J C E M O N L I N E

Hot Topics in Translational Endocrinology—Endocrine Research

Adipose Tissue and Metabolic Alterations: Regional

Differences in Fat Cell Size and Number Matter, But
Differently: A Cross-Sectional Study

Mikael Rydén, Daniel P. Andersson, Ingrid B. Bergström, and Peter Arner

Department of Medicine, Karolinska Institutet, Karolinska University Hospital, Huddinge, 141 86,
Stockholm, Sweden

Objective: White adipose tissue can expand by increasing the size and/or number of fat cells.
Although increased sc and visceral fat cell size associates with an adverse metabolic profile, the
relationship with fat cell number in either depot is unknown. We hypothesized that adipocyte
number and size displayed different relationships with clinically relevant metabolic variables.

Methods: This was a cross-sectional study of 204 patients scheduled for gastric bypass surgery. Fat
cell size and number were determined in visceral and abdominal sc adipose tissue and related to
insulin sensitivity (by hyperinsulinemic euglycemic clamp), fasting plasma levels of insulin, triglyc-
erides and high-density lipoprotein (HDL) cholesterol.

Results: Visceral and sc fat cell volumes were positively correlated with insulin and triglyceride
levels and negatively with insulin sensitivity and HDL-cholesterol (P ⫽ .0020 or better). In contrast,
although visceral fat cell number did not associate with any metabolic parameter, sc adipocyte
number displayed a positive association with insulin sensitivity and HDL-cholesterol and a negative
relationship with insulin and triglyceride levels (P ⫽ .0014 or better). All results were independent
of body fat mass.

Conclusions: Variations in fat cell size and number correlate differently with metabolic parameters
in obesity. Increased fat cell size in visceral and sc depots associates with a pernicious metabolic
profile, whereas increased sc, but not visceral, fat cell number correlates with a more beneficial
phenotype. Whether determination of sc fat cell number, in addition to adipocyte size, may have
a predictive value for the risk of type 2 diabetes needs to be demonstrated in prospective or
mechanistic studies. (J Clin Endocrinol Metab 99: E1870 –E1876, 2014)

besity is a major risk factor for metabolic distur- duction accomplished by bariatric surgery, T2DM is not
O bances, in particular dyslipidemia and type 2 dia-
betes mellitus (T2DM). Not surprisingly, the current
reversed in all subjects, and in many cases, there is a relapse
despite maintenance of reduced body weight status (5).
worldwide increase in T2DM parallels the accelerated Interestingly, longitudinal studies have shown that T2DM
prevalence of overweight/obesity (1). Pronounced weight is rapidly reversed after bariatric surgery, long before
reduction is hard to achieve, and to date, the most effective maximum weight loss has been achieved (reviewed in Ref.
long-term treatment of obesity and its metabolic compli- 6). Furthermore, surgical removal of large amounts of ab-
cations is bariatric surgery (2, 3). However, large random- dominal sc adipose tissue from obese patients does not
ized controlled trials with long follow-up periods to con- improve insulin resistance or other metabolic abnormal-
firm sustained improvements in metabolic complications ities (7, 8). Similarly, removal of visceral adipose tissue in
and mortality are still lacking (4). Despite the weight re- connection with bariatric surgery does not improve insu-

ISSN Print 0021-972X ISSN Online 1945-7197 Abbreviations: BMI, body mass index; CT, computed tomography; DEXA, dual-energy x-ray
Printed in U.S.A. absorptiometry; ESAT, estimated sc abdominal adipose tissue; EVAT, estimated visceral
Copyright © 2014 by the Endocrine Society adipose tissue; HDL, high-density lipoprotein; LBM, lean body mass; T2DM, type 2 diabetes
Received February 21, 2014. Accepted June 9, 2014. mellitus
First Published Online June 17, 2014

E1870 jcem.endojournals.org J Clin Endocrinol Metab, October 2014, 99(10):E1870 –E1876 doi: 10.1210/jc.2014-1526

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 14 January 2015. at 04:08 For personal use only. No other uses without permission. . All rights reserved.
doi: 10.1210/jc.2014-1526 jcem.endojournals.org E1871

lin sensitivity or other metabolic abnormalities beyond hypertension. An abdominal sc fat biopsy was obtained by nee-
weight loss itself (9 –12). Taken together, these data dle aspiration under local anesthesia. A second fat biopsy was
obtained from the greater omentum (visceral fat) at the begin-
strongly suggest that the amount or distribution of adipose
ning of general surgery. The study was approved by the local
tissue is not per se of major importance for the develop- committee on ethics and explained to each participant. Written
ment of T2DM in obesity. informed consent was obtained.
Although factors not related to adipose tissue, eg, gut
hormones and intestinal microbiota, may play a role in the Examinations
development of T2DM and other metabolic complica- The patients were investigated in the morning after an over-
tions, intrinsic adipose factors may also be of importance. night fast. Height and weight were determined. A venous blood
One such aspect is adipose cellularity (13). Fat tissue can sample was obtained, and plasma levels of insulin, glucose, trig-
lycerides, and HDL-cholesterol were measured as described (14).
expand by primarily increasing either the size or number
Lean body mass (LBM), total body fat, and android fat mass
of its fat cells. It is well established that a phenotype char- were measured by dual-energy x-ray absorptiometry (DEXA)
acterized by large (hypertrophic) adipocytes, is associated using a GE Lunar iDXA with the software EnCore (version
with hyperinsulinemia, insulin resistance, and dyslipide- 14.10.022) with the CoreScan feature provided by the manu-
mia (14 –18). Several of these studies have shown that this facturer (GE Medical Systems). CoreScan is an automated
association is independent of body fatness as measured by method for segmenting total adipose fat into sc fat and visceral
fat within the android region. The estimation of visceral fat with
body mass index (BMI) (14, 15, 17, 18). Two prospective this software has been approved for clinical use by the U.S. Food
studies have confirmed that adipose hypertrophy is asso- and Drug Administration (21). Automatic calibration checks of
ciated with increased risk for developing T2DM, indepen- the DEXA were performed daily throughout the study, and
dently of BMI (19, 20). However, the relationship between thrice-weekly calibrations using a spine phantom (for bone min-
adipocyte number and insulin resistance as well as other eral density, provided by the manufacturer) were performed. The
coefficient of variation for the spine phantom testing was 1.5%.
metabolic variables associated with T2DM risk is
No hardware or software changes were made during the course
unknown. of the trial. The subjects were scanned using standard imaging
We hypothesized that adipocyte number, in compari- and positioning protocols and the same scan mode (set for obese
son with fat cell size, could display different relationships subjects) was used throughout the study. CoreScan was used to
with clinical parameters and investigated the regional im- calculate estimated visceral adipose tissue (EVAT) mass. The
pact of adipose cellularity for insulin sensitivity and clas- amount of estimated sc abdominal adipose tissue (ESAT) mass,
which corresponds to the region for the sc fat biopsy, was cal-
sical metabolic risk factors related to obesity/T2DM. In a
culated from the following formula: total adipose fat mass in the
cross-sectional cohort, fat cell size and number were de- android region ⫺ EVAT ⫽ estimated sc adipose tissue in the
termined in abdominal sc and visceral (greater omentum) android region ⫺ ESAT, as previously described (21). Determi-
adipose tissue of obese subjects and set in relation to in- nation of EVAT with this method shows a strong correlation (r2
sulin sensitivity (hyperinsulinemic-euglycemic clamp) and ⱖ 0.95) with measures using computed tomography (CT) (21).
circulating levels of insulin, triglycerides, and high-density We subsequently validated this in an independent cohort of
obese women and men, comparing the weight of the greater
lipoprotein (HDL) cholesterol.
omentum with visceral adipose tissue mass estimated using the
presently used DEXA equipment/software and CT (22). Because
only total android fat mass and EVAT are used to determine
Subjects and Methods ESAT and both are valid measures, it follows that also the cal-
culation of ESAT should be valid. Fat cell volume and weight
Subjects were determined using a standard method in the field that has
Between May 2006 and September 2013, 204 patients re- been described in detail previously (14, 22). In brief, fat cells were
ferred to our surgical units (Ersta Hospital, Södertälje Hospital, isolated, and the diameter of 100 cells was measured. The di-
Karolinska University Hospital Huddinge, and Danderyd Hos- ameters were distributed in a unimodal way, and it has been
pital in Stockholm, Sweden) for Roux-en-Y gastric bypass sur- shown in several independent studies that counting 100 or 300
gery for obesity were included in the study. Inclusion criteria cells by microscopy is a reliable estimate of fat cell size that is
were an age between 18 and 60 years and BMI ⬎30 kg/m2 with comparable with results obtained using either manual or auto-
concomitant obesity-related complications or BMI ⬎35 kg/m2. mated image analysis software (23). Fat cell volume (expressed
Exclusion criteria were treatment with insulin or glitazones, oral in picoliters) was calculated as (/6) where is the cell diameter in
or parenteral steroid treatment, complicated psychiatric disease, micrometers. As previously discussed (24, 25), the diameter (d)
or warfarin use. To avoid possible influences of catabolism/ is a normally distributed variable but its cube (d3) is skewed and
weight loss, none of the participants were put on preoperative the arithmetic mean of d3 can therefore not be used to calculate
caloric restriction, and subjects who reported a hypocaloric diet mean fat cell volume. The average fat cell volume is instead much
before surgery (ie, at their own initiative) were excluded. Subjects better approximated by using the formula (where is the mean
were weight stable (less than ⫾2 kg weight change) for at least 1 diameter and ␴ is the standard deviation of the diameter). As-
year before the first visit. Twelve subjects were diagnosed with suming a mean density of fat cells equal to that of human adipose
T2DM and treated with diet and metformin, whereas 57 had triglycerides, mean fat cell weight was obtained by multiplying

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 14 January 2015. at 04:08 For personal use only. No other uses without permission. . All rights reserved.
E1872 Rydén et al Adipocyte Size and Number J Clin Endocrinol Metab, October 2014, 99(10):E1870 –E1876

the volume by 0.915 (ie, the density in grams per milliliter of ESAT/EVAT values from 188 subjects could be obtained.
triolein). Estimated fat depot weight was then divided by mean In 14 patients, EVAT and ESAT values could not be mea-
fat cell weight in the corresponding biopsy to obtain fat cell
sured because of the severe obesity. One subject was ex-
number in the respective depot. After 45 minutes of rest, the
subjects underwent a hyperinsulinemic-euglycemic clamp as de- amined using different DEXA equipment, and in one in-
scribed (12). In brief, after an iv bolus dose of insulin (1600 dividual, DEXA measurements were missing. The clinical
mU/m2 body surface area, corresponding to 9600 pmol/m2 when characteristics of the cohort are described in Table 1. The
using the conversion factor 6.0 as described on www.soc- results of Spearman correlation between metabolic vari-
bdr.org/rds/authors/unit_tables_conversions_and_genetic_dic- ables and adipocyte size and number are summarized in
tionaries/conversion_in_si_units/index_en.ht/mL) insulin was
infused iv at a rate of 120 mU/m2/min (720 pmol/m2/min) for 2
Table 2. In visceral adipose tissue, fat cell volume, but not
hours and a variable iv infusion of glucose (200 mg/mL) was used number, correlated significantly with insulin sensitivity
to maintain euglycemia between 81 and 99 mg/dL (4.5–5.5 and plasma levels of insulin, triglycerides, or HDL-cho-
mmol/L). The infusion rate of glucose during the last 60 minutes lesterol. In sc adipose tissue, similar correlations were
of the clamp, when insulin levels are in a steady state, was used observed regarding fat cell volume as for visceral adi-
to calculate whole-body glucose disposal rates (M-value). The
pose tissue (Table 2 and Figure 2). In contrast, although
average values of plasma insulin at steady state during the clamp
were 242 ⫾ 64 mU/L (1450 ⫾ 390 pmol/L). sc fat cell number correlated significantly with all vari-
ables, it did so in an opposite way compared with cell
Statistics volume (Table 2 and Figure 2). Thus, there was a pos-
Group values are mean and range. Adipose and metabolic itive relationship with insulin sensitivity and HDL-cho-
parameters were compared using Spearman correlation, analysis lesterol and a negative relationship with insulin and
of covariance, and multiple linear regression. In the multiple triglycerides. After Bonferroni correction of P values in
regression analyses, normal distribution of the residual error was
Table 2 (16 comparisons allowing P ⬍ .0031 to be sig-
determined using the Shapiro-Wilk W test and the constant vari-
ance of the errors was further checked by rando/mLy distributed nificant), all significant correlations remained. Further-
patterns in residual-by-predicted plots. A P-value ⬍ 0.05 was more, numerically similar ␳ and P values were obtained
considered to be statistically significant in all analyses. The Bon- if insulin sensitivity was corrected for LBM (M/LBM,
ferroni test was used to correct for multiple comparison statistics values not shown).
in Tables 2 and 3. Statistical analyses were performed using the
The correlations between sc fat cell volume/number,
JMP 10.0 software (SAS Institute Inc, NC).
visceral adipocyte volume, and insulin sensitivity (also as
M/LBM, values not shown) or plasma levels of insulin,
triglycerides, and HDL-cholesterol were analyzed using
Results
multiple linear regression (Table 3). All correlations re-
A flowchart for inclusion in the study is depicted in Figure mained statistically significant after correction for total
1. In total, 204 subjects were included in the study where body fat expressed in kilograms (or as percentage of total
body weight, values not shown). The same was true if
the Bonferroni correction was used (allowing P ⬍ .0042
to be significant). We also analyzed the influence of

Table 1. Clinical Characteristicsa

Mean (range)
Age, y (n ⫽ 204) 43 (18 – 64)
BMI, kg/m2 (n ⫽ 204) 41 (32.6 – 62.3)
Plasma insulin, mU/L (n ⫽ 203) 16.2 (2.5– 47.6)
Plasma HDL-cholesterol, mmol/L (n ⫽ 204) 1.2 (0.5–2.3)
Plasma triglycerides, mmol/L (n ⫽ 203) 1.4 (0.4 –3.6)
M-value, mg/kg/min (n ⫽ 177) 4.2 (0.0 –7.8)
LBM, kg (n ⫽ 201) 53.9 (36.1– 85.2)
Total body fat, kg (n ⫽ 202) 57.2 (34.9 –90.1)
Android fat, kg (n ⫽ 201) 5.9 (3.2–10.1)
EVAT, kg (n ⫽ 188) 2.3 (0.4 –5.3)
ESAT, kg (n ⫽ 188) 3.5 (0.8 – 6.3)
EVAT fat cell volume, pL (n ⫽ 203) 594 (167–1330)
EVAT fat cell number ⫻ 109 (n ⫽ 188) 4.3 (1.3–9.2)
ESAT fat cell volume, pL (n ⫽ 204) 923 (476 –1438)
ESAT fat cell number ⫻ 109 (n ⫽ 188) 4.3 (1.3– 8.6)
a
Figure 1. Flowchart for inclusion of subjects. Subjects included 183 women and 21 men.

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 14 January 2015. at 04:08 For personal use only. No other uses without permission. . All rights reserved.
doi: 10.1210/jc.2014-1526 jcem.endojournals.org E1873

Table 2. Correlation Between Adipose and Metabolic Variablesa

Adipose Tissue Parameters

Visceral Subcutaneous
Fat Cell Number Fat Cell Volume Fat Cell Number Fat Cell Volume
Metabolic Parameters ␳ P ␳ P ␳ P ␳ P
Plasma insulin ⫺0.080 .28 0.45 ⬍.0001 ⫺0.26 .0003 0.44 ⬍.0001
Insulin sensitivity (M) ⫺0.0038 .96 ⫺0.42 ⬍.0001 0.25 .0014 ⫺0.42 ⬍.0001
Plasma triglycerides ⫺0.027 .71 0.47 ⬍.0001 ⫺0.35 ⬍.0001 .33 ⬍.0001
Plasma HDL-cholesterol ⫺0.054 .47 ⫺0.22 .0020 0.29 ⬍.0001 ⫺0.28 .0002
a
Values were compared using Spearman’s correlation, with ␳ and P values shown. Numerically similar ␳ and P values were obtained if M was
corrected for total LBM (M/LBM, values not shown).

gender, diagnosed T2DM, or hypertension on the data Discussion

presented in Table 3 using analysis of covariance. All
correlations remained significant with nominal P This study was performed to elucidate how fat cell num-
value ⬍ .05 after adjustment for gender, diabetes, or ber, in comparison with fat cell size, associates with insulin
hypertension. resistance and other obesity-linked metabolic abnormal-
ities. It is well established that enlarged fat cell size is as-
sociated with several of the metabolic abnormalities ob-
served in conditions with excess body fat (14 –18). This
was confirmed in the present study, demonstrating that
individuals with large sc or visceral fat cells displayed un-
favorable values for insulin sensitivity and plasma levels of
insulin, triglycerides, and HDL-cholesterol. A novel and
intriguing observation was the opposite findings with sc
fat cell number. Thus, the number of fat cells in this depot
was positively correlated with insulin sensitivity and
HDL-cholesterol and negatively associated with fasting
plasma insulin and triglyceride levels. These relationships,
which were independent of total body fat mass, suggest
that the size and number of fat cells in different adipose
depots play different roles in obesity-associated metabolic
complications. Large fat cells in both sc and visceral adi-
pose tissue are associated with a more pernicious meta-
bolic phenotype. In abdominal sc fat, more adipocytes is
related to a more favorable state, whereas fat cell number
in the visceral depot does not correlate with any of the
studied metabolic parameters. Admittedly, although the
correlations were highly significant, the fact that the ␳
values ranged from ⫺0.5 to ⫹0.5 suggest that it takes
relatively large interindividual differences in adipocyte
size/number to observe differences in the studied meta-
bolic parameters. However, our assessments were per-
formed in a rather homogeneous cohort of obese subjects
consisting predominantly of women. It is possible that
inclusion of a larger number of individuals, including nor-
mal-weight subjects and more men would result in differ-
ent (and possibly stronger) correlations. More impor-
Figure 2. The relationship between sc fat cell volume or number and
plasma insulin (P-insulin), insulin sensitivity (M-value), plasma tantly, the predictive role and possible causal relationship
triglycerides (TG), or plasma HDL-cholesterol. between adipocyte size/number and metabolic outcome

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 14 January 2015. at 04:08 For personal use only. No other uses without permission. . All rights reserved.
E1874 Rydén et al Adipocyte Size and Number J Clin Endocrinol Metab, October 2014, 99(10):E1870 –E1876

Table 3. Relationship Between Adipose and Metabolic Variables After Correction for Total Body Fata
Adipose Variables

Visceral Fat Cell Subcutaneous Fat Subcutaneous Fat

Volume Cell Volume Cell Number
Metabolic Variable Std ␤ P Std ␤ P Std ␤ P
Plasma insulin b
0.42 ⬍.0001 0.46 ⬍.0001 ⫺0.34 ⬍.0001
Insulin sensitivity (M) ⫺0.35 ⬍.0001 ⫺0.35 ⬍.0001 0.31 .0002
Plasma triglyceridesb 0.48 ⬍.0001 0.36 ⬍.0001 ⫺0.38 ⬍.0001
Plasma HDL-cholesterolb ⫺0.23 .0015 ⫺0.31 ⬍.0001 0.35 ⬍.0001
a
Values were compared by multiple linear regression analysis using one adipose variable plus total body fat (in kilograms) together as independent
variables. Standardized ␤-values (Std ␤) and P values for the adipose variable is given. A negative Std ␤ indicates an inverse relationship between
the adipose and the metabolic variable. Numerically similar Std ␤ and P values were obtained if M was corrected for total LBM (M/LBM) or if total
fat mass was expressed in percentage of total body weight (values not shown).
b
Values were log10-transformed to obtain normally distributed residual errors.

must be validated in prospective and/or mechanistic clinical data and murine models, demonstrated that re-
studies. duced activity of the transcription factor early B cell fac-
It is important to note that calculation of fat cell num- tor-1, possibly downregulated by increased local expres-
ber is based on the quotient between 2 independent pa- sion of TNF␣, could be an important causal factor
rameters; the size of a specific fat depot and the mean promoting the development of adipose hypertrophy via
weight of its adipocytes. Although variations in adipocyte altered expression of genes involved in adipogenesis and
size between different sc depots are rather small (26), fat lipid turnover (31). These results suggest that early B cell
cell volume (and thereby weight) can differ significantly factor-1 may constitute a causal link between TNF␣ ex-
between other adipose depots. This implies 1) that our pression and fat cell size and number, a notion that needs
determinations of adipose cellularity are only relevant for to be validated in future studies. A related possibility ex-
ESAT and EVAT and 2) that the number of fat cells is not plaining the association between adipose cellularity and
a simple function of fat cell diameter. The latter notion was metabolic changes may be that the capacity of adipose
corroborated by the fact that despite the observed asso- tissue to store lipids is limited (32). Consequently, when
ciation with adipocyte volume in both regions, only sc fat adipose expansion reaches a maximum, lipids are stored
cell number was related to the studied metabolic ectopically in skeletal muscle, liver, and pancreatic islets,
parameters. which in turn leads to lipotoxic effects including abnormal
Which mechanisms could explain differences in adi- lipoprotein metabolism, insulin dysfunction, and altered
pose cellularity and their potential influence on hormonal insulin secretion. The major contributor (⬎95%) to adi-
and metabolic homeostasis? Although this study was not pocyte volume is the lipid droplet. As measured with the
designed to answer this question, some speculations can be method adopted in the present study, fat cell volume in
offered. Adipose inflammation has emerged as an impor- adult humans cannot become larger than ⬃1400 pL (33).
tant factor contributing to insulin resistance (27). Inter- When this size limit is reached, the tissue must generate
estingly, adipocyte size correlates closely with the expres- more fat cells to be able to expand further (13, 32). This
sion and secretion of the proinflammatory cytokine TNF␣ is supported by the observation that total fat cell number
in lean healthy women (28) and insulin resistance/hyper- is approximately doubled in obese compared with age-
lipidemia in morbidly obese women (14). In these studies, and gender-matched lean subjects (33). Interindividual
subjects with many small sc fat cells displayed lower TNF␣ differences in the weight of the greater omentum are de-
secretion and better insulin sensitivity, respectively. Thus, termined by variations in fat cell number and less by adi-
although the mechanisms are not clear, increased fat cell pocyte size (22), suggesting that visceral adipose expan-
size appears to alter several transcriptional regulatory net- sion may primarily be determined by increased adipocyte
works resulting in increased inflammation, altered lipid number. It is not yet known how fat cell number and size
metabolism, and insulin resistance. As discussed (29, 30), contribute to variations in sc adipose tissue mass. Never-
proinflammatory adipokines and lipids secreted from en- theless, together with the data presented herein, it could be
larged fat cells in visceral white adipose tissue may be of speculated that subjects who can increase their sc fat cell
particular importance because this depot is drained via the number may develop less ectopic lipid deposition and/or
portal vein into the liver, thereby impacting directly on that newly formed sc adipocytes display a less pernicious
hepatocyte function. A recent publication, combining functional profile. The fact that the sc depot is consider-

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 14 January 2015. at 04:08 For personal use only. No other uses without permission. . All rights reserved.
doi: 10.1210/jc.2014-1526 jcem.endojournals.org E1875

ably larger than the visceral, may explain why fat cell to measure fat cell size in small sc needle biopsies (40), and
number associates with metabolic variables only in the fat mass in different regions can be assessed by DEXA,
former region. with the use of recently developed software.
Our calculations of fat cell number rely on estimations In conclusion, this study shows that fat cell size and
of abdominal sc and visceral adipose depots using DEXA number, independently of body fat mass, associate differ-
and the recently developed CoreScan tool. Although as- ently with metabolic parameters in obese subjects. In-
sessments with this method show very good correlations creased sc or visceral fat cell size correlates with an adverse
with values obtained using CT (21, 22), they remain in- insulin/lipid profile, whereas increased sc fat cell number
direct determinations that should warrant some caution in is coupled to a less pernicious phenotype.
their interpretation. Admittedly, there have been concerns
about the accuracies regarding DEXA measurements of
adipose mass in obese individuals, but these pertain to Acknowledgments
previously used DEXA equipment and software (34). Nev-
We thank research nurses Britt-Marie Leijonhufvud, Yvonne
ertheless, a limitation of our study is that no visceral ad-
Widlund, and Katarina Hertel as well as Eva Sjölin, Kerstin
ipose tissue phantom was used for calibrations. However,
Wåhlén, and Elisabeth Dungner from Karolinska Institutet for
the precision of the CoreScan tool to measure visceral ad- excellent technical assistance.
ipose tissue mass has recently been estimated and shown
to have relatively low error in both phantoms and obese Address all correspondence and requests for reprints to: Mi-
women (35). kael Rydén, Professor, MD, PhD, Karolinska Institutet, Depart-
We investigated only a limited set of relevant metabolic ment of Medicine, Karolinska University Hospital Huddinge
C2–94, The Clinic for Endocrinology, Metabolism, and Diabe-
phenotypes linked to excess body fat, ie, insulin sensitivity
tes, SE-141 86 Stockholm, Sweden. E-mail: mikael.ryden@ki.se.
and plasma insulin, triglycerides, and HDL-cholesterol. This study was supported by grants from the Swedish Re-
This approach was chosen first because these are the most search Council, Novo Nordisk Foundation, Swedish Diabetes
commonly investigated parameters and second to avoid Association, EASD/Lilly Foundation, and Strategic Research
the inherent problems of using too many variables in mul- Program in Diabetes at Karolinska Institutet.
P.A. and M.R. designed the study and wrote the first version
tiple testing. Hence, we have not established how other
of the manuscript. All 3 authors contributed equally to data
obesity-associated metabolic variables such as lipopro- collection and analysis, writing of the manuscript, and approving
teins and proinflammatory factors as well as other lipid the final version.
and glucose measures are related to adipose cellularity. Disclosure Summary: None of the authors have any conflicts
Moreover, our cohort included participants with non–in- of interest to declare.
sulin-dependent diabetes (T2DM) and/or hypertension,
and there was a clear imbalance in gender. This reflects
that more than two-thirds of the subjects undergoing bari- References
atric surgery in Sweden are women and many have devel- 1. Smyth S, Heron A. Diabetes and obesity: the twin epidemics. Nat
oped obesity-associated metabolic complications upon ac- Med. 2005;12:75– 80.
ceptance for surgery. The subjects studied herein could 2. Mingrone G, Panunzi S, De Gaetano A, et al. Bariatric surgery versus
conventional medical therapy for type 2 diabetes. N Engl J Med.
therefore in theory not be representative for obese indi-
2012;366:1577–1585.
viduals in the general population. Nevertheless, when cor- 3. Carlsson LM, Peltonen M, Ahlin S, et al. Bariatric surgery and pre-
rected for gender, diabetes, or hypertension, the observed vention of type 2 diabetes in Swedish obese subjects. N Engl J Med.
correlations remained statistically significant with nomi- 2012;367:695–704.
4. Gloy VL, Briel M, Bhatt DL, et al. Bariatric surgery versus non-
nal P values ⬍ .05. surgical treatment for obesity: a systematic review and meta-analysis
It has been suggested that the classical categorization of of randomised controlled trials. BMJ. 2013;347:f5934.
body fat status based on BMI and adipose distribution is 5. Arterburn DE, Bogart A, Sherwood NE, et al. A multisite study of
long-term remission and relapse of type 2 diabetes mellitus following
insufficient (36). Up to one-third of overweight or obese gastric bypass. Obes Surg. 2013;23:93–102.
subjects are metabolically healthy (37). Furthermore, the 6. Knop FK, Taylor R. Mechanism of metabolic advantages after bari-
association between body fat distribution and insulin re- atric surgery: it’s all gastrointestinal factors versus it’s all food re-
striction. Diabetes Care. 2013;36(Suppl 2):S287–S291.
sistance/T2DM is controversial (38, 39). The present
7. Klein S, Fontana L, Young VL, et al. Absence of an effect of lipo-
study suggest that classifying overweight/obese patients suction on insulin action and risk factors for coronary heart disease.
according to their fat cell size and number might be clin- N Engl J Med. 2004;350:2549 –2557.
ically useful to estimate the risk of T2DM and related 8. Mohammed BS, Cohen S, Reeds D, Young VL, Klein S. Long-term
effects of large-volume liposuction on metabolic risk factors for cor-
metabolic complications and thereby determine treatment onary heart disease. Obesity (Silver Spring). 2008;16:2648 –2651.
strategies, at least in specialized clinics. It is relatively easy 9. Fabbrini E, Tamboli RA, Magkos F, et al. Surgical removal of omen-

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 14 January 2015. at 04:08 For personal use only. No other uses without permission. . All rights reserved.
E1876 Rydén et al Adipocyte Size and Number
tal fat does not improve insulin sensitivity and cardiovascular risk
factors in obese adults. Gastroenterology. 2010;139:448 – 455.
10. Dillard TH, Purnell JQ, Smith MD, et al. Omentectomy added to
Roux-en-Y gastric bypass surgery: a randomized, controlled trial.
Surg Obes Relat Dis. 2013;9:269 –275.
11. Herrera MF, Pantoja JP, Velázquez-Fernández D, et al. Potential
additional effect of omentectomy on metabolic syndrome, acute-
phase reactants, and inflammatory mediators in grade III obese pa-
tients undergoing laparoscopic Roux-en-Y gastric bypass: a ran-
domized trial. Diabetes Care. 2010;33:1413–1418.
12. Andersson DP, Thorell A, Löfgren P, et al. Omentectomy in addition
to gastric bypass surgery and influence on insulin sensitivity: a ran-
domized double blind controlled trial [published online January 12,
2014]. Clin Nutr. doi:10.1016/j.clnu.2014.01.004.
13. Hirsch J, Batchelor B. Adipose tissue cellularity in human obesity.
Clin Endocrinol Metab. 1976;5:299 –311.
14. Hoffstedt J, Arner E, Wahrenberg H, et al. Regional impact of ad-
ipose tissue morphology on the metabolic profile in morbid obesity.
Diabetologia. 2010;53:2496 –2503.
15. Veilleux A, Caron-Jobin M, Noël S, Laberge PY, Tchernof A. Vis-
ceral adipocyte hypertrophy is associated with dyslipidemia inde-
pendent of body composition and fat distribution in women. Dia-
betes. 2011;60:1504 –1511.
16. Salans LB, Knittle JL, Hirsch J. The role of adipose cell size and
adipose tissue insulin sensitivity in the carbohydrate intolerance of
human obesity. J Clin Invest. 1968;47:153–165.
17. Lundgren M, Svensson M, Lindmark S, Renström F, Ruge T, Er-
iksson JW. Fat cell enlargement is an independent marker of insulin
resistance and ’hyperleptinaemia’. Diabetologia. 2007;50:625–
633.
18. Arner E, Westermark PO, Spalding KL, et al. Adipocyte turnover:
relevance to human adipose tissue morphology. Diabetes. 2010;59:
105–109.
19. Weyer C, Foley JE, Bogardus C, Tataranni PA, Pratley RE. Enlarged
subcutaneous abdominal adipocyte size, but not obesity itself, pre-
dicts type II diabetes independent of insulin resistance. Diabetolo-
gia. 2000;43:1498 –1506.
20. Lönn M, Mehlig K, Bengtsson C, Lissner L. Adipocyte size predicts
incidence of type 2 diabetes in women. Faseb J. 2010;24:326 –331.
21. Kaul S, Rothney MP, Peters DM, et al. Dual-energy x-ray absorp-
tiometry for quantification of visceral fat. Obesity (Silver Spring).
2012;20:1313–1318.
22. Arner P, Andersson DP, Thörne A, et al. Variations in the size of the
major omentum are primarily determined by fat cell number. J Clin
Endocrinol Metab. 2013;98:E897–E901.
23. Tchoukalova YD, Harteneck DA, Karwoski RA, Tarara J, Jensen
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 14 January 2015. at 04:08 For personal use only. No other uses without permission. . All rights reserved.
J Clin Endocrinol Metab, October 2014, 99(10):E1870 –E1876
MD. A quick, reliable, and automated method for fat cell sizing. J
Lipid Res. 2003;44:1795–1801.
24. Goldrick RB. Morphological changes in the adipocyte during fat
deposition and mobilization. Am J Physiol. 1967;212:777–782.
25. Hirsch J, Gallian E. Methods for the determination of adipose cell
size in man and animals. J Lipid Res. 1968;9:110 –119.
26. Krotkiewski M, Björntorp P, Sjöström L, Smith U. Impact of obesity
on metabolism in men and women. Importance of regional adipose
tissue distribution. J Clin Invest. 1983;72:1150 –1162.
27. Rydén M, Arner P. Tumour necrosis factor-alpha in human adipose
tissue – from signalling mechanisms to clinical implications. J Intern
Med. 2007;262:431– 438.
28. Arner E, Ryden M, Arner P. Tumor necrosis factor alpha and reg-
ulation of adipose tissue. N Engl J Med. 2010;362:1151–1153.
29. Wree A, Kahraman A, Gerken G, Canbay A. Obesity affects the liver
- the link between adipocytes and hepatocytes. Digestion. 2011;83:
124 –133.
30. Wree A, Mayer A, Westphal S, et al. Adipokine expression in brown
and white adipocytes in response to hypoxia. J Endocrinol Invest.
2012;35:522–527.
31. Gao H, Mejhert N, Fretz JA, et al. Early B cell factor 1 regulates
adipocyte morphology and lipolysis in white adipose tissue. Cell
Metab. 2014;19(6):981–992.
32. Virtue S, Vidal-Puig A. Adipose tissue expandability, lipotoxicity
and the metabolic syndrome–an allostatic perspective. Biochim Bio-
phys Acta. 2010;1801:338 –349.
33. Spalding KL, Arner E, Westermark PO, et al. Dynamics of fat cell
turnover in humans. Nature. 2008;453:783–787.
34. Park YW, Heymsfield SB, Gallagher D. Are dual-energy x-ray ab-
sorptiometry regional estimates associated with visceral adipose tis-
sue mass? Int J Obes Relat Metab Disord. 2002;26:978 –983.
35. Rothney MP, Xia Y, Wacker WK, et al. Precision of a new tool to
measure visceral adipose tissue (VAT) using dual-energy x-ray ab-
sorptiometry (DXA). Obesity (Silver Spring). 2013;21:E134 –E136.
36. Ahima RS, Lazar MA. Physiology. The health risk of obesity– better
metrics imperative. Science. 2013;341:856 – 858.
37. Blüher M. The distinction of metabolically ’healthy’ from ’un-
healthy’ obese individuals. Curr Opin Lipidol. 2010;21:38 – 43.
38. Miles JM, Jensen MD. Counterpoint: visceral adiposity is not caus-
ally related to insulin resistance. Diabetes Care. 2005;28:2326 –
2328.
39. Lebovitz HE, Banerji MA. Point: visceral adiposity is causally re-
lated to insulin resistance. Diabetes Care. 2005;28:2322–2325.
40. Smith U, Sjöström L, Björnstorp P. Comparison of two methods for
determining human adipose cell size. J Lipid Res. 1972;13:822–
824.