The Missing Risk: MRI and MRS Phenotyping of Abdominal Adiposity and Ectopic Fat

Abstract

Individual compartments of abdominal adiposity and lipid content within the liver and muscle are differentially associated with metabolic risk factors, obesity and insulin resistance. Subjects with greater intra-abdominal adipose tissue (IAAT) and hepatic fat than predicted by clinical indices of obesity may be at increased risk of metabolic diseases despite their "normal" size. There is a need for accurate quantification of these potentially hazardous depots and identification of novel subphenotypes that recognize individuals at potentially increased metabolic risk. We aimed to calculate a reference range for total and regional adipose tissue (AT) as well as ectopic fat in liver and muscle in healthy subjects. We studied the relationship between age, body-mass, BMI, waist circumference (WC), and the distribution of AT, using whole-body magnetic resonance imaging (MRI), in 477 white volunteers (243 male, 234 female). Furthermore, we used proton magnetic resonance spectroscopy (MRS) to determine intrahepatocellular (IHCL) and intramyocellular (IMCL) lipid content. The anthropometric variable which provided the strongest individual correlation for adiposity and ectopic fat stores was WC in men and BMI in women. In addition, we reveal a large variation in IAAT, abdominal subcutaneous AT (ASAT), and IHCL depots not fully predicted by clinically obtained measurements of obesity and the emergence of a previously unidentified subphenotype. Here, we demonstrate gender- and age-specific patterns of regional adiposity in a large UK-based cohort and identify anthropometric variables that best predict individual adiposity and ectopic fat stores. From these data we propose the thin-on-the-outside fat-on-the-inside (TOFI) as a subphenotype for individuals at increased metabolic risk.

Full text

76 VOLUME 20 NUMBER 1 | JANUARY 2012 | www.obesityjournal.org
articles nature publishing group
integrative Physiology
INTRODUCTION
Obesity is a major risk factor for insulin resistance, type 2 dia-
betes mellitus, and cardiovascular disease (CVD). However,
not every obese patient is insulin resistant or at high risk of
diabetes mellitus and CVD (1). Abdominal obesity is associ-
ated with an increased predisposition to metabolic and CVD
risk with the compartmental distribution of adipose tissue (AT)
linked to dierences in susceptibility (2,3). Intra-abdominal
AT (IAAT) appears to play a major role in the pathogenesis
of insulin resistance, diabetes, dyslipidemia, inammation,
hypertension, and CVD, whereas the metabolic consequences
of subcutaneous AT (SAT) are less clear (1,4–6).
SAT has been linked to features of the metabolic syndrome
independently of intra-abdominal fat, including obesity-related
insulin resistance (7,8). However, data suggests that while both
IAAT and SAT are correlated with metabolic risk factors, IAAT
remains more strongly associated with an adverse metabolic
risk prole even aer accounting for anthropometric indices
(9,10). Loss of IAAT following diet and exercise is associ-
ated with improvements in insulin sensitivity, blood pressure,
and circulating lipid levels, whereas comparable loss of SAT
by liposuction does not result in amelioration of these meta-
bolic abnormalities (7,11,12). Interestingly, human and animal
studies have indicated a possible protective role for SAT; in
humans, increased SAT in the leg is associated with decreased
risk of perturbed glucose and lipid metabolism (13), whereas
in mice, transplantation of SAT into intra-abdominal compart-
ments results in improved glucose metabolism and a reduction
in body mass and total fat mass (14). Additional variables such
as age, gender, and ethnicity may also be an important con-
founding factors in the relationship between adiposity stores
and metabolic risk.
The Missing Risk: MRI and MRS Phenotyping
of Abdominal Adiposity and Ectopic Fat
E. Louise Thomas1, James R. Parkinson1, Gary S. Frost2, Anthony P. Goldstone1, Caroline J. Doré3,
John P. McCarthy1, Adam L. Collins4, Julie A. Fitzpatrick1,5, Giuliana Durighel1,
Simon D. Taylor-Robinson5 and Jimmy D. Bell1
Individual compartments of abdominal adiposity and lipid content within the liver and muscle are differentially
associated with metabolic risk factors, obesity and insulin resistance. Subjects with greater intra-abdominal adipose
tissue (IAAT) and hepatic fat than predicted by clinical indices of obesity may be at increased risk of metabolic
diseases despite their “normal” size. There is a need for accurate quantification of these potentially hazardous
depots and identification of novel subphenotypes that recognize individuals at potentially increased metabolic risk.
We aimed to calculate a reference range for total and regional adipose tissue (AT) as well as ectopic fat in liver and
muscle in healthy subjects. We studied the relationship between age, body-mass, BMI, waist circumference (WC),
and the distribution of AT, using whole-body magnetic resonance imaging (MRI), in 477 white volunteers (243 male,
234 female). Furthermore, we used proton magnetic resonance spectroscopy (MRS) to determine intrahepatocellular
(IHCL) and intramyocellular (IMCL) lipid content. The anthropometric variable which provided the strongest individual
correlation for adiposity and ectopic fat stores was WC in men and BMI in women. In addition, we reveal a large
variation in IAAT, abdominal subcutaneous AT (ASAT), and IHCL depots not fully predicted by clinically obtained
measurements of obesity and the emergence of a previously unidentified subphenotype. Here, we demonstrate
gender- and age-specific patterns of regional adiposity in a large UK-based cohort and identify anthropometric
variables that best predict individual adiposity and ectopic fat stores. From these data we propose the thin-on-the-
outside fat-on-the-inside (TOFI) as a subphenotype for individuals at increased metabolic risk.
Obesity (2012) 20, 76–87. doi:10.1038/oby.2011.142
1Metabolic and Molecular Imaging Group, MRC Clinical Sciences Centre, Imperial College London, Hammersmith Hospital, London, UK; 2Nutrition and Dietetic
Research Group, Department of Investigative Medicine, Imperial College London, Hammersmith Hospital, London, UK; 3MRC Clinical Trials Unit, London, UK;
4Nutritional Sciences Division, Faculty of Health and Medical Sciences, University of Surrey, Surrey, UK; 5Division of Diabetes Endocrinology and Metabolism,
Department of Medicine, Imperial College London, London, UK. Correspondence: Elizabeth L. Thomas (louise.thomas@csc.mrc.ac.uk)
Received 24 February 2010; accepted 5 April 2011; published online 9 June 2011. doi:10.1038/oby.2011.142

OBESITY | VOLUME 20 NUMBER 1 | JANUARY 2012 77
articles
integrative Physiology
BMI is the current benchmark for obesity classication, but
like all anthropometric measurements, it only oers a proxy
measure of body adiposity (15). Waist circumference (WC) is
widely used as a surrogate of central fat distribution, but while
easily obtainable, it is unable to distinguish between IAAT and
abdominal SAT (ASAT) deposition (16). Magnetic resonance
imaging (MRI) is a noninvasive technique that allows accurate
measurement of whole-body fat and specic internal stores
of AT. MRI studies have demonstrated signicant variation
among individual AT compartments that is not predicted by
total body or trunk fat or standard anthropomorphic charac-
teristics; such as skin-fold measurements, BMI, and waist-to-
hip ratio (WHR) (6,17,18). Ectopic fat in organs has also been
linked to obesity, insulin resistance, type 2 diabetes mellitus,
and in particular, the lipids within the muscle cells (intramyo-
cellular lipids (IMCL)) and the liver (IHCL) (19–22). Recent
data have suggested a dierence between individual ectopic fat
depots with IHCL, but not with IMCL being linked to insulin
resistance (23).
In this study, we have employed magnetic resonance tech-
niques to accurately measure the patterns of fat deposition
in 477 UK-based white volunteers. In addition to providing
gender- and age-specific reference range data, we aimed to
describe the relationships between individual anthropomet-
ric, adiposity, and ectopic fat measurements. The variation
observed in metabolically adverse internal fat deposition
led to our proposal of the thin-on-the outside fat-on-the-
inside (TOFI) subphenotype; a ratio of IAAT and ASAT
calculated from the range of abdominal adiposity stores of
a defined healthy subset of volunteers, as a simple, quanti-
tative means of identifying those that may be at increased
metabolic risk.
METHODS AND PROCEDURES
Subjects
Written, informed consent was acquired from all volunteers. Ethical
approval permission for this study was obtained from the research
ethics committee of Hammersmith and Queen Charlotte’s and Chelsea
Research Ethics Committee Hospital, London (Rec: 07Q04011/19).
In total, 477 unpaid volunteers (243 male, 234 female) were recruited
via advertisements in newspapers, websites, and academic newslet-
ters, inviting male and female volunteers of white ethnicity from the
general public. No age constraints were placed on recruitment in
order to generate cross-sectional data. Self-reported exclusion crite-
ria included subjects suering from chronic disease (including dia-
betes, cardiovascular or liver disease, metabolic conditions, anyone
taking prescribed medication and women on the contraceptive pill).
Volunteers underwent anthropometric assessment, total body MRI
scanning and in vivo proton (
1
H) magnetic resonance spectroscopy
(MRS) of liver and calf muscle.
Anthropometric measurements
Body mass (kg), height (cm), WC (cm), and hip circumference (cm)
were measured in each subject by a single experienced observer. WC
was measured at the (WHO recommended) midpoint (24) between
the distal border of the lowest rib and the superior border of the iliac
crest. From these values, BMI (kg/m2), WHR (waist/hip), and waist-to-
height ratio (WHtR, waist/height) were calculated. BMI grouping cor-
responded to the following ranges: 1: 18.5 <25 kg/m2, 2: 25 <30 kg/ m2,
3: 30 <40 kg/m2, 4: 40+ kg/m2.
MRI scanning: Total body and regional AT content
Rapid T
1
-weighted MR images were acquired using a 1.5T Phillips
Achiva scanner (Phillips, Best, the Netherlands), as previously
described (17). Subjects lay in a prone position with arms straight above
the head, and were scanned from ngertips to toes, acquiring 10-mm
thick contiguous transverse images throughout the body. Images were
analyzed using SliceOmatic (Tomovision, Montreal, Quebec, Canada).
Regional volumes were recorded in liters (l); comprising; ASAT, non-
ASAT (NASAT), IAAT, and non-abdominal internal-AT (NAIAT)
which includes internal AT in the head, neck, chest, pelvis, arms, and
legs as previously described (17). e abdominal region was dened
as the image slices from the slice containing the femoral heads, to the
slice containing the top of the liver/bottom of the lungs; therefore the
measurement of IAAT contains a mixture of visceral, perirenal, and
retroperitoneal AT (17). Total AT was calculated from the sum of SAT
and internal adipose stores: TAT = SAT + Internal. SAT was subdivided
into ASAT and non-NASAT: SAT = ASAT + NASAT. Total internal AT
(Internal) was subdivided into IAAT and non-abdominal internal AT
(NAIAT): Internal = IAAT + NAIAT. In order to gauge abdominal adi-
posity as a whole, “trunk” fat was derived from the sum of IAAT and
ASAT: Trunk = IAAT + ASAT.
MRS of liver and muscle fat
During the same scanning session, 1H MR spectra were also
acquired at 1.5T, using a surface coil. Transverse images of the
liver were used to ensure accurate positioning of the (20 × 20 × 20 mm)
voxel in the liver, avoiding blood vessels, the gall bladder, and fatty
tissue. Spectra were obtained from the right lobe of the liver using a
PRESS sequence (repetition time 1,500 ms, echo time 135 ms) with-
out water saturation and with 128 signal averages. Intrahepatocellular
lipids (IHCL) were measured relative to liver water content, as previ-
ously described (21). IMCL were measured in the soleus (S-IMCL)
and tibialis (T-IMCL) muscles by 1H MRS. Proton MR spectra were
acquired from 20 × 20 × 20 mm voxels localized to the soleus and
tibialis muscles of the le calf using a PRESS sequence (repetition
time 1,500 ms, echo time 135 ms). IMCL were subsequently mea-
sured, relative to total muscle creatine signal, as previously described
(25). Spectra from both S-IMCL and T-IMCL muscles were obtained
because their dierences in ber composition and fuel requirements
(tibialis primarily utilizes carbohydrate; soleus predominantly uti-
lizes lipid) result in dierent lipid levels and metabolism. MRS data
(IHCL, S-IMCL, and T-IMCL) are presented as the geometric mean,
whereas statistical analysis was performed on loge transformed vari-
ables, due to the positively skewed distribution of these datasets (21).
Of the male volunteers (total = 243) IHCL data was available for 234
individuals, S-IMCL for 239, T-IMCL for 239 subjects; whereas for
females (total = 234): IHCL was available for 169, S-IMCL for 179,
and T-IMCL for 178. We have previously published data concerning
the reproducibility of the MRI (26) and MRS protocols for IHCL (21)
and IMCL (25) measurements implemented here.
Definition of healthy controls
To identify subjects who had a fat distribution which deviated from
“normal” it was necessary to dene a healthy control group from our
larger population. e following criteria were used to dene white indi-
viduals within our test population as healthy controls: (i) Absence of
disease/metabolic condition; (ii) BMI: 18.5 <25 kg/m2 (WHO guide-
lines (27)); (iii) WC: male ≤94 cm, female ≤80 cm (WHO guidelines
(28)); (iv) WHR: male ≤0.90, female ≤0.80 (29); (v) age: male (18–50
years), female (18–39 years): the younger age group in female subjects
was chosen in order to eliminate the eects of menopause from our con-
trol group, as there is a signicant increase in obesity related-metabolic
disorders aer menopause, which has been linked to alterations in body
adiposity, notably an increase in IAAT. (vi) Activity level: sedentary
subjects were identied using Baecke (30) and/or I-PAQ questionnaires
(31). Both tests calculate weekly physical activity, classifying individuals
into low, moderate, or high categories based on specic criteria; I-PAQ

78 VOLUME 20 NUMBER 1 | JANUARY 2012 | www.obesityjournal.org
articles
integrative Physiology
assesses the duration and frequency of walking, moderate intensity, and
vigorous intensity activity. We included subjects with a high or mod-
erate score in our denition of “healthy” as this level of activity was
considered sucient to maintain cardiovascular health while individu-
als reported in the “low” physical activity group from either test were
excluded. Subjects with a fat distribution that deviated by 2 s.d. from
that found in healthy controls were identied by comparing the ratio of
intra-abdominal to ASAT (IAAT/ASAT) and against BMI.
Age- and BMI-related reference ranges
Age group (16–25 years, 26–35 years, 36–45 years, and 56+ years) and
BMI group reference ranges were calculated for each anthropometric
and adipose variable for male and female subjects. For each depend-
ent variable a fractional polynomial model was tted to predict the
mean curve estimated using generalized least squares (32). e s.d. was
assumed to be constant and estimated using the residual mean square
from the model. e reference ranges were calculated on the logarith-
mic scale and back transformed to the original scale of measurement
for presentation. A logarithmic scale was used for the y axis in each
graph. For IHCL the y axis contains (IHCL + 0.03) as it is not possible
to plot IHCL values of zero on a logarithmic scale as the log of zero is
not dened. Residuals from each tted model were assessed for nor-
mality using normal plots and Shapiro and Francia’s W’-test of normal-
ity (33). Loge transformations were used for all six dependent variables
to improve the assumption of normality of residuals of the models. As
there were some values of zero for IHCL, the statistical analysis used a
log (IHCL + 0.03) transformation. For T-IMCL a log (T-IMCL + 0.8)
transformation was used to improve the normality of the residuals.
Multiple regression analysis
Regression models predicting AT were tted to 11 variables: TAT,
SAT, ASAT, NASAT, total internal, IAAT, NAIAT, trunk AT, S-IMCL,
T-IMCL, and IHCL. Separate regression models were tted to males
and females. Loge transformations were used for all dependent variables
to improve the assumption of normality of residuals of the models. As
there were some values of zero for IHCL, the statistical analysis used a
log (IHCL + 0.03) transformation. For T-IMCL a log (T-IMCL + 0.8)
transformation was used to improve the normality of the residuals. For
each dependent variable six regression models were tted. Independent
variables considered were WC, BMI, hip, age, WHtR, and weight. e
six regression models were: 1a: WC BMI; 1b: WC BMI hip; 1c: WC
BMI hip age; 2a: WHtR weight; 2b: WHtR weight hip; 2c: WHtR weight
hip age. e Bayesian Information Criterion was used to compare the
goodness-of-t of these six models (34).
Statistical analysis
Gender dierences were analyzed using the Student’s t-test. Signicance
is taken as P < 0.05. All data are presented as mean ± s.d. e statisti-
cal analysis was performed using Stata Release 11. Age- and BMI-
related reference ranges were tted to the data using established Stata
routines (32).
RESULTS
Descriptive statistics
e mean age of all subjects was 37 years (range 18–71 years)
with 25.0% of all subjects classied as overweight (32.6% of
Table 1 Gender-specific variable data
Male (n = 243) Female (n = 234) P
Mean ± s.d. Range Mean ± s.d. Range M vs. F
Age (years) 40.3 ± 13 17–70 34.5 ± 12.4 17–71 <0.001
Weight (kg) 87.6 ± 16.2 59.0–146.6 71.4 ± 17.6 40.7–146.8 <0.001
BMI (kg/m2) 27.3 ± 4.8 18.6–44.5 26.2 ± 6.6 15.5–57.3 <0.05
WC (cm) 95.4 ± 13.3 70.0–131 81.4 ± 13.8 56.5–131 <0.001
Hip (cm) 103.5 ± 8.4 85.4–136 102.3 ± 10.7 76–134 0.23
Height (cm) 179 ± 7.3 143–199 165.2 ± 6.6 145.5–182 <0.001
WHR 0.92 ± 0.07 0.75–1.11 0.8 ± 0.07 0.58–1.04 <0.001
WHtR 0.53 ± 0.08 0.37–0.85 0.49 ± 0.09 0–0.78 <0.001
IHCLa6.8 ± 14.0 0–89.6 2.8 ± 8.5 0–65.0 <0.001
S-IMCLa15.5 ± 9.7 2.9–100.1 11.5 ± 6.9 2.28–51.0 <0.001
T-IMCLa6.3 ± 3.8 0.25–30.5 6.7 ± 4.1 0.96–35.4 0.25
TAT (l) 24.9 ± 10.9 6–67.7 31.2 ± 15.9 8.3–106.2 <0.001
SAT (l) 18.6 ± 8.4 4.2–58.2 26.6 ± 13.5 7.3–90.5 <0.001
ASAT (l) 5.3 ± 3.0 0.7–20.2 7.6 ± 4.9 1.5–29.7 <0.001
NASAT (l) 13.3 ± 5.6 1.3–38 19.0 ± 8.8 5.8–60.9 <0.001
Internal (l) 6.3 ± 3.3 0.7–15.8 4.6 ± 2.8 1–15.7 <0.001
IAAT (l) 3.5 ± 2.1 0.2–9.4 2.3 ± 1.8 0.4–9.6 <0.001
NAIAT (l) 2.8 ± 1.4 0.5–7.9 2.4 ± 1.1 0.6–6.2 <0.001
Trunk (l) 8.8 ± 4.7 1.0–25.5 9.9 ± 6.4 1.9–39.3 <0.05
IAAT/ASAT 0.7 ± 0.3 0.18–1.64 0.3 ± 0.1 0.09–0.97 <0.001
Mean and range variable data. Adipose tissue deposits are in liters (l).
ASAT, abdominal subcutaneous adipose tissue; IAAT, intra-abdominal adipose tissue; IHCL, intrahepatocellular lipid; IMCL, intramyocellular lipid (S, soleus, T, tibialis);
internal, total internal; NAIAT, non-abdominal internal adipose tissue; NASAT, non-abdominal subcutaneous adipose tissue; SAT, subcutaneous adipose tissue; TAT, total
adipose tissue; WC, waist circumference; WHR, waist-to-hip ratio; WHtR, waist-to-height ratio.
aMRS data (IHCL (M: 234, F: 169), S-IMCL (M: 239, F: 179), and T-IMCL (M: 239, F: 178)) is presented as the geometric mean, while statistical analysis was performed
on log10 transformed variables. All data are presented as mean ± s.d. Male vs. female data analyzed by Student’s t-test.

OBESITY | VOLUME 20 NUMBER 1 | JANUARY 2012 79
articles
integrative Physiology
men, 17.0% women, BMI: 25< 30), 24.7% qualifying as obese
(26.8% of men, 22.6% of women, BMI: 30< 40) and 2.9%
morbidly obese (2.1% of men, 3.8% of women, BMI: 40+).
Gender-specic characteristics are shown in Table 1. Overall,
female subjects were characterized by greater TAT, SAT, ASAT,
NASAT, and lower internal, IAAT and lower IAAT/ASAT ratio
than males (P < 0.001 for all, Table 1). e gender-specic range
observed in IAAT and ASAT stores is shown in Figure 1a–f.
Given that males were slightly older than females (male: 40 ±
13 years vs. female: 36 ± 12 years, P < 0.01, Table 1) correction
for age by multiple linear regression analysis was performed
in order to examine gender dierences (Supplementary Table
S1). Males had a signicantly greater weight, WC, height and
WHR, but lower NASAT and NAIAT than females, when
correcting for age (all P < 0.05, Supplementary Table S1).
Male volunteers also demonstrated a trend toward increased
IHCL (P = 0.08) compared to females (adjusting for age)
(Supplementary Table S1). e range of IHCL, S-IMCL, and
T-IMCL values, as categorized by age and BMI group, are illus-
trated in Figure 2a–f. All variables categorized by BMI and age
group can be found in Supplementary Tables S2–S4.
Age was positively correlated with all anthropometric vari-
ables (apart from a negative inuence on height), ectopic fat
stores (IHCL, S-IMCL, and T-IMCL) and adiposity depots col-
lapsed across gender (Supplementary Table S1). NAIAT and
IAAT/ASAT ratio showed a signicantly positive interaction
between age and gender (gender × age P: NAIAT: 0.01, IAAT/
ASAT: <0.001) (Supplementary Table S1). Age was associated
20.0
10.0
5.0
2.0
1.0
20.0
10.0
5.0
2.0
1.0
0.5
0.2
20.0
2.0
1.0
0.5
0.2
0.1
10.0
5.0
2.0
1.0
0.5
0.2
20.0
10.0
5.0
2.0
1.0
20.0
10.0
5.0
2.0
1.0
ASAT (litres)
20.0
10.0
5.0
2.0
1.0
ASAT (litres) IAAT (litres)
IAAT/ASAT
2.0
1.0
0.5
0.2
0.1
FemaleMale
FemaleMale
FemaleMale
20.0
10.0
5.0
2.0
1.0
0.5
0.2
20.0
10.0
5.0
2.0
1.0
0.5
0.2
IAAT (litres)
FemaleMale
FemaleMale
2.0
1.0
0.5
0.2
0.1
IAAT/ASAT
2.0
1.0
0.5
0.2
0.1
FemaleMale
20 30 40 50 60 70
Age (years)
20 30 40 50 6010
BMI (kg/m2)
20 30 40 50 6010
BMI (kg/m2)
20 30 40 50 6010
BMI (kg/m2)
20 30 40 50 6010
BMI (kg/m2)
20 30 40 50 6010
BMI (kg/m2)
20 30 40 50 6010
BMI (kg/m2)
20 30 40 50 60 70
Age (years)
20 30 40 50 60 70
Age (years)
20 30 40 50 60 70
Age (years)
20 30 40 50 60 70
Age (years)
20 30 40 50 60 70
Age (years)
ab
cd
ef
Figure 1 Abdominal subcutaneous (ASAT), intra-abdominal adipose tissue (IAAT), and IAAT/ASAT ratio by age and BMI group in male and female
volunteers. (a–c) Age group and (d–f) BMI group-specific variation in (a, d) abdominal subcutaneous (ASAT) and (b, e) intra-abdominal adipose
tissue (IAAT) in male and female volunteers. Adiposity stores are presented in liters. The ratio of IAAT/ASAT is also presented by (c) age and (f) BMI
groups. The graphs present the fitted mean curve, and the 2.5th and 97.5th centiles, calculated assuming normal errors.

80 VOLUME 20 NUMBER 1 | JANUARY 2012 | www.obesityjournal.org
articles
integrative Physiology
with increased IAAT/ASAT in both genders, but had a greater
eect in males (β: 0.005 ± 0.001, Supplementary Table S1).
Correlation analysis
Gender-specic correlation analysis for all anthropometric
variables, AT (in l) and ectopic fat stores are shown in Table 2
(men) and Table 3 (women). Apart from a few exceptions, all
parameters correlated with each other to a signicant degree (P
< 0.01). In male subjects, WC was the variable which correlated
to the greatest degree with individual adiposity stores (TAT:
r = 0.915, SAT: r = 0.878, ASAT: r = 0.850, NASAT: r = 0.868,
internal: r = 0.804, IAAT: r = 0.815, NAIAT: r = 0.7147, trunk:
r = 0.919, P < 0.01, Table 2), whereas in female subjects BMI
had the strongest correlation with individual adiposity stores
(TAT: r = 0.9514, SAT: r = 0.944, ASAT: r = 0.937, NASAT:
r = 0.927, internal: r = 0.850, IAAT: r = 0.839, NAIAT: r = 0.777,
trunk: r = 0.949, P < 0.01, Table 3).
IHCL correlated most strongly with WC in male subjects
(r = 0.712, P < 0.01, Ta bl e 2) and WHtR in female subjects
(r = 0.644, P < 0.01, Ta b le 3). In male subjects, WC was also
the strongest correlate of S-IMCL (r = 0.504, P < 0.01) and
T-IMCL (r = 0.389, P < 0.01) (Ta b le 2), whereas in females
S-IMCL and T-IMCL correlated most strongly with BMI
(S-IMCL: r = 0.450, T-IMCL: r = 0.265, P < 0.01 for both,
Table 3). IAAT was the depot which correlated most strongly
with ectopic fat in both genders (male IAAT: IHCL r = 0.716,
S-IMCL; r = 0.473, T-IMCL; r = 0.417, P < 0.01 for all, Table 2;
female IAAT: IHCL; r = 0.720, S-IMCL; r = 0.506, T-IMCL;
r = 0.319, P < 0.01 for all, Ta bl e 3). In both male and female
subjects, mass, BMI, WC, and WHtR correlated more strongly
100.0
10.0
1.0
0.1
0.01
50.0
20.0
10.0
5.0
2.0
50.0
20.0
10.0
5.0
2.0
50.0
20.0
10.0
5.0
1.0
2.0
50.0
20.0
10.0
5.0
1.0
2.0
100.0
10.0
1.0
0.1
0.01
IHCL
100.0
10.0
1.0
0.1
0.01
100.0
10.0
1.0
0.1
0.01
IHCL S-IMCL
T-IMCL
FemaleMale
FemaleMale
FemaleMale
50.0
20.0
10.0
5.0
2.0
50.0
20.0
10.0
5.0
2.0
S-IMCL
FemaleMale
FemaleMale
50.0
20.0
10.0
5.0
1.0
2.0
50.0
20.0
10.0
5.0
1.0
2.0
T-IMCL
FemaleMale
20 30 40 50 60 70
Age (years)
20 30 40 50 6010
BMI (kg/m2)
20 30 40 50 6010
BMI (kg/m2)
20 30 40 50 6010
BMI (kg/m2)
20 30 40 50 6010
BMI (kg/m2)
20 30 40 50 6010
BMI (kg/m2)
20 30 40 50 6010
BMI (kg/m2)
20 30 40 50 60 70
Age (years)
20 30 40 50 60 70
Age (years)
20 30 40 50 60 70
Age (years)
20 30 40 50 60 70
Age (years)
20 30 40 50 60 70
Age (years)
ab
cd
ef
Figure 2 Ectopic lipid stores by BMI and age group in male and female volunteers. (a–c) Age group and (d–f) BMI group-specific variation in (a, d)
intrahepatocellular lipid (IHCL), (b, e) intramyocellular muscle in the soleus (S-IMCL), and (c, f) tibialis (T-IMCL) in male and female volunteers. The
graphs present the fitted mean curve, and the 2.5th and 97.5th centiles, calculated assuming normal errors.

OBESITY | VOLUME 20 NUMBER 1 | JANUARY 2012 81
articles
integrative Physiology
Table 2 Linear correlation analysis between anthropometric measurements, lipid stores, and body fat stores in male subjects
Male Age Anthropometric variable Ectopic fat storeaAdiposity depot (liters)
n = 243 Age Weight BMI WC Hip Height WHR WHtR IHCL S-IMCL T-IMCL TAT SAT ASAT NASAT Internal IAAT NAIAT Trunk
Weight 0.288**
BMI 0.378** 0.883**
WC 0.478** 0.881** 0.901**
Hip 0.311** 0.914** 0.841** 0.861**
Height −0.124 0.352** −0.120 0.090 0.355**
WHR 0.510** 0.603** 0.706** 0.856** 0.479** −0.108
WHtR 0.493** 0.745** 0.912** 0.950** 0.766** −0.216* 0.885**
IHCL 0.405** 0.551** 0.625** 0.712** 0.559** −0.070 0.663** 0.707**
S-IMCL 0.389** 0.377** 0.378** 0.504** 0.437** 0.054 0.427** 0.463** 0.440**
T-IMCL 0.314** 0.293** 0.343** 0.389** 0.300** −0.051 0.375** 0.384** 0.417** 0.590**
TAT 0.376** 0.836** 0.832** 0.915** 0.863** 0.098 0.712** 0.855** 0.663** 0.427** 0.386**
SAT 0.271** 0.836** 0.823** 0.878** 0.876** 0.116 0.632** 0.816** 0.590** 0.368** 0.338** 0.975**
ASAT 0.231** 0.829** 0.841** 0.850** 0.858** 0.074 0.599** 0.806** 0.598** 0.358** 0.359** 0.922** 0.944**
NASAT 0.280** 0.801** 0.775** 0.868** 0.862** 0.133* 0.632** 0.799** 0.567** 0.357** 0.311** 0.958** 0.984** 0.870**
Internal 0.550** 0.635** 0.659** 0.804** 0.642** 0.028 0.750** 0.761** 0.692** 0.451** 0.414** 0.825** 0.678** 0.647** 0.663**
IAAT 0.548** 0.629** 0.669** 0.815** 0.632** −0.004 0.779** 0.778** 0.716** 0.473** 0.417** 0.804** 0.659** 0.639** 0.639** 0.976**
NAIAT 0.501** 0.585** 0.580** 0.714** 0.598** 0.075 0.638** 0.664** 0.589** 0.462** 0.370** 0.781** 0.642** 0.598** 0.637** 0.943** 0.847**
Trunk 0.396** 0.822** 0.848** 0.919** 0.840** 0.047 0.740** 0.874** 0.711** 0.435** 0.421** 0.962** 0.911** 0.939** 0.854** 0.859** 0.865** 0.769**
IAAT/
ASAT
0.460** −0.026 0.012 0.187** −0.013 −0.071 0.369** 0.182** 0.330** 0.195** 0.169** 0.125 −0.064 −0.132* −0.024 0.571** 0.603** 0.467** 0.185**
All data are presented as Pearson partial correlation r values. Bold typeface indicates a significant correlation; shaded boxes indicate the anthropometric variable with the strongest correlation.
ASAT, abdominal subcutaneous adipose tissue; IAAT, intra-abdominal adipose tissue; IHCL, intrahepatocellular lipid; IMCL, intramyocellular lipid (S, soleus, T, tibialis); internal, total internal; NAIAT, non-abdominal internal
adipose tissue; NASAT, non-abdominal subcutaneous adipose tissue; SAT, subcutaneous adipose tissue; TAT, total adipose tissue; WC, waist circumference; WHR, waist-to-hip ratio; WHtR, waist-to-height ratio.
aStatistical analysis of IHCL (M: 234), S-IMCL (M: 239), and T-IMCL (M: 239) performed on log10 transformed variables.
*P < 0.05, **P < 0.01.

82 VOLUME 20 NUMBER 1 | JANUARY 2012 | www.obesityjournal.org
articles
integrative Physiology
Table 3 Linear correlation analysis between anthropometric measurements, lipid stores, and body fat stores in female subjects
Female Age Anthropometric variable Ectopic fat store a Adiposity depot (liters)
n = 234 Age Weight BMI WC Hip Height WHR WHtR IHCL S-IMCL T-IMCL TAT SAT ASAT NASAT Internal IAAT NAIAT Trunk
Weight 0.276**
BMI 0.357** 0.945**
WC 0.477** 0.862** 0.892**
Hip 0.350** 0.882** 0.874** 0.848**
Height −0.261* 0.114 −0.208* −0.049 0.052
WHR 0.456** 0.488** 0.530** 0.825** 0.405** −0.158*
WHtR 0.462** 0.734** 0.832** 0.973** 0.801** −0.227* 0.832**
IHCL 0.477** 0.501** 0.622** 0.623** 0.510** −0.309* 0.539** 0.644**
S-IMCL 0.501** 0.397** 0.450** 0.436** 0.402** −0.133 0.333** 0.402** 0.404**
T-IMCL 0.371** 0.196* 0.265** 0.278** 0.191* −0.162* 0.295** 0.239** 0.399** 0.448**
TAT 0.292** 0.959** 0.951** 0.854** 0.879** −0.032 0.461** 0.762** 0.584** 0.426** 0.174*
SAT 0.256** 0.959** 0.944** 0.834** 0.878** −0.013 0.431** 0.738** 0.532** 0.387** 0.138 0.995**
ASAT 0.227** 0.945** 0.937** 0.845** 0.874** −0.038 0.455** 0.753** 0.545** 0.384** 0.151 0.970** 0.974**
NASAT 0.266** 0.946** 0.927** 0.803** 0.852** −0.001 0.408** 0.706** 0.509** 0.378** 0.127 0.987** 0.992** 0.939**
Internal 0.424** 0.821** 0.850** 0.772** 0.704** −0.117 0.532** 0.721** 0.686** 0.505** 0.299** 0.879** 0.828** 0.807** 0.822**
IAAT 0.461** 0.796** 0.839** 0.787** 0.690** −0.165* 0.577** 0.745** 0.720** 0.506** 0.319** 0.852** 0.802** 0.799** 0.786** 0.974**
NAIAT 0.322** 0.773** 0.777** 0.657** 0.643** −0.031 0.408** 0.596** 0.552** 0.445** 0.232** 0.828** 0.781** 0.734** 0.790** 0.935** 0.829**
Trunk 0.302** 0.943** 0.949** 0.879** 0.874** −0.075 0.511** 0.797** 0.637** 0.448** 0.212** 0.978** 0.967** 0.986** 0.936** 0.887** 0.888** 0.792**
IAAT/
ASAT
0.446** 0.053 0.122 0.232** 0.075 −0.184* 0.335** 0.267** 0.373** 0.303** 0.304** 0.122 0.038 0.009 0.0535 0.509** 0.539** 0.409** 0.158*
All data are presented as Pearson partial correlation r values. Bold typeface indicates a significant correlation; shaded boxes indicate the anthropometric variable with the strongest correlation.
ASAT, abdominal subcutaneous adipose tissue; IAAT, intra-abdominal adipose tissue; IHCL, intrahepatocellular lipid; IMCL, intramyocellular lipid (S, soleus, T, tibialis); internal, total internal; NAIAT, non-abdominal internal
adipose tissue; NASAT, non-abdominal subcutaneous adipose tissue; SAT, subcutaneous adipose tissue; TAT, total adipose tissue; WC, waist circumference; WHR, waist-to-hip ratio; WHtR, waist-to-height ratio.
aStatistical analysis of IHCL (F: 169), S-IMCL (F: 179), and T-IMCL (F: 178) performed on log10 transformed variables.
*P < 0.05, **P < 0.01.

OBESITY | VOLUME 20 NUMBER 1 | JANUARY 2012 83
articles
integrative Physiology
with each other than individual percentage adiposity stores
(Tables 2 and 3).
Multiple regression analysis—Bayesian Information
Criterion
e goodness-of-t of six regression models using Bayesian
Information Criterion is shown in Supplementary Table S5.
WC provides the best model for the majority of adipose and
ectopic variables in male subjects. In female subjects, BMI con-
tributes the most for each dependent variable.
The TOFI phenotype
e gender-specic variable data from healthy, active indi-
viduals used to dene the TOFI phenotype can be found in
Supplementary Table S6. e mean IAAT/ASAT ratio for
healthy white individuals was 0.59 (male) and 0.25 (female).
Two standard deviations above the mean IAAT/ASAT of
healthy individuals (+2 s.d. male: 1.04, female: 0.45) was used
to dene the cuto for TOFI classication. us, individuals
with a BMI 18.5 < 25 kg/m2 with an IAAT/ASAT ratio above
1.0 (males) and 0.45 (females) were classied as TOFI; this cor-
responds to 14% of men (15/106) and 12% of women (17/132)
in our cohort. Gender-specic variable data for TOFI and
non-TOFI individuals can be found in Supplementary Table
S7. Signicantly greater IHCL, S-IMCL, and T-IMCL were
observed in both male and female TOFI female volunteers
(Supplementary Table S7). Subjects classied with the TOFI
phenotype based on their IAAT, could not be predicted from
BMI, WHR or WHtR or trunk fat because there was no signi-
cant dierence in these variables between TOFI and non-TOFI
healthy subjects (Supplementary Table S7).
WHR was the anthropometric variable which correlated most
strongly with the IAAT/ASAT ratio in both men (r = 0.369, P
< 0.01) (Table 2) and women (r = 0.335, P < 0.01) (Table 3).
e variation in IAAT/ASAT in male subjects with either an
identical BMI (24.0 kg/m2) or WC (84.0 cm) is illustrated in
Figure 3. Axial MRI scan data in Figure 4 demonstrates the
variation in ASAT and IAAT from two individuals with identi-
cal trunk fat (IAAT + ASAT).
DISCUSSION
ere is a growing recognition that the increased health risks
of obesity and metabolic syndrome are more strongly associ-
ated with central rather than total adiposity, with an excess in
IAAT and liver fat being the key determinants (1). In this study,
we have employed MRI-based techniques in order to determine
patterns of fat distribution in a large heterogeneous cohort and
used the ratio of intra-abdominal (IAAT) and subcutaneous
(ASAT) AT in healthy subjects as a means of identifying white
individuals at potentially increased metabolic risk. In addition,
we demonstrate a large variation in IAAT, ASAT, and ectopic fat
deposition in the liver and skeletal muscle that is not fully pre-
dicted by conventionally used anthropometric measurements.
BMI: 24kg/m2
WC: 84.5cm
TAT: 13.2 (I)
ASAT: 2.5 (I)
IAAT/ASAT: 0.43
IAAT: 1.07 (I)
BMI: 24kg/m2
WC: 88.0cm
TAT: 16.8 (I)
ASAT: 3.2 (I)
IAAT/ASAT: 0.69
IAAT: 2.2 (I)
BMI: 24kg/m2
WC: 92.0cm
TAT: 21.8 (I)
ASAT: 3.5 (I)
IAAT/ASAT: 1.03
IAAT: 3.6 (I)
BMI: 25.5kg/m2
WC: 84cm
TAT: 13.6 (I)
ASAT: 2.9 (I)
IAAT/ASAT: 0.24
IAAT: 0.5 (I)
BMI: 24.2kg/m2
WC: 84cm
TAT: 13.6 (I)
ASAT: 2.8 (I)
IAAT/ASAT: 0.42
IAAT: 1.2 (I)
BMI: 23.7kg/m2
WC: 84cm
TAT: 25.3 (I)
ASAT: 3.8 (I)
IAAT/ASAT: 1.14
IAAT: 4.3 (I)
Figure 3 Anthropometric variation in abdominal adiposity. Umbilical MRI images obtained from six different male subjects with a (a–c) BMI of 24.0,
or a (d–f) WC of 84.0 cm. ASAT, subcutaneous abdominal adipose tissue (l); IAAT, intra-abdominal adipose tissue (l); TAT, total adipose tissue (liters);
WC, waist circumference. In male subjects, the TOFI phenotype is attributed to individuals with BMI of 18.5–24.99 kg/m2 and a IAAT/ASAT ratio
greater or equal to 1.0.
Tr unk fat: 12.8 (I)
ASAT: 8.2 (I)
IAAT: 4.6 (I)
IAAT/ASAT: 0.56
Tr unk fat: 12.8 (I)
ASAT: 6.5 (I)
IAAT: 6.3 (I)
IAAT/ASAT: 0.97
Figure 4 Truncal variation in abdominal adiposity. Umbilical magnetic
resonance imaging (MRI) images obtained from two different male
subjects with an equal amount truncal fat (liters), but markedly different
IAAT and ASAT deposits. ASAT, subcutaneous abdominal adipose tissue
(l); IAAT, intra-abdominal adipose tissue (l).

84 VOLUME 20 NUMBER 1 | JANUARY 2012 | www.obesityjournal.org
articles
integrative Physiology
Age and gender effects on fat depots
Gender dierences in energy regulation and body fat distribu-
tion are well established; with women demonstrating propor-
tionally higher total AT and greater fat deposition in the lower
body, whereas men are predisposed to increased upper body fat
accumulation. Our results reect these ndings, with women
demonstrating signicantly higher percentage total and sub-
cutaneous fat stores than the males. ere is conicting data in
the literature regarding gender dierences in intra-abdominal
deposition, with some studies suggesting that men have greater
IAAT than women (18), while others reveal no dierence (35).
When correcting for age in our cohort we found signicantly
greater IAAT in males compared to females. In agreement with
previous studies, we reveal signicantly greater subcutane-
ous fat depots in male subjects as age increases but a variable
picture in female subjects (18,36). Our MRI data corroborate
previous observations indicating an abrupt reduction in sub-
cutaneous fat stores with the menopause as recorded in white
females in the 46–55 years compared to 36–45 years groups
(37) (Supplementary Table S4).
Ectopic fat stores
In addition to the major subcutaneous and abdominal fat com-
partments, ectopic fat depots in the liver and skeletal muscle
are implicated in the pathogenesis of insulin resistance, a key
facet of the metabolic syndrome (1). e ectopic fat hypoth-
esis suggests that a lack of sucient adipocytes and/or limited
capacity results in excess adipose storage around tissues and
organs such as the liver, heart, and kidneys (38). e exact
mechanism by which ectopic fat accumulation aects tissue
and organ function is unknown, but may include physical
compression, altering local secretory proles, and lipotoxicity
(1). More tangible is the strong association between increased
ectopic fat storage and obesity, type 2 diabetes mellitus, and
insulin resistance (39).
Here, we report male subjects demonstrate signicantly
higher levels of IHCL, similar levels of S-IMCL and lower lev-
els of T-IMCL compared to women, correcting for age. ere
are contradictory reports concerning the eects of age on
ectopic fat deposition, with one study reporting an age-related
increase in IHCL and IMCL (40) whereas others have found
no association (18,41). In our cohort, correlation analysis indi-
cates a signicant positive relationship between age and IHCL
and both IMCL ectopic fat depots in both male and female
subjects. In agreement with previous data showing a signi-
cant relationship between ectopic fat deposition and internal
adiposity (18,21) we demonstrate IHCL, S-IMCL, and T-IMCL
all correlate strongest to abdominal adiposity stores, in partic-
ular IAAT. As such, separate analysis with multiple methods of
data presentation would seem prudent when analyzing cohort
body composition data.
Anthropometric variables as markers of fat deposition
Anthropometric measurements are easily obtainable, inexpen-
sive, and commonly used determinants of both obesity and
the metabolic syndrome (27,42). We found anthropometric
variables generally correlated with total and subcutaneous
stores better than with internal depots or ectopic fat stores.
Previous studies have demonstrated that both WC and WHtR
correlate well with abdominal fat mass (both subcutaneous
and intra-abdominal) and cardiometabolic disease (43). ere
is a growing body of evidence endorsing WHtR as the best
measure of obesity and metabolic risk, regardless of ethnic-
ity (44,45) with the incorporation of adjustment for stature
the rationale behind its improvement in correlation over WC
(46). In our analysis, we found that both WC and WHtR cor-
related strongly with ectopic fat depots and percentage adipos-
ity stores. However, as illustrated in Figure 3 there is a large
amount of variation in abdominal fat at a given WC measure-
ment. Both correlation and multiple regression analysis reveal
a clearly dened segregation between the strongest anthro-
pometric determinants for either subcutaneous or internal
adipose stores, specic to gender. In agreement with a recent
study by Flegal et al. we found BMI to be the best anthropo-
metric predictor for individual adiposity stores in women (47).
By contrast, WC was the best predictor of individual adiposity
depot volumes in males, while WHtR was the best predictor
of IHCL in both genders. Overall, we found anthropometric
variables to be more closely related to each other than indi-
vidual adiposity or ectopic fat stores but generally correlated
well within gender.
Analyzing adipose distribution as a percentage of body mass
(%kg) or as a percentage of total AT (%TAT) is oen imple-
mented as it provides an insight into both fat-free mass and the
distribution of specic adipose stores, respectively, Correlation
and multiple regression analysis revealed similar gender pat-
terns of signicance between adiposity data transformed via
these two approaches and data analyzed as an absolute volume
in liters (data not shown).
Individuals matched for ASAT, with high levels of IAAT
demonstrate signicantly increased insulin resistance and
decreased glucose tolerance compared to those with less IAAT
(48–50). Several more recent studies have also shown that
IAAT is a stronger correlate of metabolic syndrome associated
risks than either ASAT or anthropometric variables (4,6,10).
Indeed, there is conicting data in the literature concerning
the use of multiple anthropometric variables to predict cardio-
vascular risk. It has been suggested that combining BMI with
WC increases the cardiovascular risk prediction than either
measure (51) whereas other data suggests BMI reduces the
discriminatory power for CVD risk factors, supporting the use
of WHtR as the sole measure of obesity (52).
Multiple regression analysis performed in our cohort sug-
gests that specic combinations of anthropometric variables
can be used to predict 60–70% of ASAT and IAAT values.
However, despite this signicant degree of correlation we reveal
a wide range of IAAT and ASAT values by BMI and age group.
In addition, we illustrate a signicant variation in IAAT and
ASAT in subjects with identical BMI’s or WC measurements
(Figure 3). Anthropometric variables such as WC and WHtR
can give no indication of the proportion of IAAT or ASAT in
seemingly “lean” subjects and are therefore inappropriate for

OBESITY | VOLUME 20 NUMBER 1 | JANUARY 2012 85
articles
integrative Physiology
classifying individuals that may be at increased metabolic risk
within a “normal” BMI range.
The “thin outside fat inside (TOFI)” subphenotype
Methods of measuring abdominal obesity such as Viscan and
dual-energy X-ray absorptiometry oer a faster, less expensive
alternative to the MRI protocol implemented here (53,54).
However, while there is no doubt that a strong correlation
exists between abdominal fat and metabolic risk, these meth-
ods are unable to dierentiate between individual abdominal
adipose stores. Here, we propose using a ratio of IAAT and
ASAT (IAAT/ASAT) to identify individuals at potentially
increased metabolic risk. We have determined the range of
IAAT and ASAT in healthy individuals in order to dene the
limits by which individuals with a proportionally elevated
IAAT, or TOFI (thin outside fat inside) can be classied. When
considering our entire cohort, this healthy control group rep-
resents 10–20% of individuals; the same proportion of healthy
individuals observed in the normal population (55). For ease
of reference, we propose using an IAAT/ASAT ratio cutos of
>1.0 in white male subjects and 0.45 in white female subjects
to dene this phenotype.
e components of the IAAT/ASAT ratio were selected in
order to identify patterns of abnormal body fat distribution.
We employed ASAT due to its abdominal location and poten-
tially protective role against cardiovascular and metabolic
risk compared to IAAT (9), as opposed to whole-body SAT
or NASAT. Anthropometric measurements, such as WHR
and WC, identify individuals with larger waists and hence
more total abdominal fat. As a predictor for abnormal body
fat distribution these measurements rely on a strong relation-
ship between total abdominal fat and visceral fat across a wide
range of abdominal fat. However, we found a notable varia-
tion in visceral adiposity observed in our healthy population,
who represent individuals with “normal” WCs, highlighting
the fact that on an individual basis visceral adiposity may vary
despite similar waist or abdominal fat. Indeed, as illustrated
in Figure 4, individuals can show identical amounts of trunk
fat and yet have entirely dierent amounts of IAAT and ASAT.
Other measurements that could potentially reect the TOFI
phenotype, such as fat-free mass or percentage body fat, would
only reect excess total adiposity either relative to body size
(fat-free mass index, fat mass index), lean tissue (FM:FFM
ratio) or as a percentage of weight (% body fat) and therefore
not accurately represent fat distribution. If used in combina-
tion with waist measurements these ratios may imply excess
adiposity around the abdomen, however this does not neces-
sarily reect visceral fat.
Correlation analysis revealed relatively weak associations
between the IAAT/ASAT ratio that we have used to dene TOFI
individuals and other physiological characteristics. Indeed, it
was age, a nonanthropometric variable, which provided the
strongest correlation to IAAT/ASAT. ere was a notable lack
of statistical dierence in anthropometric variables between
TOFI and non-TOFI individuals save for changes in internal
fat depots. ese data suggest that MRI analysis is currently
the only means of successfully identifying individuals with a
disproportionately high amount of intra-abdominal fat.
Our data and that of other studies have previously reported a
signicant positive relationship between liver fat and IAAT con-
tent (18,21). Furthermore, we and others have also found that
individuals with a phenotype opposite to that of the TOFI (the
so called “fat-t”), have reduced intra-abdominal and IHCL
compared to weight matched individuals (56,57). In addition
to a strong correlation between the IAAT/ASAT ratio and indi-
vidual ectopic fat depots, we also observed a signicant increase
in IHCL, S-IMCL, and T-IMCL in TOFI compared to non-TOFI
individuals, for both sexes. When our proposed IAAT/ASAT
cuto values were applied to individuals with an increased BMI
(greater than the 18.5 <25 kg/m2 range) we found that 16% of
females and 23% of males were registered as TOFI. is increase
in male classication is likely a reection of the increased propor-
tion and deposition of IAAT observed at a higher BMI in males.
Further work will be required to characterize healthy control
individuals within increased BMI ranges to accurately dene
those with excessive IAAT. Here, we attribute the TOFI phenotype
to 12–13% of European white volunteers that fall within a normal
BMI range (28). Additional studies may also reveal the applica-
bility of the TOFI index to additional ethnic populations given
the established dierences in body fat distribution between racial
groups (58,59). Currently, the TOFI index provides a quantitative
means of comparing intra-abdominal fat deposition. Clearly, the
utility of the TOFI phenotype to classify this “at risk” group of
individuals will only be fully realized once it has been correlated
with markers of the metabolic syndrome. Further work will also
be required to determine the physiological basis for the wide var-
iation in abdominal fat partitioning we have recorded here. e
mechanism is likely to be complex, with a multitude of genetic-,
environmental-and age-related determinants.
In summary, we reveal the pattern of regional adiposity in
a large cohort of UK-based volunteers, providing gender- and
age-specic reference range data and elucidate the relation-
ships between individual fat depots. We found anthropometric
variables to be more closely related to each other than adiposity
or ectopic fat stores but overall correlated well within gender.
Furthermore, our data demonstrate that specic anthropomet-
ric variables should be used to best predict individual adiposity
stores and ectopic fat stores for each gender; WC in men, and
BMI in women. Finally, we have used the ratio of IAAT and
ASAT in a dened “healthy” subset of our cohort to dene the
TOFI subphenotype, a potential means of evaluating abdomi-
nal obesity and identifying individuals at potentially increased
metabolic risk.
SUPPLEMENTARY MATERIAL
Supplementary material is linked to the online version of the paper at http://
www.nature.com/oby
ACKNOWLEDGMENTS
The authors are grateful to the United Kingdom Medical Research Council
(MRC) for financial support and the Statistical Advisory Service of Imperial
College London for statistical advice. We acknowledge infrastructure
support from the NIHR Biomedical Research Centre funding scheme and
would like to thank Prof J. Hajnal, Dr Larkman, Dr Morin, and Prof E. Leen for

86 VOLUME 20 NUMBER 1 | JANUARY 2012 | www.obesityjournal.org
articles
integrative Physiology
useful discussions and practical advice and Luis Alberto Gaete Balmaceda
por inspiratio.
DISCLOSURE
The authors declared no conflict of interest.
© 2011 The Obesity Society
REFERENCES
1. Després JP, Lemieux I. Abdominal obesity and metabolic syndrome. Nature
2006;444:881–887.
2. Mokdad AH, Ford ES, Bowman BA et al. Prevalence of obesity, diabetes,
and obesity-related health risk factors, 2001. JAMA 2003;289:76–79.
3. Miyazaki Y, Glass L, Triplitt C et al. Abdominal fat distribution and peripheral
and hepatic insulin resistance in type 2 diabetes mellitus. Am J Physiol
Endocrinol Metab 2002;283:E1135–E1143.
4. Demerath EW, Reed D, Rogers N et al. Visceral adiposity and its anatomical
distribution as predictors of the metabolic syndrome and cardiometabolic
risk factor levels. Am J Clin Nutr 2008;88:1263–1271.
5. Kuk JL, Katzmarzyk PT, Nichaman MZ et al. Visceral fat is an independent
predictor of all-cause mortality in men. Obesity (Silver Spring) 2006;14:
336–341.
6. Goldstone AP, Thomas EL, Brynes AE et al. Visceral adipose tissue and
metabolic complications of obesity are reduced in Prader-Willi syndrome
female adults: evidence for novel influences on body fat distribution. J Clin
Endocrinol Metab 2001;86:4330–4338.
7. Uusitupa M, Lindi V, Louheranta A et al.; Finnish Diabetes Prevention Study
Group. Long-term improvement in insulin sensitivity by changing lifestyles
of people with impaired glucose tolerance: 4-year results from the Finnish
Diabetes Prevention Study. Diabetes 2003;52:2532–2538.
8. Goodpaster BH, Thaete FL, Simoneau JA, Kelley DE. Subcutaneous
abdominal fat and thigh muscle composition predict insulin sensitivity
independently of visceral fat. Diabetes 1997;46:1579–1585.
9. Porter SA, Massaro JM, Hoffmann U et al. Abdominal subcutaneous
adipose tissue: a protective fat depot? Diabetes Care 2009;32:1068–1075.
10. Fox CS, Massaro JM, Hoffmann U et al. Abdominal visceral and
subcutaneous adipose tissue compartments: association with metabolic risk
factors in the Framingham Heart Study. Circulation 2007;116:39–48.
11. Langendonk JG, Kok P, Frölich M, Pijl H, Meinders AE. Decrease in visceral
fat following diet-induced weight loss in upper body compared to lower
body obese premenopausal women. Eur J Intern Med 2006;17:465–469.
12. Shojaee-Moradie F, Baynes KC, Pentecost C et al. Exercise training
reduces fatty acid availability and improves the insulin sensitivity of glucose
metabolism. Diabetologia 2007;50:404–413.
13. Snijder MB, Visser M, Dekker JM et al.; Health ABC Study. Low
subcutaneous thigh fat is a risk factor for unfavourable glucose and
lipid levels, independently of high abdominal fat. The Health ABC Study.
Diabetologia 2005;48:301–308.
14. Tran TT, Yamamoto Y, Gesta S, Kahn CR. Beneficial effects of subcutaneous
fat transplantation on metabolism. Cell Metab 2008;7:410–420.
15. Prentice AM, Jebb SA. Beyond body mass index. Obes Rev 2001;2:141–147.
16. Janssen I, Heymsfield SB, Allison DB, Kotler DP, Ross R. Body mass index
and waist circumference independently contribute to the prediction of
nonabdominal, abdominal subcutaneous, and visceral fat. Am J Clin Nutr
2002;75:683–688.
17. Thomas EL, Saeed N, Hajnal JV et al. Magnetic resonance imaging of total
body fat. J Appl Physiol 1998;85:1778–1785.
18. Machann J, Thamer C, Schnoedt B et al. Age and gender related effects
on adipose tissue compartments of subjects with increased risk for type 2
diabetes: a whole body MRI/MRS study. MAGMA 2005;18:128–137.
19. Consitt LA, Bell JA, Houmard JA. Intramuscular lipid metabolism, insulin
action, and obesity. IUBMB Life 2009;61:47–55.
20. Boden G. Effects of free fatty acids (FFA) on glucose metabolism:
significance for insulin resistance and type 2 diabetes. Exp Clin Endocrinol
Diabetes 2003;111:121–124.
21. Thomas EL, Hamilton G, Patel N et al. Hepatic triglyceride content and
its relation to body adiposity: a magnetic resonance imaging and proton
magnetic resonance spectroscopy study. Gut 2005;54:122–127.
22. Hwang JH, Stein DT, Barzilai N et al. Increased intrahepatic triglyceride
is associated with peripheral insulin resistance: in vivo MR imaging and
spectroscopy studies. Am J Physiol Endocrinol Metab 2007;293:
E1663–E1669.
23. Kotronen A, Seppälä-Lindroos A, Bergholm R, Yki-Järvinen H. Tissue
specificity of insulin resistance in humans: fat in the liver rather than muscle
is associated with features of the metabolic syndrome. Diabetologia
2008;51:130–138.
24. Lean ME, Han TS, Deurenberg P. Predicting body composition by
densitometry from simple anthropometric measurements. Am J Clin Nutr
1996;63:4–14.
25. Rico-Sanz J, Thomas EL, Jenkinson G et al. Diversity in levels of intracellular
total creatine and triglycerides in human skeletal muscles observed by (1)
H-MRS. J Appl Physiol 1999;87:2068–2072.
26. Thomas EL, Brynes AE, McCarthy J et al. Preferential loss of visceral fat
following aerobic exercise, measured by magnetic resonance imaging. Lipids
2000;35:769–776.
27. Physical status: the use and interpretation of anthropometry. Report of a WHO
Expert Committee. World Health Organ Tech Rep Ser 1995;854:1–452.
28. Obesity: preventing and managing the global epidemic. Report of a WHO
consultation. World Health Organ Tech Rep Ser 2000;894:i-253.
29. Dobbelsteyn CJ, Joffres MR, MacLean DR, Flowerdew G. A comparative
evaluation of waist circumference, waist-to-hip ratio and body mass index as
indicators of cardiovascular risk factors. The Canadian Heart Health Surveys.
Int J Obes Relat Metab Disord 2001;25:652–661.
30. Baecke JA, Burema J, Frijters JE. A short questionnaire for the
measurement of habitual physical activity in epidemiological studies. Am J
Clin Nutr 1982;36:936–942.
31. Hagströmer M, Bergman P, De Bourdeaudhuij I et al.; HELENA Study
Group. Concurrent validity of a modified version of the International Physical
Activity Questionnaire (IPAQ-A) in European adolescents: The HELENA
Study. Int J Obes (Lond) 2008;32 Suppl 5:S42–S48.
32. Royston P, Sauerbrei W. Building multivariable regression models with
continuous covariates in clinical epidemiology–with an emphasis on
fractional polynomials. Methods Inf Med 2005;44:561–571.
33. Royston P. A simple method for evaluating the Shapiro-Francia W’ test of
non-normality. Statistician 1983;32:297–300.
34. Schwartz G. Estimating the dimension of a model. Ann Stat 1978;6:461–464.
35. Westerbacka J, Cornér A, Tiikkainen M et al. Women and men have similar
amounts of liver and intra-abdominal fat, despite more subcutaneous fat in
women: implications for sex differences in markers of cardiovascular risk.
Diabetologia 2004;47:1360–1369.
36. Shen W, Punyanitya M, Silva AM et al. Sexual dimorphism of adipose tissue
distribution across the lifespan: a cross-sectional whole-body magnetic
resonance imaging study. Nutr Metab (Lond) 2009;6:17.
37. Tchernof A, Calles-Escandon J, Sites CK, Poehlman ET. Menopause,
central body fatness, and insulin resistance: effects of hormone-replacement
therapy. Coron Artery Dis 1998;9:503–511.
38. Heilbronn L, Smith SR, Ravussin E. Failure of fat cell proliferation,
mitochondrial function and fat oxidation results in ectopic fat storage, insulin
resistance and type II diabetes mellitus. Int J Obes Relat Metab Disord
2004;28 Suppl 4:S12–S21.
39. Rasouli N, Molavi B, Elbein SC, Kern PA. Ectopic fat accumulation and
metabolic syndrome. Diabetes Obes Metab 2007;9:1–10.
40. Cree MG, Newcomer BR, Katsanos CS et al. Intramuscular and liver
triglycerides are increased in the elderly. J Clin Endocrinol Metab
2004;89:3864–3871.
41. Tarasów E, Siergiejczyk L, Panasiuk A et al. MR proton spectroscopy in
liver examinations of healthy individuals in vivo. Med Sci Monit 2002;8:
MT36–MT40.
42. Alberti KG, Zimmet P, Shaw J. Metabolic syndrome–a new world-wide
definition. A Consensus Statement from the International Diabetes
Federation. Diabet Med 2006;23:469–480.
43. Klein S, Allison DB, Heymsfield SB et al.; Association for Weight Management
and Obesity Prevention; NAASO, The Obesity Society; American Society
for Nutrition; American Diabetes Association. Waist circumference and
cardiometabolic risk: a consensus statement from Shaping America’s Health:
Association for Weight Management and Obesity Prevention; NAASO,
The Obesity Society; the American Society for Nutrition; and the American
Diabetes Association. Am J Clin Nutr 2007;85:1197–1202.
44. Sayeed MA, Mahtab H, Latif ZA et al. Waist-to-height ratio is a better obesity
index than body mass index and waist-to-hip ratio for predicting diabetes,
hypertension and lipidemia. Bangladesh Med Res Counc Bull 2003;29:1–10.
45. Hsieh SD, Muto T. The superiority of waist-to-height ratio as an
anthropometric index to evaluate clustering of coronary risk factors among
non-obese men and women. Prev Med 2005;40:216–220.

OBESITY | VOLUME 20 NUMBER 1 | JANUARY 2012 87
articles
integrative Physiology
46. Ashwell M, Hsieh SD. Six reasons why the waist-to-height ratio is a rapid
and effective global indicator for health risks of obesity and how its use
could simplify the international public health message on obesity. Int J Food
Sci Nutr 2005;56:303–307.
47. Flegal KM, Shepherd JA, Looker AC et al. Comparisons of percentage body
fat, body mass index, waist circumference, and waist-stature ratio in adults.
Am J Clin Nutr 2009;89:500–508.
48. Pouliot MC, Després JP, Nadeau A et al. Visceral obesity in men.
Associations with glucose tolerance, plasma insulin, and lipoprotein levels.
Diabetes 1992;41:826–834.
49. Ross R, Aru J, Freeman J, Hudson R, Janssen I. Abdominal adiposity
and insulin resistance in obese men. Am J Physiol Endocrinol Metab
2002;282:E657–E663.
50. Ross R, Freeman J, Hudson R, Janssen I. Abdominal obesity, muscle
composition, and insulin resistance in premenopausal women. J Clin
Endocrinol Metab 2002;87:5044–5051.
51. Zhu S, Heshka S, Wang Z et al. Combination of BMI and Waist
Circumference for Identifying Cardiovascular Risk Factors in Whites. Obes
Res 2004;12:633–645.
52. Lee CM, Huxley RR, Wildman RP, Woodward M. Indices of abdominal
obesity are better discriminators of cardiovascular risk factors than BMI:
a meta-analysis. J Clin Epidemiol 2008;61:646–653.
53. Thomas EL, Collins AL, McCarthy J et al. Estimation of abdominal fat
compartments by bioelectrical impedance: the validity of the ViScan
measurement system in comparison with MRI. Eur J Clin Nutr 2010;64:
525–533.
54. Wells JC, Fewtrell MS. Measuring body composition. Arch Dis Child
2006;91:612–617.
55. Finucane MM, Stevens GA, Cowan MJ et al. National, regional, and
global trends in body-mass index since 1980: systematic analysis
of health examination surveys and epidemiological studies with 960
country-years and 9.1 million participants. Lancet 2011;12:557–567.
56. O’Donovan G, Thomas EL, McCarthy JP et al. Fat distribution in men
of different waist girth, fitness level and exercise habit. Int J Obes (Lond)
2009;33:1356–1362.
57. Stefan N, Kantartzis K, Machann J et al. Identification and characterization
of metabolically benign obesity in humans. Arch Intern Med 2008;168:
1609–1616.
58. Chandalia M, Lin P, Seenivasan T et al. Insulin resistance and body fat
distribution in South Asian men compared to Caucasian men. PLoS ONE
2007;2:e812.
59. Modi N, Thomas EL, Uthaya SN et al. Whole body magnetic resonance
imaging of healthy newborn infants demonstrates increased central adiposity
in Asian Indians. Pediatr Res 2009;65:584–587.