Professional Documents
Culture Documents
Estefania Oliveros
a
, Virend K. Somers
a
, Ondrej Sochor
a, b
,
Kashish Goel
a
, Francisco Lopez-Jimenez
a,
a
Division of Cardiovascular Diseases, Mayo Clinic, Rochester, Minnesota
b
International Clinical Research Center, St. Annes University Hospital, Brno, Czech Republic
A R T I C L E I N F O A B S T R A C T
Individuals with normal body weight by body mass index (BMI) and high body fat
percentage show a high degree of metabolic dysregulation. This phenomenon, defined as
normal weight obesity, is associated with a significantly higher risk of developing metabolic
syndrome, cardiometabolic dysfunction and with higher mortality. Recently, we have also
shown that coronary artery disease patients with normal BMI and central obesity have the
highest mortality risk as compared to other adiposity patterns. Therefore, it is important to
recognize these high-risk groups for better adiposity-based risk stratification. There is a
need for an updated definition of obesity based on adiposity, not on body weight.
2014 Elsevier Inc. All rights reserved.
Keywords:
Normal weight obesity
Obesity
Metabolically obese normal weight
Central obesity
Body fat percentage
The Merriam-Webster dictionary defines obesity as a condi-
tion characterized by the excessive accumulation and storage
of fat in the body.
1
Obesity is a chronic metabolic disorder
characterized by an increase in the number and/or the size of
fat cells. Global prevalence of obesity has almost doubled
since 1980
2
and has now become an epidemic
3,4
threatening
public health. In 2008, more than 1.4 billion adults in the
world were overweight, from which approximately 200
million men and 300 million women were obese.
2
Overweight
and obesity represent the fifth leading risk for global deaths.
2
Formerly thought as a problem of high-income countries,
overweight and obesity are becoming more prevalent in low-
and middle-income countries.
The American Heart Association and the American College
of Cardiology guidelines labeled obesity as a major modifiable
cardiovascular disease (CVD) risk factor. Obesity is associated
with higher rates of insulin resistance, type 2 diabetes
mellitus (DM), hypertension (HTN), dyslipidemia, coronary
heart disease (CHD), gallbladder disease, obstructive sleep
apnea, non-alcoholic fatty liver disease and some malignan-
cies including endometrial, breast, and colon cancer.
3,4
Obesity is considered an independent risk factor for CVD
5
and is associated with increased mortality in general healthy
populations.
3
Historic perspective of the concept of obesity
Measurement of height and weight was the initial step in the
clinical assessment of overweight and obesity. In 1908,
Symonds reported the results of a large prospective study of
weight and mortality in New Jersey. He registered weight for a
given height and age, and the influence of excess weight on
vitality.
6
Subsequently, obesity was defined in relation to
desirable weight, taking in consideration the actuarial tables
from the Metropolitan Life Insurance Company.
7
The concept
used was percent of ideal body weight. Later on, the Diet and
Health report questioned the approach of using ideal body
weight
8
and suggested that terms like healthy or good weight
ranges were associated with decreased mortality.
9
After-
wards, body mass index (BMI) substituted the assessment of
obesity, calculated as body weight (in kg) divided by height (in
P R O G R E S S I N C A R D I O V A S C U L A R D I S E A S E S 5 6 ( 2 0 1 4 ) 4 2 6 4 3 3
Statement of Conflict of Interest: see page 432.
Address reprint requests to Francisco Lopez-Jimenez, MD, MSc, Division of Cardiovascular Diseases Mayo Clinic. 200 First St SW
Rochester, MN 55905.
E-mail address: lopez@mayo.edu (F. Lopez-Jimenez).
0033-0620/$ see front matter 2014 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.pcad.2013.10.003
Avai l abl e onl i ne at www. sci encedi r ect . com
ScienceDirect
www. onl i nepcd. com
meters) squared. The
i ni ti al use of the
index was in 1842 by
Quetelet a Belgian
mathematician who
noticed that in people
he considered to have
a normal frame the
weight was propor-
tional to the height
squared.
10
In the last
3 decades of the 20th
century, several epi-
demiologic studies
used BMI to prove the
association between
adiposity and mortali-
ty, CVD, DMand many
other obesity-related
comorbidities. With
the help of experts
from around the world
who formed The Inter-
national Obesity Task
Force, i n 1997 the
World Health Organi-
zation (WHO) came up with the definition of obesity as a
BMI 30 kg/m
2
.
11,12
The definition also described other
degrees of adiposity like overweight and obesity grade 1 and
2, although the cutoffs chosen were arbitrary. Fig 1
shows the classification of obesity modified by the National
Heart, Lung and Blood Institute task force, along with
the associated disease risk with increasing BMI.
13
Obesity can be also measured using direct and indirect
measures of fatness other than BMI. The methods to estimate
body fat (BF) include bioelectrical impedance, hydrostatic
plethysmography, isotope dilution techniques, dual x-ray
absorptiometry, skinfold method, body impedance measures
with over the counter scales, and air displacement
plethysmography.
14
Epidemiologic studies have also demon-
strated that central fat distribution, measured with waist
circumference (WC), waist-to-hip ratio (WHR) and weight-
to-height ratio, is also an important measure of adiposity-
related risk.
15
Pitfalls of current BF mass measurements
BMI as a measure of BF became popular and widely used
because of its simplicity and validation in multiple epidemi-
ologic studies. Surprisingly, even though obesity is defined as
excessive adiposity, there is no consensus on how to define
obesity using fat mass calculation or fat percentage, other
than the effort by the American Society of Endocrinologists
who defined obesity by BF percent as >35% in women and
>25% in men.
16
Direct measurement of adipose tissue using
methods like water-displacement plethysmography or mag-
netic resonance is too cumbersome to be used in large
populations or in clinical practice. Newer methods to measure
fat content like air-displacement plethysmography, DEXA or
electrical bioimpedance have shown to be valid and not
necessarily expensive.
Although several studies have demonstrated a high
correlation between BMI and directly-measured BF, the
diagnostic performance of BMI is not optimal to identify
leanness or excessive BF. We have tested the accuracy of BMI
for diagnosing obesity in the adult general population using
data from 13,601 individuals from the Third National Health
and Nutrition Examination Survey.
17
Using bioimpedance to
calculate BF and a BMI >30 kg/m
2
to define obesity, BMI had a
very high specificity (97%) but poor sensitivity (42%) to detect
obesity
17,18
Therefore, more than half of the individuals with
increased BF percentage may be misclassified by BMI. In
individuals with BMI of 25 kg/m
2
, the index had 86%
sensitivity and a specificity of 73%. A meta-analysis of the
diagnostic performance of BMI to detect excessive adiposity
using different techniques as the gold standard showed
similar results.
18
A recent study demonstrated a wide range
of BF % using dual energy x-ray absorptiometry in people with
normal BMI, ranging from 5.6 to 31.2% in men and from 4.6 to
51.1% among women.
19
The main limitation of BMI is that it
cannot differentiate BF from lean mass, and central from
peripheral fat. Therefore, athletes with enhanced body
muscle mass may be misclassified as obese when using only
BMI to diagnose obesity, whereas people with low lean mass
but high BF content may still have a normal BMI.
20
Challenging the simplistic concept of obesity as
defined by BMI
Over the last 30 years, there have been several new concepts
challenging the simplistic concept that obesity can be
diagnosed based on weight and height. Numerous studies
have proposed definitions of the obesity subtypes (Table 1).
Firstly, Ruderman et al
21
challenged the notion that
standard weightheight tables were the proper way to
determine high-risk groups for obesity associated disorders.
They observed normal weight individuals suffering from type
2 DM, premature CHD, HTN and hypertriglyceridemia with
associated hyperinsulinemia. They pointed out that these
abnormalities could not be explained by skinfold thickness
or adipose mass and hypothesized that it was due to larger
fat cells. The identified metabolically obese, normal weight
individuals had benefits when they went through programs
of energy restriction and weight loss. If patients were
challenged to a 412 week period of diet and exercise there
was metabolic improvement.
22
Some studies suggested that
the main issue to explain the metabolic abnormalities in
individuals not particularly overweight was fat distribution.
On the basis of these studies, it was proposed a scoring
method to identify a metabolically obese normal weight
individual. Depending on the presence of associated diseases
or biochemical abnormalities related to insulin resistance,
individuals would be assigned a score to base the diagnosis
of metabolically obese normal weight. All of these mentioned
disturbances predispose the individual to suffer from,
23
as well as making them a susceptible population to suffer
Abbreviations and Acronyms
BF = body fat
BMI = body mass index
CHD = coronary heart disease
CVD = cardiovascular disease
DM = diabetes mellitus
HDL = high-density lipoprotein
cholesterol
HF = heart failure
HOMA = Homeostasis Model
Assessment
HTN = hypertension
LDL = low-density lipoprotein
NWO = normal weight obesity
WC = waist circumference
WHO = World Health
Organization
WHR = waist-to-hip ratio
427 P R O G R E S S I N C A R D I O V A S C U L A R D I S E A S E S 5 6 ( 2 0 1 4 ) 4 2 6 4 3 3
from type 2 DM with a three to four-fold time higher risk
than control non-obese individuals.
24
Specific reversible
factors appear to predispose metabolically-obese normal-
weight individuals, like decreased aerobic fitness, physical
activity energy expenditure
25,26
and a lower level of dietary
restraint.
23
Karelis et al
27
and Dvorak et al
25
realized that insulin
sensitivity was related to body composition and that body
composition could be a major determinant for the metabolic
behavior of individuals (Table 2). In their review Karelis et al
27
addressed the differences among the metabolically healthy
obese, the metabolically obese normal weight, the individual
at risk of obesity and the metabolically healthy individual
(Table 3). Metabolically healthy obese subjects when com-
pared with obese insulin resistant adults have a healthier
metabolic risk profile and higher disposition index (insulin
sensitivity insulin secretion).
28
These findings challenge the
concept that obesity itself produces -cell dysfunction. Most
metabolically obese normal weight subjects can be identified
with risk factors in the familial background, birth weight,
adult onset weight gain and central adiposity, physical
activity status, and the presence of related pathologies.
22
In addition to the metabolically obese normal weight
phenotype, other subclassifications of obesity have been
described, like metabolically normal obesity and NWO.
Normal weight obesity: a distinct phenotype linked
to metabolic dysregulation and inflammation
In 2006, De Lorenzo et al
29
described the association between
normal weight and high fat content with metabolic abnor-
malities. The term NWO described those individuals with
normal body weight and BMI (<25 kg/m
2
) with increased BF %
(>30%). Despite the fact that obesity is defined as excessive
body adiposity according to the etymology of the word, there
is no consensus about how to define obesity based on BF
content or %. Some investigators have proposed sex- and age-
adjusted cut-off values for NWO. These were defined as the
combination of a normal BMI and increased BF%: 2039 years,
>19% and >32%; 4059 years, >21% and 33% and 6079 years,
> 24% and 35% for men and women, respectively.
30
A current
challenge when evaluating BF is that there is no consensus
about the best cutoff for percent of BF to define excess fatness.
The different proposed cutoff points of BF vary between 20
and 25% for men and 30 and 37% for women.
20,31,32
Importantly, metabolically obese normal weight individ-
uals likely represent a subset of all NWOpeople. It is clear that
not everybody with NWO has the cluster of metabolic
abnormalities. However, what has not been clearly deter-
mined is what extent of the metabolic dysregulation in
metabolically obese normal weight people could be solely
explained by increased amounts of total fat, a high BF percent
or increased visceral fat.
Relationship between NWO, metabolic
dysregulation and inflammation
There is strong evidence linking NWO and metabolic
dysregulation. De Lorenzo et al
29
evaluated 74 women and
analyzed anthropometric variables, body composition, resting
metabolic rate and biohumoral variables. They found signif-
icant differences in high-density lipoprotein (HDL) between
normal weight obese and pre-obese-obese individuals. There
were significant correlations among CVD risk indexes, lean of
the right part of the trunk and total cholesterol/HDL (R =
0.69, p < 0.001) and LDL/HDL (R = 0.72, p < 0.001) and lean
Fig 1 Classification of obesity developed by the National Heart, Lung and Blood Institute task force, along with the associated
disease risk with increasing BMI, waist circumference and waist to hip ratio.
428 P R O G R E S S I N C A R D I O V A S C U L A R D I S E A S E S 5 6 ( 2 0 1 4 ) 4 2 6 4 3 3
and resting metabolic rate (R = 0.44, p = 0.022) in normal
weight obese women. Twenty-eight women with high BF
(>30%) had significantly lower metabolic rate and oxygen
consumption compared to the 20 women with normal BMI
and no excess in BF defined as <30% of fat.
29
We studied 6,171 participants with normal BMI of the
Third National Health and Nutrition Examination Survey III
from a representative sample of the US population.
20
This
study demonstrated that BF % was significantly associated
with markers of metabolic dysregulation. Men and women in
the highest tertile of BF measured with bioimpedance had
lower levels of serum HDL-cholesterol and higher levels of
LDL-cholesterol, triglycerides, apoB/A-I ratio, high sensitivity
C-reactive protein, and leptin levels. The results were similar
when people were classified as NWO using arbitrary cut-offs
of BF of 20% in men and 30% in women. In individuals with
fasting morning blood samples, insulin sensitivity (HOMA-S)
decreased as BF increased ( = 2.78, p < 0.0001). Triglycer-
ides ( = 1.93 mg/dL, p < 0.0001) and insulin ( = 0.97 mg/L,
p < 0.0001) had the same trend.
In another study, plasma inflammatory interleukin-1,
interleukin-1, interleukin-6, interleukin-8, and tumor necro-
sis factor- were significantly higher in the pre-obese-obese
women and in normal weight obese compared to the non-
Table 1 Chronological perspective of the definition of the subtypes of obesity.
Publication Year Subtype of Obese Definition Comments
Ruderman
et al
21
1981 Metabolically obese
normal weight
Subtle increase in adiposity and/or Hyperinsulinism
creating obese associated diseases in normal
weight (by standard weight tables)
Possibly and increase in fat cell size
Ruderman
et al
22
1998 Metabolically obese
normal weight
CHD, DM and other disorders associated with
obesity + normal weight (<115% of ideal body
weight or BMI <28 kg/m
2
)
Role of central fat distribution, inactivity
and low VO
2max.
Scoring method
Dvorak et al
25
1999 Metabolically obese
normal weight
Impaired insulin sensitivity: Hyperinsulinemic/
euglycemic clamp (>8.0 mg/min * kg lean body
mass) + BMI <26.3 kg/m
2
total fat mass (+20%), BF percentage
(+16%) (dual x-ray absorptiometry),
subcutaneous fat (+33%), visceral
fat (+26%)
Karelis et al
27
2004 Metabolically
healthy obese
Increased fat mass + Normal Metabolic profile +
High levels of insulin sensitivity
Metabolically obese
normal weight
Normal weight (by BMI) + Risk factors
(DM, MetS, CVD)
Possibly due to higher fat mass,
triglycerides, visceral fat, and liver content.
Conus et al
23
2004 Metabolically obese
normal weight
Insulin sensitivity determined by HOMA >1.69 with
normal weight (by BMI <25 kg/m
2
)
BF % (32.24 8.16 vs. 25.04 5.84) (dual
x-ray absorptiometry), fat-free mass,
physical activity energy expenditure,
peak oxygen uptake
St-Onge et al
23
2004 Metabolically obese
normal weight
BMI normal to slightly elevated 18.5-26.9 kg/m
2
+
MetS criteria
De Lorenzo
et al
29
2005 NWO Normal weight (by BMI) + BF % >30 (dual x-ray
absorptiometry) + lean body composition of the
left leg. DO NOT have MetS (normal glycemia,
triglycerides, Total cholesterol and HDL)
CVD risk indexes relate to metabolic
variables and BF mass distribution.
Meigs et al
24
2006 Metabolically obese
normal weight
BMI <25 kg/m
2
+ MetS criteria/insulin resistance
Marques-Vidal
et al
32
2008 NWO Normal weight (by BMI) + BF % or fat mass
index 8.3 kg/m
2
(men) or 11.8 kg/m
2
(women)
BF according to age and gender
(bio-electrical impedance)
Sucurro et al
28
2008 Metabolically obese
normal weight
Normal weight (by BMI) + Metabolic dysfunction
(Impaired insulin sensitivity, visceral adiposity,
HDL, fasting glucose, triglycerides, HTN)
Metabolically
healthy obese
Insulin sensitive with favorable CVD risk profile
Di Renzo
et al
35
2010 NWO Normal weight (Body weight and BMI) + Fat mass
>30%
Presence of oxidative stress. BF
percentage (dual x-ray absorptiometry)
Romero-
Corral
et al
20
2010 NWO Normal Weight (By BMI) + BF% (Highest sex-
specific tertile 23.1 in men and >33.3% in women)
Sex-specific tertile of BF %
(bio-electrical impedance). WC
associated with CVD risk as
sex-specific tertile of BF %
Shea et al
36
2012 Metabolically obese
normal weight
body fat % + Normal Weight (by BMI) +
Metabolically abnormal (Insulin resistance, HDL,
fasting glucose, triglycerides, hsCRP, HTN)
BF % (dual x-ray absorptiometry)
Kim et al
44
2012 NWO Normal weight (by BMI) + BF%
Madeira
et al
40
2013 NWO Normal weight (by BMI <25) + Sum of triceps and
subscapular skinfolds >90th percentiles.
Abbreviations: BF, Body Fat; BMI, Body Mass Index; NWO, normal weight obese; CVD, Cardiovascular Disease; DM, diabetes mellitus; CVD,
cardiovascular disease; MetS, Metabolic Syndrome; HOMA, Homeostasis Model Assessment; HDL, high-density lipoprotein cholesterol; HTN,
Hypertension; WC, Waist circumference; hs-CRP, high sensitivity C reactive protein.
429 P R O G R E S S I N C A R D I O V A S C U L A R D I S E A S E S 5 6 ( 2 0 1 4 ) 4 2 6 4 3 3
obese group, and the concentrations of interleukin-6 and
tumor necrosis factor- were correlated with fat mass
percentage in the normal weight obese.
33
The cytokines
concentrations had an intermediate value in the NWO
individuals suggesting that this group was in an early
inflammatory state.
33
Tumor necrosis factor-, interleukin-1
and interleukin-6 are related to the adipose mass; conse-
quently they are increased in the obese individuals.
34
Di
Renzo et al
35
studied 60 white Italian women subdivided as
follows: 20 normal-weight individuals (BMI <25 kg/m
2
; FM%
<30%); 20 normal weight obese (BMI <25 kg/m
2
; FM% >30%);
and 20 pre-obese-obese (BMI >25 kg/m
2
; FM% >30%). They
demonstrated that NWO women were in early inflammatory
status, as well as in an oxidative stress related to metabolic
dysfunction related to obesity. Reduced glutathione and nitric
oxide metabolites were significantly lower in pre-obese-obese
(BMI > 25 kg/m
2
; Fat mass > 30%) and normal weight obese
compared to normal weight individuals.
35
Glutathione
showed a protective role against oxidative and free radical-
mediated injury.
There also was a strong association with reduced gluta-
thione levels and body weight, BMI and fat mass percentage
(R = 0.45, at least p < 0.05); WC (R = 0.33, p < 0.05); free fat
mass (FFM) % (R = 0.45; p < 0.01); interleukin-1, interleukin-
6, interleukin-10, interleukin-15 (R = 0.39, 0.33, 0.36,
0.34, respectively, p < 0.05); and triglycerides (R = 0.416,
p < 0.05). Lipid peroxide levels negatively correlated to FFM %
(R = 0.413, p < 0.05) and positively correlated to fat mass
percentage (R = 0.408), interleukin-15 (R = 0.502), tumor ne-
crosis factor- (R = 0.341), insulin (R = 0.412), total cholesterol
(R = 0.4036), low-density lipoprotein cholesterol (R = 0.405),
and triglycerides (R = 0.405) [p < 0.05].
35
There is a known association between left ventricular
dysfunction and heart failure (HF). However, Kosmala et al
36
realized that most of the studies relating HF and left
ventricular dysfunction did not assess individuals with
normal BMI but with increased BF content. They designed
a study with 168 individuals, from which the normal
weight obese subjects demonstrated left ventricle systolic
and diastolic dysfunction, increased fibrosis intensity
(assessed through procollagen type I carboxyl-terminal
propeptide), impaired insulin sensitivity, and augmented
pro-inflammatory activation in comparison to the individuals
with normal BF. Those results were consistent with older
studies assessing the link between obesity as measured with
BMI, and changes in myocardial performance.
37
Marques-Vidal et al
38
conducted a cross-sectional study in
3,213 women and 2,912 men and defined normal weight
obesity as a BMI <25 kg/m
2
and a BF % 66th gender-specific
percentiles. They found an association between NWO and CV
risk factors and early inflammation. They demonstrated that
women with NWO had higher blood pressure, lipids and
glycemia than lean women, but showed no difference for
hepatic or inflammation markers. Interestingly, there were no
differences between NWO and overweight females. In the
same study NWO women had larger waists than lean
women,
38
so it is not clear if the differences in metabolic
dysregulation were due to differences in total BF content or
specifically to increased abdominal fat among those with
NWO. Furthermore, increased leptin levels in NWO women
correlate with BMI and BF amount. Leptin levels are
Table 2 Diagnosis of the metabolically abnormal
phenotype through cardiometabolic markers.
Metabolic Abnormality Values Considered
Increased blood pressure
130/85 mm Hg
Increased triglycerides