Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

ORIGINAL INVESTIGATIONS

Pathogenesis and Treatment of Kidney Disease and Hypertension

A Central Body Fat Distribution Is Related to Renal Function


Impairment, Even in Lean Subjects
Sara-Joan Pinto-Sietsma, MD, PhD, Gerjan Navis, MD, PhD, Wilbert M.T. Janssen, MD, PhD,
Dick de Zeeuw, MD, PhD, Reinhold O.B. Gans, MD, PhD, and
Paul E. de Jong, MD, PhD, for the PREVEND Study Group

● Background: Overweight and obesity are believed to be associated with renal damage. Whether this depends on
fat distribution is not known. We hypothesize that in addition to overweight, fat distribution may be associated with
renal function abnormalities. Methods: We studied the relation between body weight and fat distribution and
microalbuminuria and elevated or diminished filtration in 7,676 subjects without diabetes. Microalbuminuria is
defined as urinary albumin excretion (UAE) of 30 to 300 mg/24 h. Elevated and diminished filtration are defined as
filtration plus or minus 2 SDs of a nondiabetic lean group with a peripheral fat distribution and UAE of 0 to 15 mg/24
h, corrected for age and sex. The total population was divided into six groups according to body weight (overweight
is defined as body mass index [BMI] > 25 and < 30 kg/m2; obesity, as BMI > 30 kg/m2) and fat distribution. Results:
In logistic regression analysis, obese subjects with central fat distribution had a greater risk for microalbuminuria
(relative risk, 1.7; 95% confidence interval, 1.19 to 2.35). Obese subjects with either peripheral or central fat
distribution had a greater risk for elevated filtration (relative risk, 3.2; 95% confidence interval, 1.19 to 8.47; relative
risk, 2.6; 95% confidence interval, 1.59 to 4.28, respectively). Furthermore, subjects with central fat distribution,
either lean, overweight, or obese, had a greater risk for diminished filtration (relative risk, 1.9; 95% confidence
interval, 1.19 to 3.12; relative risk, 2.0; 95% confidence interval, 1.19 to 3.19; and relative risk, 2.7; 95% confidence
interval, 1.46 to 4.85, respectively). Finally, by dividing waist-hip ratio (WHR) into quartiles, greater WHR was
associated with a greater risk for diminished filtration, even when corrected for BMI. Conclusion: Not only
overweight and obese subjects, but also lean subjects with central fat distribution are at risk for diminished
filtration. Therefore, a central pattern of fat distribution, not overweight or obesity by itself, seems to be important
for renal impairment. Am J Kidney Dis 41:733-741.
© 2003 by the National Kidney Foundation, Inc.

INDEX WORDS: Overweight; obesity; central fat distribution; renal function; renal impairment; microalbuminuria

T HE PREVALENCE of overweight and obe-


sity, defined according to World Health
Organization criteria1 as a body mass index (BMI)
Also from human studies, it has been argued
that overweight may lead to end-stage renal
disease. For instance, focal glomerular sclerosis
more than 25 kg/m2 and more than 30 kg/m2, and/or diabetic nephropathy have been observed
respectively, appears to be steadily increasing.2 in massive obese patients who presented with
Because excess weight gain initiates a cascade of proteinuria.11 Recently, Praga et al12 showed that
metabolic, endocrine, cardiovascular, and renal
changes, obesity is a major health problem.
The impact of obesity on renal function has From the Department of Internal Medicine, Division of
been investigated over the years. Obesity is asso- Nephrology, University Hospital Groningen; and Depart-
ment of Clinical Pharmacology, University Groningen, and
ciated with both renal hyperfiltration3-5 and hyper- Groningen University Institute for Drug Exploration, Gro-
perfusion6 and microalbuminuria.7 Moreover, ningen, The Netherlands
experimental observations of an increased glo- Received June 14, 2002; accepted December 17, 2002.
merular filtration rate, mainly in superficial For the Prevention of Renal and Vascular End-Stage
Disease (PREVEND) Study Group (see Appendix).
nephrons and in association with expansion of Supported in part by grant no. E033 from the Dutch
glomerular area and mesangial matrix, were made Kidney Foundation.
in Zucker rats with genetic obesity.8 Later, glo- Address reprint requests to Sara-Joan Pinto-Sietsma, MD,
merular filtration rate tends to normalize and PhD, Department of Internal Medicine, University Hospital
Maastricht, P Debeyelaan 25, 6202 AZ Maastricht, The
subsequently decreases, together with the devel- Netherlands. E-mail: pintosj@hotmail.com
opment of progressive albuminuria and glomeru- © 2003 by the National Kidney Foundation, Inc.
losclerosis.9,10 0272-6386/03/4104-0001$30.00/0

American Journal of Kidney Diseases, Vol 41, No 4 (April), 2003: pp 733-741 733
734 PINTO-SIETSMA ET AL

obese subjects were more at risk to develop committee and conducted in accordance with the guidelines
proteinuria and renal failure after unilateral ne- of the Declaration of Helsinki. All participants attending the
outpatient clinic gave written informed consent.
phrectomy than nonobese subjects.
However, there is evidence that not over- Study Design
weight or obesity per se, but distribution of body The screening program in the outpatient clinic consisted
fat, is related to a variety of physiological aberra- of two visits. At the first visit, participants completed a
tions. Central body fat distribution is related to self-administered questionnaire regarding demographics, car-
hyperinsulinemia, hypertension, hyperlipidemia, diovascular and renal history, and drug use. After removal of
and atherosclerosis.13,14 Interestingly, in a small shoes and heavy clothing, weight was measured to the
nearest 0.5 kg using a Seca balance scale (Seca Vogel &
study, Scaglione et al15 also showed that subjects Halke GmbH & Co, Hamburg, Germany). Height was mea-
with central fat distribution had reduced renal sured to the nearest 0.5 cm. Minimal waist circumference
plasma and blood flow and increased filtration was measured on bare skin at the natural indentation be-
fraction and albuminuria, whereas this could not tween the 10th rib and iliac crest. Hip circumference was
be observed in subjects with peripheral fat distri- measured at the maximum circumference of the buttocks.17
At both visits, blood pressure was measured in supine
bution. Conversely, these were both normoten- position every minute for 10 and 8 minutes (respectively)
sive and hypertensive subjects, and no compari- using an automatic Dinamap XL model 9300 series monitor
son was made with lean subjects with central fat (Critikon, Tampa, FL).
distribution. Subjects were asked to collect 24-hour urine samples on 2
We therefore hypothesized that central fat dis- consecutive days in the last week before the second visit.
Subjects were given oral and written instructions on how to
tribution, whether in lean, overweight, or obese collect 24-hour urine samples and instructed to postpone
subjects, is related to such renal abnormalities as urine collection in case of fever, urinary tract infection, or
microalbuminuria and elevated and diminished menstruation and refrain as far as possible from heavy
filtration in a nondiabetic population. To test this exercise during the collection period. Furthermore, subjects
hypothesis, we studied the cross-sectional rela- were asked to store the urine cold (4°C) for a maximum of 4
days before the second visit. Measurements of urinary vol-
tion between body weight and body fat distribu- ume and albumin and creatinine concentrations were per-
tion with albuminuria and creatinine clearance in formed on each collection. At the second visit, blood was
a large cohort collected to study causes and drawn after an overnight fast for determination of plasma
consequences of microalbuminuria in subjects glucose and serum creatinine and cholesterol levels.
without diabetes.
Calculations
Systolic and diastolic blood pressures were calculated as
METHODS the mean of the last two measurements of the two visits
(systolic blood pressure range, 85 to 239 mm Hg; diastolic
Study Population blood pressure range, 47 to 114 mm Hg). BMI was calcu-
This study is part of the ongoing Prevention of Renal and lated as the ratio between weight (kilograms) and the square
Vascular End-Stage Disease (PREVEND) study in the city of height (in meters; weight/height2; BMI range, 16.31 to
of Groningen, The Netherlands. All inhabitants aged 28 to 63.61 kg/m2). Waist-hip ratio (WHR) was calculated as the
75 years (n ⫽ 85,421) were asked to send in a morning urine ratio between minimal waist circumference (centimeters)
sample and fill out a short questionnaire on demographics and hip circumference (centimeters; WHR range, 0.45 to
and cardiovascular history. A total of 40,856 subjects (47.8%) 1.39). Body surface area was calculated according to DuBois
responded. Our actual study cohort was drawn from this and DuBois.18 Creatinine clearance is defined as the mean of
cohort based on all subjects with a urinary albumin concen- two creatinine clearances based on 24-hour urinary creati-
tration of 0.10 g/dL or greater (ⱖ1.0 g/L; n ⫽ 7,768) in their nine excretions divided by plasma creatinine level and
morning urine sample and a randomly selected control group corrected for body surface area (creatinine clearance range,
with a urinary albumin concentration less than 10 mg/L (n ⫽ 12.72 to 257.34 mL/min/1.73 m2 [0.21 to 4.30 mL/s/1.73
3,395). Details of this protocol have been described else- m2]). Urinary albumin excretion (UAE) is given as the mean
where.16 of two 24-hour urine excretions (UAE range, 1.0 to 4,710.0
In total, 11,163 subjects were invited to the outpatient mg/24 h).
clinic, of whom 8,592 subjects completed the screening
program. Subjects on insulin therapy or pregnant women Laboratory Methods
were excluded from participation in this screening program. Urinary albumin concentration was determined by neph-
Our total population consisted mainly of Caucasian subjects. elometry, with a threshold of 0.023 g/dL (0.23 g/L) and
Only 68 subjects (0.9%) were black and 163 subjects (2.1%) intra-assay and interassay coefficients of variation of 4.3%
were Asian. These groups were too small for a separate or less and 4.4% or less, respectively (Dade Behring Diag-
analysis. The study was approved by the medical ethics nostic, Marburg, Germany). Plasma glucose, serum choles-
CENTRAL FAT DISTRIBUTION AND RENAL FUNCTION 735

terol, and serum and urinary creatinine levels were deter- diminished filtration was performed in the following way.
mined using Kodak Ektachem dry chemistry (Eastman First, of the total population, we selected two control or
Kodak, Rochester, NY). Urinary leukocyte and erythrocyte healthy subgroups that consisted of either male or female
measurements were performed using Nephur-test⫹leuco nondiabetic lean control subjects with peripheral fat distribu-
sticks (Boehringer Mannheim, Mannheim, Germany). tion and UAE of 0 to 15 mg/24 h. In a linear regression
model with creatinine clearance corrected for body surface
Data Handling and Definitions area as the dependent variable and age as the independent
variable, we constructed a regression line of creatinine
In the present analysis, we excluded 433 subjects because
clearance for men and women separately. Because each of
of erythrocyturia or leukocyturia (erythrocytes ⬎ 50 cells/
these two subgroups showed a normal distribution concern-
&muL, leukocytes ⬎ 75 cells/&muL, or leukocytes of 75
ing creatinine clearance corrected for body surface area, we
cells/&muL and erythrocytes ⬎ 5 cells/&muL). In addition,
used the two times SD borders of this regression line of
301 subjects were excluded because they had non–insulin-
creatinine clearance to define elevated and diminished filtra-
dependent diabetes. Diabetes is defined as a fasting plasma
tion.20
glucose level of 126.13 mg/dL or greater (ⱖ7.0 mmol/L),
To investigate the risk between overweight in combina-
nonfasting plasma glucose level of 200 mg/dL or greater
tion with fat distribution and microalbuminuria and both
(ⱖ11.1 mmol/L), or use of oral antidiabetic drugs. Finally,
elevated and diminished filtration, we used a logistic regres-
182 subjects were excluded because of missing data. Alto-
sion model. We first performed a crude analysis. Second,
gether, 7,676 subjects were eligible for this analysis.
because fat distribution is highly related to sex, we tested
We differentiated subjects with central or peripheral fat
whether there was a sex-independent relation between over-
distribution by using WHR. Central fat distribution is de-
weight in combination with fat distribution and microalbu-
fined as a WHR of 0.9 or greater for men and 0.8 or greater
minuria and both elevated and diminished filtration. There-
for women, and peripheral fat distribution, a WHR less than
fore, we analyzed data only by adjusting for sex to be able to
0.9 for men and less than 0.8 for women.19 Lean is defined as
get insight on the influence of sex in the model. Third, odds
a BMI of 25 kg/m2 or less, overweight is defined1 as a BMI
ratios were calculated after adjustment for other possible
greater than 25 and 30 kg/m2 or less, and obesity is defined
confounding factors, such as age, smoking, systolic and
as a BMI greater than 30 kg/m2.
diastolic blood pressure, use of antihypertensive medication,
We divided our study population into six groups accord-
serum cholesterol level, lipid-lowering medication, and
ing to BMI and WHR. Group 1 included lean subjects with
plasma glucose level. Calculated odds ratios and 95% confi-
peripheral fat distribution; group 2, lean subjects with cen-
dence intervals are expressed as an approximation of relative
tral fat distribution; group 3, overweight subjects with periph-
risk.
eral fat distribution; group 4, overweight subjects with
central fat distribution; group 5, obese subjects with periph-
RESULTS
eral fat distribution; and group 6, obese subjects with central
fat distribution. Table 1 lists unadjusted population characteris-
Furthermore, microalbuminuria is defined according to tics according to body weight and fat distribu-
the classic definition of albumin of 30 to 300 mg/24 h.
Elevated and diminished filtration are defined as a creati-
tion. All groups with central fat distribution were
nine clearance ⫾ two times the SD of the creatinine clear- more often men, older, had higher blood pressure
ance regression line of a group of nondiabetic, peripheral, and glucose levels, a greater percentage of sub-
lean control subjects with UAE of 0 to 15 mg/24 h.16 The jects with hypertension, and a greater percentage
creatinine clearance regression line with the two times SD of subjects with hypercholesterolemia, com-
borders was obtained by means of linear regression analysis
adjusted for age and sex. Hypercholesterolemia is defined as
pared with lean control subjects with peripheral
a plasma cholesterol level greater than 250.97 mg/dL (6.5 fat distribution. Overweight subjects with central
mmol/L) or greater than 193.05 mg/dL (5.0 mmol/L) in case fat distribution were more often men, older, had
of a previous myocardial infarction. Hypertension is defined higher blood pressure and glucose levels, and a
as systolic blood pressure of 160 mm Hg or greater, diastolic greater percentage of subjects with hypertension
blood pressure of 95 mm Hg or greater, or use of antihyper-
tensive medication. Smokers are defined as currently smok-
or hypercholesterolemia compared with lean sub-
ing, and nonsmokers, as never having smoked. jects with central fat distribution. Obese subjects
with central fat distribution had a greater glucose
Statistical Analysis and cholesterol level and greater percentage of
All calculations were performed using SPSS, version 9.0 subjects with hypertension compared with lean
(SPSS Inc, Chicago, IL) software. Continuous data are and overweight subjects with central fat distribu-
reported as mean ⫾ SD. In case of skewed distribution, tion.
median with 25th and 75th percentiles was presented. Differ- Concerning renal parameters, lean and over-
ences among the six groups were assessed using chi-square
analysis or analysis of variance. All P are two tailed. P less weight subjects with central fat distribution had
than 0.05 is considered statistically significant. greater UAE, a greater percentage of subjects
The linear regression model used to define elevated and with microalbuminuria, and a lower creatinine
736 PINTO-SIETSMA ET AL

Table 1. Population Characteristics According to Body Weight and Fat Distribution

Lean Overweight Obese


Fat
Distribution Peripheral Central Peripheral Central Peripheral Central

No. of patients (%) 2,033 (26.8) 1,406 (18.5) 654 (8.6) 2,398 (31.6) 103 (1.4) 993 (13.1)
Men (%) 42.9 47.4* 48.9* 63.2*†‡ 20.4* 51.2*§㛳
Age (y) 42 ⫾ 10 48 ⫾ 12* 45 ⫾ 12* 53 ⫾ 12*†‡ 49 ⫾ 13* 53 ⫾ 12*†‡
BMI (kg/m2) 22.17 ⫾ 1.75 23.04 ⫾ 1.50* 26.65 ⫾ 1.27* 27.26 ⫾ 1.39*†‡ 32.86 ⫾ 3.19* 33.22 ⫾ 3.46*†‡§
WHR 0.79 ⫾ 0.06 0.90 ⫾ 0.07* 0.81 ⫾ 0.06* 0.93 ⫾ 0.07*†‡ 0.78 ⫾ 0.06 0.95 ⫾ 0.08*†‡§㛳
SBP (mm Hg) 118 ⫾ 15 124 ⫾ 18* 127 ⫾ 17* 135 ⫾ 20*†‡ 131 ⫾ 20* 140 ⫾ 20*†‡§㛳
DBP (mm Hg) 69 ⫾ 9 72 ⫾ 9* 73 ⫾ 9* 77 ⫾ 9*†‡ 72 ⫾ 10* 78 ⫾ 10*†‡㛳
Hypertension (%) 10.2 21.3* 24.2* 42.4*†‡ 34.0* 54.9*†‡§㛳
Serum cholesterol 201.16 ⫾ 40.54 215.44 ⫾ 42.86* 214.67 ⫾ 42.47* 226.25 ⫾ 42.86*†‡ 218.92 ⫾ 42.08* 230.50 ⫾ 41.31*†‡§
(mg/dL)
Hypercholesterolemia (%) 12.7 25.5* 18.2* 32.1*†‡ 21.4* 33.8*†‡㛳
Glucose (mg/dL) 79.28 ⫾ 9.01 82.88 ⫾ 10.81* 82.88 ⫾ 10.81* 88.23 ⫾ 10.81*†‡ 86.47 ⫾ 12.61* 91.89 ⫾ 12.61*†‡§㛳
Smoking (%) 53.7 62.2* 41.4* 53.6†‡ 35.5* 50.3†‡㛳
UAE (mg/24 h) 7.9 (5.8–12.5) 8.4 (6.0–13.5)* 8.1 (6.1–12.4)* 10.2 (6.5–20.6)*†‡ 8.5 (6.1–16.3) 12.6 (7.7–29.2)*†‡§
Microalbumaria (%) 6.7 8.9* 6.3 15.9*†‡ 13.6* 21.2*†‡§
Serum creatinine 0.92 ⫾ 0.19 0.93 ⫾ 0.18 0.95 ⫾ 0.16* 0.98 ⫾ 0.19*†‡ 0.89 ⫾ 0.16 0.97 ⫾ 0.17*†㛳
(mg/dL)
Creatinine clearance 94.97 ⫾ 19.02 92.20 ⫾ 20.40* 95.02 ⫾ 21.07* 91.14 ⫾ 21.38*‡ 91.81 ⫾ 21.12 93.00 ⫾ 22.97
mL/min/1.73 (m2)
EF (%) 2.9 4.0 4.6* 3.8* 4.9 6.8*†§
DF (%) 2.8 3.5 2.8 3.4 1.9 3.9

NOTE. Values expressed as mean ⫾ SD, except for UAE, expressed as median (25th to 75th percentiles). Conversion
factors from conventional to SI units for cholesterol, glucose, and serum creatinine are 0.0259 0.0555, and 88.4, respectively.
Abbreviations: EF, elevated filtration; DF, diminished filtration; SBP, systolic blood pressure; DBP, diastolic blood
pressure; UAE, urinary albumin excretion.
*P ⬍ 0.05 versus lean control subjects with peripheral fat distribution.
†P ⬍ 0.05 versus lean subjects with central fat distribution.
‡P ⬍ 0.05 versus overweight subjects with peripheral fat distribution.
§P ⬍ 0.05 versus overweight subjects with central fat distribution.
㛳P ⬍ 0.05 versus obese subjects with peripheral fat distribution.

clearance compared with lean and overweight among the six groups. To correct for these differ-
subjects with peripheral fat distribution. Concern- ences, we also analyzed our data in a logistic
ing obese subjects with central fat distribution, regression model with microalbuminuria and both
except for serum creatinine level, there was no elevated and diminished filtration as dependent
significant difference compared with obese sub- variables.
jects with peripheral fat distribution. Further- First, we analyzed the crude relation between
more, overweight subjects with central fat distri- overweight and renal abnormalities, such as mi-
bution had greater UAE and a greater percentage croalbuminuria and both elevated and dimin-
of subjects with microalbuminuria compared with ished filtration (Tables 2, 3, and 4). Second,
lean subjects with central fat distribution. Obese because it seems obvious that sex acts as a
subjects with central fat distribution had greater confounder on overweight and fat distribution,
UAE, a greater percentage of subjects with mi- which also can be concluded from Table 1, we
croalbuminuria, and elevated filtration compared assessed the impact of sex on overweight and fat
with both the lean and overweight group with distribution in a logistic regression analysis. We
central fat distribution. therefore analyzed our data with and without
Because Table 1 shows only the crude differ- adjusting for sex, which did not severely change
ence among the six groups without statistical the model (Tables 2, 3, and 4). Furthermore, sex
corrections, we are not allowed to draw conclu- did not act as an interaction term when used as
sions at this point. Table 1 only provides us such in the models and therefore did not influ-
insight in possible differences or confounders ence results.
CENTRAL FAT DISTRIBUTION AND RENAL FUNCTION 737

Table 2. Adjusted Relative Risks for Microalbuminuria According to Body Weight and Fat Distribution

Microalbuminuria

Crude Relative Risk Sex Relative Risk Confounder Relative Risk

Lean
Peripheral 1.0* 1.0* 1.0*
Central 1.4 (1.09–1.81)† 1.4 (1.06–1.77) 0.7 (0.51–1.00)
Overweight
Peripheral 0.9 (0.65–1.35) 0.9 (0.63–1.30) 0.6 (0.38–1.02)
Central 3.0 (2.40–3.64)* 2.6 (2.14–3.25)* 1.0 (0.73–1.31)
Obese
Peripheral 2.2 (1.23–4.03)† 2.7 (1.46–4.86)† 1.2 (0.50–2.85)
Central 4.6 (3.62–5.80)* 4.5 (3.52–5.66)* 1.7 (1.19–2.35)†

NOTE. Adjusted for the confounders age, sex, systolic and diastolic blood pressure, antihypertensive medication, serum
cholesterol level, lipid-lowering medication, plasma glucose level, and smoking. Relative risks and 95% confidence intervals
are for lean subjects with central fat distribution, overweight subjects with a peripheral fat distribution, overweight subjects
with central fat distribution, obese subjects with peripheral fat distribution, and obese subjects with central fat distribution
compared with lean control subjects with peripheral fat distribution. The significance given in the group with lean control
subjects with peripheral fat distribution refers to the overall significance of the adjusted model, with symbols representing the
significance as stated before.
*P ⬍ 0.001.
†P ⬍ 0.01.

Finally, we assessed whether there was an glucose level, and smoking. These were all con-
independent relation between overweight and founders, as observed from Table 1. Subjects
renal abnormalities. Therefore, in addition to with central fat distribution, whether they were
sex, we adjusted for age, systolic and diastolic lean, overweight, or obese, showed an indepen-
blood pressure, antihypertensive medication, se- dent greater risk for diminished filtration (Table
rum cholesterol level, lipid-lowering medication, 4). Obese subjects with and without central fat

Table 3. Adjusted Relative Risks for Elevated Filtration According to Body Weight and Fat Distribution

Elevated Filtration

Crude Relative Risk Sex Relative Risk Confounder Relative Risk

Lean
Peripheral 1.0* 1.0* 1.0†
Central 1.4 (0.98–2.07) 1.4 (0.99–2.08) 1.2 (0.77–1.92)
Overweight
Peripheral 1.6 (1.04–2.56)‡ 1.7 (1.05–2.59)‡ 1.7 (1.00–2.98)
Central 1.3 (0.96–1.88) 1.4 (0.99–1.96) 1.4 (0.87–2.12)
Obese
Peripheral 1.7 (0.67–4.35) 1.6 (0.64–4.19) 3.2 (1.19–8.47)‡
Central 2.5 (1.74–3.58)* 2.5 (1.77–3.64)* 2.6 (1.59–4.28)*

NOTE. Adjusted for the confounders age, sex, systolic and diastolic blood pressure, antihypertensive medication, serum
cholesterol level, lipid-lowering medication, plasma glucose level, and smoking. Relative risks and 95% confidence intervals
are for lean subjects with central fat distribution, overweight subjects with peripheral fat distribution, overweight subjects with
central fat distribution, obese subjects with peripheral fat distribution, and obese subjects with central fat distribution
compared with lean control subjects with peripheral fat distribution. The significance given in the group with lean control
subjects with peripheral fat distribution refers to the overall significance of the adjusted model, with the symbols representing
the significance as stated before.
*P ⬍ 0.001.
†P ⬍ 0.01.
‡P ⬍ 0.05.
738 PINTO-SIETSMA ET AL

Table 4. Adjusted Relative Risks for Diminished Filtration According to Body Weight and Fat Distribution

Diminished Filtration

Crude Relative Risk Sex Relative Risk Confounder Relative Risk

Lean
Peripheral 1.0 1.0 1.0‡
Central 1.3 (0.87–1.91) 1.3 (0.87–1.90) 1.9 (1.19–3.12)
Overweight
Peripheral 1.0 (0.59–1.74) 1.0 (0.59–1.73) 1.3 (0.67–2.70)
Central 1.2 (0.88–1.76) 1.2 (0.86–1.74) 2.0 (1.19–3.19)†
Obese
Peripheral 0.7 (0.17–2.94) 0.7 (0.17–2.99) 0.01 (0.00–⬎100)
Central 1.5 (0.97–2.23) 1.5 (0.96–2.22) 2.7 (1.46–4.85)‡

NOTE. Adjusted for the confounders age, sex, systolic and diastolic blood pressure, antihypertensive medication, serum
cholesterol level, lipid-lowering medication, plasma glucose level, and smoking. Relative risks and 95% confidence intervals
are for lean subjects with central fat distribution, overweight subjects with peripheral fat distribution, overweight subjects with
central fat distribution, obese subjects with peripheral fat distribution, and obese subjects with central fat distribution
compared with lean control subjects with peripheral fat distribution. The significance given in the group with lean control
subjects with peripheral fat distribution refers to the overall significance of the adjusted model, with symbols representing the
significance as stated before.
†P ⬍ 0.01.
‡P ⬍ 0.05.

distribution showed an independent greater risk ished filtration (Wald statistic, 5.38; P ⬍ 0.02)
for elevated filtration (Table 3). Furthermore, than BMI (Wald statistic, 2.83; P ⫽ not signifi-
obese subjects with central fat distribution also cant), which is in agreement with our results.
showed an independent greater risk for microalbu-
minuria (Table 2). All analyses were performed DISCUSSION
in comparison to lean control subjects with per- This study shows that subjects with central fat
ipheral fat distribution. distribution, whether they are lean, overweight,
Because from these data, central fat distribu- or obese, have a greater risk for diminished
tion, not just overweight or obesity, seems impor- filtration. This effect even seems dose dependent
tant for diminished filtration, we were interested in a way that with increasing WHR, risk for
in whether there would be a linear relation be- diminished filtration increases. Obese subjects
tween WHR and diminished filtration indepen- with and without central fat distribution have a
dent of obesity. Therefore, we analyzed our data greater risk for elevated filtration. Furthermore,
in a logistic regression analysis with diminished obese subjects with central fat distribution also
filtration as the dependent variable and WHR in showed a greater risk for microalbuminuria. In-
quartiles. Figure 1 shows a linear relation be- terestingly, fat distribution and diminished filtra-
tween WHR and diminished filtration, even when tion appear to be linearly related to one another.
adjusted for such confounders as age, sex, sys- From our data, it seems that fat distribution may
tolic and diastolic blood pressure, antihyperten- be of more importance for renal impairment than
sive medication, serum cholesterol level, lipid- overweight or obesity by itself. By dividing our
lowering medication, glucose level, smoking, data into six groups, we lose power, especially in
and BMI (P ⬍ 0.01 for the overall model). the group of obese subjects with peripheral fat
Again, sex could act as a major confounder, distribution, making it harder to interpret.
but by analyzing our model with and without What could be the underlying mechanism that
correcting for sex, results remained similar. In makes subjects with central fat distribution more
addition, we analyzed our data split for both sex prone to impaired renal function? There is strong
groups, which gave similar results. We also ana- evidence that central body fat distribution is
lyzed our data in a linear regression model that related to insulin resistance.13,21 Furthermore,
showed WHR was more important for dimin- greater insulin and glucagon levels were found in
CENTRAL FAT DISTRIBUTION AND RENAL FUNCTION 739

Fig 1. Relative risk and


95% confidence intervals for
diminished filtration among
quartiles of WHR. P < 0.01
for the overall model, cor-
rected for such confounders
as age, sex, systolic and
diastolic blood pressure, an-
tihypertensive medication,
serum cholesterol level,
lipid-lowering medication,
plasma glucose level, smok-
ing, and BMI.

patients with central compared with peripheral addition, as listed in Tables 3 and 4, the model
fat distribution.22 It has been suggested that both remained similar with and without correcting for
insulin and glucagon, by means of their effects sex. Finally, we also analyzed our data in a split
on renal hemodynamics, may influence glomeru- database for men and women separately. There
lar filtration rate and albuminuria.23-25 Because was no difference in outcomes between analysis
we did not measure insulin levels, we are not of the split database and multivariate analysis.
entitled to make a statement about insulin resis- We therefore conclude that sex is no true con-
tance in our subjects with central fat distribution. founder in this analysis.
Conversely, clustering of such features of meta- One might argue that by using categorical
bolic syndrome as hypertension, high plasma variables, arbitrary cutoff points are being intro-
glucose level, hypercholesterolemia, and greater duced. However, we used generally accepted
UAE could be observed, particularly in our sub- cutoff levels according to the literature and clini-
jects with central fat distribution. cal practice. We therefore prefer to present the
Hyperfiltration is a frequent early finding in data in this way because it is easier to interpret
patients with diabetes26 and a possible predictor and also accessible for the more clinically inter-
of the subsequent development of microalbumin- ested reader. We also analyzed WHR linearly by
uria and ultimate diabetic nephropathy.27 Hyper- analyzing our data with WHR in quartiles; this
filtration also has been shown in obese sub- showed a linear relationship between WHR and
jects,3-5 suggesting that also in obesity, diminished filtration.
hyperfiltration might precede renal function loss. The question is whether these renal function
However, in lean subjects with central fat distri- abnormalities in relation to fat distribution are
bution, no such increased risk for hyperfiltration caused by the accompanying hypertension and
could be detected. diabetes so frequently present in obese subjects.
An important question is whether sex effects Subjects with central fat distribution had higher
on obesity and fat distribution could influence blood pressure and glucose levels. No clinical
our results. Because we did not find a significant studies to date have tried to establish an indepen-
univariate relation between renal function and dent relation between overweight and renal abnor-
sex, we doubted whether sex would yield a major malities. Our study gives us the opportunity to
influence on renal function in our analysis. In investigate the independent relation between cen-
740 PINTO-SIETSMA ET AL

tral fat distribution and renal function abnormali- ties, not just whether a subject is overweight or
ties because it is a large population-based study obese.
and therefore suitable for correcting for such
confounders as hypertension and diabetes. We ACKNOWLEDGMENT
chose to correct for diabetes by selecting a non- The authors thank research nurses Roelie Dijkstra-
diabetic population. Oppenhuizen and Erika Konneman-van Zalk for assistance
at the outpatient clinic; Winie G.A. de Jonge-Hovenkamp
Conversely, because diabetes is a continuum for secretarial assistance; Dr Enno van der Veur (General
and there exists a state of prediabetes, often Practitioners Laboratory, Groningen) and Jacko J. Duker
associated with central obesity, we also corrected (Nephrology Laboratory, University Hospital, Groningen)
for plasma glucose level. Furthermore, we cor- for laboratory assistance; and Drs Frank P.G. Lambert and
rected not only for blood pressure level, but also Robert J. Bieringa (Trial Coordination Centre, Groningen)
for data handling.
for antihypertensive treatment. Interestingly, our
study showed a positive relation between fat APPENDIX
distribution and renal function that was indepen-
dent of hypertension and diabetes. The PREVEND Study Group
We corrected creatinine clearance for body In addition to the authors, PREVEND investi-
surface area because the impact of body compo- gators are: Andries J. Smit, MD, PhD (Depart-
sition on renal function is known. It has been ment of Internal Medicine); Harry J.G.M. Crijns,
suggested that correcting for body surface area MD, PhD, Ad J. van Boven, MD, PhD (Depart-
might give inappropriately low values of renal ment of Cardiology); Wiek H. Van Gilst, PhD,
plasma flow in obese subjects.15 Therefore, part Gilles F.H. Diercks, MD (Department of Clinical
of the obese group might be in the group with Pharmacology); Lolkje T.W. de Jong-van den
diminished filtration, thereby influencing results. Berg, PhD, Maarten Postma, PhD, Taco B.M.
We do not believe this is the case because obese Monster, MSc (Social Pharmacy and Pharmaco-
subjects tend to have a greater filtration rate than epidemiology); Gerard J. te Meerman, PhD (De-
normal, and by correcting their renal function for partment of Medical Genetics), all from Univer-
body surface area, their renal function will return sity Hospital Groningen, The Netherlands; and
to normal. Furthermore, the group with dimin- Jan Broer, MD, PhD (Municipal Health Depart-
ished filtration had a mean creatinine clearance ment), Groningen, The Netherlands.
of approximately 50 mL/min/1.73 m2, which
shows that their renal function really is compro- REFERENCES
mised. 1. World Health OrganizationPhysical Status: The Use
As described by Turner and Reilly,28 we also and Interpretation of Anthropometry. Report of a WHO
Technical Expert Committee. WHO Technical Report Series
analyzed our data by introducing weight and 854. Geneva, Switzerland, World Health Organization, 1995
height as dependent variables in the model. This 2. Stamler J: Epidemic obesity in the United States. Arch
gave similar results. Also, by changing our model Intern Med 153:1040-1144, 1993
by omitting or reentering one of the other vari- 3. Stockholm KH, Brœchner-Mortensen J, Hoilund-
ables, results remained similar. Carlsen PF: Increased glomerular filtration rate and adreno-
cortical function in obese women. Int J Obes 4:57-63, 1980
Caution should be applied when interpreting 4. Reisin E, Messerli FG, Ventura HO, Frohlich ED:
our data because this is only a cross-sectional Renal hemodynamic studies in obesity hypertension. J Hy-
study. Cross-sectional studies render the opportu- pertens 5:397-400, 1987
nity to investigate more risky hypotheses without 5. Hall JE: Mechanisms of abnormal renal sodium han-
the efforts and costs of follow-up studies. At dling in obesity hypertension. Am J Hypertens 10:49S-55S,
1997 (suppl, part 2)
present, these subjects are followed up for 6. Ribstein J, du Cailar G, Mimran A: Combined renal
changes in renal function. Such data will be effects of overweight and hypertension. Hypertension 26:610-
needed to confirm our present observation. 615, 1995
In conclusion, subjects with central fat distri- 7. Basedevant A, Cassuto D, Gibault T, Raison J, Guy-
bution are at risk for renal function impairment Grand B: Microalbuminuria and body fat distribution in
obese subjects. Int J Obes Relat Metab Disord 18:806-811,
independent of blood pressure and plasma glu- 1994
cose levels. This might indicate that fat distribu- 8. O’Donnell MP, Kasiske BL, Cleary MP, Keane WF:
tion is important for renal function abnormali- Effects of genetic obesity on renal structure and function in
CENTRAL FAT DISTRIBUTION AND RENAL FUNCTION 741

the Zucker rat. II: Micropuncture studies. J Lab Clin Med 18. DuBois DF, DuBois EF: A formula to estimate the
106:605-610, 1985 approximate surface area if height and weight are known.
9. Kasiske BL, Cleary MP, O’Donnell MP, Keane WF: Arch Intern Med 17:863-871, 1916
Effects of genetic obesity on renal structure and function in 19. Bray GA: Overweight is risking fate. Definition,
the Zucker rat. J Lab Clin Med 106:598-604, 1985 classification, prevalence, and risks. Ann N Y Acad Sci
10. Schmitz PG, O’Donnell MP, Kasiske BL, Katz SA, 499:14-28, 1987
Keane WF: Renal injury in obese Zucker rats: Glomerular 20. Vora JP, Dolben J, Dean JD, et al: Renal haemodynam-
hemodynamics alterations and the effect of enalapril. Am J ics in newly presenting non-insulin-dependent diabetes mel-
Physiol 263:F496-F502, 1992 litus. Kidney Int 41:829-835, 1992
11. Kasiske BL, Crosson JT: Renal disease in patients 21. Krotkiewski M, Björntrop P, Sjöström L, Smith U:
with massive obesity. Arch Intern Med 146:1105-1109, 1986 Impact of obesity on metabolism in men and women: Impor-
12. Praga M, Hernandez E, Herrero JC, et al: Influence of tance of regional adipose tissue distribution. J Clin Invest
obesity on the appearance of proteinuria and renal insuffi- 72:1150-1162, 1983
ciency after unilateral nephrectomy. Kidney Int 58:2111- 22. Solerte SB, Fioravanti M, Severgnini S, et al: Hyper-
2118, 2000 insulinemia and glucagon serum concentrations influence
13. Haffner SM, Stern MP, Hazuda HP, Pugh J, Patterson renal hemodynamics and urinary protein loss in normoten-
JK: Do upper-body and centralized adiposity measure differ- sive patients with central obesity. Int J Obes Relat Metab
ent aspects of regional body-fat distribution? Relationship to Disord 24:S122-S123, 2000 (suppl 2)
non-insulin-dependent diabetes mellitus, lipids, and lipopro- 23. Ahloulay M, Dechaux M, Laborde K, Bankir L:
teins. Diabetes 36:43-51, 1987 Influence of glucagon on GFR and on urea and electrolyte
14. Seidell JC, Bakx JC, De-Boer E, Deurenberg P, excretion: Direct and indirect effects. Am J Physiol 269:F225-
Hautvast JG: Fat distribution of overweight persons in F235, 1995
relation to morbidity and subjective health. Int J Obes 24. Nestler JE, Barlascini CO, Tetrault GA: Increased
9:363-374, 1985 transcapillary escape rate of albumin in non-diabetic men in
15. Scaglione R, Ganguzza A, Corrao S, et al: Central response to hyperinsulinemia. Diabetes 39:1212-1217, 1990
obesity and hypertension: Pathophysiologic role of renal 25. Kubo M, Kiyohara Y, Kato I, et al: Effect of hyperin-
hemodynamics and function. Int J Obes Relat Metab Disord sulinemia on renal function in a general Japanese popula-
19:403-409, 1995 tion: The Hisayama study. Kidney Int 55:2450-2406, 1999
16. Pinto-Sietsma SJ, Janssen WMT, Hillege HL, Navis 26. Rudberg S, Persson B, Dahlquist G: Increased glomer-
G, de Zeeuw D, de Jong PE: Urinary albumin excretion is ular filtration rate as a predictor of diabetic nephropa-
associated with renal functional abnormalities in a non- thy—An 8-year prospective study. Kidney Int 41:822-828,
diabetic population. J Am Soc Nephrol 11:1882-1888, 2000 1992
17. Karter AJ, Mayer-Davis EJ, Selby JV, et al: Insulin 27. Mogensen CE: Microalbuminuria as a predictor of
sensitivity and abdominal obesity in African-American, His- clinical diabetic nephropathy. Kidney Int 31:673-789, 1987
panic and non-Hispanic white men and women. The Insulin 28. Turner ST, Reilly SL: Fallacy of indexing renal and
Resistance and Atherosclerosis study. Diabetes 45:1547- systemic hemodynamic measurements for body surface area.
1555, 1996 Am J Physiol 268:R978-R988, 1995

You might also like