Inflammatory Cytokines and the Risk to Develop Type 2

Diabetes
Results of the Prospective Population-Based European
Prospective Investigation into Cancer and Nutrition
(EPIC)-Potsdam Study
Joachim Spranger,1,2 Anja Kroke,3 Matthias Mohlig,1,2 Kurt Hoffmann,3 Manuela M. Bergmann,3
Michael Ristow,1,2 Heiner Boeing,3 and Andreas F.H. Pfeiffer1,2

A subclinical inflammatory reaction has been shown to vated levels of IL-6 had an independently increased risk
precede the onset of type 2 (noninsulin-dependent) to develop type 2 diabetes (3.3, 1.7 6.8), whereas indi-
diabetes. We therefore examined prospectively the ef- viduals with increased concentrations of IL-6 but unde-
fects of the central inflammatory cytokines interleukin tectable levels of IL-1 had no significantly increased
(IL)-1, IL-6, and tumor necrosis factor- (TNF-) on risk, both compared with the low-level reference group.
the development of type 2 diabetes. We designed a These results were confirmed in an analysis including
nested case-control study within the prospective popu- only individuals with HbA1c <5.8% at baseline. Our data
lation-based European Prospective Investigation into suggest that the pattern of circulating inflammatory
Cancer and Nutrition (EPIC)-Potsdam study including cytokines modifies the risk for type 2 diabetes. In
27,548 individuals. Case subjects were defined to be particular, a combined elevation of IL-1 and IL-6,
those who were free of type 2 diabetes at baseline and rather than the isolated elevation of IL-6 alone, inde-
subsequently developed type 2 diabetes during a 2.3- pendently increases the risk of type 2 diabetes. These
year follow-up period. A total of 192 cases of incident data strongly support the hypothesis that a subclinical
type 2 diabetes were identified and matched with 384 inflammatory reaction has a role in the pathogenesis of
non disease-developing control subjects. IL-6 and type 2 diabetes. Diabetes 52:812 817, 2003
TNF- levels were found to be elevated in participants
with incident type 2 diabetes, whereas IL-1 plasma
levels did not differ between the groups. Analysis of
single cytokines revealed IL-6 as an independent pre-

L
ow physical activity and hyperalimentation are
dictor of type 2 diabetes after adjustment for age, sex,
BMI, waist-to-hip ratio (WHR), sports, smoking status,
lifestyle factors associated with an increased risk
educational attainment, alcohol consumption, and of type 2 diabetes (1). Despite more than 100
HbA1c (4th vs. the 1st quartile: odds ratio [OR] 2.6, 95% million patients affected worldwide and a dra-
CI 1.25.5). The association between TNF- and future matic socioeconomic burden due to vascular complica-
type 2 diabetes was no longer significant after adjust- tions, the etiology of type 2 diabetes is not yet completely
ment for BMI or WHR. Interestingly, combined analysis understood.
of the cytokines revealed a significant interaction be- It has been hypothesized that type 2 diabetes is a
tween IL-1 and IL-6. In the fully adjusted model, manifestation of an ongoing acute-phase response that is
participants with detectable levels of IL-1 and ele- primarily characterized by alterations of the so-called
acute-phase proteins, such as C-reactive protein (CRP)
From the 1Department of Endocrinology, Diabetes and Nutrition, Benjamin (2,3). Cross-sectional and prospective studies demon-
Franklin Medical Center, Free University Berlin, Berlin, Germany; the 2De- strated increased concentrations of markers of the acute-
partment of Clinical Nutrition, German Institute of Human Nutrition Potsdam-
Rehbrucke, Potsdam-Rehbrucke, Germany; and the 3Department of
phase response (including CRP, serum amyloid-A, and
Epidemiology, German Institute of Human Nutrition Potsdam-Rehbrucke, sialic acid) in patients with type 2 diabetes (29). In one of
Potsdam-Rehbrucke, Germany. these studies, elevated levels of IL-6, which is known to be
Address correspondence and reprint requests to Joachim Spranger, MD,
Department of Clinical Nutrition, German Institute of Human Nutrition, a main stimulator of the production of most acute-phase
Potsdam-Rehbrucke, Arthur-Scheunert-Allee 114-116, 14558 Bergholz- proteins (10,11), were shown to increase the risk of
Rehbrucke, Germany. E-mail: spranger@mail.dife.de or joachim.spranger@ diabetes (9). However, in addition to IL-6, other cytokines,
medizin.fu-berlin.de.
Received for publication 26 July 2002 and accepted in revised form 19 such as interleukin (IL)-1 or tumor necrosis factor-
November 2002. (TNF-), are central mediators of inflammatory reactions.
J.S., A.K., and M.M. contributed equally to this article.
A.K. is currently affiliated with the Research Institute of Child Nutrition
It is well known that cytokines operate as a network in
Dortmund, Dortmund, Germany. stimulating the production of acute-phase proteins. For
CRP, C-reactive protein; CV, coefficient of variation; ELISA, enzyme-linked example, the effects of IL-6 on CRP synthesis largely
immunosorbent assay; EPIC, European Prospective Investigation into Cancer
and Nutrition; IL, interleukin; LOQ, limit of quantification; OR, odds ratio; depend on an interaction with IL-1 (11). The acute-phase
TNF-, tumor necrosis factor-; WHR, waist-to-hip ratio. response in various artificial inflammatory models requires
812 DIABETES, VOL. 52, MARCH 2003
 SPRANGER AND ASSOCIATES

both IL-6 and IL-1, as demonstrated in the respective assays (ELISAs; R&D Systems, Minneapolis, MN). CRP was determined by
high-sensitivity ELISA (Immun Diagnostik, Bensheim, Germany). All assay
knockout mouse models (12,13). These data strongly procedures were performed as described by the manufacturer. Blood samples
suggest that inflammatory reactions do not depend on were analyzed in random order to exclude systemic bias due to interassay
single mediators, but rather that the pattern of various variation. Control specimens were analyzed simultaneously on each plate for
cytokines is crucially important for the perpetuation of an every marker. The intra-assay coefficient of variation (CV) ranged between 6.4
acute-phase response. and 10.2% for IL-1, 3.8 and 11.1% for IL-6, 8.7 and 14.8% for TNF-, and 4.9 and
6.2% for CRP. The interassay CV was 10.3% for IL-1, 9.9% for IL-6, 16.1% for
Until now, there has been neither prospective evidence TNF-, and 13.2% for CRP. The limits of detection were 0.1 pg/ml for IL-1,
concerning the individual regulation of the inflammatory 0.094 pg/ml for IL-6, 0.12 pg/ml for TNF-, and 0.124 ng/ml for CRP. HbA1c was
cytokines IL-1 or TNF- nor prospective evidence about determined by enzyme immunoassay (DAKO Diagnostika, Hamburg, Germa-
the combined role of IL-1, IL-6, and TNF- preceding type ny). The intra-assay CV ranged from 2 to 3%, and the interassay CV was 1.9%.
Values of subjects without diabetes have been shown to range from 4.8 to 6.9%
2 diabetes, although the combined effects of these cyto- in this assay, according to the information of the manufacturer. Diabetes-
kines are likely to be more important than the circulating associated antibodies GAD65 and IA-2 were analyzed by radioimmunoassay
levels of the single cytokines. We therefore designed a (Medipan Diagnostica, Selchow, Germany), which was performed as de-
nested case-control study within the prospective popula- scribed by the manufacturer.
tion-based European Prospective Investigation into Can- Statistical analyses. For all analyses, we used SAS software release 8.0 (SAS
Institute, Cary, NC). In a first step, measurement values of the laboratory
cer and Nutrition (EPIC)-Potsdam cohort of 27,548 parameters (IL-1, TNF-, and CRP) below the limit of quantification (LOQ)
individuals to further evaluate the role of CRP, IL-1, IL-6, were set at 0.7 the respective LOQ (18). Means, standard deviations, and
and TNF- in the development of type 2 diabetes. proportions of baseline characteristics of case and control subjects were
calculated. The means SD are reported. Significance was considered at
two-tailed 0.05.
RESEARCH DESIGN AND METHODS The nonparametric Wilcoxons rank-sum test was used to test for differ-
Study population. The EPIC-Potsdam study, as part of the multicenter ences in continuous variables between case and control subjects, and a 2 test
population-based cohort study EPIC, aims to explore the relation between with 1 degree of freedom (Mantel-Haenszel test) was used to describe
dietary and lifestyle factors and the development of complex diseases. Details differences in proportions between case and control subjects. Spearman
of the recruitment procedure of EPIC-Potsdam have been published (14). In correlation coefficients were used to test the association between anthropo-
brief, 27,548 subjects (women aged 35 65 years and men aged 40 65 years) metric, lifestyle variables, and cytokines.
were recruited from the general population. Baseline examinations, including Associations were initially investigated separately for IL-1, IL-6, and
anthropometric measurements, blood sampling, a self-administered food TNF-. IL-6 and TNF- were divided into quartiles, and CRP was dichoto-
frequency questionnaire, and a personal interview on lifestyle habits and mized. Because of the high numbers of undetectable values, IL-1 was
medical history were conducted between 1994 and 1998. Follow-up question- dichotomized. Because of case and control exclusions due to missing values,
naires are sent to the study participants every 23 years to obtain information primarily unconditional logistic regression analysis was used to estimate odds
on, among other things, current medication and newly developed diseases, ratios (ORs) and corresponding 95% CIs. As previously demonstrated, the
including diabetes. estimation of the OR approximates the relative risk given an infrequent
Anthropometry and lifestyle characteristics. Anthropometric measure- disease occurrence (19). We therefore calculated the OR to estimate the
ments (body height and weight, waist and hip circumference) were performed relative risk to develop type 2 diabetes that is associated with increasing
by trained personnel, with the participants wearing only light underwear and categories of the investigated cytokine. Because the design of the study also
without shoes (15). BMI was calculated as body weight (in kilograms) divided allows conditional logistic analysis, study results were recalculated using
by body height (in meters) squared. Waist-to-hip ratio (WHR) was calculated conditional logistic regression analysis. In the article, the results of the
as waist divided by hip circumference. unconditional regression and major results of the conditional regression
Information on lifestyle characteristics were obtained from self-adminis- analysis are shown. Estimates of relative risk were first obtained from
tered questionnaires and a personal, computer-guided interview by trained age-adjusted (continuous) and sex-adjusted (categorical: female and male)
and quality-monitored personnel (16). Sporting activities (hours per week) analyses; this analysis was followed by further adjustment for BMI (continu-
were calculated as the average of hours of sports per week during the summer ous). The subsequent model also considers sex-normalized WHR (continu-
and the winter season. ous), sporting activities in hours/week (continuous), smoking status
End points and disease status. We studied 27,548 participants of the (categorical: current smoker and nonsmoker), alcohol consumption in grams/
population-based EPIC-Potsdam study. The follow-up procedure was success- day (continuous), and educational attainment (categorical: basic training,
ful in receiving completely filled-in questionnaires from 96% of all cohort technical school, and university). In the last model, we additionally added
participants attending the baseline examination (17). HbA1c (continuous) to the fully adjusted model to reduce potential bias
Case subjects were those who were free of type 2 diabetes at baseline and caused by undetected cases of prevalent diabetes.
developed type 2 diabetes during the first 2- to 3-year follow-up, depending on To investigate the role of cytokine patterns, formal interaction terms
the time of recruitment. Potential cases of incident diabetes were identified including the cytokines in question were analyzed. Product terms were built
from self-reports on incident disease, current medications, and/or current from dichotomized subgroups. IL-1 was therefore dichotomized as previ-
dietary treatment for diabetes (n 399). For each potentially incident subject, ously described (setting individuals with undetectable levels as 0 and those
a special questionnaire was sent to the primary care physician. The study with detectable levels as 1), whereas IL-6 and TNF- were dichotomized by
subject was considered as a case subject only if the diagnosis of newly using the 75th percentile as the cutoff point (setting individuals who were
developed diabetes was confirmed by this physician. A total of 201 cases of 75th percentile for both IL-6 and TNF- as 0 and those who were 75th
incident diabetes were identified by this verification process until 1 November quartile as 1). Quartile cut points were estimated from the combined group of
2001. At that time, another 10 potential incident cases were pending confir- control and case subjects. In addition, ORs and 95% CIs were estimated for
mation by the primary care physician. These subjects were not further each combination, including two of these cytokines with the low-level
considered. The biochemical analysis regarding diabetes-associated antibod- category (75th percentile for IL-6 and TNF- and nondetectable values of
ies GAD65 and IA-2 revealed that nine of the case subjects should be IL-1, respectively) as the reference category.
considered as case subjects with type 1 diabetes, leaving 192 medically To reduce bias by individuals with undetected prevalent type 2 diabetes at
confirmed cases of type 2 diabetes. Each of the 192 individuals with confirmed baseline, we repeated the analysis, including only those case and control
type 2 diabetes was matched with two control subjects by age (1 year) and subjects with HbA1c 5.8%.
sex (n 384). For statistical analysis, those individuals with missing values in
one of the variables being used in the statistical models were not considered
(n 4 for case subjects, n 7 for control subjects), thus leaving 188 case RESULTS
subjects and 377 control subjects for the final analysis. Clinical parameters. Baseline characteristics of the par-
Laboratory procedures. Peripheral venous citrate-blood samples were ticipants are shown in Table 1. As expected, case subjects
taken at enrollment into the study. The blood samples were centrifuged at
1,000g for 10 min at 4C. Plasma was then removed and stored in aliquots in
had higher BMI and WHR, and they exercised less (Table
freezers at 80C until assays of the markers of interest were performed. 1). Elevated glucose levels, such as impaired fasting
IL-1, IL-6, and TNF- were measured by enzyme-linked immunosorbent glucose or impaired glucose tolerance, are well known to
DIABETES, VOL. 52, MARCH 2003 813
 INFLAMMATORY CYTOKINES AND TYPE 2 DIABETES

TABLE 1
Baseline characteristics of the participants
Characteristic Case subjects Control subjects P
n 188 377
Age (years) 56 7 56 7 Matching variable
BMI (kg/m2) 30.7 4.8 26.7 3.5 0.0001
WHR 0.95 0.09 0.89 0.09 0.0001
Sports (h/week) 0.5 1.1 0.9 1.6 0.0100
Alcohol consumption (g/day) 18.5 28.2 16.1 16.4 0.4678
IL-1 (pg/ml) 0.57 0.93 0.47 0.79 0.1959
IL-6 (pg/ml) 2.45 1.80 1.67 1.59 0.0001
TNF- (pg/ml) 2.04 1.51 1.79 1.28 0.0094
CRP (g/ml) 4.14 5.1 2.45 4.38 0.0001
Men 111 (59) 222 (59) Matching variable
Current smokers 36 (19) 80 (21) 0.5661
Less than high school education 83 (44) 142 (38) 0.1383
Prevalence of hypertension 148 (79) 195 (52) 0.0001
Prevalence of hyperlipoproteinemia 81 (43) 120 (32) 0.0085
Data are means SD or n (%).

be a risk factor for future type 2 diabetes (20), which is results were observed in analyses restricted to case sub-
reflected in our study group by significantly elevated jects with HbA1c 5.8% only. Within this subgroup analy-
HbA1c levels in case subjects (6.39 2.16%) compared sis, the risk of individuals within the 4th quartile of IL-6
with control subjects (4.73 0.74%). Of the participants, was also substantially increased (3.1, 1.37.4 [uncondition-
2.8% had undetectable levels of CRP (5% for TNF- and al]; 3.7, 1.16 11.9 [conditional]) in the fully adjusted
62% for IL-1, respectively). IL-6 levels were detectable in model.
all participants. As demonstrated in previous studies (8,9), Elevated levels of TNF- and IL-1 alone are not
elevated levels of CRP were found to be associated with an independently associated with future type 2 diabe-
increased risk of type 2 diabetes in the fully adjusted tes. Mean concentrations of TNF- were higher in case
model (OR 1.9, 95% CI 1.23.2). With respect to clinical subjects (2.04 1.51 pg/ml) compared with control sub-
parameters and CRP, subgroup analysis (HbA1c 5.8%) jects (1.79 1.28 pg/ml, P 0.01). The risk of type 2
yielded results comparable to the analysis that included all diabetes increased with increasing quartiles of TNF-
participants. For example, the risk of individuals with (Table 3). However, after adjustment for BMI or WHR, this
elevated CRP levels to develop type 2 diabetes was 2.1 association was no longer significant. Correlation analysis
(95% CI 1.23.7) in the fully adjusted model in this sub- showed a mild correlation between TNF- and BMI (r
group. The mean HbA1c was 4.5 0.5% for control and 0.17, P 0.001) or WHR (r 0.15, P 0.001). Again,
4.9 0.5% for case subjects (P 0.001) within the analysis of participants with HbA1c 5.8% yielded a similar
restricted subcohort. Correlations of cytokines, CRP, BMI, picture as analysis of all participants. For example, the
WHR, and HbA1c are demonstrated in Table 2. ORs in the fully adjusted model including HbA1c were 1.0,
Effects of single cytokines on diabetes risk 1.8 (95% CI 0.8 3.8), 0.9 (0.4 1.9), and 1.3 (0.6 3.1) for the
IL-6 independently predicts the risk of type 2 diabe- increasing quartiles. Using conditional regression analysis
tes. Mean baseline levels of IL-6 were higher among case also yielded no significant results regarding TNF- depen-
subjects compared with control subjects (P 0.0001). dent relative risks after adjustment for BMI and/or WHR.
Elevated levels of IL-6 were associated with an increased We found that 41.0% of case subjects and 36.6% of
risk of type 2 diabetes (risk estimates of individuals control subjects had detectable levels of IL-1, which was
according to IL-6 quartiles are demonstrated in Table 3). not a statistically significant difference. According to these
After adjustment for all covariables (BMI, WHR, sports, data, the relative risk of developing type 2 diabetes was
age, sex, smoking status, educational attainment, alcohol not associated with circulating levels of IL-1, indepen-
consumption, and HbA1c), IL-6 was found to be an inde- dent of the model or type of analysis applied (Table 3).
pendent predictor of type 2 diabetes (OR 2.57, 95% CI Combined effects of cytokines modify the risk of
1.24 5.47) This result was confirmed in conditional regres- future type 2 diabetes. Participants with a combined
sion analysis (2.6, 1.25.9). Comparable and significant elevation of IL-6 levels and detectable levels of IL-1 were

TABLE 2
Correlations of cytokines, CRP, BMI, WHR, and HbA1c
HbA1c BMI WHR IL-6 TNF-
HbA1c
BMI 0.224 (P 0.001)
WHR 0.239 (P 0.001) 0.425 (P 0.001)
IL-6 0.099 (P 0.019) 0.302 (P 0.001) 0.264 (P 0.001)
TNF- 0.031 (NS) 0.17 (P 0.001) 0.15 (P 0.001) 0.177 (P 0.001)
CRP 0.1 (P 0.017) 0.242 (P 0.001) 0.11 (P 0.01) 0.51 (P 0.001) 0.135 (P 0.001)

814 DIABETES, VOL. 52, MARCH 2003

TABLE 3
Risk estimates of individuals according to circulating cytokine levels
Age- and sex- BMI-adjusted Adjusted for all Adjusted for all risk
Median adjusted analysis analysis risk factors* factors and HbA1c
IL-1
Category 1 0.09 (0.090.09) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference)
Category 2 0.84 (0.135.52) 1.20 (0.841.72) 1.18 (0.791.75) 1.19 (0.781.79) 1.20 (0.741.96)
CRP
Category 1 0.62 (0.091.49) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference)
Category 2 3.57 (1.5039.7) 3.5 (2.45.1) 2.2 (1.53.3) 1.9 (1.22.9) 1.9 (1.23.2)
IL-6
Category 1 0.66 (0.110.90) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference)
Category 2 1.19 (0.901.45) 1.77 (0.973.29) 1.36 (0.722.6) 1.22 (0.642.38) 1.14 (0.542.45)
Category 3 1.80 (1.452.24) 3.88 (2.217.03) 2.17 (1.184.07) 1.92 (1.023.67) 1.72 (0.833.60)
Category 4 3.30 (2.2410.8) 7.34 (4.1813.32) 3.26 (1.766.19) 2.75 (1.445.34) 2.57 (1.245.47)
TNF-
Category 1 0.79 (0.351.11) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference)
Category 2 1.34 (1.111.56) 1.85 (1.113.11) 1.66 (0.942.95) 1.62 (0.902.94) 1.39 (0.953.85)
Category 3 1.82 (1.562.26) 1.40 (0.832.39) 1.05 (0.581.89) 1.04 (0.571.92) 1.06 (0.512.21)
Category 4 2.94 (2.2615.15) 2.30 (1.373.90) 1.70 (0.963.04) 1.52 (0.842.78) 1.57 (0.783.18)
Data are OR (95% CI), unless otherwise indicated. Median (category limits), Median (interquartile range). Individuals were divided into
quartiles according to their baseline plasma concentrations of IL-6 or TNF-. They were dichotomized according to their levels of IL-1 and
CRP; *adjusted for age, sex, BMI, WHR, sporting activities, smoking status, alcohol consumption, and educational attainment.

found to have an increased risk of future type 2 diabetes compared with the low-level reference group. These ef-
(OR 3.3, 95% CI 1.7 6.8 [unconditional]; 3.479, 1.078 fects of a combined elevation of IL-6 and TNF- (OR 3.2,
11.222 [conditional]) compared with the low-level refer- 95% CI 1.6 6.4) or IL-1 and TNF- (2.3, 1.1 4.9) was still
ence group in the fully adjusted model. In contrast, significant in the analysis restricted to participants with
individuals with elevated levels of IL-6 but nondetectable HbA1c 5.8%. However, as described for TNF- alone, this
levels of IL-1 had no significantly increased risk to effect did not remain significant in the fully adjusted
develop type 2 diabetes (1.2, 0.6 2.5 [unconditional]; 1.5, model. Results were again confirmed using conditional
0.5 4.4 [conditional]) compared with the low-level refer- regression analysis and by calculation of formal interac-
ence group in the fully adjusted model. These results were tion terms (Table 4).
confirmed by the inclusion of a formal interaction term
between detectable IL-1 and elevated IL-6; this term was DISCUSSION
3.3 (95% CI 1.19.7) and was significant (Table 4). Similar
We evaluated the effects of various inflammatory cyto-
results were obtained including only individuals with
kines on the risk of type 2 diabetes. Participants with a
HbA1c 5.8%; here, the interaction term was 2.3 (1.15.1).
In crude analysis, individuals with a combined elevation combined elevation of IL-6 and IL-1 had a roughly three-
of IL-6 and TNF- or with a combined elevation of TNF- fold increased risk of developing type 2 diabetes compared
and IL-1 had a substantially increased risk compared with the low-level reference group. In contrast, partici-
with individuals with elevated levels of IL-6 alone or pants with elevated levels of IL-6 alone (and undetectable
levels of IL-1) had no substantial increase of their diabe-
TABLE 4 tes risk. In this regard, IL-1 appears to have a permissive
Interaction between IL-1, IL-6, and TNF- on diabetes risk role in the IL-6 mediated acute-phase response preceding
the onset of type 2 diabetes. Elevated levels of TNF- were
OR (95% CI) associated with an increased diabetes risk in the crude
Interaction TNF-/IL-1 analysis. However, this TNF- dependent effect was no
Reference (TNF- low/IL-1 undetectable) 1 longer significant after adjustment for BMI or WHR. Fur-
TNF- (high) 0.89 (0.451.73) thermore, we found no significant effects of TNF- on
IL-1 (undetectable) 0.95 (0.541.67) IL-1 or IL-6 dependent risk estimates.
Interaction term TNF*IL-1 (high/detectable) 2.51 (0.847.57) To the best of our knowledge, this is the first study to
Interaction TNF-/IL-6 describe the combined effects of the three inflammatory
Reference (TNF- low/IL-6 low) 1 cytokines IL-1, IL-6, and TNF- on the risk to develop
TNF- (high) 1.32 (0.682.52)
IL-6 (high) 2.15 (1.124.11)
type 2 diabetes. However, some limitations of this study
Interaction term TNF*IL-6 (both high) 0.7 (0.232.16) need to be considered. It is well known that there is a
Interaction IL-1/IL-6 relatively high proportion of individuals with undiagnosed
Reference (IL-6 low/IL-1 undetectable) 1 type 2 diabetes among the general population (21,22). We
IL-1 (undetectable) 0.8 (0.431.44) therefore adjusted for blood glucose control by including
IL-6 (high) 1.14 (0.552.32) HbA1c into the fully adjusted model. To reduce the remain-
Interaction term IL-1*IL-6 (detectable/high) 3.31 (1.149.87) ing potential bias of prevalent diabetes in case or control
Data shown were calculated after adjustment for age, sex, BMI, subjects at baseline, we also performed separate analyses
WHR, sporting activities, HbA1c smoking status, alcohol consump- among participants (case and control subjects) with HbA1c
tion, and educational attainment. 5.8%. A total of 94 individuals in our cohort provided
DIABETES, VOL. 52, MARCH 2003 815
 INFLAMMATORY CYTOKINES AND TYPE 2 DIABETES

fasting blood specimens. Within this subgroup, all case cally increased the expression of the acute-phase proteins,
individuals with HbA1c 5.8% had a baseline fasting glu- compared with the effect of each cytokine alone (11).
cose 7.0 mmol/l, which is sufficient to exclude diabetes Another potential molecular mechanism how inflamma-
in epidemiological studies, according to American Diabe- tion may be involved in the pathogenesis of type 2 diabetes
tes Association criteria. Thus, the number of individuals has been elucidated in recent elegant studies showing that
with prevalent diabetes at baseline (in case and control sensitizing of insulin signaling by salicylates is induced via
subjects) appears to be small in the analyses restricted to inhibition of the activity of IB kinase (2325). IL-1 is
individuals with HbA1c 5.8%. However, physician-diag- well known to activate the IB kinase and might thereby
nosed cases (as in this study) are likely to represent more induce insulin resistance.
progressed stages of diabetes compared with those diag- In conclusion, our data support the concept that sub-
nosed, for example, by an oral glucose tolerance test. clinical activation of the immune system is involved in the
Given that cytokine changes at baseline may be associated pathogenesis of type 2 diabetes. We demonstrated that a
with the stage of diabetes, one might expect weaker specific pattern of cytokines was associated with an
associations if earlier stages of diabetes are investigated. increased risk of type 2 diabetes, rather than isolated
In addition, some of the control subjects may have devel- elevation of the respective cytokines.
oped diabetes by the end of follow-up but were not
diagnosed by a physician. ACKNOWLEDGMENTS
The robustness of results was additionally confirmed by
inclusion of further covariates (preexisting hypertension This project was supported by grants from Gottfried-
Wilhelm-Leibnitz-Gesellschaft. Further grants to the authors
and hyperlipidemia) into the model, although there is no
were from the German Diabetes Association (104/03/2001
clear functional evidence that these factors influence the
[to J.S. and M.M.] and 103/03/2001 [to M.R.]), Fritz-Thyssen-
development of type 2 diabetes. It is important to note that
Stiftung (10.01.2.102), Deutsche Forschungsgemeinschaft (RI
all of these additional analyses confirmed the findings
1076/1-1), the Eli-Lilly International Foundation (to J.S.
described. Although BMI and WHR are more common and A.F.H.P.), the European Union (SOC 95 201408
clinical measures of obesity and central obesity, respec- OSF02), and the Deutsche Krebshilfe (70-2488-HAI).
tively, they may not fully account for the metabolic We thank K. Sprengel and S. Richter for laboratory
consequences of obesity and residual confounding even assistance and U. Fiddicke and W. Bernigau for assistance
though adjustment for these parameters may exist. We with the study data. The HbA1c analyses were conducted
analyzed various components of the metabolic syndrome, at the Department of Clinical Biochemistry, University of
and cytokine effects were found to be independent of Greifswald, under the responsibility of Dr. Hans-Joachim
hypertension or hyperlipidemia. Although previous studies Rose. We thank C.A. Barth for critical discussion of the
demonstrated that inflammatory markers are associated project.
with future type 2 diabetes, even after adjustment for
fasting insulin (6,8), it remains to be elucidated whether
this holds true for the cytokine changes described here. REFERENCES
Another potential bias may result from different precision 1. Hu FB, Manson JE, Stampfer MJ, Colditz G, Liu S, Solomon CG, Willett WC:
of cytokine measurement. When two or more cytokines Diet, lifestyle, and the risk of type 2 diabetes mellitus in women. N Engl
J Med 345:790 797, 2001
are entered into the statistical models, the relative 2. Pickup JC, Mattock MB, Chusney GD, Burt D: NIDDM as a disease of the
strength of association between cytokines and disease can innate immune system: association of acute-phase reactants and interleu-
be expected to be highest for the cytokine with the least kin-6 with metabolic syndrome X. Diabetologia 40:1286 1292, 1997
measurement error. In our study, inter- and intra-assay 3. Pickup JC, Crook MA: Is type II diabetes mellitus a disease of the innate
immune system? Diabetologia 41:12411248, 1998
coefficients of variation were comparable in measure- 4. Schmidt MI, Duncan BB, Sharrett AR, Lindberg G, Savage PJ, Offenbacher
ments of the three cytokines. However, additional sources S, Azambuja MI, Tracy RP, Heiss G: Markers of inflammation and predic-
of measurement variability (related to venipuncture, blood tion of diabetes mellitus in adults (Atherosclerosis Risk in Communities
processing, or effects of long-term storage) may have study): a cohort study. Lancet 353:1649 1652, 1999
5. Vozarova B, Weyer C, Lindsay RS, Pratley RE, Bogardus C, Tataranni PA:
differed for the three cytokines. Thus, we cannot entirely
High white blood cell count is associated with a worsening of insulin
exclude that some of the above-mentioned limitations may sensitivity and predicts the development of type 2 diabetes. Diabetes
have influenced our results. 51:455 461, 2002
Our data suggest that the pattern of inflammatory cyto- 6. Festa A, DAgostino R Jr, Tracy RP, Haffner SM: Elevated levels of
kines is important in the pathogenesis of type 2 diabetes. acute-phase proteins and plasminogen activator inhibitor-1 predict the
development of type 2 diabetes: the insulin resistance atherosclerosis
These findings are in line with the fact that inflammatory study. Diabetes 51:11311137, 2002
reactions depend on a cluster of cytokines rather than on 7. Freeman DJ, Norrie J, Caslake MJ, Gaw A, Ford I, Lowe GD, OReilly DS,
single cytokines only. Patterns of cytokine production Packard CJ, Sattar N: C-reactive protein is an independent predictor of risk
differ with different inflammatory conditions, and cyto- for the development of diabetes in the West of Scotland Coronary
Prevention Study. Diabetes 51:1596 600, 2002
kines are components of a large complex signaling net-
8. Barzilay JI, Abraham L, Heckbert SR, Cushman M, Kuller LH, Resnick HE,
work (1113). Several mechanisms, such as considerable Tracy RP: The relation of markers of inflammation to the development of
influence of cytokines on lipid metabolism, may be impor- glucose disorders in the elderly: the Cardiovascular Health Study. Diabetes
tant for the effects of combined elevations of different 50:2384 2389, 2001
cytokines. For example, both IL-6 and IL-1 act on the 9. Pradhan AD, Manson JE, Rifai N, Buring JE, Ridker PM: C-reactive protein,
interleukin 6, and risk of developing type 2 diabetes mellitus. JAMA
liver to produce the characteristic dyslipidemia of the 286:327334, 2001
metabolic syndrome, with increased VLDL and decreased 10. Gauldie J, Richards C, Harnish D, Lansdorp P, Baumann H: Interferon beta
HDL (3). Combined elevation of IL-6 and IL-1 dramati- 2/B-cell stimulatory factor type 2 shares identity with monocyte-derived

816 DIABETES, VOL. 52, MARCH 2003

hepatocyte-stimulating factor and regulates the major acute phase protein 19. Cornfield J: A method of estimating comparative rates from clinical data:
response in liver cells. Proc Natl Acad Sci U S A 84:72517255, 1987 applications to cancer of the lung, breast and cervix. J Natl Cancer Inst
11. Gabay C, Kushner I: Acute-phase proteins and other systemic responses to 11:1269 1275, 1951
inflammation. N Engl J Med 340:448 454, 1999 20. de Vegt F, Dekker JM, Jager A, Hienkens E, Kostense PJ, Stehouwer CD,
12. Fattori E, Cappelletti M, Costa P, Sellitto C, Cantoni L, Carelli M, Faggioni Nijpels G, Bouter LM, Heine RJ: Relation of impaired fasting and postload
R, Fantuzzi G, Ghezzi P, Poli V: Defective inflammatory response in glucose with incident type 2 diabetes in a Dutch population: the Hoorn
interleukin 6-deficient mice. J Exp Med 180:12431250, 1994 Study. JAMA 285:2109 2113, 2001
13. Zheng H, Fletcher D, Kozak W, Jiang M, Hofmann KJ, Conn CA, Soszynski 21. Warram JH, Kopczynski J, Janka HU, Krolewski AS: Epidemiology of
D, Grabiec C, Trumbauer ME, Shaw A, et al.: Resistance to fever induction
non-insulin-dependent diabetes mellitus and its macrovascular complica-
and impaired acute-phase response in interleukin-1 beta-deficient mice.
tions: a basis for the development of cost-effective programs. Endocrinol
Immunity 3:9 19, 1995
Metab Clin North Am 26:165188, 1997
14. Boeing H, Korfmann A, Bergmann MM: Recruitment procedures of EPIC-
Germany: European Investigation into Cancer and Nutrition. Ann Nutr 22. Harris MI, Flegal KM, Cowie CC, Eberhardt MS, Goldstein DE, Little RR,
Metab 43:205215, 1999 Wiedmeyer HM, Byrd-Holt DD: Prevalence of diabetes, impaired fasting
15. Klipstein-Grobusch K, Georg T, Boeing H: Interviewer variability in anthro- glucose, and impaired glucose tolerance in U.S. adults: the Third National
pometric measurements and estimates of body composition. Int J Epide- Health and Nutrition Examination Survey 1988 1994. Diabetes Care
miol 26:S174 S180, 1997 21:518 524, 1998
16. Kroke A, Bergmann MM, Lotze G, Jeckel A, Klipstein-Grobusch K, Boeing 23. Kim JK, Kim YJ, Fillmore JJ, Chen Y, Moore I, Lee J, Yuan M, Li ZW, Karin
H: Measures of quality control in the German component of the EPIC M, Perret P, Shoelson SE, Shulman GI: Prevention of fat-induced insulin
study: European Prospective Investigation into Cancer and Nutrition. Ann resistance by salicylate. J Clin Invest 108:437 446, 2001
Nutr Metab 43:216 224, 1999 24. Yuan M, Konstantopoulos N, Lee J, Hansen L, Li ZW, Karin M, Shoelson SE:
17. Bergmann MM, Bussas U, Boeing H: Follow-up procedures in EPIC- Reversal of obesity- and diet-induced insulin resistance with salicylates or
Germany data quality aspects: European Prospective Investigation Into targeted disruption of Ikkbeta. Science 293:16731677, 2001
Cancer and Nutrition. Ann Nutr Metab 43:225234, 1999 25. Hundal RS, Petersen KF, Mayerson AB, Randhawa PS, Inzucchi S, Shoel-
18. Hallez S, Derouane A: Novelle methode de traitement de series de donnees son SE, Shulman GI: Mechanism by which high-dose aspirin improves
tronquees dans letude de la pollutions atmospherique. Sci Total Environ glucose metabolism in type 2 diabetes. J Clin Invest 109:13211326,
22:115123, 1982 2002

DIABETES, VOL. 52, MARCH 2003 817