er
an
Joh
pbe
Department of Medicine, Flinders Medical Centre, Flinders University, Bedford Park 5042, South Australia
b
ABSAT and VAT depots but possibly similar pathways for storage and mobilization of fat in the liver and viscerally.
adipose tissue (ABSAT) depots. The relative degree of
adipose tissue loss from one site over another after a weight
Available online at www.sciencedirect.com
Metabolism Clinical and ExperimentaCrown Copyright © 2009 Published by Elsevier Inc. All rights reserved.
1. Introduction
The significance of adiposity varies according to its
location. Abdominal obesity, indicated by an increased waist
to hip ratio, is a feature of the metabolic syndrome and
increases the risk of cardiovascular disease, type 2 diabetes
mellitus, and other metabolic abnormalities such as hyperli-
pidemia and obstructive sleep apnea. On the other hand, a
predominance of subcutaneous adiposity over visceral
adiposity, for example, in subjects with a low waist to hip
ratio, is associated with insulin sensitivity and is seen in
those treated with thiazolidinediones [1].
Weight loss is associated with variable reductions in both
visceral adipose tissue (VAT) and abdominal subcutaneousDepartment of General and Digestive Surgery, Flinders Medical Centre, Flinders University, Bedford Park 5042, South Australia
cDepartment of Medical Imaging, Flinders Medical Centre, Flinders University, Bedford Park 5042, South Australia
Received 5 November 2007; accepted 27 May 2008
Abstract
Weight loss after laparoscopic adjustable gastric banding surgery (LAGB) is associated with mobilization of adipose tissue from a variety
of depots. We sought to evaluate and relate abdominal and hepatic lipid deposition in an obese female population 3 and 12 months after
LAGB. We related changes in these depots to markers of insulin sensitivity. Eighteen female obese subjects underwent magnetic resonance
imaging and spectroscopy before and 3 and 12 months after LAGB for the quantification of abdominal subcutaneous (ABSAT) and visceral
(VAT) adipose tissue areas and liver fat content (LFAT). Fasting blood free fatty acids (FFA) were analyzed. Insulin sensitivity was assessed
by the homeostasis model assessment of insulin resistance index (HOMA-R). Mean weight loss 3 and 12 months after LAGB was 9.8 ±
1.1 kg and 20.0 ± 2.2 kg, respectively. Postoperatively, VAT area loss exceeded ABSAT area loss in the cohort as a whole and when divided
according to preoperative liver fat stores. Three months after LAGB, reductions had occurred in VAT and ABSAT areas (both P b .01) and in
FFA (P b .05). Twelve months after LAGB, further significant reductions (P b .01) occurred in VAT and ABSAT areas but not in FFA. No
significant reduction occurred in LFAT at either time point in the group as a whole. In those with preoperative hepatic steatosis (LFAT N∼5%,
n = 7), LFAT fell by 42% (P = .036) 3 months after LAGB, with a total reduction of 50% (P = .027 cf baseline) occurring by 12 months.
There was an improvement in HOMA-R at 12 months (1.9 ± 0.3 cf 2.9 ± 0.5 at baseline, P = .04) but not 3 months (2.7 ± 0.4).
Preoperatively, LFAT related significantly to VAT area (r = 0.67, P = .003) and HOMA-R (r = 0.497, P = .04) but not ABSAT area.
Postoperatively at both 3 and 12 months, LFAT continued to relate to VAT area (r = 0.63, P b .01 at both time points) but not HOMA-R. The
changes in LFAT and VAT area were unrelated postoperatively. Abdominal adipose tissue loss was greater from the visceral than
subcutaneous depots, suggesting that insulin sensitivity may not be an important determinant of selective lipid depot loss. The lack of a
significant change in liver fat in the group as a whole may relate to low preoperative liver fat stores and to high postoperative dietary fat
intakes. Preoperative liver fat stores did not influence insulin sensitivity or abdominal lipid changes during weight loss. Liver fat content and
VAT area interrelated more closely than either related to ABSAT area, suggesting differing regulatory pathways for fat mobilization fromAbdominal adiposity and liv
after gastric b
Madeleine L. Heatha,⁎, Lilian Kowb,
Jim Tooulib, Cam
a⁎ Corresponding author. Tel.: +61 8 8204 6249; fax: +61 8 8204 6255.
E-mail address: madeleine.heath@fmc.sa.gov.au (M.L. Heath).
0026-0495/$ – see front matter. Crown Copyright © 2009 Published by Elsevier
doi:10.1016/j.metabol.2008.05.021fat content 3 and 12 months
ding surgery
n P. Slavotinekc, Robin Valentinec,
ll H. Thompsona
l 58 (2009) 753–758
www.metabolismjournal.comloss intervention has been studied by several groups,
although the results conflict [2-4]. The preferential loss of
Inc. All rights reserved.

from loss of adipose tissue from one depot over another has
754 M.L. Heath et al. / Metabolism Clinical and Experimental 58 (2009) 753–758not been well explored [10-12].
In recent years, the accumulation of triglyceride within
hepatocytes (liver fat content [LFAT]) has been associated
with features of the metabolic syndrome; and the liver is
another location for lipid deposition that must be considered
when evaluating the relative importance of lipid depots in
determining disease risk. With the advent of noninvasive
techniques to quantitate LFAT, studies have examined the
relationship between LFATand other lipid depots, again with
conflicting results [13,14].
The purposes of the present study were several. Firstly, we
aimed to distinguish the effect over time of a bariatric surgical
procedure on weight loss from 3 important lipid depots (liver,
abdominal subcutaneous layer, and viscera). As secondary
aims, we examined the effect of progressive weight loss upon
each patient's homeostasis model assessment (HOMA)
score, a marker of their insulin sensitivity; and we sought to
explain the fat loss from each depot on the basis of the
subject's HOMA score and their preoperative LFAT stores.
2. Research methods and procedures
Ethics approval to conduct this study was received from
the clinical research ethics committee of our institution.
2.1. Subjects
Eighteen obese female subjects (mean age, 44 years;
range, 20-57) scheduled for laparoscopic adjustable gastric
banding surgery (LAGB) gave their written consent to
participate in this study.
Subjects that met the clinical criteria for LAGB (body
mass index [BMI] N35 kg/m2 or N30 kg/m2 with a recognized
comorbidity of obesity) were considered for enrolment in the
study. Patients with a known history of diabetes mellitus or a
baseline fasting blood glucose level greater than 6.0 mmol/L
were excluded from the study. Because most patients who
undergo this procedure are female, men were excluded fromadipose tissue from one depot over another may relate to
lifestyle habits such as physical activity and diet as well as
familial and sex influences [5] and the mechanism of weight
loss [6]. Preexisting hepatic steatosis and insulin sensitivity
may also be relevant. Insulin has an antilipolytic effect in
adipose tissue that is greater in subcutaneous than visceral
adipose tissue such that the latter lipid depot is more
vulnerable to lipolysis [5,7]. In insulin-resistant states, as
occur for example with hepatic steatosis, the relative
insensitivity of VAT to the actions of insulin could be more
pronounced, resulting in the observed tendency toward
greater loss of VAT than ABSAT during weight loss [2,8,9].
Metabolic improvements occur with weight loss and the
accompanying reductions in VAT and ABSAT volumes.
Although there is general consensus that visceral adiposity is
responsible for many of the metabolic disturbances asso-
ciated with obesity, any beneficial metabolic effect arisingthis study to improve sample homogeneity. Other reasons for
exclusion were chronic conditions that were poorly con-
trolled and contraindications to magnetic resonance studies
including excessive size, claustrophobia, or the presence of
metallic foreign bodies or incompatible implants. Assess-
ments of the study cohort occurred preoperatively and 3 and
12 months post-LAGB.
2.2. Magnetic resonance
Magnetic resonance images and spectra were obtained to
quantify lipid contained within the abdominal subcutaneous
and visceral compartments and the liver using a Philips Intera
1.5-T magnet (Best, the Netherlands). With the subject lying
supine in the magnet, a series of contiguous 10-mm–thick,
T1-weighted axial images of the abdomen was taken from
above the diaphragm to the symphysis pubis. Using one of
these images with the liver clearly visible, an 8-cm3 voxel
was placed over a homogenous portion of liver, near a 7.6-cm
surface coil that had been positioned laterally over the
subject's right upper abdominal quadrant.
Liver spectra were acquired, with and without water sup-
pression, using a predefined point-resolved spectroscopy
sequence(echotime,31milliseconds; repetition time,3000milli-
seconds; 80measurements; 32measurements 1024 data points).
2.2.1. Hepatic lipid
Proton magnetic resonance spectroscopy was used for
determining hepatic fat content [15,16]. Hepatic fat content
was calculated from the ratio of the area under the lipid
resonance in the water-suppressed sequence to that of the
water resonance in the unsuppressed sequence [17] using the
AMARES algorithm contained within the MRUI software
program (EU project TMR, FMRX-CT97-0160, Barcelona,
Spain). This method provides a numerical value for liver fat,
the units of which are arbitrary (AU) but may be defined in
percentage terms. An LFAT greater than 0.056 AU (5.6%)
was used to identify those subjects with hepatic steatosis [17].
Sixteen complete sets of spectra were available for analysis.
2.2.2. Visceral and subcutaneous lipid
Images were transferred to a local workstation for
volumetric analysis using Philips EasyVision software as
previously described [18]. Briefly, a single axial image
acquired at the level of the L4-5 intervertebral space was
used to measure the change in abdominal adipose tissue [19].
The cross-sectional areas of both the visceral and
subcutaneous fat depots were automatically computed after
manual perimetry of each fat compartment. A total of 17 sets
of images were analyzed. Repeated abdominal subcutaneous
and visceral fat estimation using this technique gave rise to
coefficients of variation of 3% and 6%, respectively [18].
2.3. Anthropometry
Body weight was measured to the nearest 0.1 kg with the
subject in light clothing (shoes removed) on an upright
pedestal digital scale (Seca, Birmingham, UK). Height was