Ensrud KE, et al. 2001 Older
women with diabetes have an increased risk of fracture: a pro-
spective study. J Clin Endocrinol Metab 86(1):32–38.
11. Leslie WD, Aubry-Rozier B, Lamy O, et al. 2013 TBS (tra-
becular bone score) and diabetes-related fracture risk. J Clin
Endocrinol Metab 98(2):602–609.
12. Kim JH, Choi HJ, Ku EJ, et al. 2015 Trabecular bone score
as an indicator for skeletal deterioration in diabetes. J Clin
Endocrinol Metab 100(2):475–482.
13. Strotmeyer ES, Cauley JA, Schwartz AV, et al. 2004 Diabe-
tes is associated independently of body composition with BMD
and bone volume in older white and black men and women:
the health, aging, and body composition study. J Bone Miner
Res 19(7):1084–1091.
14. Seeman E, Delmas PD. 2006 Bone quality—the material and
structural basis of bone strength and fragility. N Engl J Med
354(21):2250–2261.
15. Giangregorio LM, Leslie WD, Lix LM, et al. 2012 FRAX un-
derestimates fracture risk in patients with diabetes. J Bone
Miner Res 27(2):301–308.
16. Schwartz AV, Vittinghoff E, Bauer DC, et al. 2011 Associa-
tion of BMD and FRAX score with risk of fracture in older
adults with type 2 diabetes. JAMA 305(21):2184–2192.
17. Hans D, Barthe N, Boutroy S, et al. 2011 Correlations between
trabecular bone score, measured using anteroposterior dual-
energy X-ray absorptiometry acquisition, and 3-dimensional
parameters of bone microarchitecture: an experimental study
on human cadaver vertebrae. J Clin Densitom 14(3):302–
312.
18. Boutroy S, Bouxsein ML, Munoz F, Delmas PD. 2005 In vivo
assessment of trabecular bone microarchitecture by high-
resolution peripheral quantitative computed tomography. J
Clin Endocrinol Metab 90(12):6508–6515.
19. Genant HK, Engelke K, Prevrhal S. 2008 Advanced CT bone
imaging in osteoporosis. Rheumatology (Oxford) 47(Suppl
4):iv9–iv16.
20. Krug R, Carballido-Gamio J, Banerjee S, et al. 2008 In vivo
ultra-high-field magnetic resonance imaging of trabecular bone
microarchitecture at 7 T. J Magn Reson Imaging 27(4):854–
859.
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Downloaded for Budi Lakukua (googledotbudi@yahoo.co.id) at Universitas Tarumanagara from ClinicalKey.com by Elsevier on October 01, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
Xue et al.
21. Silva BC, Leslie WD, Resch H, et al. 2014 Trabecular bone
score: a noninvasive analytical method based upon the DXA
image. J Bone Miner Res 29(3):518–530.
22. Piveteau T, Winzenrieth R, Hans D. 2011 Trabecular bone
score (TBS) the new parameter of 2D texture analysis for the
evaluation of 3D bone micro architecture status. Bone
48:S176.
23. Leib ES, Winzenrieth R. 2016 Bone status in glucocorticoid-
treated men and women. Osteoporos Int 27(1):39–48.
24. De Laet C, Kanis JA, Odén A, et al. 2005 Body mass index
as a predictor of fracture risk: a meta-analysis. Osteoporos
Int 16(11):1330–1338.
25. Strotmeyer ES, Cauley JA, Schwartz AV, et al. 2005
Nontraumatic fracture risk with diabetes mellitus and im-
paired fasting glucose in older white and black adults: the
health, aging, and body composition study. Arch Intern Med
165(14):1612–1617.
26. Schwartz AV, Sellmeyer DE. 2007 Diabetes, fracture, and bone
fragility. Curr Osteoporos Rep 5(3):105–111.
27. Hernandez CJ, Tang SY, Baumbach BM, et al. 2005 Trabecu-
lar microfracture and the influence of pyridinium and non-
enzymatic glycation-mediated collagen cross-links. Bone
37(6):825–832.
28. Schwartz AV, Garnero P, Hillier TA, et al. 2009 Pentosidine
and increased fracture risk in older adults with type 2 dia-
betes. J Clin Endocrinol Metab 94(7):2380–2386.
29. Lekamwasam S, Adachi JD, Agnusdei D, et al. 2012 A frame-
work for the development of guidelines for the manage-
ment of glucocorticoid-induced osteoporosis. Osteoporos Int
23(9):2257–2276.
30. Spreafico A, Frediani B, Francucci CM, et al. 2008 Role of
apoptosis in osteoporosis induced by glucocorticoids.
J Endocrinol Invest 31(7 Suppl):22–27.
31. van Staa TP, Leufkens HG, Abenhaim L, et al. 2000 Oral cor-
ticosteroids and fracture risk: relationship to daily and cu-
mulative doses. Rheumatology (Oxford) 39(12):1383–1389.
32. Paggiosi MA, Peel NF, Eastell R. 2015 The impact of gluco-
corticoid therapy on trabecular bone score in older women.
Osteoporos Int 26(6):1773–1780.
33. Saag KG, Agnusdei D, Hans D, et al. 2016 Trabecular bone
score in patients with chronic glucocorticoid therapy-induced
osteoporosis treated with alendronate or teriparatide. Ar-
thritis Rheumatol 68(9):2122–2128.
34. Short R. 2007 Fracture risk a class effect of thiazolidinediones
in women. BMJ 334(7593):551.
35. Grey A, Bolland M, Gamble G, et al. 2007 The peroxisome
proliferator-activated receptor-gamma agonist rosiglitazone
decreases bone formation and bone mineral density in healthy
postmenopausal women: a randomized, controlled trial. J Clin
Endocrinol Metab 92(4):1305–1310.
Volume 21, 2018