obesity and body fat classification in the metabolic ... article_62.pdf · obesity and body fat...
TRANSCRIPT
Obesity and Body Fat Classification in theMetabolic Syndrome: Impact onCardiometabolic Risk MetabotypeCatherine M. Phillips1,2, Audrey C. Tierney1, Pablo Perez-Martinez3,Catherine Defoort4, Ellen E. Blaak5, Ingrid M. F. Gjelstad6,7,Jose Lopez-Miranda3, Malgorzata Kiec-Klimczak8, Malgorzata Malczewska-Malec8,Christian A. Drevon6, Wendy Hall9, Julie A. Lovegrove9, Brita Karlstrom10,Ulf Ris�erus10 and Helen M. Roche1
Objective: Obesity is a key factor in the development of the metabolic syndrome (MetS), which is
associated with increased cardiometabolic risk. We investigated whether obesity classification by BMI
and body fat percentage (BF%) influences cardiometabolic profile and dietary responsiveness in 486
MetS subjects (LIPGENE dietary intervention study).
Design and Methods: Anthropometric measures, markers of inflammation and glucose metabolism, lipid
profiles, adhesion molecules, and hemostatic factors were determined at baseline and after 12 weeks of
four dietary interventions (high saturated fat (SFA), high monounsaturated fat (MUFA), and two low fat
high complex carbohydrate (LFHCC) diets, one supplemented with long chain n-3 polyunsaturated fatty
acids (LC n-3 PUFAs)).
Results: About 39 and 87% of subjects classified as normal and overweight by BMI were obese according to
their BF%. Individuals classified as obese by BMI (�30 kg/m2) and BF% (�25% (men) and �35% (women))
(OO, n ¼ 284) had larger waist and hip measurements, higher BMI and were heavier (P < 0.001) than those
classified as nonobese by BMI but obese by BF% (NOO, n ¼ 92). OO individuals displayed a more
proinflammatory (higher C reactive protein (CRP) and leptin), prothrombotic (higher plasminogen activator
inhibitor-1 (PAI-1)), proatherogenic (higher leptin/adiponectin ratio) and more insulin resistant (higher HOMA-IR)
metabolic profile relative to the NOO group (P < 0.001). Interestingly, tumor necrosis factor-a (TNF-a)concentrations were lower post-intervention in NOO individuals compared with OO subjects (P < 0.001).
Conclusions: In conclusion, assessing BF% and BMI as part of a metabotype may help to identify
individuals at greater cardiometabolic risk than BMI alone.
Obesity (2013) 21, E154-E161. doi:10.1038/oby.2012.188
Introduction
The prevalence of obesity is increasing worldwide, with the condi-
tion predicted to affect more than one billion people by the year
2020 (1). Excess adiposity, particularly central adiposity, is a key
causal factor in the development of insulin resistance, the hallmark
of the metabolic syndrome (MetS). In addition to abdominal obe-
sity the MetS is characterized by dyslipidemia and hypertension,
which are associated with increased risk of type 2 diabetes melli-
tus (T2DM) and cardiovascular disease (CVD) (2). A number of
adiposity measures are currently used as diagnostic tools in over-
weight and obesity classification including waist circumference
1 Nutrigenomics Research Group, UCD School of Public Health and Population Science, UCD Conway Institute, University College Dublin, Dublin, Ireland2 Department of Epidemiology and Public Health, University College Cork, Cork, Ireland 3 Lipid and Atherosclerosis Unit, IMIBIC/Reina Sofia UniversityHospital/University of Cordoba, and CIBER Fisiopatologia Obesidad y Nutricion (CIBEROBN), Instituto de Salud Carlos III, Cordoba, Spain 4 INSERM, 476Human Nutrition and Lipids, INRA, 1260, University M�editerran�ee Aix-Marseille 2, Marseille, France 5 Department of Human Biology, Nutrition andToxicology Research Institute Maastricht (NUTRIM), Maastricht,The Netherlands 6 Department of Nutrition, Institute of Basic Medical Sciences, University ofOslo, Oslo, Norway 7 Department of Clinical Endocrinology, Oslo University Hospital Aker, Oslo, Norway 8 Department of Clinical Biochemistry, JagiellonianUniversity Medical College, Krakow, Poland 9 Hugh Sinclair Unit of Human Nutrition and Institute for Cardiovascular and Metabolic Research, Department ofFood and Nutritional Sciences, University of Reading, Reading, UK 10 Department of Public Health and Caring Sciences/Clinical Nutrition and Metabolism,Uppsala University, Uppsala, Sweden. Correspondence: Helen M. Roche ([email protected])
Disclosure: The authors declared no conflict of interest. See the online ICMJE Conflict of Interest Forms for this article.
Received: 13 January 2012 Accepted: 31 May 2012 First published online by Nature Publishing Group on behalf of The Obesity Society 9 August 2012.
doi:10.1038/oby.2012.188
E154 Obesity | VOLUME 21 | NUMBER 1 | JANUARY 2013 www.obesityjournal.org
Original ArticleEPIDEMIOLOGY/GENETICS
Obesity
(WC), BMI, and body fat percentage (BF%). WC is the only adi-
posity measure included in the current International Diabetes Fed-
eration and National Cholesterol Education Program’s Adult Treat-
ment Panel III report (NCEP ATP III) MetS definitions. However,
WC does not take whole body fat distribution into consideration.
Moreover, prevalence of the MetS has been shown to increase
across BMI categories with approximately twofold higher preva-
lence in the severely obese compared with nonobese (3). However
BMI, the traditional diagnostic tool, is also limited because it does
not discriminate between lean and fat body mass. Recent data
from a large cross-sectional study suggest that using BMI may
under estimate obesity prevalence defined as excess body fat, par-
ticularly in overweight individuals (4). Simultaneous comparison
of the association between WC, BMI and BF% with CVD risk
showed that WC and BF% were more strongly associated with
MetS and CVD risk, respectively (5). Furthermore, recent examina-
tion of markers of glucose metabolism according to obesity classi-
fication revealed that BF% may be a better determinant for pre-di-
abetes and T2DM development (6).
Ideally, obesity prevention would reduce risk of associated cardiometa-
bolic conditions, although several current approaches are ineffective,
probably due, at least in part, to lack of prompt identification, diagnosis,
and appropriate treatment of obese individuals, together with genetic het-
erogeneity and differences in dietary responsiveness. Thus, there is a
need to improve obesity diagnosis and to develop new preventative strat-
egies and evidence-based public health measures to attenuate disease de-
velopment and reduce dependence on medical care, particularly among
individuals with increased cardiometabolic risk. Comparative data on
whether obesity classification by BMI and BF% influence the cardiome-
tabolic profile of individuals with the MetS is currently unavailable. Con-
sidering the increasing prevalence of the MetS and its associated cardio-
metabolic risk, the main objective of this paper was to examine a
comprehensive panel of risk factors in MetS individuals comparing those
classified as nonobese by BMI and obese by BF% (NOO) to subjects
classified as obese by both BMI and BF% (OO). Another novel aim of
this work was to assess whether obesity classification influences dietary
responsiveness in the MetS. Examination of whether the complementary
use of BF% and BMI to define the obese metabotype, or metabolic phe-
notype, in MetS is more effective in detecting individuals at greater cardi-
ometabolic risk than BMI alone may have public health implications in
terms of improving obesity classification in high-risk groups.
Methods and ProceduresSubjects aged 35-70 years and BMI 20-40 kg/m2 were recruited for
the LIPGENE dietary intervention study from eight European coun-
tries (Ireland, UK, Norway, France, The Netherlands, Spain, Poland,
and Sweden) all conforming to the Helsinki Declaration of 1975 as
revised in 1983. The study was registered with The US National
Library of Medicine Clinical Trials registry (NCT00429195). Sub-
ject eligibility was determined using a modified version of the
NCEP criteria for MetS (7), where subjects were required to fulfill
at least three of the following five criteria: waist circumference
>102 cm (men) or >88 cm (women); fasting glucose 5.5-7.0 mmol/
l; triglycerides �1.5 mmol/l; high-density lipoprotein cholesterol
(HDL-C) <1.0 mmol/l (men) or <1.3 mmol/l (women); blood pres-
sure �130/85 mmHg or treatment of previously diagnosed hyperten-
sion. We used the preintervention data for 486 subjects and the post-
intervention data for the 417 subjects completing the intervention.
Detailed characteristics of this cohort have been published (8).
Dietary interventionParticipants were recruited to a 12-week dietary intervention after
being randomly allocated to one of the four following diets: high-
fat (38% energy) SFA-rich diet (16% SFA, 12% MUFA, 6%
PUFA (HSFA); high-fat (38% energy), MUFA-rich diet (8% SFA,
20% MUFA, 6% PUFA) (HMUFA); isoenergetic low fat (28%
energy), high complex carbohydrate diet (8% SFA, 11% MUFA,
6% PUFA), with 1 g/day high-oleic sunflower oil supplement
(LFHCC); isoenergetic low-fat (28% energy), high complex carbo-
hydrate diet (8% SFA, 11% MUFA, 6% PUFA), with 1.24 g/day
LC n-3 PUFA supplement (LFHCC n-3). Randomization was per-
formed using age, gender, and fasting plasma glucose concentra-
tion as matching variables, applying Minimisation Programme for
Allocating patients to Clinical Trials (Department of Clinical Epi-
demiology, The London Hospital Medical College, UK). The LC
n-3 PUFA supplement (Marinol C-38; 1.24 g per day LC n-3
PUFA) and control high-oleic acid sunflower seed oil supplement
were supplied by Lipid Nutrition, Loders Croklaan (Wormerveer,
The Netherlands). More details about dietary models have been
published elsewhere (9).
Anthropometric and clinical measurementsAnthropometric measurements were recorded according to a standar-
dized protocol for the LIPGENE study. Bio-electric impedance
measures of body composition were performed by a multi-frequency
tetra-polar device (Tanita BIA machine; Tanita, Arlington Heights,
IL) (10). The subjects were placed in the supine position with arms
comfortably abducted from the body at 15� and legs spread comfort-
ably. Two current-injection electrodes were placed at the right hand
and foot on the dorsal surfaces proximal to the metacarpal-phalan-
geal and metatarsal-phalangeal joints, respectively. The centers of
two voltage-detector electrodes were placed on the midline between
the prominent ends of the right radius and ulna of the wrist, and
midline between the medial and lateral malleoli of the right ankle.
The black current-injection and red voltage electrode detectors were
at least 5 cm apart, respectively. The black current-injection lead
alligator clips and the red voltage-detector lead alligator clips were
connected to the electrodes placed on the right hand and foot and
right wrist and ankle, respectively. The most frequently used cutoff
points for BF% defining obesity (�25% in men and �35% in
women) were used (11-13). Blood pressure was measured according
to the European Society of Hypertension Guidelines.
Biochemical measurementsPlasma and serum were prepared from 12-h fasting blood samples in
each subject. Serum insulin was measured by solid-phase, two-site
fluoroimmunometric assay on a 1235 automatic immunoassay sys-
tem (AutoDELFIA kits; Wallac Oy, Turku, Finland). Plasma glucose
concentrations were measured using the IL Test Glucose Hexokinase
Clinical Chemistry kit (Instrumentation Laboratories, Warrington,
UK). Homeostasis model assessment of insulin resistance (HOMA-
IR) was derived from fasting glucose and insulin concentrations as
follows ((fasting plasma glucose � fasting serum insulin)/22.5) (14).
Quantitative insulin-sensitivity check index, a measure of insulin
sensitivity, was calculated as ¼ (1/(log fasting insulin þ log fasting
glucose þ log fasting free fatty acid)) (15). An insulin-modified in-
travenous glucose tolerance test was performed. Measures of insulin
sensitivity (sensitivity index) were determined using the MINMOD
Millenium Program (version 6.02, Richard N. Bergman). The acute
insulin response to glucose (AIRg ¼ first phase insulin response)
Original Article ObesityEPIDEMIOLOGY/GENETICS
www.obesityjournal.org Obesity | VOLUME 21 | NUMBER 1 | JANUARY 2013 E155
was defined as the incremental area under the curve from time 0-8
min. Disposition index (DI) was calculated as the product of acute
insulin response to glucose and sensitivity index. Cholesterol and tri-
glycerides were quantified using the IL TestCholesterol kit and IL
Tes Triglycerides kit (Instrumentation Laboratories). The IL Test
HDL-C Kit (Instrumentation Laboratories) was used for direct quan-
tification of HDL-cholesterol. The WAKO NEFA C enzymatic color
kit (Alpha Laboratories, Hampshire, UK) was used to quantify
plasma non-esterified fatty acids concentration. Plasma concentra-
tions of adiponectin, leptin, and resistin were measured by enzyme-
linked immunosorbent assay (ELISA) (DuoSet ELISA Development
System DY1065, DY398, AND DY1359; R&D Systems, Minneapo-
lis, MN). Plasma concentrations of C reactive protein (CRP) were
determined by high-sensitivity ELISA (BioCheck, Foster City, CA).
Tumor necrosis factor-a (TNF-a) and interleukin 6 were measured
by ultra sensitive ELISA (R&D Systems, Abingdon, UK and
Biosource International, Camarillo, CA). Intracellular and vascular
adhesion molecules were measured by ELISA (R&D Systems, Abing-
don, UK). Plasminogen activator inhibitor-1 (PAI-1) was determined
by the immunoactivity assay Chromolize PAI-1 (Trinity Biotech,
Bray, Ireland) and tissue plasminogen activator (tPA) was measured
by ELISA (Affinity Biologicals, Ancaster, Ontario, Canada).
Statistical analysisData are presented as means 6 s.e.m. Statistical analyses were car-
ried out using SPSS version 18.0 for Windows (SPSS, Chicago, IL).
Biochemical variables were assessed for normality of distribution,
and skewed variables were normalized by log10 or square root trans-
formation as appropriate. Cutoff points for BF% defining obesity in
adult populations (�25% in men and �35% in women) are those
most frequently used in the literature, which include examination of
a number of European populations and a meta-analysis (11-13).
Individuals identified as being obese by BMI (�30 kg/m2) and
obese according to their BF% (�25% in men and �35% in women)
were classified as OO (n ¼ 284). Individuals identified as nonobese
by BMI (BMI <30 kg/m2) and as obese by their BF% were classi-
fied as NOO (n ¼ 92). Differences between groups were analyzed
by two-tailed Student’s t-tests. To examine dietary responsiveness,
post-intervention changes (post-intervention minus baseline) for each
group were also compared. ANOVA-based models (with Bonferroni
correction) were then used to test for associations in each of the
four dietary arms to detect specific effects of the different dietary
interventions. Correlations between two variables were computed by
Spearman correlation coefficient. For all analyses a P value of
<0.05 was considered significant.
ResultsAnthropometric measures and clinicalcharacteristics of MetS subjectsAccording to their BMI, 2.8%, 27.5%, and 69.7% of the MetS sub-
jects participating in this study were classified as normal, over-
weight, and obese. When BF% was used to classify individuals
5.9%, 10.9%, and 83.2% of the study population were identified as
normal, overweight, and obese. Clinical and anthropometric charac-
teristics of the study population according to both obesity classifica-
tions are presented in Table 1. In addition to greater anthropometric
TABLE 1 Anthropometric and clinical characteristics of the study population according to their BMI and percentage body fat
Obesity Obesity and BF Classification in MetS Phillips et al.
E156 Obesity | VOLUME 21 | NUMBER 1 | JANUARY 2013 www.obesityjournal.org
measures (P < 0.001), obese individuals displayed raised CRP, lep-
tin, and insulin concentrations and were more insulin resistant (P <0.005) compared with nonobese subjects regardless of which classi-
fication was used to define obesity. Use of BF% alone identified
higher blood concentration of TNF-a, resistin, and fibrinogen con-
centrations in the obese individuals (Table 2) (P < 0.05). Use of
BMI alone identified higher PAI-1 and tPA concentrations, and
lower insulin sensitivity in the obese subjects (Table 3) (P < 0.005).
TABLE 2 Inflammatory markers of the study population according to their BMI and percentage body fat
TABLE 3 Measures of glucose homeostasis and plasma lipid profiles of the study population according to their BMI andpercentage body fat
Original Article ObesityEPIDEMIOLOGY/GENETICS
www.obesityjournal.org Obesity | VOLUME 21 | NUMBER 1 | JANUARY 2013 E157
Obesity classification of MetS subjectsExamination of the use of both body composition tools revealed that
38.5% of the MetS cases classified as normal weight by BMI were
actually obese when classified by BF%. This observation was unique
to the female subjects (46% classified as lean by BMI were actually
obese according to BF%). Although it might be expected that
women would have higher BF% for a given BMI than men, it
should also be noted that this is a MetS only cohort and the numbers
of individuals classified as normal weight is small according to their
BMI. Of all MetS individuals classified as overweight by BMI, 87%
were actually obese when classified by BF%. Again, this discrep-
ancy in classification was higher for women (84% of those classified
as overweight were actually obese when classified by BF%) than for
men (53%). In contrast, none of the subjects classified as obese by
BMI were normal weight according to BF%.
BMI showed strong positive correlations with body weight (r ¼0.66, P < 0.0001), waist circumference (r ¼ 0.62, P < 0.0001) and
to a lesser extent with BF% (r ¼ 0.38, P < 0.0001) in the whole
population. Interestingly, following stratification by gender, stronger
correlations were observed in the male subjects between BMI and
BF% (r ¼ 0.64 and r ¼ 0.36, P < 0.0001, for men and women,
respectively) and waist circumference (r ¼ 0.83, and r ¼ 0.60, P <0.0001, for men and women, respectively), with similar correlations
between BMI and body weight in both men and women (r ¼ 0.78
and r ¼ 0.78, P < 0.0001).
Impact of combined BMI and BF% obesityclassification on cardiometabolic riskCharacteristics of the study population stratified by obesity classifi-
cation are presented in Table 4. Individuals classified as obese by
both BMI and BF% (OO, n ¼ 2 84) were younger and comprised
more male subjects compared with individuals classified as
nonobese by BMI and obese by BF% (NOO, n ¼ 92). OO individu-
als had larger waist and hip measurements, higher BMI, and were
heavier due to greater lean and fat mass (kg) and body water (liters)
(P < 0.001) compared with the NOO subjects. OO individuals
displayed a more insulin resistant, proinflammatory, prothrombotic
and proatherogenic profile characterized by higher CRP, leptin and
PAI-1 concentrations and a greater leptin/adiponectin ratio (Table 5)
and lower insulin-sensitivity and higher insulin-resistance indexes
(Table 6) relative to the NOO group (P < 0.001). Interestingly, OO
subjects had more favorable plasma lipids with lower total and low-
density lipoprotein cholesterol compared with the NOO subjects
(Table 6) (P < 0.05). However, this did not translate into significant
differences between groups with respect to atherogenic lipid indexes
(low-density lipoprotein cholesterol/HDL, Log (triglycerides/HDL-
cholesterol)) and total cholesterol/HDL-C; not shown) probably due
to lower HDL cholesterol concentrations in the OO subjects (P <0.05). Despite the gender difference for obesity classification
according to BMI and BF% and between OO and NOO groups, it is
worthwhile to note that separate comparisons of OO vs. NOO
groups in the male and female subjects mirrored the findings for the
entire cohort (data not shown), with the exception of BF% which
was higher in both OO male (32.7 6 0.4 vs. 29.3 6 0.6, P < 0.05)
TABLE 4 Clinical and anthropometric characteristics accordingto combined BMI and percentage body fat obesityclassification
TABLE 5 Concentrations of inflammatory markers, adhesionmolecules and haemostatic factors according to combinedBMI and percentage body fat obesity classification
Obesity Obesity and BF Classification in MetS Phillips et al.
E158 Obesity | VOLUME 21 | NUMBER 1 | JANUARY 2013 www.obesityjournal.org
and female subjects (44.1 6 0.3 vs. 42.7 6 0.5, P < 0.05) relative
to their NOO counterparts.
Obesity classification and dietary responsivenessChanges (post-intervention minus baseline) in each of the cardiome-
tabolic profile parameters for the NOO and OO individuals were
compared. Following the intervention, the NOO subjects demon-
strated a significant reduction in TNF-a concentrations (P < 0.001)
compared with the OO individuals. When the individual dietary
interventions were analyzed separately to ascertain whether this
finding was a diet-specific effect, 52% and 31% reductions (com-
pared with baseline) in TNF-a concentrations were observed in the
NOO subjects following the HSFA (P < 0.01) and HMUFA (P <0.05) interventions, respectively ( Figure 1). Moreover, compared
with pre-intervention, NOO individuals demonstrated post-interven-
tion reductions in plasma concentrations of CRP (4.21 6 0.37 vs
3.51 6 0.39 mg/l, P < 0.05) and resistin (9.40 6 1.04 vs 7.06 6
1.05 mg/ml, P < 0.05) following the LFHCC LC n-3 PUFA diet and
a BF% loss following the LFHCC diet (38.9 6 1.8 vs 37.0 6 1.9%,
P < 0.05). No changes in markers of glucose homeostasis, adhesion
molecules, and haemostatic factors or lipids were noted in either
group after 12 weeks of dietary intervention.
DiscussionThe National Health and Nutrition Examination Survey (1999-2004)
revealed that 24% of normal weight adults were metabolically
abnormal whereas 51% of overweight and 32% obese adults were
metabolically healthy (16). There has been much interest in the
paradoxical finding of individuals considered inappropriately healthy
for their degree of obesity and subsequently several phenotype sub-
groups of obesity have been described including metabolically
healthy or insulin-sensitive obese, metabolically obese but normal
weight and more recently taking BF% into account, normal weight
obese (17,18,19). The aim of the current work was to examine cardi-
ometabolic risk metabotype in obese and nonobese adults with the
MetS and BF% in the obese range. We found that 39% and 87% of
the MetS cases classified as normal and overweight by BMI had a
BF% in the obese range, suggesting that use of BMI alone to diag-
nose obesity underestimates BF%, particularly in overweight MetS
subjects. These data support earlier findings from a large cross-sec-
tional study which reported that 29% of individuals classified as
normal weight (BMI �24.9 kg/m2) and 80% of individuals charac-
terized as overweight (BMI 25-29.99 kg/m2) had a BF% within the
obese range (4). The discrepancy in classification of normal weight
and overweight by BMI as obese by BF% reported in our study was
higher for women. We also report stronger correlations between
BMI and both BF% and waist in the male subjects. Whether inclu-
sion of BF% with BMI in defining physiologically relevant obesity
FIGURE 1 Plasma concentrations of tumor necrosis factor-a (TNF-a) among meta-bolic syndrome subjects in the LIPGENE study. A significant change (post-interven-tion minus baseline) in TNF-a concentrations was noted for the NOO compared tothe OO individuals (P < 0.001). Post-intervention reductions in plasma concentra-tion of TNF-a were observed among the NOO subjects (a) following the HSFA (P <0.01) and HMUFA diets (P < 0.05). (b) No significant changes were noted in theOO individuals following any of the four dietary interventions. Pre-intervention TNF-a concentrations are depicted as black bars and post-intervention TNF-a concen-trations are shown as white bars. LFHCC, low fat high complex carbohydrate.
TABLE 6 Indexes of glucose metabolism and plasma lipidprofiles according to combined BMI and percentage bodyfat obesity classification
Original Article ObesityEPIDEMIOLOGY/GENETICS
www.obesityjournal.org Obesity | VOLUME 21 | NUMBER 1 | JANUARY 2013 E159
is more important in women is unknown. It should be noted that the
number of normal weight individuals in this cohort was small (2.8%)
and that these findings may be a reflection of the greater number of
females in the study, the limitations of the BMI tool and gender dif-
ferences in BF% and fat/lean tissue mass distribution. When BMI and
BF% were used in conjunction individuals classified as obese by both
tools (OO) displayed a more insulin resistant, proinflammatory, pro-
thrombotic, and proatherogenic profile compared with subjects classi-
fied as nonobese by BMI with BF% in the obese range (NOO). These
findings were mirrored in both male and female subjects. Thus, com-
plementary use of both diagnostic tools has the potential to detect
individuals at greater cardiometabolic risk.
The NCEP ATP III identified a proinflammatory state as an impor-
tant MetS characteristic (20). Chronic low-grade inflammation plays
a role in the pathogenesis of insulin resistance, with elevated circu-
lating levels of CRP and the proinflammatory cytokines such as
TNF-a associated with greater risk of having T2DM and MetS
(21,22,23). In normal weight obese women without the MetS, con-
centrations of proinflammatory cytokines were higher than in the
nonobese group and intermediate to a preobese/obese group, sug-
gesting that these biomarkers might be prognostic indicators of the
risk of obesity, MetS, and CVD in normal weight obese women
(24). Given the central role of obesity in the pathogenesis of these
cardiometabolic diseases, the adipose tissue-derived inflammatory
mediators adiponectin and leptin may also be particularly important.
Circulating plasma levels of adiponectin are reduced in obese and
T2DM subjects (25). In contrast, plasma leptin levels increase pro-
portionally with fat mass and have been shown to be a predictor of
CVD in both case-control and prospective studies (26,27). In recent
years, the leptin/adiponectin ratio has been suggested as an athero-
sclerotic index and as a useful parameter to assess insulin resistance
in patients with and without T2DM (28,29).
We demonstrated that MetS individuals with both BMI and BF% in
the obese range were more insulin resistant, had higher plasma con-
centrations of CRP, leptin and PAI-1 and a greater leptin/adiponectin
compared with subjects classified as obese by BF% with a normal
BMI. We did not observe any differences in adiponectin levels
between obese and nonobese MetS subjects or between NOO and OO
individuals. However, considering that the adiponectin concentrations
reported in our study are low in all subjects, it may be that the obe-
sity-related reduction in adiponectin levels per se is diminished
against a background of numerous metabolic perturbations which also
contribute to reduced adiponectin levels. While the Gomez-Ambrosi
et al., study did not measure adiponectin, they did show higher leptin
concentrations in men and women, and higher HOMA-IR values in
women with BMI and BF% in the obese range as compared with
those with normal BMI and BF% (4). Plasma CRP concentrations
were not different between these groups. It should be noted that that
study was a cross-sectional investigation and used an air displacement
plethysmographic method to estimate BF%, whereas our data relate
to a MetS only cohort wherein BF% was determined by bioelectrical
impedance. Our method provides a cost-effective and direct determi-
nation of total body composition, which is comparable in terms of
accuracy of BF% determination with dual-energy X-ray absorptiome-
try (30). Our data support the notion of BF% determination by
bioelectrical impedance as a valuable additional diagnostic tool.
Surprisingly, the HOMA-IR values for the NOO subjects were
below the cutoff point for insulin resistance (>2.61) (11,14); thus,
these individuals might be considered as insulin-sensitive obese.
Investigation of insulin signaling and inflammatory pathways in in-
sulin-sensitive and insulin-resistant severely obese (IRMO) subjects,
support the concept that insulin-sensitive severely obese subjects
have a lower inflammatory response than insulin-resistant morbidly
obese patients (31). In a recent study of obese (by BMI) 70-79 year
individuals with and without the MetS, the metabolically healthy (or
non MetS) obese subjects had a more favorable inflammatory profile
(lower plasma concentrations of TNF-a and PAI-1) and body fat dis-
tribution than the obese MetS individuals, despite both groups hav-
ing BMI and BF% in the obese range (32). Examination of the waist
to hip ratio in our current study revealed that OO subjects had a
higher waist to hip ratio, suggesting that they carried more abdomi-
nal weight than the NOO individuals. Leg fat has been associated
with more favorable metabolic and inflammatory profiles (33,34)
and visceral, but not abdominal subcutaneous fat, has been linked
with higher plasma concentrations of IL-6 and CRP (35). It would
be interesting to determine whether body fat depots were different
between the NOO and OO groups in the current work.
A novel finding in our study is the difference in dietary responsive-
ness between the NOO and OO subjects. No changes in any plasma
measurements were noted after intervention in the OO subjects. In
contrast, TNF-a concentrations were significantly reduced in the
NOO subjects. When each of the four dietary arms were analyzed
separately, reductions in plasma concentrations of TNF-a were
observed following the HSFA and HMUFA interventions, whereas
CRP and resistin concentrations were reduced following the LFHCC
LC n-3 PUFA diet. NOO subjects also experienced a BF% reduction
following the LFHCC diet. Cross-sectional, intervention and experi-
mental data suggest that high-fat diets promote obesity, insulin re-
sistance and inflammation, driving the development of MetS,
T2DM, and CVD (36,37). Epidemiological studies also demonstrate
anti-inflammatory effects of dietary fish, fish oil, and/or LC n-3
PUFA consumption (38,39). We recently reported that the LFHCC
LC n-3 PUFA diet reduced triglycerides-related MetS phenotypes
and the risk of having the MetS in this cohort (9,40). While the
reduction in TNF-a concentrations following the HSFA and
HMUFA diets contradicts the literature, the beneficial effects
observed after the low-fat interventions in the NOO group were not
entirely unexpected. However, why the NOO, and not the OO group,
appear to be responsive remains unclear. Although speculative, it
may be that NOO subjects who are more insulin sensitive and have
less proinflammatory, prothrombotic, and proatherogenic profiles
compared with the OO subjects have greater metabolic flexibility to
adapt to changes in dietary fat. Perhaps coordination of the path-
ways involved in nutrient handling, insulin signaling, inflammation,
and lipid metabolism is less disturbed than in the OO subjects who
are simply metabolically overburdened and no longer dietary respon-
sive. Whatever the explanation, these data suggest that not only are
the OO subjects, who represent almost 60% of our MetS cohort, at
greater cardiometabolic risk but that they are less responsive to die-
tary intervention. Whether these individuals would have more reduc-
tion of cardiometabolic risk by lifestyle and behavioral intervention
alone or in combination with dietary changes is unknown but would
be worth examining further.
To our knowledge, this is the first study to investigate whether obe-
sity classification by both BMI and BF% may influence cardiometa-
bolic risk metabotypes and dietary responsiveness in the MetS. Our
study has a number of strengths including relatively large subject
Obesity Obesity and BF Classification in MetS Phillips et al.
E160 Obesity | VOLUME 21 | NUMBER 1 | JANUARY 2013 www.obesityjournal.org
numbers, comprehensive determination of insulin and glucose me-
tabolism by static (glucose and insulin plasma concentrations) and
dynamic (disposition index, Si, HOMA-IR, and acute insulin
response to glucose) indexes, and a 12-week dietary intervention.
Despite these strengths our study presents some limitations. First,
more comprehensive examination of body fat distribution would be
advantageous. Although bioelectrical impedence tends to overesti-
mate BF% in normal weight subjects but tends to underestimate
BF% in obese individuals, such potential misclassification would
however, if anything, result in underestimating the degree of body
fatness in some of the ‘‘true’’ obese subjects and thus merely under-
estimate the present associations. Second, the lack of a follow-up
assessment to determine if post-intervention changes observed fol-
lowing a 12-week intervention might be altered after long-term
intervention. Finally, the cross-sectional study design does not allow
causality to be established. In conclusion, we have demonstrated
that the combined use of BF% and BMI may be more useful in
identifying individuals with a greater cardiometabolic risk metabo-
type than BMI alone. This finding may be particularly important in
the MetS considering the prevalence of obesity and increased CVD
risk associated with this condition.O
AcknowledgmentsThis work was supported by the European Commission, Framework
Programme 6 (LIPGENE), contract number FOOD-CT-2003-505
944; Johan Throne Holst Foundation for Nutrition Research, Freia
Medical Foundation. The CIBEROBN is an initiative of the Instituto
de Salud Carlos III, Madrid, Spain.
VC 2012 The Obesity Society
References1. Flier JS. Obesity wars: molecular progress confronts an expanding epidemic. Cell
2004;116:337-350.
2. Moller DE, Kaufman KD. Metabolic syndrome: a clinical and molecular perspec-tive. Annu Rev Med 2005;56:45-62.
3. Esteghamati A, Khalilzadeh O, Anvari M et al. Metabolic syndrome and insulin re-sistance significantly correlate with body mass index. Arch Med Res 2008;39:803-808.
4. Gomez-Ambrosi J, Silva C, Galofre JC, et al. Body mass index classification missessubjects with increased cardiometabolic risk factors related to elevated adiposity. IntJ Obes (Lond) 2012;36:286-294.
5. Dervaux N, Wubuli M, Megnien JL, Chironi G, Simon A. Comparative associationsof adiposity measures with cardiometabolic risk burden in asymptomatic subjects.Atherosclerosis 2008;201:413-417.
6. G�omez-Ambrosi J, Silva C, Galofr�e JC et al. Body adiposity and type 2 diabetes:increased risk with a high body fat percentage even having a normal BMI. Obesity(Silver Spring) 2011;19:1439-1444.
7. Executive Summary of The Third Report of The National Cholesterol EducationProgram (NCEP) Expert Panel on Detection, Evaluation, And Treatment of HighBlood Cholesterol In Adults (Adult Treatment Panel III). JAMA 2001;285:2486-97.
8. Shaw DI, Tierney AC, McCarthy S, et al. LIPGENE food-exchange model for alter-ation of dietary fat quantity and quality in free-living participants from eight Euro-pean countries. Br J Nutr 2009:101:750-759.
9. Tierney AC, McMonagle J, Shaw DI et al. Effects of dietary fat modification on in-sulin sensitivity and on other risk factors of the metabolic syndrome-LIPGENE: aEuropean randomized dietary intervention study. Int J Obes (Lond) 2011;35:800-809.
10. Matthie J. In: technologies X, (ed). Hydra ECF/ICF Bio-Impedance Analyzer(Mode 4200) Operational Manual Revision 1.01. Xitron Technologies: San Diego,CA, 1997, pp 44-48.
11. Bosy-Westphal A, Geisler C, Onur S et al. Value of body fat mass vs anthropomet-ric obesity indices in the assessment of metabolic risk factors. Int J Obes (Lond)2006;30:475-483.
12. Deurenberg P, Andreoli A, Borg P et al. The validity of predicted body fat percent-age from body mass index and from impedance in samples of five European popu-lations. Eur J Clin Nutr 2001;55:973-979.
13. Okorodudu DO, Jumean MF, Montori VM et al. Diagnostic performance of bodymass index to identify obesity as defined by body adiposity: a systematic reviewand meta-analysis. Int J Obes (Lond) 2010;34:791-799.
14. Matthews DR, Hosker JP, Rudenski AS et al. Homeostasis model assessment:insulin resistance and beta-cell function from fasting plasma glucose and insulinconcentrations in man. Diabetologia 1985;28:412-419.
15. Perseghin G, Caumo A, Caloni M, Testolin G, Luzi L. Incorporation of the fastingplasma FFA concentration into QUICKI improves its association with insulin sensi-tivity in nonobese individuals. J Clin Endocrinol Metab 2001;86:4776-4781.
16. Wildman RP, Muntner P, Reynolds K et al. The obese without cardiometabolic riskfactor clustering and the normal weight with cardiometabolic risk factor clustering:prevalence and correlates of 2 phenotypes among the US population (NHANES1999-2004). Arch Intern Med 2008;168:1617-1624.
17. De Lorenzo A, Martinoli R, Vaia F, Di Renzo L. Normal weight obese (NWO)women: an evaluation of a candidate new syndrome. Nutr Metab Cardiovasc Dis2006;16:513-523.
18. Grundy SM, Brewer HB Jr, Cleeman JI, Smith SC Jr, Lenfant C. Definition of met-abolic syndrome: Report of the National Heart, Lung, and Blood Institute/AmericanHeart Association conference on scientific issues related to definition. Circulation2004;109:433-438.
19. Karelis AD, Messier V, Brochu M, Rabasa-Lhoret R. Metabolically healthy butobese women: effect of an energy-restricted diet. Diabetologia 2008;51:1752-1754.
20. Third Report of the National Cholesterol Education Program (NCEP) Expert Panelon Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults(Adult Treatment Panel III) final report. Circulation 2002;106: 3143-421.
21. Hu FB, Meigs JB, Li TY, Rifai N, Manson JE. Inflammatory markers and risk ofdeveloping type 2 diabetes in women. Diabetes 2004;53:693-700.
22. Mosca L. C-reactive protein-to screen or not to screen? N Engl J Med 2002;347:1615-1617.
23. Spranger J, Kroke A, M€ohlig M et al. Inflammatory cytokines and the risk todevelop type 2 diabetes: results of the prospective population-based EuropeanProspective Investigation into Cancer and Nutrition (EPIC)-Potsdam Study.Diabetes 2003;52:812-817.
24. De Lorenzo A, Del Gobbo V, Premrov MG et al. Normal-weight obese syndrome:early inflammation? Am J Clin Nutr 2007;85:40-45.
25. Wannamethee SG, Lowe GD, Rumley A et al. Adipokines and risk of type 2 diabe-tes in older men. Diabetes Care 2007;30:1200-1205.
26. S€oderberg S, Ahr�en B, Jansson JH et al. Leptin is associated with increased risk ofmyocardial infarction. J Intern Med 1999;246:409-418.
27. Wallace AM, McMahon AD, Packard CJ et al. Plasma leptin and the risk of cardio-vascular disease in the west of Scotland coronary prevention study (WOSCOPS).Circulation 2001;104:3052-3056.
28. Inoue M, Maehata E, Yano M, Taniyama M, Suzuki S. Correlation between theadiponectin-leptin ratio and parameters of insulin resistance in patients with type 2diabetes. Metab Clin Exp 2005;54:281-286.
29. Inoue M, Yano M, Yamakado M, Maehata E, Suzuki S. Relationship between theadiponectin-leptin ratio and parameters of insulin resistance in subjects withouthyperglycemia. Metab Clin Exp 2006;55:1248-1254.
30. Bolanowski M, Nilsson BE. Assessment of human body composition using dual-energy x-ray absorptiometry and bioelectrical impedance analysis. Med Sci Monit2001;7:1029-1033.
31. Barbarroja N, L�opez-Pedrera R, Mayas MD et al. The obese healthy paradox: isinflammation the answer? Biochem J 2010;430:141-149.
32. Koster A, Stenholm S, Alley DE et al. Body fat distribution and inflammationamong obese older adults with and without metabolic syndrome. Obesity (SilverSpring) 2010;18:2354-2361.
33. Snijder MB, Dekker JM, Visser M et al. Associations of hip and thigh circumfer-ences independent of waist circumference with the incidence of type 2 diabetes: theHoorn Study. Am J Clin Nutr 2003;77:1192-1197.
34. Snijder MB, Dekker JM, Visser M et al. Trunk fat and leg fat have independentand opposite associations with fasting and postload glucose levels: the Hoorn study.Diabetes Care 2004;27:372-377.
35. Beasley LE, Koster A, Newman AB et al. Inflammation and race and gender differ-ences in computerized tomography-measured adipose depots. Obesity (Silver Spring)2009;17:1062-1069.
36. Kennedy A, Martinez K, Chuang CC, LaPoint K, McIntosh M. Saturated fatty acid-mediated inflammation and insulin resistance in adipose tissue: mechanisms ofaction and implications. J Nutr 2009;139:1-4.
37. Vessby B. Dietary fat, fatty acid composition in plasma and the metabolic syn-drome. Curr Opin Lipidol 2003;14:15-19.
38. Lopez-Garcia E, Schulze MB, Manson JE et al. Consumption of (n-3) fatty acids isrelated to plasma biomarkers of inflammation and endothelial activation in women.J Nutr 2004;134:1806-1811.
39. Madsen T, Skou HA, Hansen VE et al. C-reactive protein, dietary n-3 fatty acids,and the extent of coronary artery disease. Am J Cardiol 2001;88:1139-1142.
40. Paniagua JA, P�erez-Martinez P, Gjelstad IM et al. A low-fat high-carbohydrate dietsupplemented with long-chain n-3 PUFA reduces the risk of the metabolic syn-drome. Atherosclerosis 2011;218:443-450.
Original Article ObesityEPIDEMIOLOGY/GENETICS
www.obesityjournal.org Obesity | VOLUME 21 | NUMBER 1 | JANUARY 2013 E161