Beruflich Dokumente
Kultur Dokumente
& 2004 Nature Publishing Group All rights reserved 0307-0565/04 $25.00
www.nature.com/ijo
PAPER
Obesity is the major determinant of elevated
C-reactive protein in subjects with the metabolic
syndrome
D Aronson1*, P Bartha2, O Zinder3, A Kerner1, W Markiewicz1, O Avizohar4,5, GJ Brook4,5 and
Y Levy2
1
Department of Cardiology; 2Department of Internal Medicine D; 3Department of Laboratory Medicine; 4Center for
Preventive Medicine, Rambam Medical Center, Haifa, Israel; and 5Rappaport Faculty of Medicine, Rambam Medical Center,
Haifa, Israel
OBJECTIVE: To investigate the relationship between C-reactive protein (CRP) and various characteristics of the metabolic
syndrome.
DESIGN: Population-based cross-sectional study.
SUBJECTS: A total of 1929 subjects undergoing a medical examination in a preventive medicine clinic (age, 50710 y; 63%
males).
RESULTS: The proportion of subjects with CRP levels above the cut point generally used to indicate an obvious source of
infection or inflammation (410 mg/l) was 3, 7, and 15% in subjects who were normal weight, overweight, and obese,
respectively. Subjects with obesity had markedly higher CRP level compared to patients without obesity regardless of whether
they had the metabolic syndrome. However, there was no significant difference in CRP levels between nonobese subjects
without the metabolic syndrome and subjects in whom the diagnosis of the metabolic syndrome was based on criteria other
than obesity (adjusted geometric mean CRP 1.75 vs 2.08 mg/l, P ¼ 0.79). Similarly, CRP levels did not differ among obese
subjects with and without the metabolic syndrome (adjusted geometric mean CRP 3.22 vs 3.49 mg/l, P ¼ 0.99). There was a
linear increase in CRP levels with an increase in the number of metabolic disorders (Ptrend o0.0001), which was substantially
diminished after controlling for body mass index (BMI) (Ptrend ¼ 0.1). Stepwise multivariate linear regression analysis identified
BMI, triglyceride levels, HDL cholesterol levels (inversely), and fasting glucose as independently related to CRP levels. However,
BMI accounted for 15% of the variability in CRP levels, whereas triglycerides, HDL cholesterol and fasting glucose levels
accounted for only B1% of the variability in CRP levels.
CONCLUSION: Obesity is the major factor associated with elevated CRP in individuals with the metabolic syndrome. CRP levels
in the range suggesting a source of infection or inflammation (410 mg/l) are more common among obese subjects than in
nonobese subjects.
International Journal of Obesity (2004) 28, 674–679. doi:10.1038/sj.ijo.0802609
Published online 2 March 2004
Figure 2 Adjusted geometric means CRP and 95% confidence intervals for
subjects without both the metabolic syndrome and obesity (MS, Ob),
subjects with the metabolic syndrome based on components other than
obesity (MS þ , Ob), subjects with obesity but without the metabolic
Figure 1 (a) Box and whisker plots of CRP level according to BMI category. syndrome (MS, Ob þ ) and subjects with the diagnosis of the metabolic
Normal BMI o25 kg/m2; overweight BMI 25.0–29.9 kg/m2; class I obese BMI syndrome when one of the components is obesity (MS þ , Ob þ ). CRP levels
30.0–34.9 kg/m2; class II obese BMI 35.0–39.9 kg/m2; and class III obese BMI were adjusted for age, sex, WBC count, serum albumin, physical activity,
Z40.0 kg/m2. The line within the box denotes the median and the box spans smoking status, and HRT using ANCOVA. P-values are based on the Bonferroni
the interquartile range (25–75th percentiles). Whiskers extend from the 10– adjustment for multiple comparisons. *Po0.0001 compared with MS, Ob;
w
90th percentiles. (b) Percentages of patients with CRP levels Z10 mg/l P ¼ 0.005 compared with MS þ , Ob; z Po0.0001 compared with MS þ ,
according to BMI category. Trend was evaluated by w2 test. Ob;
Characteristic N (%) Absent Present P-value* Variable R Regression coefficient (s.e.) 95% CI
Obesity 349 (18) 1.5 (0.8–3.0) 3.4 (1.8–5.5) o0.0001 Coefficient (s.e.)
Hypertension 323 (17) 1.7 (0.8–3.4) 2.3 (1.2–4.5) o0.0001 BMI 0.39* 0.10 (0.01) 0.09, 0.11
Hypertriglyceridemia 632 (33) 1.5 (0.8–3.2) 2.2 (1.3–2.2) o0.0001 Hypertension 0.12* 0.34 (0.06) 0.21, 0.46
Low HDL 202 (10) 1.7 (0.8–3.4) 2.7 (1.4–4.9) o0.0001 Triglycerides (mmol/l) 0.18* 0.19 (0.02) 0.14, 0.24
Glucose intolerance 219 (11) 1.7 (0.8–3.5) 2.3 (1.3–4.7) o0.0001 HDL cholesterol (mmol/l) 0.12* 0.36 (0.07) 0.50, 0.23
Fasting glucose (mmol/l) 0.16* 0.13 (0.02) 0.09, 0.17
Data are median (interquartile range); *Mann-Whitney Rank Sum test HRT 0.08** 0.53 (0.15) 0.23, 0.82
Smoking habit 0.08* 0.19 (0.05) 0.09, 0.30
Physical activity 0.08** 0.12 (0.04) 0.18, 0.05
WBC count 0.31* 0.20 (0.01) 0.17, 0.22
Albumin 0.22* 1.1 (0.11) 1.29, 0.87
Discussion
In this sample of middle-aged subjects, components of the
metabolic syndrome were strongly related to plasma con-
centrations of CRP, independent of other factors known to
influence CRP levels, such as smoking, HRT, and physical
activity. Levels of CRP increased continuously with increas-
ing number of metabolic disorders. However, BMI was the
dominant contributor to CRP levels among subjects with the
metabolic syndrome. In addition, CRP levels consistent with
active inflammation were common in obese subjects.
Previous studies have shown that components of the
metabolic syndrome cluster with elevation of acute phase
Figure 3 Adjusted geometric means CRP and 95% confidence intervals
(error bars) according to the number of metabolic disorders including (1)
reactants.11–15,23 Our results confirm recent observations
obesity (BMI Z30 kg/m2), (2) hypertension (blood pressure Z135/ that in middle-aged subjects, CRP levels are strongly
85 mmHg), (3) triglyceride Z150 mg/dl, (4) low HDL cholesterol r40 mg/ associated with various characteristics of the metabolic
dl for men and r50 mg/dL for women, and (5) fasting glucose Z110 mg/dl. syndrome. However, the association between measures of
(a) Geometric mean CRP adjusted for age, sex, WBC count, serum albumin,
physical activity, smoking status, and HRT using ANCOVA. (b) After further
metabolic risk and CRP was mainly related to overall obesity.
adjustment for BMI. First, subjects with the metabolic syndrome had higher CRP
levels than subjects without the metabolic syndrome only if
they were obese. Subjects with obesity had markedly higher
metabolic syndrome, as well as with hormone replacement CRP level compared to patients without obesity regardless of
therapy (HRT), smoking, and leisure time physical activity whether they had the metabolic syndrome, and nonobese
(Table 2). The strongest correlation (r ¼ 0.39) was observed patients had comparable CRP levels regardless of whether
between CRP and BMI. After adjustment for potential they had the metabolic syndrome. Second, the effect of
confounders in a stepwise multivariate linear regression increasing number of metabolic disorders on CRP levels
described in previous studies12 was markedly attenuated between measures of insulin resistance and CRP.12,28
after accounting for BMI. Third, multivariate regression By contrast, other studies failed to identify a relationship
analysis indicated that among the components of the between serum markers of inflammation and measures
metabolic syndrome, BMI contributes most of the variance of insulin resistance after controlling for the effect of
in CRP levels. obesity.29–31
Recent guideline suggest that if CRP level of 410 mg/l is Inflammation is a modifiable risk factor amenable to
identified, there should be a search initiated for an obvious improvement by drugs4 and lifestyle modifications. Weight
source of infection or inflammation, and that the result of loss has been associated with a concomitant decrease in CRP
the test should then be discarded.22 Although only 7.6% of and cytokine levels.32,33 In the context of the present study,
subjects had CRP values greater than the generally accepted it is important to emphasize that weight loss is probably the
cutoff for inflammation, our results suggest that these values most useful intervention to ameliorate the proinflammatory
are most frequently associated with overweight and espe- state associated with the metabolic syndrome.
cially with obesity. Therefore, CRP results should always be
interpreted in the context of the subject’s BMI, as very high
levels may be expected in severely obese subjects. Study limitations
One of the explanations for increased CRP concentrations The prevalence of the metabolic syndrome has varied
in subjects with the metabolic syndrome relates to the between different studies, mostly because of the lack of
potential effect of body fat. Both tumor necrosis factor-a and accepted criteria for the definition of the syndrome.
interleukin-6 are expressed in, and released by adipose Recently, both the World Health Organization20 and the
tissue.14,16 There is a close relationship between circulating ATP–III19 proposed unifying definitions for the syndrome.
CRP and cytokine concentrations14 and between CRP We defined components of the metabolic syndrome based
concentrations and anthropometric measures of obesity.14,15 on a combination of the ATP-III and WHO criteria. This
Since the synthesis of CRP by the liver is predominantly definition differs from the full ATP-III criteria in that obesity
regulated by inteleukin-6,17 it is believed that interleukin-6 is defined based on BMI rather than waist circumference.19,20
originating from adipose tissue contributes to the raised CRP However, Ridker et al34 have recently found almost identical
concentrations in obese insulin-resistance subjects. Tumor results when using different definitions of the metabolic
necrosis factor-a does not induce CRP directly, but can syndrome for predicting cardiovascular disease. Further-
potentiate the induction of CRP by interleukin-6.24 more, Lemieux et al15 have recently performed a study with
This hypothesis is consistent with the strong relationship particular attention to the relation between CRP and
between BMI and CRP levels.11–15,25 However, the existence measures of abdominal fat accumulation including waist
of an association between CRP and insulin resistance girth, waist-to-hip ratio, and visceral adipose tissue accumu-
independent of body mass is controversial. As obesity lation assessed by computed tomography. Measures of
induces insulin resistance and inflammatory cytokines are abdominal fat accumulation were positively associated with
secreted by adipose tissue, it is difficult to exclude the CRP in a univariate analysis. However, multivariate regres-
possibility that the relationship between markers of systemic sion analysis reveled that after including body fat mass in the
inflammation and measures of insulin resistance reflects model, no other variable contributed to the variance in CRP
residual confounding of the stronger relationship between levels. In a large study by Festa et al,12 waist circumference
obesity and insulin resistance. For example, inflammatory was positively associated with CRP in a univariate analysis
cytokines originating from adipose tissue induce insulin but not in a multivariate analysis. Therefore, we believe that
resistance by direct action on the insulin-signaling cas- inclusion of waist circumference instead of BMI in our
cade26,27 and stimulate CRP production by the liver.17,24 analysis would not likely have produced materially different
Some studies describe an independent inverse association results.