Content uploaded by Corey M Gill
Author content
All content in this area was uploaded by Corey M Gill on Sep 29, 2015
Content may be subject to copyright.
ORIGINAL RESEARCH
n
Musculoskeletal IMagIng
534 radiology.rsna.org
n
Radiology: Volume 269: Number 2—November 2013
Ectopic and Serum Lipid Levels
Are Positively Associated with
Bone Marrow Fat in Obesity
1
Miriam A. Bredella, MD
Corey M. Gill, BS
Anu V. Gerweck, NP
Melissa G. Landa, BA
Vidhya Kumar, MA
Scott M. Daley, BA
Martin Torriani, MD
Karen K. Miller, MD
Purpose:
To investigate the associations between ectopic and serum
lipid levels and bone marrow fat, as a marker of stem cell
differentiation, in young obese men and women, with the
hypothesis that ectopic and serum lipid levels would be
positively associated with bone marrow fat.
Materials and
Methods:
The study was institutional review board approved and
complied with HIPAA guidelines. Written informed con-
sent was obtained. The study group comprised 106 healthy
young men and women (mean age, 33.7 years 6 6.8 [stan-
dard deviation]; range, 19–45 years; mean body mass in-
dex (BMI), 33.1 kg/m
2
6 7.1; range, 18.1–48.8 kg/m
2
)
who underwent hydrogen 1(
1
H) magnetic resonance (MR)
spectroscopy by using a point-resolved spatially localized
spectroscopy sequence at 3.0 T of L4 for bone marrow
fat content, of soleus muscle for intramyocellular lipids
(IMCL), and liver for intrahepatic lipids (IHL), serum cho-
lesterol level, serum triglyceride level, and measures of
insulin resistance (IR). Exercise status was assessed with
the Paffenbarger activity questionnaire.
Results:
There was a positive correlation between bone marrow
fat and IHL (r = 0.21, P = .048), IMCL (r = 0.27, P = .02),
and serum triglyceride level (r = 0.33, P = .001), inde-
pendent of BMI, age, IR, and exercise status (P , .05).
High-density lipoprotein cholesterol levels were inversely
associated with bone marrow fat content, independent of
BMI, age, IR, and exercise status (r = 20.21, P = .019).
Conclusion:
Results of this study suggest that ectopic and serum lipid
levels are positively associated with bone marrow fat in
obese men and women.
q
RSNA, 2013
1
From the Department of Radiology (M.A.B., C.M.G., V.K.,
S.M.D., M.T.) and Neuroendocrine Unit (A.V.G., M.G.L.,
K.K.M.), Massachusetts General Hospital and Harvard
Medical School, Yawkey 6E, 55 Fruit St, Boston, MA 02114.
Received February 12, 2013; revision requested March
26; final revision received April 9; accepted April 26; final
version accepted April 30. Address correspondence to
M.A.B. (e-mail: mbredella@partners.org).
q
RSNA, 2013
Note: This copy is for your personal non-commercial use only. To order presentation-ready
copies for distribution to your colleagues or clients, contact us at www.rsna.org/rsnarights.
Radiology: Volume 269: Number 2—November 2013
n
radiology.rsna.org 535
MUSCULOSKELETAL IMAGING: Lipid Levels and Bone Marrow Fat in Obesity Bredella et al
selective serotonin reuptake inhibitors,
and proton pump inhibitors; and con-
traindications to magnetic resonance
(MR) imaging, such as the presence of
a pacemaker or metallic implant. None
of the patients had a history of liver
disease.
Each participant underwent pro-
ton MR spectroscopy (hydrogen 1 [
1
H]
MR spectroscopy) for quantification of
bone marrow fat content, IMCL, and
IHL. All subjects underwent a 75-g,
2-hour oral glucose tolerance test; and
laboratory tests included fasting tri-
glyceride levels and total, high-density
lipoprotein (HDL), and low-density
lipoprotein (LDL) cholesterol levels.
Level of activity, including exercise,
was assessed by using the Paffenbarg-
er questionnaire, a self-administered
questionnaire that measures current
levels of activity.
Clinical characteristics, ectopic
lipid levels, and bone marrow fat data
have been previously reported in 50
women (12–17), and bone marrow fat
Obesity also influences stem cell dif-
ferentiation into the adipocyte lineage
at the expense of osteoblastogenesis
(11), and concordant with this factor,
we previously demonstrated an inverse
association between bone marrow fat
and BMD and a positive association
between bone marrow fat and visceral
fat in obese premenopausal men and
women (3,12). However, the associa-
tions between intrahepatic lipids (IHL),
IMCL, serum lipids, and bone marrow
fat have not been studied.
The purpose of our study was to
investigate the associations between ec-
topic lipid levels, serum lipid levels, and
bone marrow fat, as a marker of stem
cell differentiation, in young, obese but
otherwise healthy men and women. We
hypothesized that ectopic and serum
lipid levels would be positively associ-
ated with bone marrow fat.
Materials and Methods
The study was approved by our insti-
tutional review board and complied
with Health Insurance Portability and
Accountability Act guidelines. Written
informed consent was obtained from all
subjects after the nature of the proce-
dures had been fully explained.
Subjects
The study group comprised 106
healthy premenopausal women and
men of similar mean age who were
recruited from the community through
advertisements. Inclusion criteria
were an age that ranged from 18 to
45 years and eumenorrhea in women.
Exclusion criteria included hypotha-
lamic or pituitary disorders, diabetes
mellitus, or other chronic illnesses;
smoking; use of osteoporosis medica-
tion or medication that could influence
bone metabolism, such as estrogen
or glucocorticoids, anticonvulsants,
A
lthough obesity is traditionally
viewed as protective against os-
teoporosis, recent studies have
linked obesity to osteoporosis and
increased fracture risk (1,2). It has
been suggested that specific fat com-
partments have differential effects on
bone mineral density (BMD), with
visceral fat exerting potential detri-
mental effects on bone health (3–5).
In addition, the metabolic syndrome,
which includes abdominal obesity, dys-
lipidemia, and impaired glucose toler-
ance, among other cardiovascular risk
factors, is associated with osteoporotic
fractures (6).
The pathophysiologic basis of
obesity-associated bone loss is in-
completely understood. Lipids and
lipoproteins are emerging as impor-
tant regulators of skeletal physiologic
characteristics and have been shown
to inhibit osteoblast and to enhance
osteoclast differentiation and survival
(7,8). Findings in a recent study (9) in
postmenopausal women have revealed
decreased BMD in subjects with non-
alcoholic fatty liver disease, which is
common in patients with the metabolic
syndrome and obesity, suggesting an
etiologic relationship between liver fat
and bone loss. Intramyocellular lipids
(IMCL) are ectopic lipids deposited
in skeletal muscle that are increased
in obesity and type 2 diabetes melli-
tus, and accumulation of IMCL is im-
plicated as an etiologic factor in the
development of muscle insulin resis-
tance (10). However, the relationship
between IMCL and bone has not been
studied.
Implication for Patient Care
n
Increased ectopic and serum lipid
levels may be detrimental to
bone, and
1
H MR spectroscopy
can be used to identify patients
at risk for bone loss.
Advances in Knowledge
n
Mean intrahepatic lipids (IHL)
and intramyocellular lipids
(IMCL), as well as mean serum
triglyceride levels, are on average
significantly positively associated
with bone marrow fat in obese
men and women.
n Significant positive correlation
between mean IHL and IMCL
with bone marrow fat is indepen-
dent of insulin resistance and
exercise status.
Published online before print
10.1148/radiol.13130375 Content code:
Radiology 2013; 269:534–541
Abbreviations:
BMD = bone mineral density
BMI = body mass index
CRLB = Cramer-Rao lower bounds
HDL = high-density lipoprotein
IHL = intrahepatic lipids
IMCL = intramyocellular lipids
IR = insulin resistance
LDL = low-density lipoprotein
Author contributions:
Guarantors of integrity of entire study, M.A.B., K.K.M.;
study concepts/study design or data acquisition or data
analysis/interpretation, all authors; manuscript drafting or
manuscript revision for important intellectual content, all
authors; approval of final version of submitted manuscript,
all authors; literature research, M.A.B., C.M.G., M.G.L.,
S.M.D., K.K.M.; clinical studies, M.A.B., C.M.G., A.V.G.,
M.G.L., S.M.D., K.K.M.; experimental studies, M.G.L.,
S.M.D., M.T., K.K.M.; statistical analysis, M.A.B., C.M.G.,
M.G.L.; and manuscript editing, M.A.B., C.M.G., M.G.L.,
S.M.D., M.T., K.K.M.
Funding:
This research was supported by the National Institutes of
Health (grants R01 HL-077674, UL1 RR-025758, and K23
RR-23090).
Conflicts of interest are listed at the end of this article.
536 radiology.rsna.org
n
Radiology: Volume 269: Number 2—November 2013
MUSCULOSKELETAL IMAGING: Lipid Levels and Bone Marrow Fat in Obesity Bredella et al
1 mm; matrix, 512; number of signals
acquired, one; and field of view, 22 cm.
A voxel measuring 15 3 15 3 15 mm
3
(3.4 cm
3
) was placed on the axial T1-
weighted section with largest cross-sec-
tional area of soleus muscle, avoiding
visible interstitial tissue, fat, or vessels.
Single-voxel
1
H MR spectroscopy data
were acquired by using a point-resolved
spatially localized spectroscopy pulse
sequence, with the following parame-
ters: 3000/30; number of acquisitions,
64; data points, 1024; and receiver
bandwidth, 1000 Hz. Frequency-selec-
tive water signal suppression (chemi-
cal shift-selective, or CHESS, water
suppression) was used for metabolite
acquisition (bandwidth for water sup-
pression, 50 Hz), and unsuppressed
water spectra of the same voxel were
obtained for each image, with the same
parameters as were used for the metab-
olite acquisition, except for the use of
eight acquisitions. For each voxel place-
ment, automated optimization of gradi-
ent shimming, water suppression, and
transmit-receive gain were performed,
followed by manual adjustment of gra-
dient shimming, targeting water line
widths of 12–14 Hz.
1
H MR Spectroscopy Data Analysis
Fitting of all
1
H MR spectroscopy data
was performed by using LCModel (ver-
sion 6.3–0 K; Stephen Provencher,
Oakville, Ontario, Canada) (18). Data
were transferred from the imaging unit
to a Linux workstation (Canonical, Lex-
ington, Mass), and metabolite quanti-
fication was performed by using eddy
current correction and water scaling.
Fitting algorithms specific for bone
marrow and liver lipid estimates were
scaled to unsuppressed water peak (4.7
ppm) and expressed as lipid-to-water
ratio. For liver and marrow analysis,
the lipid peaks included resonances at
0.9, 1.3, and 2.0 ppm. For soleus mus-
cle, IMCL (1.3 ppm) and extramyocellu-
lar lipid (1.5 ppm) methylene estimates
and a combined estimate for lipid peaks
(total muscle lipid levels) (0.9, 1.1, 1.3,
1.5, 2.1, and 2.3 ppm) were automati-
cally scaled to unsuppressed water peak
(4.7 ppm) and expressed as lipid-to-wa-
ter ratio. The jMRUI software package
the pulse durations for the used voxels,
the calculated chemical shift was 3.6%
per parts per million in the excitation
direction and 10.6% per parts per mil-
lion in the refocusing direction. Careful
positioning of the voxel took into ac-
count the fact that the shifted fat-water
voxels would still contain the same tis-
sue types. For each voxel placement,
automated optimization of gradient
shimming was performed.
1
H MR spectroscopy of the liver.—
For
1
H MR spectroscopy of the liver,
subjects were positioned supine and
feet first in the magnet bore. A body
matrix phased-array coil was posi-
tioned over the abdomen. A triplane
gradient-echo localizer pulse sequence
was performed in the abdomen to
localize the liver, with the following
parameters: 15/5; section thickness,
3 mm. Subsequently, a breath-hold
true fast imaging with steady-state
precession sequence in the liver was
performed, with the following param-
eters: 3.8/1.9; section thickness, 10
mm; and imaging time, 11 seconds. A
voxel measuring 20 3 20 3 20 mm
3
(8
cm
3
) was placed within the right he-
patic lobe, avoiding vessels or artifacts.
For each voxel placement, automated
optimization of gradient shimming was
performed. Single breath-hold single-
voxel
1
H MR spectroscopy data were
acquired by using a point-resolved
spatially localized spectroscopy pulse
sequence without water suppres-
sion, with the following parameters:
1500/30; number of acquisitions,
eight; number of data points, 1024;
and receiver bandwidth, 2000 Hz.
1
H MR spectroscopy of soleus mus-
cle.—For
1
H MR spectroscopy of so-
leus muscle, subjects were positioned
feet first in the magnet bore, and the
right calf was placed in a commercially
available transmit-receive quadrature
extremity coil with an 18-cm diameter
(USA Instruments, Aurora, Ohio). A
triplane gradient-echo localizer pulse
sequence was performed, with the
following parameter: 15/5. Axial T1-
weighted images of the proximal two-
thirds of the calf were obtained, with
the following parameters: 400/11; sec-
tion thickness, 4 mm; intersection gap,
data have been reported in 35 men (3).
However, none of these have been re-
ported in the entire cohort, and the
associations between ectopic lipids, se-
rum lipid levels, and bone marrow fat
have not been described in any of the
subjects.
Proton MR Spectroscopy (
1
H MR
Spectroscopy)
All subjects underwent
1
H MR spec-
troscopy of the fourth lumbar verte-
bral body to determine bone marrow
lipid content, of soleus muscle to de-
termine IMCL and total lipid levels,
and of the liver to determine IHL.
1
H
MR spectroscopy studies and analyses
were performed with the supervision
of two musculoskeletal radiologists with
8 years of experience (M.A.B.) and
16 years of experience (M.T.). Stud-
ies were performed after an 8-hour
overnight fast. All studies were per-
formed by using a 3.0-T MR imaging
system (Siemens Trio; Siemens Medical
Systems, Erlangen, Germany). Coeffi-
cient of variation for measurements at
our institution is 3% for bone marrow
fat quantification, 6% for intramuscular
fat quantification, and 8% for intrahe-
patic fat quantification.
1
H MR spectroscopy of bone mar-
row.—For L4 marrow
1
H MR spec-
troscopy, subjects were positioned feet
first in the magnet bore in the prone
position. A body matrix phased-array
coil was positioned over the lumbar
region. A triplane gradient-echo local-
izer pulse sequence of the lumbar spine
was performed to localize the L4 ver-
tebral body, with the following param-
eters: repetition time msec/echo time
msec, 15/5; section thickness, 3 mm.
A voxel measuring 15 3 15 3 15 mm
3
(3.4 cm
3
) was placed within the L4 ver-
tebral body. Single-voxel
1
H MR spec-
troscopy data were acquired by using a
point-resolved spatially localized spec-
troscopy pulse sequence without water
suppression and the following parame-
ters: 3000/30, eight acquisitions, 1024
data points, and receiver bandwidth of
2 KHz. The bandwidth of the 90° exci-
tation pulse was 3.24 KHz, and that of
the 180° refocusing pulse was 1.2 KHz.
On the basis of reasonable estimates of
Radiology: Volume 269: Number 2—November 2013
n
radiology.rsna.org 537
MUSCULOSKELETAL IMAGING: Lipid Levels and Bone Marrow Fat in Obesity Bredella et al
was 4.8 hours per week 6 5.9. None of
the subjects were trained athletes.
Obese subjects had higher serum
lipid levels, lower HDL cholesterol, and
higher IMCL, total muscle lipid content,
IHL, and marrow fat (trend) compared
with data in normal-weight control sub-
jects (Table 2).
Effects of Ectopic and Serum Lipid Levels
on Bone Marrow Fat
All spectra of bone marrow fat, liver,
and soleus muscle were of diagnostic
quality and were used for analyses.
Mean signal-to-noise ratio and mean
full width at half maximum for marrow
spectra were 224 Hz 6 108 and 49 Hz
6 10, respectively and for liver spec-
tra were 201 Hz 6 62 and 30 Hz 6 11,
respectively. Mean CRLBs for marrow,
liver, and muscle spectra are provided
in Table 1.
to denote a trend. Data are presented
as means 6 standard deviation.
Results
Clinical Characteristics of Study Subjects
Subject characteristics are shown in
Table 1. The study group included 60
women and 46 men. The age of study
participants ranged from 19 to 45 years,
with a mean age of 33.7 years 6 6.8
(standard deviation). Women ranged in
age from 19 to 45 years, with a mean
age of 33.9 years 6 7.2. Men ranged in
age from 21 to 45 years, with a mean
age of 33.7 years 6 6.3. BMI of study
participants ranged from 18.1 to 48.8
kg/m
2
, with a mean BMI of 33.1 kg/m
2
6 7.1. Eighty-eight subjects were over-
weight or obese (BMI . 25 kg/m
2
) and
18 subjects were of normal weight (BMI
, 25 kg/m
2
). Average vigorous activity
(A. van den Boogaart, Katholieke Uni-
versiteit Leuven, Leuven, Belgium) was
used for measurement of line widths
(full width at half maximum) and signal
to noise ratio of unsuppressed water
peaks, analyzing each time domain di-
rectly from free induction decays. Full
width at half maximum was calculated
automatically by using the Hankel-
Lanczos single-variable decomposition
fitting algorithm. Signal-to-noise ra-
tio was determined by using the ratio
between maximal unsuppressed water
amplitude at 4.7 ppm measured with
Hankel-Lanczos single-variable decom-
position and noise root mean square
(standard deviation) extracted from the
final 10% of free induction decay (102
data points).
Spectra were plotted by using soft-
ware (iNMR 3.6.2; Nucleomatica/Mes-
trelab Research, Molfetta, Italy) after
postprocessing with a 1-Hz Gaussian fil-
ter and automatic baseline correction.
Statistical Analysis
Statistical database software (JMP,
version 5.0.1; SAS Institute, Cary,
NC) was used for statistical analyses.
Variables were tested for normality of
distribution by using the Shapiro-Wilk
test. Variables that were not normally
distributed were log transformed. Lin-
ear regression analysis was performed.
Because bone marrow fat increases
with age and our aim was to study body
mass index (BMI)-independent effects
of ectopic and serum lipid levels on
bone marrow fat, we used multivariate
standard least squares regression mod-
eling to control for BMI and age. As
IMCL increase with insulin resistance
and as IMCL can also be increased in
insulin-sensitive athletes, we controlled
for insulin resistance by using the
2-hour glucose level from the oral glu-
cose tolerance test and exercise status
(number of hours of vigorous exercise
per week) by using the Paffenbarger
questionnaire. Forward stepwise re-
gression modeling was also performed
to determine predictors of bone mar-
row fat. The normal-weight and obese
groups were compared by using the
Student t test. P .05 was used to de-
note significance, and P .1 was used
Table 1
Clinical Characteristics of All Study Subjects
Variable Subjects (n = 106)
Age (y) 33.7 6 6.8 (19–45)
Weight (kg) 97.8 6 26.0 (45.8–159.0)
BMI (kg/m
2
) 33.1 6 7.1 (18.1–48.8)
Vigorous activity (h/wk) 4.8 6 5.9 (0–32)
Triglyceride level (mg/dL)* 110.6 6 57.5 (32.0–275.0)
Cholesterol level (mg/dL)
†
Total 170.6 6 35.3 (81.0–280.0)
LDL 105.7 6 30.4 (48.0–191.0)
HDL 43.7 6 11.4 (16.0–72.0)
Glucose level (mg/dL)
‡
Fasting 85.2 6 7.3 (65.0–102.0)
2-hour 114.8 6 28.8 (53.0–215.0)
IHL
Lipid-water ratio 0.12 6 0.21 (0.0003–1.148)
CRLB (%) 7.5 6 7.6 (0–28)
IMCL soleus muscle
Lipid-water ratio 0.06 6 0.30 (0.0026–3.1100)
CRLB (%) 4.5 6 1.4 (0–8)
Total soleus muscle intramuscular lipids
Lipid-water ratio 0.17 6 0.63 (0.04–6.54)
CRLB (%) 0.25 6 0.46 (0–2)
Bone marrow fat
Lipid-water ratio 0.71 6 0.37 (0.26–2.43)
CRLB (%) 0.53 6 0.54 (0–2)
Note.—Data are means 6 standard deviations. Numbers in parentheses are ranges. CRLB = Cramer-Rao lower bounds.
* To convert to Système International units in millimoles per liter, multiply by 0.0113.
†
To convert to Système International units in millimoles per liter, multiply by 0.0259.
‡
To convert to Système International units in millimoles per liter, multiply by 0.0555.
538 radiology.rsna.org
n
Radiology: Volume 269: Number 2—November 2013
MUSCULOSKELETAL IMAGING: Lipid Levels and Bone Marrow Fat in Obesity Bredella et al
There was a positive correlation be-
tween log IHL and log bone marrow fat
(r = 0.21, P = .048) (Fig 1a), which re-
mained significant after controlling for
BMI, age, insulin resistance (IR), and
exercise status (P = .027). There was a
positive correlation between log IMCL
and log bone marrow fat (r = 0.24, P
= .022) (Fig 1b), which remained sig-
nificant after controlling for age, BMI,
IR, and exercise status (P = .05). There
was a positive correlation between log
total muscle lipid content and log bone
marrow fat (r = 0.25, P = .014) (Fig
1c), which remained significant after
controlling for IR (P = .02) but became
a trend after controlling for age and
BMI (P = .054) and exercise status (P
= .066). There were positive associa-
tions between IMCL and 2-hour glucose
levels (r = 0.21, P = .027). There were
no significant associations between fast-
ing or 2-hour glucose levels or exercise
status and bone marrow fat (P = .29
to P = .76). Figures 2 and 3 show rep-
resentative subjects with high and low
IHL, respectively, and corresponding
vertebral marrow fat spectra. Figures 4
and 5 show representative IMCL spectra
of subjects with high (Fig 4) and low (Fig
5) bone marrow fat content. Spectra for
Figures 2–5 were plotted by using soft-
ware (iNMR 3.6.2; Nucleomatica/Mes-
trelab Research).
There was a positive correlation be-
tween log serum triglyceride levels and
log bone marrow fat (r = 0.33, P = .001)
(Fig 1d), which remained significant af-
ter controlling for BMI, age, IR, and
exercise status (P = .008) and a trend
toward an inverse association between
HDL cholesterol level and bone marrow
fat (r = 20.20, P = .06), which became
significant after controlling for BMI,
age, IR, and exercise status (P = .019).
There was a positive correlation be-
tween bone marrow fat and age (r =
0.27. P = .008) and weight (r = 0.27, P
= .009), while there was no association
between bone marrow fat and BMI (P =
.153), total cholesterol level (P = .447),
and LDL cholesterol level (P = .202).
Predictors of Bone Marrow Fat
When bone marrow fat was entered as a
dependent variable in a forward stepwise
Table 2
Body Composition and Serum Lipid Levels in Obese and Normal-weight Control
Subjects
Variable Obese Subjects (n = 88)* Control Subjects (n = 18)* P Value
Triglyceride level (mg/dL)
†
117.7 6 57.7 74.2 6 41.3 .004
Cholesterol level (mg/dL)
‡
Total 174.1 6 35.1 152.6 6 31.7 .02
LDL 109.4 6 30.0 87.1 6 25.9 .005
HDL 42.3 6 11.1 50.8 6 10.3 .004
Log IHL
§
22.9 6 1.4 24.5 6 1.7 .0001
Log IMCL soleus muscle
§
23.5 6 0.09 24.3 6 0.19 ,.0001
Log total intramuscular lipid levels
§
22.2 6 0.62 22.8 6 0.28 ,.0001
Bone marrow fat
§
0.75 6 0.38 0.57 6 0.20 .07
* Data are the means 6 standard deviations.
†
To convert to Système International units in millimoles per liter, multiply by 0.0113.
‡
To convert to Système International units in millimoles per liter, multiply by 0.0259.
§
Values are lipid-water ratio.
Figure 1
Figure 1: (a–d) Nonadjusted regression analysis between bone marrow fat and ectopic and serum lipid
levels. There are positive correlations between bone marrow fat and (a) IHL, (b) IMCL, (c) total muscle lipid
levels, and (d) serum triglyceride levels.
Radiology: Volume 269: Number 2—November 2013
n
radiology.rsna.org 539
MUSCULOSKELETAL IMAGING: Lipid Levels and Bone Marrow Fat in Obesity Bredella et al
in bone marrow adipogenesis, inde-
pendent of obesity. Therefore, patients
with increased liver fat may be at risk
for low BMD, and
1
H MR spectroscopy
can be used to identify these patients at
risk for bone loss.
We observed a positive correlation
between vertebral bone marrow fat and
IMCL of soleus muscle. IMCL are in-
creased in subjects with obesity and are
thought to play an etiologic role in the
development of insulin resistance (10).
However, IMCL can also be increased
between IHL and BMD in postmeno-
pausal women with nonalcoholic fatty
liver disease (9). Potential mechanisms
for low BMD in patients with chronic
liver disease include decreased levels
of insulin-like growth factor 1 (25) and
detrimental effects of bilirubin level on
osteoblast proliferation (26). In our
healthy obese subjects without a his-
tory of liver disease, IHL were posi-
tively associated with bone marrow fat,
even after adjusting for BMI, age, and
IR, suggesting that IHL may be involved
regression model and IHL, IMCL, triglyc-
eride levels, and BMI were entered as in-
dependent variables, serum triglyceride
levels were significant predictors of bone
marrow fat. Triglyceride levels explained
10% (r
2
= 0.10, P = .003) of the variability
of bone marrow fat.
Discussion
Our study in obese young men and
women showed positive associations
between IHL, IMCL, serum lipids, and
bone marrow fat, a marker of stem cell
differentiation, independent of BMI,
age, IR, and exercise status. Further-
more, we demonstrated an inverse as-
sociation between HDL cholesterol level
and bone marrow fat.
Accumulation of bone marrow fat
has traditionally been thought of as
a space filler (19); however, findings
in recent studies have suggested an
important link between bone and fat
(11,20,21). Both bone and fat cells
arise from the same mesenchymal stem
cell within bone marrow, capable of
differentiation into osteoblasts, adipo-
cytes, and marrow fat (11,21), and in-
creased bone marrow fat content has
been shown to be an indicator of bone
weakening (22). We were able to quan-
tify bone marrow fat content noninva-
sively in all subjects with use of
1
H MR
spectroscopy, and
1
H MR spectroscopy
might be added to routine spinal MR
imaging to identify subjects at risk for
bone loss.
Stem cell differentiation is under
the control of several transcription fac-
tors and signaling pathways (23,24),
and obesity has been found to cause a
shift into the adipocyte lineage (11,20).
In our study, obese subjects had higher
bone marrow fat content compared
with normal-weight subjects, confirm-
ing the role of obesity as a regulator of
stem cell differentiation.
Obesity is associated with in-
creased IHL and IMCL (13,14,17). We
found a positive correlation between
bone marrow fat and IHL, independent
of BMI, age, and IR. These results are
concordant with data in a prior study
in which the researchers demonstrated
that there was an inverse association
Figure 2
Figure 2:
1
H MR spectroscopy of liver and bone
marrow in a 35-year-old obese man (BMI, 37.4 kg/
m
2
) with high IHL content. For purposes of visual
comparison, the amplitude of unsuppressed water
in Figures 2 and 3 were scaled identically. (a)
1
H
MR spectrum of liver shows lipid (1.3 ppm) and
unsuppressed water (4.7 ppm) resonances. (b)
1
H
MR spectrum of bone marrow at L4 shows lipid
(1.3 ppm) and unsuppressed water (4.7 ppm)
resonances.
Figure 3
Figure 3:
1
H MR spectroscopy of liver and bone
marrow in a 37-year-old obese man with similar
BMI as subject in Figure 2 (BMI, 37.2 kg/m
2
) but
with lower IHL content. Despite similar age and BMI,
this obese man has lower bone marrow fat content
(0.3 vs 0.93 lipid-water ratio). (a)
1
H MR spectrum
of liver shows lipid (1.3 ppm) and unsuppressed
water (4.7 ppm) resonances. (b)
1
H MR spectrum
of bone marrow at L4 shows lipid (1.3 ppm) and
unsuppressed water (4.7 ppm) resonances.
540 radiology.rsna.org
n
Radiology: Volume 269: Number 2—November 2013
MUSCULOSKELETAL IMAGING: Lipid Levels and Bone Marrow Fat in Obesity Bredella et al
in mouse models and in vitro (7,8).
Statin therapy to lower hyperlipidemia
is associated with increased BMD and
decreased fracture risk (31,32), and
bisphosphonates have been shown to
lower serum LDL cholesterol (33). We
found a positive correlation between
serum triglyceride levels and bone mar-
row fat, confirming negative effects of
serum lipid levels on bone health. HDL
cholesterol levels were inversely asso-
ciated with bone marrow fat, which is
consistent with findings in prior stud-
ies in which researchers demonstrated
HDL levels to be a positive predictor of
BMD in postmenopausal women (34).
Our study had several limitations.
First, the cross-sectional study design
limited our ability to ascertain causal-
ity. Second, we studied only premeno-
pausal women and men of comparable
age, and our findings cannot be extrap-
olated to postmenopausal women or
older men. Third, the relationship be-
tween bone and fat is complex and is
mediated by additional factors. Fourth,
we used exercise questionnaires to
evaluate exercise status and did not
measure aerobic capacity directly.
A strength of our study is the large
number of subjects and detailed as-
sessment of bone marrow fat, IHL, and
IMCL levels by using
1
H MR spectros-
copy at 3.0 T. Furthermore, this is the
first study, to our knowledge, in which
the association among IHL and IMCL,
serum lipid levels, and bone marrow fat
has been examined, providing more in-
sights into the bone-fat connection and
the pathophysiologic characteristics of
obesity-associated bone loss.
In conclusion, ectopic lipid levels,
such as IHL and IMCL, and serum
lipid levels are positively associated
with bone marrow fat in young obese
men and women, independent of IR
and exercise status. Because bone
marrow fat is known to be inversely
related to BMD, these results support
the notion that ectopic and serum
lipid levels are influenced by the same
additional factors as bone marrow or
may exert negative effects on bone.
1
H MR spectroscopy might be used
to identify patients at risk for bone
loss. Further studies are needed to
osteoblastogenesis of stem cells while
enhancing adipogenesis (29). These
results suggest that insulin resistance
leads to increased marrow fat forma-
tion and concordant inhibition of os-
teoblast formation. This phenomenon
might serve as an explanation for our
observed correlation between marrow
adiposity and IMCL. However, further
studies are necessary to study the effect
of bone marrow fat on glucose homeo-
stasis and to use
1
H MR spectroscopy
to identify patients at risk for bone loss.
We also found a positive association
between total muscle lipid content and
bone marrow fat. This is concordant
with results in prior studies in which
researchers showed fatty infiltration of
skeletal muscle measured by using com-
puted tomography to be a negative pre-
dictor of BMD (4) and a risk factor for
hip fractures (30). Therefore, patients
with insulin resistance and elevated
muscle fat content might benefit from
evaluation of BMD.
Researchers in prior studies have
proposed that serum lipids are dam-
aging to bone. Elevated serum lipid
and lipoprotein levels inhibit osteo-
blast differentiation and enhance os-
teoclast differentiation and survival
in insulin-sensitive athletes (27). We
therefore controlled for measures of IR
and exercise status in our correlational
analyses. We have previously shown
significant correlations among IMCL,
soleus muscle, different fat depots (vis-
ceral adipose tissue, subcutaneous adi-
pose tissue, abdomen, and thigh), se-
rum lipid levels, and markers of insulin
resistance (14,17).
Lee et al (28) have suggested that
bone may exert an endocrine regulation
of glucose homeostasis. In their study,
mice lacking the osteoblast-secreted
molecule osteocalcin showed decreased
beta-cell proliferation, glucose intoler-
ance, and insulin resistance (28). Ad-
ditionally, a study using cell cultures of
mesenchymal stem cells from diabetic
and nondiabetic donors found that
high glucose concentrations reduce
Figure 4
Figure 4:
1
H MR spectroscopy of soleus muscle
in a 35-year-old obese woman (BMI, 41.0 kg/m
2
)
with high IMCL and high bone marrow fat content.
For purposes of visual comparison, the amplitude of
total creatine peaks in Figures 4 and 5 were scaled
identically.
1
H MR spectrum of soleus muscle shows
the following metabolite peaks: A, IMCL methylene
protons (2CH2) at 1.3 ppm; B, extramyocellular
lipid methylene protons (2CH2) at 1.5 ppm; C,
total creatine (2CH3) resonance at 3.0 ppm; D,
trimethylamines peak at 3.2 ppm; E, creatine
(2CH2) resonance at 3.96 ppm; F, residual water
peak at 4.7 ppm; and G, olefinic proton (HC5CH)
resonances at 5.3–5.5 ppm.
Figure 5
Figure 5:
1
H MR spectroscopy of soleus muscle
in a 34-year-old obese woman (BMI, 38.0 kg/m
2
)
with low IMCL content. Despite similar age and
BMI, this obese woman had lower bone marrow
fat content (0.33 vs 1.06 lipid-water ratio).
1
H MR
spectrum of soleus muscle shows the following
metabolite peaks: A, IMCL methylene protons
(2CH2) at 1.3 ppm; B, extramyocellular lipid meth-
ylene protons (2CH2) at 1.5 ppm; C, total creatine
(2CH3) resonance at 3.0 ppm; D, trimethylamines
peak at 3.2 ppm; E, creatine (2CH2) resonance at
3.96 ppm; F, residual water peak at 4.7 ppm; and
G, olefinic proton (HC5CH) resonances at 5.3–5.5
ppm.
Radiology: Volume 269: Number 2—November 2013
n
radiology.rsna.org 541
MUSCULOSKELETAL IMAGING: Lipid Levels and Bone Marrow Fat in Obesity Bredella et al
of bone and fat mass and its serum level is
modulated by growth hormone. Endocrinol-
ogy 2007;148(7):3111–3121.
24. Takada I, Suzawa M, Matsumoto K, Kato S.
Suppression of PPAR transactivation switch-
es cell fate of bone marrow stem cells from
adipocytes into osteoblasts. Ann N Y Acad
Sci 2007;1116:182–195.
25. Gallego-Rojo FJ, Gonzalez-Calvin JL, Mu-
ñoz-Torres M, Mundi JL, Fernandez-Perez
R, Rodrigo-Moreno D. Bone mineral den-
sity, serum insulin-like growth factor I, and
bone turnover markers in viral cirrhosis.
Hepatology 1998;28(3):695–699.
26. Janes CH, Dickson ER, Okazaki R, Bonde S,
McDonagh AF, Riggs BL. Role of hyperbili-
rubinemia in the impairment of osteoblast
proliferation associated with cholestatic
jaundice. J Clin Invest 1995;95(6):2581–
2586.
27. Goodpaster BH, He J, Watkins S, Kelley
DE. Skeletal muscle lipid content and in-
sulin resistance: evidence for a paradox in
endurance-trained athletes. J Clin Endocri-
nol Metab 2001;86(12):5755–5761.
28. Lee NK, Sowa H, Hinoi E, et al. Endocrine
regulation of energy metabolism by the skel-
eton. Cell 2007;130(3):456–469.
29. Cramer C, Freisinger E, Jones RK, et al. Per-
sistent high glucose concentrations alter the
regenerative potential of mesenchymal stem
cells. Stem Cells Dev 2010;19(12):1875–
1884.
30. Lang T, Cauley JA, Tylavsky F, et al. Com-
puted tomographic measurements of thigh
muscle cross-sectional area and attenuation
coefficient predict hip fracture: the health,
aging, and body composition study. J Bone
Miner Res 2010;25(3):513–519.
31. Edwards CJ, Hart DJ, Spector TD. Oral
statins and increased bone-mineral den-
sity in postmenopausal women. Lancet
2000;355(9222):2218–2219.
32. Wang PS, Solomon DH, Mogun H, Avorn J.
HMG-CoA reductase inhibitors and the risk
of hip fractures in elderly patients. JAMA
2000;283(24):3211–3216.
33. Adami S, Braga V, Guidi G, Gatti D, Gerardi
D, Fracassi E. Chronic intravenous amino-
bisphosphonate therapy increases high-den-
sity lipoprotein cholesterol and decreases
low-density lipoprotein cholesterol. J Bone
Miner Res 2000;15(3):599–604.
34. Jeong IK, Cho SW, Kim SW, et al. Lipid
profiles and bone mineral density in pre-
and postmenopausal women in Korea. Cal-
cif Tissue Int 2010;87(6):507–512.
11. Rosen CJ, Bouxsein ML. Mechanisms of
disease: is osteoporosis the obesity of bone?
Nat Clin Pract Rheumatol 2006;2(1):35–43.
12. Bredella MA, Torriani M, Ghomi RH, et al.
Vertebral bone marrow fat is positively as-
sociated with visceral fat and inversely asso-
ciated with IGF-1 in obese women. Obesity
(Silver Spring) 2011;19(1):49–53.
13. Bredella MA, Ghomi RH, Thomas BJ, Miller
KK, Torriani M. Comparison of 3.0 T proton
magnetic resonance spectroscopy short and
long echo-time measures of intramyocellular
lipids in obese and normal-weight women. J
Magn Reson Imaging 2010;32(2):388–393.
14. Bredella MA, Ghomi RH, Thomas BJ, et
al. Breath-hold 1H-magnetic resonance
spectroscopy for intrahepatic lipid quantifi-
cation at 3 Tesla. J Comput Assist Tomogr
2010;34(3):372–376.
15. Bredella MA, Lin E, Brick DJ, et al. Effects
of GH in women with abdominal adipos-
ity: a 6-month randomized, double-blind,
placebo-controlled trial. Eur J Endocrinol
2012;166(4):601–611.
16. Bredella MA, Torriani M, Ghomi RH, et
al. Adiponectin is inversely associated with
intramyocellular and intrahepatic lipids in
obese premenopausal women. Obesity (Sil-
ver Spring) 2011;19(5):911–916.
17. Bredella MA, Torriani M, Thomas BJ, et al.
Peak growth hormone-releasing hormone-
arginine-stimulated growth hormone is in-
versely associated with intramyocellular and
intrahepatic lipid content in premenopausal
women with obesity. J Clin Endocrinol
Metab 2009;94(10):3995–4002.
18. Provencher SW. Estimation of metabo-
lite concentrations from localized in vivo
proton NMR spectra. Magn Reson Med
1993;30(6):672–679.
19. Tavassoli M. Marrow adipose cells and he-
mopoiesis: an interpretative review. Exp He-
matol 1984;12(2):139–146.
20. Rosen CJ, Klibanski A. Bone, fat, and
body composition: evolving concepts in the
pathogenesis of osteoporosis. Am J Med
2009;122(5):409–414.
21. Takeda S, Elefteriou F, Karsenty G. Com-
mon endocrine control of body weight, re-
production, and bone mass. Annu Rev Nutr
2003;23:403–411.
22. Schellinger D, Lin CS, Hatipoglu HG,
Fertikh D. Potential value of vertebral
proton MR spectroscopy in determining
bone weakness. AJNR Am J Neuroradiol
2001;22(8):1620–1627.
23. Abdallah BM, Ding M, Jensen CH, et al.
Dlk1/FA1 is a novel endocrine regulator
investigate this hypothesis and other
potential mediators of the effects of
obesity on bone.
Disclosures of Conflicts of Interest: M.A.B. No
relevant conflicts of interest to disclose. C.M.G.
No relevant conflicts of interest to disclose.
A.V.G. No relevant conflicts of interest to dis-
close. M.G.L. No relevant conflicts of interest to
disclose. V.K. No relevant conflicts of interest to
disclose. S.M.D. No relevant conflicts of interest
to disclose. M.T. No relevant conflicts of interest
to disclose. K.K.M. No relevant conflicts of in-
terest to disclose.
References
1. Goulding A, Jones IE, Taylor RW, Williams
SM, Manning PJ. Bone mineral density and
body composition in boys with distal forearm
fractures: a dual-energy x-ray absorptiom-
etry study. J Pediatr 2001;139(4):509–515.
2. Nielson CM, Marshall LM, Adams AL, et al.
BMI and fracture risk in older men: the os-
teoporotic fractures in men study (MrOS). J
Bone Miner Res 2011;26(3):496–502.
3. Bredella MA, Lin E, Gerweck AV, et al. De-
terminants of bone microarchitecture and
mechanical properties in obese men. J Clin
Endocrinol Metab 2012;97(11):4115–4122.
4. Bredella MA, Torriani M, Ghomi RH, et
al. Determinants of bone mineral density
in obese premenopausal women. Bone
2011;48(4):748–754.
5. Gilsanz V, Chalfant J, Mo AO, Lee DC, Do-
rey FJ, Mittelman SD. Reciprocal relations
of subcutaneous and visceral fat to bone
structure and strength. J Clin Endocrinol
Metab 2009;94(9):3387–3393.
6. Kim KC, Shin DH, Lee SY, Im JA, Lee DC.
Relation between obesity and bone min-
eral density and vertebral fractures in Ko-
rean postmenopausal women. Yonsei Med J
2010;51(6):857–863.
7. Parhami F, Tintut Y, Beamer WG, Gharavi N,
Goodman W, Demer LL. Atherogenic high-
fat diet reduces bone mineralization in mice.
J Bone Miner Res 2001;16(1):182–188.
8. Tintut Y, Morony S, Demer LL. Hyperlip-
idemia promotes osteoclastic potential of
bone marrow cells ex vivo. Arterioscler
Thromb Vasc Biol 2004;24(2):e6–e10.
9. Moon SS, Lee YS, Kim SW. Association
of nonalcoholic fatty liver disease with low
bone mass in postmenopausal women. En-
docrine 2012;42(2):423–429.
10. Shulman GI. Cellular mechanisms of insulin
resistance. J Clin Invest 2000;106(2):171–176.
