Profile of the Immune andInflammatory Response inIndividuals With Prediabetes andType 2 DiabetesDiabetes Care 2015;38:1356–1364 | DOI: 10.2337/dc14-3008

OBJECTIVE

The inflammatory and immune systems are altered in type 2 diabetes. Here, theaim was to profile the immune and inflammatory response in subjects with pre-diabetes and diabetes in a large population-representative sample.

RESEARCH DESIGN AND METHODS

In total, 15,010 individuals were analyzed from the population-based GutenbergHealth Study. Glucose status was classified according to HbA1c concentration andhistory of diagnosis. All samples were analyzed for white blood cells (WBCs),granulocytes, lymphocytes, monocytes, platelets, C-reactive protein (CRP), albu-min, fibrinogen, and hematocrit. Interleukin-18 (IL-18), IL-1 receptor antagonist(IL-1RA), and neopterin concentrations were determined in a subcohort.

RESULTS

In total, 7,584 men and 7,426 women were analyzed (range 35–74 years), with1,425 and 1,299 having prediabetes and diabetes, respectively. Biomarkersshowed varying dynamics from normoglycemic via subjects with prediabetes tosubjects with diabetes: 1) gradual increase (WBCs, granulocytes, monocytes, IL-1RA, IL-18, and fibrinogen), 2) increase with subclinical disease only (lymphocytesand CRP), 3) increase from prediabetes to diabetes only (neopterin), and 4) novariationwith glucose status (hematocrit). The strongest relative differenceswerefound for CRP, IL-1RA, and fibrinogen concentrations. Several inflammatory andimmune markers were associated with the glucose status independent from car-diovascular risk factors and comorbidities, varied with disease severity and thepresence of disease-specific complications in the diabetes subcohort.

CONCLUSIONS

The inflammatory and immune biomarker profile varies with the developmentand progression of type 2 diabetes.Markers of inflammation and immunity enabledifferentiation between the early preclinical and clinical phases of the disease,disease complications, and progression.

Type 2 diabetes and its disease-associated complications represent an importantand increasing public health burden worldwide (1). Obesity, which leads to meta-bolic and adipocyte stress, is the most important predisposing factor for type 2diabetes (2). Obesity-associated insulin resistance, activation of the innate immunesystem, and chronic increased production of cytokines and adipokines are an

1Center for Thrombosis and Hemostasis, Univer-sity Medical Center Mainz, Mainz, Germany2Department of Medicine 2, University MedicalCenter Mainz, Mainz, Germany3Clinic for General and Interventional Cardiology,University Heart Centre Hamburg, Hamburg,Germany4German Center for Cardiovascular Research(DZHK), Partner Site Hamburg, Lubeck, Kiel,Germany5Preventive Cardiology and PreventiveMedicine,Department of Medicine 2, University MedicalCenter Mainz, Mainz, Germany6Department of Psychosomatic Medicine andPsychotherapy, University Medical CenterMainz, Mainz, Germany7Institute for Clinical Chemistry and LaboratoryMedicine, University Medical Center Mainz,Mainz, Germany8German Center for Cardiovascular Research(DZHK), Partner Site Rhine-Main, Mainz,Germany

Corresponding author: Philipp S. Wild, philipp.wild@unimedizin-mainz.de.

Received 18 December 2014 and accepted 14March 2015.

This article contains Supplementary Data onlineat http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc14-3008/-/DC1.

© 2015 by the American Diabetes Association.Readers may use this article as long as the workis properly cited, the use is educational and notfor profit, and the work is not altered.

Vera Grossmann,1 Volker H. Schmitt,2

Tanja Zeller,3,4 Marina Panova-Noeva,1

Andreas Schulz,5 Dagmar Laubert-Reh,5

Claus Juenger,5 Renate B. Schnabel,3,4

Tobias G.J. Abt,2 Rafael Laskowski,2

Jorg Wiltink,6 Eberhard Schulz,2

Stefan Blankenberg,3,4 Karl J. Lackner,7

Thomas Munzel,2,8 and Philipp S. Wild1,5,8

1356 Diabetes Care Volume 38, July 2015

PATH

OPHYS

IOLO

GY/COMPLICATIONS

important link between obesity andtype 2 diabetes (3,4). Novel data furthersuggest that chronic adipose tissue in-flammation and b-cell stress cause anactivation of the adaptive immune sys-tem as well, which may also participatein the progression of the inflammatoryresponse (5). Autoimmune and inflamma-tory mechanisms during hyperglycemia-induced glucotoxicity could favor anincreased expression of several b-cellantigens, thus increasing b-cell apoptosisthrough autoantibodies (6). Such isletautoantibodies are presented in;10% ofsubjects with diabetes, which have distinctclinical and phenotypical characteristicscompared with patients with type 2diabetes without signs of autoimmunity(5,7).With respect to inflammatory pro-

cesses, prospective studies have dem-onstrated that chronic low-gradeinflammation precedes the onset of di-abetes (8–11). For example, elevatedconcentrations of biomarkers such asC-reactive protein (CRP), white bloodcell (WBC) count, interleukin-1b (IL-1b), IL-1 receptor antagonist (IL-1RA),IL-6, IL-8, IL-18, monocyte chemoattrac-tant protein-1 (MCP-1), interferon-g–inducible protein-10 (IP-10), hapto-globin, and fibrinogen point to a chronic,often subclinical degree of inflammationand are increased already years beforetype 2 diabetes onset (11–18). The de-gree of subclinical inflammation was re-ported to be similar between subjectswith impaired fasting glucose and indi-viduals with diabetes (9,19).Subjects with diabetes are at increased

risk for atherosclerotic cardiovascular dis-ease (CVD), peripheral arterial disease(PAD), and cerebrovascular disease (20)because of the coexistence of multiplecardiovascular risk factors (CVRFs), in-cluding hypertension and dyslipidemia,that are more prevalent in individualshaving type 2 diabetes compared withsubjects with normal homeostasis (20–22). Further, the onset of diabetes leadsto severalmicro- andmacrovascular com-plications. Long-term complications in-clude retinopathy with a potential lossof vision, neuropathy with the danger offoot amputation, and nephropathy lead-ing to renal function decline and dialysis(20). In this context, IL-6 and CRP are themost discussed candidates to predictCVRFs and disease-associated complica-tions in individuals with diabetes (23,24).

Taken together, inflammatory pro-cesses are a causal link for the develop-ment of vascular complications during(pre)diabetes (8,25). However, thereis a lack of data on the overall profileof the immune and inflammatory re-sponse in the disease, particularly insubjects with prediabetes.

With the current study, we character-ized the immune and inflammatoryresponse of subjects with prediabetesand diabetes in a large population-representative sample. This representa-tive setting offers the possibility toinvestigate the complete spectrum ofthe disease from normal metabolichomeostasis over the subclinical to themanifest disease and helping to differen-tiate between biological effects versus asimple progression of the disease.

RESEARCH DESIGN AND METHODS

The Gutenberg Health Study (GHS) is apopulation-representative, prospective,observational, single-center cohortstudy in western mid-Germany andincludes a total sample size of 15,010individuals. The interdisciplinary study,including a detailed biobanking, focusespredominantly on CVD, but also meta-bolic, ophthalmological, and mental dis-ease. The sample was drawn randomlyfrom the local governmental registry of-fices. The sampling procedure was strat-ified 1:1 for sex and residence (urbanand rural) and in equal strata for de-cades of age (range 35–74 years). Thestudy design has been published else-where in more detail (26).

The baseline collection started inApril 2007 and was completed in April2012. All individuals were invited for a5-h baseline examination at the studycenter, where the clinical data assess-ment was performed. The study was de-signed according to the tenets of therevised Helsinki protocol. All participantsgave written informed consent to labora-tory analyses, clinical examinations, sam-pling of biomaterial, and the use of datarecords for research purposes.

Study Population and DiabetesDiagnosisAll individuals with unspecified or typesother than type 2 diabetes were ex-cluded from analysis (see Supplemen-tary Fig. 1). HbA1c concentration wasdetermined in all remaining subjects(n = 14,876) using a standardized high-

performance liquid chromatography as-say. Study participants were categorizedaccording to the International ExpertCommittee, i.e., ,6.0% (,42 mmol/molhemoglobin) as with normal glucose ho-meostasis, 6.0–6.4% (42–46 mmol/molhemoglobin) as having increased riskfor diabetes or prediabetes, and $6.5%($48 mmol/mol hemoglobin) as havingdiabetes (27). Analyses were performedin a centralized laboratory setup. Infor-mation on duration and complications oftype 2 diabetes, concomitant diseases,and family history were collected dur-ing a comprehensive computer-assistedpersonal interview. All individualsbrought their drugs to the study centerfor recording current drug intake and,if available, medical records. CVRFswere assessed by clinical examinationand laboratory measurements. Individ-uals with diabetes were classifiedaccording to no treatment/dietary treat-ment, insulin-dependent diabetes, andnoninsulin-dependent diabetes.

Laboratory MeasurementsVenous blood sampling was performedin lying position, and the prior fastingperiod was documented. Biosampleswere stored at2808C immediately afterblood withdrawal. Plasma CRP concen-tration (n = 14,983) was measured by ahigh-sensitivity latex enhanced immu-noturbidimetric assay (Abbott Laborato-ries, Abbott Park, IL). The limit ofdetection was #0.1 mg/L for the ultra-sensitive calibrator and #0.2 mg/L forthe wide-range calibrator. Fibrinogen(n = 14,892) was determined by a de-rived method with a detection limit of40 mg/dL (Siemens, Munich, Germany).Albumin (Abbott Laboratories; n =14,982) and blood cell counts (WBCcount, n = 14,973; platelet count, n =14,959; neutrophil granulocyte count,n = 14,973; lymphocyte count, n =14,973; monocyte count, n = 14,973)were performed by routine laboratorymethods. WBCs were counted, whereasfor granulocytes, lymphocytes, andmonocytes, relative frequencies weregiven. Relative frequencies were con-verted to absolute numbers based onthe WBC number. Hematocrit (n =14,976) was calculated as follows: MCV *erythrocytes. IL-18 (Human IL-18 ELISAKit; MBL, Woburn, MA; detection limitof 128 pg/mL; n = 4,573), IL-1RA(Quantikine; R&D Systems, Wiesbaden,

care.diabetesjournals.org Grossmann and Associates 1357

Germany; detection limit of 31.2 pg/mL;n = 4,967), and neopterin (Thermo FisherScientific, Hennigsdorf, Germany; n =3,972) were determined by commer-cially available assays. HbA1c, WBCs,granulocytes, lymphocytes, monocytes,platelets, CRP, albumin, fibrinogen, andhematocrit were determined in freshmaterial, whereas IL-1RA, IL-18, and neo-pterin were assessed in frozen material.

Data Assessment, Data Management,and Statistical AnalysisAll clinical and laboratory examinationswere performed according to standardoperating procedures by specificallytrained and certifiedmedical technical as-sistants. The definitions of CVRFs andcomorbidities are presented in theSupplementary Data. All data of thepresent investigation underwent qualitycontrol by a central data managementunit. Data were reviewed for complete-ness by predefined algorithms and plau-sibility criteria. Data are presented asabsolute numbers, percentages, andmeans with SDs or medians with 25thand 75th percentiles as appropriate.Fisher exact or x2 tests were used totest differences in categorical variables.Statistical comparisons for continuousvariables were made by Mann-WhitneyU test as well as Student t test as appro-priate. All analyses were carried out sexspecifically. For the correlation analyses,Spearman rank correlation coefficienttest was used. To assess the effects ofinflammation on the presence of predia-betes, diabetes, and disease-associatedcomplications, multivariable logistic re-gression models were used: model 1was adjusted for age and sex and model2 was additionally adjusted for CVRFs(waist-to-height ratio [WHtR] as continu-ous variable and hypertension, smoking,dyslipidemia, and family history of myo-cardial infarction [MI] or stroke as di-chotomous variables), cardiovascularcomorbidities (coronary artery disease[CAD], MI, stroke, PAD, atrial fibrillation[AF], and chronic obstructive pulmonarydisease [COPD]), arthritis, autoimmunedisease, hemostatic disorders, and acuteinfection. Adjusted odds ratios (ORs) aregiven with 95% CI and P value. The OR isgiven per one interquartile range (IQR)change of the biomarkers. In this explor-ative analysis, the OR was used to de-scribe the strength of an association. Ingeneral, P values,0.05 were considered

as relevant associations. Statistical dataanalyses were performed using the soft-ware program R, version 3.0.2 (http://www.R-project.org).

RESULTS

Study SampleIn total, 7,584 (50.5%) men and 7,426(49.5%) women were enrolled, medianage was 55.0 years (range 35–74; IQR46–65 years). From this study sample,134 individuals were excluded with un-specified diabetes (0.4%; n = 55) or otherthan type 2 diabetes (0.5%; n = 79). Fromthe remaining subjects (n = 14,876), 81.0%(n = 12,152) of individuals had normalHbA1c concentrations. In contrast, 9.5%(n = 1,425) of individuals fulfilled the cri-teria of having prediabetes (HbA1c 6.0–6.4% [42–47 mmol/mol hemoglobin]),and 8.7% (n = 1,299) of subjects had di-abetes according to HbA1c concentration($6.5% [$48 mmol/mol hemoglobin]) orwere previously diagnosed by a physician(Supplementary Fig. 1). The prevalence ofprediabetes and diabetes in the sampleaccording to decades of age and sex isprovided in Supplementary Fig. 2.

With respect to disease treatment,6.2% (n = 81) of individuals affectedwith type 2 diabetes received no or die-tary treatment and 39.1% (n = 508)showed a noninsulin-dependent diabetesand 19.4% (n = 252) an insulin-dependentdiabetes; no information on treatmentwas available for 458 subjects. The spe-cific medicinal treatment is provided inSupplementary Table 1. The disease du-ration in the subgroup of individuals withdiabetes was as follows: #1 year, 11.7%(n = 105);.1 to#2 years, 8.7% (n = 78);.2 to #5 years, 25.4% (n = 228); .5 to#10 years, 28.6% (n = 257); and .10years, 25.6% (n = 230); no informationwas available in 401 individuals.

Prevalence of CVRFs andComorbidities According to GlucoseStatusSample characteristics according to theglucose status are presented in Table 1.Compared with individuals with nor-moglycemia and prediabetes, subjectshaving type 2 diabetes were morefrequently of male sex. With the excep-tion of smoking, the prevalence of

Table 1—Characteristics of individuals according to glucose status

Normoglycemia Prediabetes Diabetes

Size (n) 12,152 1,425 1,299

Sex (women) 50.4% (6,121) 51.6% (736) 38.3% (497)

Age (years) 53.4 6 11.0 61.1 6 9.0 63.0 6 8.3

Glucose (mg/dL) 90.0 (85.0, 96.0) 97.0 (91.0, 104.0) 112.6 (99.0, 132.0)

HbA1c (%) 5.40 (5.10, 5.60) 6.10 (6.00, 6.20) 6.70 (6.30, 7.20)

HbA1c (mmol/mol) 36 (32, 38) 43 (42, 44) 50 (45, 55)

AnthropometryBMI (kg/m2) 26.7 6 4.6 29.3 6 5.6 31.5 6 5.9Waist (cm) 92.6 6 13.0 99.8 6 14.1 106.7 6 14.1

Classical CVRFsHypertension 71.5% (8,687) 84.4% (1,202) 91.1% (1,184)Smoking 19.3% (2,342) 23.6% (336) 16.2% (209)Obesity 20.5% (2,494) 37.8% (538) 55.2% (716)Dyslipidemia 39.2% (4,765) 59.9% (854) 76.4% (991)Family history of MI/stroke 21.0% (2,552) 26.5% (377) 27.5% (357)

ComorbiditiesCAD 3.0% (355) 7.6% (106) 14.0% (173)MI 2.0% (238) 5.4% (77) 9.6% (123)Stroke 1.4% (168) 3.0% (42) 4.9% (63)PAD 2.6% (316) 5.1% (72) 8.4% (107)AF 2.3% (274) 4.0% (56) 5.9% (75)CHF 1.0% (121) 2.2% (31) 3.4% (44)COPD 4.5% (541) 6.7% (95) 7.7% (100)

Participants having type 1 diabetes, other types, or an unspecified type were excluded from thecohort (n = 134). With the exception of smoking (P = 0.4), the prevalence of the depictedparameters increased significantly from individuals with normoglycemia to prediabetes and todiabetes (P, 0.001). Data are presented as relative and absolute numbers (sex, hypertension,smoking, obesity, dyslipidemia, family history of MI/stroke, CAD, MI, stroke, PAD, AF, CHF,COPD), mean and standard deviation (age, BMI, waist), median and interquartile range (glucose,HbA1c).

1358 Immune and Inflammatory Response in Diabetes Diabetes Care Volume 38, July 2015

CVRFs increased from individuals withnormoglycemia to those with prediabetesand to those with diabetes, respectively.Subjects with prediabetes had the highestprevalence of smoking compared withsubjects with diabetes or normoglycemia.With respect to cardiovascular comor-

bidities, individuals with prediabetes anddiabetes showed a higher prevalence ofCAD, MI, stroke, PAD, AF, congestiveheart failure (CHF), and COPD. Again,the frequencies of comorbidities in-creased gradually from normoglycemicsubjects to those with prediabetes and,finally, subjects with diabetes. Overall,19.8% (273 of 1,379) of individuals withprediabetes had CVD, i.e., the individualshad at least one of the following: CAD,MI,stroke, PAD, AF, CHF, or COPD.Withmoredetail, 69.6% (n = 190) of subjects hadone, 22.3% (n = 61) had two, and 8.1%(n = 22) had three or more CVDs; nodata were available for 46 subjects. Sub-jects with diabetes were more often af-fected with CVD, i.e., 30.5% (378 of 963)of subjectswith diabetes reported at leastone CVD; with more detail, 62.7% (n =237) had one, 23.5% (n = 89) had two,and 13.8% (n = 52) had three or moreCVDs, with no data available in 61 sub-jects. With respect to the diabetes sub-sample, 14.5% (n = 189) of the subjectshad neuropathy, 14.5% (n = 189) hadmacroalbuminuria, 4.9% (n = 64) had pro-teinuria, 4.3% (n = 56) had retinopathy,0.8% (n = 11) had blindness, 0.7% (n = 9)had foot amputation, and 0.7% (n = 9)had a need for dialysis.

Distribution of Inflammatory andImmune BiomarkersConcentrations of inflammatory and im-mune biomarkers compared betweennormoglycemic subjects, individualswith prediabetes and diabetes, and theirdifferences are presented in Table 2. Bio-markers were distinguishable in fourdifferent subgroups:

1. Gradual increase from normoglyce-mic via subjects with prediabetes todiabetes (WBC, granulocyte, andmonocyte count and IL-1RA, IL-18,and fibrinogen)

2. Increase with subclinical disease andapproximately stable levels despiteprogression to diabetic disease (lym-phocytes and CRP)

3. Comparable levels between indi-viduals with normoglycemia and

Table

2—Diffe

rences

inco

nce

ntratio

nofce

llsandpro

teinsreflectin

ginflammatio

nandim

munesy

stem

betw

eennorm

oglyce

mia,prediabetes,

anddiabetes

P

Norm

oglycem

iaPred

iabetes

Diab

etesFortren

d(Norm

oglycem

iavs.

pred

iabetes)

(Norm

oglycem

iavs.

diab

etes)(Pred

iabetes

vs.diab

etes)

Size(n)

12,1521,425

1,299

WBCs(10

9/L)6.80

(5.76,8.12)7.30

(6.09,8.60)7.60

(6.36,8.90),0.0001

,0.0001

,0.0001

,0.0001

Gran

ulocytes

(109/L)

4.24(3.41,5.29)

4.53(3.59,5.55)

4.74(3.85,5.81)

,0.0001

,0.0001

,0.0001

,0.0001

Lymphocytes

(109/L)

1.78(1.45,2.19)

1.89(1.51,2.35)

1.89(1.49,2.35)

,0.0001

,0.0001

,0.0001

0.53

Monocytes

(109/L)

0.40(0.32,0.48)

0.42(0.34,0.53)

0.46(0.37,0.56)

,0.0001

,0.0001

,0.0001

,0.0001

Platelets(10

9/L)269

(230,312)270

(235,325)262

(218,308)0.55

0.0013,0.0001

,0.0001

IL-1RA(pg/m

L)310.7

(233,408)350.8

(271,485)404.9

(298,545),0.0001

,0.0001

,0.0001

,0.0001

IL-18(pg/m

L)222

(178,285)233

(181,298)260

(204,334),0.0001

0.05,0.0001

,0.0001

CRP(m

g/L)1.40

(0.50,2.90)2.30

(1.20,4.40)2.40

(1.30,5.10),0.0001

,0.0001

,0.0001

0.0078

Neo

pterin

(pmol/L)

5.40(4.70,6.30)

5.40(4.70,6.30)

5.90(5.10,7.37)

,0.0001

0.76,0.0001

,0.0001

Albumin

(g/L)42

(40,44)41

(39,43)41

(39,43),0.0001

,0.0001

,0.0001

0.34

Fibrin

ogen

(mg/d

L)315

(273,366)345

(298,410)359

(308,417),0.0001

,0.0001

,0.0001

0.00038

Hem

atocrit

(%)

41.9(39.7,44.2)

42.2(39.9,44.4)

41.9(39.6,44.3)

0.570.098

0.430.08

Participantshavin

gtyp

e1diab

etes,other

types,o

rnotspecifi

edwere

excluded

(n=134).D

ataare

presen

tedas

med

ian(Q1,Q

3).

care.diabetesjournals.org Grossmann and Associates 1359

prediabetes and strong increasefrom prediabetes to diabetes (neo-pterin)

4. No influence on levels by prediabetesand diabetes (hematocrit)

The relative change between indi-viduals with diabetes and prediabetesand subjects with normal HbA1c

concentration as well as sex-specificrelative changes of biomarker concen-trations are presented in Supple-mentary Fig. 3. The largest relativedifferences in the subsample with di-abetes were observed for CRP, IL-1RA,and fibrinogen, whereas the relative in-crease in CRP levels was the highest inboth males (;60%) and females(;90%). With respect to individualswith prediabetes, the strongest in-crease was seen for CRP as well, bothin men (;50%) and women (;70%).The other biomarkers displayed minorrelative changes when subjects withprediabetes were compared with sub-jects with diabetes. Correlation analy-ses between biomarkers and betweenbiomarkers and CVRFs indicate thatthere is no impact of diabetes and pre-diabetes on correlations (see Supple-mentary Data).

Association of Biomarkers WithDisease Status

The concentrations of inflammatory andimmune biomarkers depending on thepresence of CVD within the cohortwith prediabetes and the cohort withdiabetes are presented in Supplemen-tary Table 2. As expected, the majorityof biomarkers varied clearly between in-dividuals with prediabetes with (n = 273)and without CVD (n = 1,106) and individ-uals with diabetes with (n = 378) andwithout CVD (n = 860). Using diseaseduration and the treatment form as sur-rogate markers for disease severitywithin the diabetes subcohort, an asso-ciation between the prevalence of CVDand disease severity was observed withan increasing prevalence of CVD withlonger type 2 diabetes duration andmore intense treatment (Supplemen-tary Table 3). Similar to thesedata, severalinflammatory and immune biomarkersincreased with disease severity. With re-spect to intensity of treatment, WBCs,granulocytes, monocytes, CRP, IL-1RA,IL-18, neopterin, albumin, and fibrino-gen concentrations clearly increasedfrom subjects with no/dietary treat-ment via noninsulin-dependent diabe-tes to insulin-dependent diabetes.

Focusing on differences betweennoninsulin-dependent diabetes andinsulin-dependent diabetes, a substan-tial increase was present forWBCs, gran-ulocytes, monocytes, CRP, albumin,neopterin, fibrinogen, and hematocrit(Supplementary Table 4). Of note, theconcentration of granulocytes increasedwith a longer disease duration (Supple-mentary Table 5).

In multivariable logistic regressionmodels for the presence of diabetesand prediabetes, the concentrations ofalmost all biomarkers were strongly as-sociated under adjustment for sex andage (Fig. 1, model 1). Especially forWBCs, granulocytes, monocytes, IL-1RA,IL-18, and fibrinogen, the associationincreased from prediabetes to diabetes.No or minor differences were seen forlymphocytes, CRP, and albumin. Inter-estingly, neopterin showed an inverseassociation with prediabetes but, incontrast, a weakly positive associationwith diabetes. The multivariable logisticregression model was adjusted for tra-ditional CVRFs (WHtR, hypertension,smoking, dyslipidemia, and family his-tory of MI or stroke), cardiovascular co-morbidities and arthritis, hemostaticdisorders, autoimmune disease, and

Figure 1—Impact of inflammatory and immune biomarkers on the presence of prediabetes and diabetes. A logistic model with diabetes or prediabetesvs. normoglycemia as dependent variable was calculated for each biomarker. Model 1 was adjusted for sex and age. Model 2 was adjusted additionallyfor CVRFs, including WHtR, hypertension, smoking, dyslipidemia, and comorbidities including CAD, MI, stroke, PAD, AF, and COPD as well as arthritis,hemostatic disorder, autoimmune disease, and acute infection.

1360 Immune and Inflammatory Response in Diabetes Diabetes Care Volume 38, July 2015

acute infection. Here, elevated WBC,granulocyte, and monocyte count; CRP,IL-1RA, and fibrinogen concentrations;and hematocrit (with inverse associa-tion) were still independently associ-ated with the presence of type 2diabetes (Fig. 1, model 2). Plateletsand, with an inverse association, albu-min and neopterin showed strong asso-ciations only with prediabetes, but notwith diabetes, whereas granulocytes,CRP, IL-1RA, and hematocrit werestrongly associated with diabetes butshowed no associations with the pres-ence of prediabetes. A further compari-son between subjects with prediabetesand diabetes in models 1 and 2 is pre-sented in Supplementary Fig. 4.Of note, only marginal differences

were observed when comparing regres-sion models with adjustment for CVRFsonly to models adjusting for CVRFs andcomorbidities. This suggests CVRFs assignificant confounders, whereas co-morbidities do not affect the associationbetween inflammatory and immunemarkers and the diabetic disease (datanot presented). Of note, the socioeconomic

status did not affect the association be-tween diabetes and biomarkers.

Biomarkers of Inflammation andImmunity and Disease-SpecificComplicationsFinally, the role of inflammatory and im-mune biomarkers for the prevalence ofdisease-associated complications wasevaluated in the subgroup with type 2diabetes. Regressionmodels for the fourmost frequent complications, i.e., neu-ropathy (14.5%; n = 189), macroalbumin-uria (14.5%; n = 189), nephropathy(4.9%, n = 64), and retinopathy (4.3%,n = 59), were analyzed (Fig. 2). Afteradjustment for sex and age, CVRF, co-morbidities, acute infection, arthritis,hemostatic disorders, and autoimmunedisease, higher concentrations of WBCsin general and granulocytes in particularwere associated with the prevalence ofretinopathy and elevated lymphocytesand CRP concentrations with the prev-alence of neuropathy. Further, higherhematocrit values were associatedwith a lower prevalence of proteinuria.The sample size for foot amputation,

dialysis, and blindness was too low forfurther analysis.

CONCLUSIONS

The current study represents one of thelargest single-center population-basedcohort studies (n = 15,010) profiling theinflammatory and immune response insubjects with prediabetes, diabetes, andits disease-specific complications. Besidesthe worsening of the cardiovascular riskprofile, a clear alteration of the inflamma-tory and immune response was observeddepending on the glucose status. Inter-estingly, biomarkers showed varying pro-files with disease progression (i.e.,normoglycemia vs. prediabetes vs. diabe-tes): 1) an increase with subclinical dis-ease (i.e., prediabetes) and stable levelsdespite progression to overt type 2 dia-betes (lymphocytes and CRP), 2) compa-rable levels for normoglycemia andprediabetes with an increase in type 2diabetes for neopterin, 3) a continuousincrease with worsening of endogenousglycemic control (WBCs, granulocytes,monocytes, IL-1RA, IL-18, and fibrinogen),and 4) biomarkers where concentrations

Figure 2—Inflammatory and immune biomarkers and presence of disease-associated complications. Logistic regression analysis with the presence ofcomplications as dependent variable. As in Fig. 1, model 1 was adjusted for sex and age and model 2 additionally for CVRFs, including WHtR,hypertension, smoking, dyslipidemia, and comorbidities including CAD,MI, stroke, PAD, AF, and COPD aswell as acute infection, arthritis, hemostaticdisorder, and autoimmune disease. Boldface text indicates ORs with a P value,0.05 and are highlighted in red (for OR,1) or blue (for OR.1).

care.diabetesjournals.org Grossmann and Associates 1361

were not influenced by the presence ofprediabetes or diabetes (albumin andhematocrit).The current study underlines that

type 2 diabetes comes along with an on-going cytokine-mediated acute phaseresponse initiated by the innate and,as novel data suggest, the adaptive im-mune system (5,8,10,11). Obesity andb-cell stress have been demonstratedto be involved in increased expressionof proinflammatory cytokines in theliver and adipose tissue (3,28). These cy-tokines, including TNF-a, IL-6, and IL-1b,may promote insulin resistance in thetissues, where they are produced andfurther affect more distant sites via cir-culation such as vessel walls, skeletaland cardiac muscle, kidney, and circulat-ing leukocytes, respectively (8). Further,new data suggest that cytokines such asIL-1R, IL-1b, IL-6, or TNF-a contributeto an islet cell autoimmunity in type 2diabetes through activation of T cells,B cells, and macrophages, increased ex-pression of b-cell antigens, and subse-quent b-cell apoptosis (5).In the present project, two proteins

from the IL-1 superfamily have been in-vestigated. IL-1RA is an anti-inflammatorycytokine that inhibits the effect of theproinflammatory IL-1b, an importantcytokine in the context of type 2 diabetesdue to its relation to insulin resistanceand b-cell dysfunction (25). In contrast,IL-18 is a potent proinflammatorycytokine that plays a central role inthe inflammatory cascade. In this study,IL-1RA was more strongly increased insubjects with prediabetes versus sub-jects with normoglycemia than IL-18.Yet, the concentration of both cytokinessubstantially increased further from theprecursor to themanifest disease. In sev-eral studies, elevated IL-1RA concentra-tions were reported to discriminatebetween individuals who finally devel-oped type 2 diabetes compared withlower values in individuals who re-mained diabetes free (9,14,25,29,30).Therefore, not only proinflammatorybut also anti-inflammatory markers,even though not sufficient to preventfurther diabetes development, are ele-vated in individuals with prediabetesand diabetes and may reflect the body’sresponse to counterbalance increasedIL-1b activity (25).WBCs, including lymphocytes, mono-

cytes, and granulocytes, are essential in

the innate and adaptive immune re-sponse and influenced by stress, infec-tion, and inflammation. In this study,WBCs, granulocytes, and monocytesgradually increased from normoglyce-mic subjects to subjects with diabetes,whereas the lymphocyte concentrationimmediately increased with the subclin-ical disease and was stable despite thedisease progression. Interestingly, in themodel adjusted for CVRFs and diseases,granulocytes and monocytes showed astronger association for diabetes thanprediabetes compared with normogly-cemia in the sense of a dose relation-ship. Of note, neopterin, a product ofinterferon-g–activated monocytes/macrophages and a sensitive indicatorof cell-mediated immune activation(31), was inversely associated with thesubclinical disease in the multivariablelinear regression model but positivelyassociated with the prevalence of theestablished disease. The reason for thisis still unknown.

Moreover, WBCs, monocytes, andgranulocytes also increased with dis-ease severity, i.e., there was a gradientfrom individuals with insulin-dependentdiabetes to noninsulin-dependent dia-betes to dietary or no treatment.Thereby, .90% of the granulocytes areneutrophils, which play an importantrole in the early stages of inflammatoryresponses (32). A recent publication inmice presented evidence that neutro-phils are involved in insulin resistancethrough a secreted elastase (32). Thiselastase appears to cause a reductionof insulin signaling, enhancement of glu-cose production, and derangement ofthe lipid metabolism, all of them con-tributing to an increased cellular insulinresistance (32). Further, granulocyte ag-gregation is enhanced in subjects withdiabetes influenced by the intense met-abolic disturbances (33). Activated gran-ulocytes can also trigger vasculardamage, which might contribute to thedevelopment of atherosclerosis (33).

In addition to cytokines and immunecells, acute-phase proteins were ele-vated as well in this study. CRP is oneof the best-investigated epidemiologicalbiomarkers for prediabetes, diabetes,and type 2 diabetes–associated CVD(8,14,17,34) and plays a crucial role innatural host defense (11). Immunoregu-latory functions of CRP include enhance-ment of leukocyte reactivity, complement

fixation, and modulation of platelet acti-vation (11,35,36). In the current study,CRP was only weakly associated withprediabetes and diabetes prevalenceafter adjusting for other CVRFs and co-morbidities, although CRP was themarker with the highest relative changein these individuals. CRP strongly in-creased from normoglycemic subjectsto subjects with prediabetes (1.4 vs.2.3 mg/L), whereas only a small increasewas observed between subjects withprediabetes and diabetes (2.3 vs. 2.4mg/L), reflecting a very early activationof the immune system. Fibrinogen, an-other acute-phase protein participatingin the systemic response to inflamma-tion, contributes to blood viscosity,platelet aggregation, and fibrin forma-tion; modulates coagulation activationand fibrinolysis; and may enhance pla-que progression (37,38). Higher concen-trations of fibrinogen in subjects withdiabetes can therefore possibly contrib-ute to atherosclerosis in the advancedstage of disease. Similar to CRP, fibrino-gen strongly increased in subjects withnormal HbA1c concentrations to sub-jects with subclinical disease (315 vs.345 mg/L) and only weakly increasedfrom subclinical to clinical disease (345vs. 359 mg/L). In contrast to CRP, it wasstrongly associated with prediabetesand diabetes independently from CVRFand CVD and may therefore point to animportant role in the pathophysiology oftype 2 diabetes.

Inflammatory processes are also de-terminants for the presence of diabetescomplications (15). Rectified associa-tions were found in the present workbetween elevated CRP/lymphocyte con-centrations and neuropathy and WBCcount/granulocytes and retinopathy,whereas lower hematocrit levels werecorrelated with proteinuria. Lowerlevels of hemoglobin, hematocrit, anderythrocytes have been reported to beassociated with impaired glomerularfiltration rate in patients with type 1 di-abetes in the absence of nephropathycompared with subjects with normalglomerular filtration rate (39). Recently,hematocrit, serum urea concentration,and sex have been found useful in eval-uating renal function in patients withdiabetic nephropathy as part of theHUGE formula (40). The current dataconfirm the association of hematocritlevel with renal function and indicate

1362 Immune and Inflammatory Response in Diabetes Diabetes Care Volume 38, July 2015

an independent association of subclini-cal inflammation with the developmentof diabetes complications.Limitations of this study are as fol-

lows: 1) data on diabetes and CVD anddisease-associated complications wereassessed based on medical diagnosis re-corded from the interview with studyparticipants and/or medical records,implying a potential misclassification;2) 384 subjects were newly diagnosedas having diabetes within the baselineexamination by elevated HbA1c ($6.5%[$48 mmol/mol hemoglobin]) concen-trations (these subjects were classifiedas having type 2 diabetes, which is themost probable type of disease in higherage, without additional tests to confirmthe diagnosis); 3) the time between di-abetes diagnosis and time of examina-tion as surrogate for disease durationmight include a potential misclassifica-tion; and 4) this investigation providesonly an indirect estimate for the dynam-ics of the inflammatory and immune sys-tem in diabetes as the association ispresented for representative subgroupsof individuals with normoglycemia, pre-diabetes, and diabetes. This study can-not report a causal interrelationbetween diabetes, inflammation, andimmunity.In summary, we demonstrated the

variation of the inflammatory and im-mune biomarker profile with the devel-opment and progression of type 2diabetes in a large and population-representative sample including 15,010subjects. The biomarkers of inflamma-tion and immunity show varying dynam-ics with the advancing disease, enablingdifferentiation of the early preclinicaland clinical phases of the disease (i.e.,prediabetes), diabetes complications,and disease progression by intensity ofmedical treatment.

Funding. The GHS is funded through thegovernment of Rhineland-Palatinate (StiftungRheinland-Pfalz fur Innovation, contract AZ 961-386261/733), the research programs Wissenschafft Zukunft and Center for TranslationalVascular Biology of the Johannes Gutenberg-University of Mainz, and its contract withBoehringer Ingelheim and Philips Medical Sys-tems, including an unrestricted grant for theGHS. V.G., M.P.-N., and P.S.W. are funded bythe Federal Ministry of Education and Research(BMBF 01EO1003). S.B., T.Z., T.M., and P.S.W.are principal investigators of the German Centerfor Cardiovascular Research.

Duality of Interest. No potential conflicts ofinterest relevant to this article were reported.Author Contributions. V.G. and P.S.W. con-ceived, designed, and performed the experi-ments; analyzed the data; wrote the manuscript;and provided results discussion and criticalreview. V.H.S., M.P.-N., R.B.S., T.G.J.A., R.L., andS.B. conceived and designed the experiments.T.Z. and K.J.L. conceived, designed, and per-formed the experiments. A.S. and D.L.-R.analyzed the data. C.J. conceived and designedthe experiments, analyzed the data, and pro-vided results discussion and critical review.J.W. and E.S. provided results discussion andcritical review. T.M. conceived and designed theexperiments and provided results discussion andcritical review. The authors are responsible for thecontents of this publication. P.S.W. is the guar-antor of this work and, as such, had full access toall the data in the study and takes responsibilityfor the integrity of the data and the accuracy ofthe data analysis.

References1. Shaw JE, Sicree RA, Zimmet PZ. Global esti-mates of the prevalence of diabetes for 2010and 2030. Diabetes Res Clin Pract 2010;87:4–142. Wild S, Roglic G, Green A, Sicree R, King H.Global prevalence of diabetes: estimates for theyear 2000 and projections for 2030. DiabetesCare 2004;27:1047–10533. DonathMY. Inflammation as a sensor ofmet-abolic stress in obesity and type 2 diabetes. En-docrinology 2011;152:4005–40064. Velloso LA, Eizirik DL, Cnop M. Type 2 diabe-tes mellitusdan autoimmune disease? Nat RevEndocrinol 2013;9:750–7555. Itariu BK, Stulnig TM. Autoimmune aspectsof type 2 diabetes mellitus - a mini-review. Ger-ontology 2014;60:189–1966. Wilkin TJ. The accelerator hypothesis: weightgain as the missing link between type I and typeII diabetes. Diabetologia 2001;44:914–9227. Hawa MI, Buchan AP, Ola T, et al. LADA andCARDS: a prospective study of clinical outcomein established adult-onset autoimmune diabe-tes. Diabetes Care 2014;37:1643–16498. Donath MY, Shoelson SE. Type 2 diabetes asan inflammatory disease. Nat Rev Immunol2011;11:98–1079. Luotola K, Pietila A, Zeller T, et al.; Health 2000and FINRISK97 Studies. Associations betweeninterleukin-1 (IL-1) gene variations or IL-1 recep-tor antagonist levels and the development of type2 diabetes. J Intern Med 2011;269:322–33210. Pickup JC, Crook MA. Is type II diabetesmellitus a disease of the innate immune sys-tem? Diabetologia 1998;41:1241–124811. Pradhan AD, Manson JE, Rifai N, Buring JE,Ridker PM. C-reactive protein, interleukin 6, andrisk of developing type 2 diabetes mellitus.JAMA 2001;286:327–33412. Cruz NG, Sousa LP, Sousa MO, Pietrani NT,Fernandes AP, Gomes KB. The linkage betweeninflammation and Type 2 diabetes mellitus. Di-abetes Res Clin Pract 2013;99:85–9213. Herder C, Baumert J, Thorand B, et al. Che-mokines as risk factors for type 2 diabetes: re-sults from the MONICA/KORA Augsburg study,1984-2002. Diabetologia 2006;49:921–92914. Herder C, Brunner EJ, Rathmann W, et al.Elevated levels of the anti-inflammatory

interleukin-1 receptor antagonist precede theonset of type 2 diabetes: the Whitehall II study.Diabetes Care 2009;32:421–42315. King GL. The role of inflammatory cytokinesin diabetes and its complications. J Periodontol2008;79(Suppl.):1527–153416. Meier CA, Bobbioni E, Gabay C,Assimacopoulos-Jeannet F, Golay A, Dayer JM.IL-1 receptor antagonist serum levels areincreased in human obesity: a possible link tothe resistance to leptin? J Clin Endocrinol Metab2002;87:1184–118817. Spranger J, Kroke A, Mohlig M, et al. Inflam-matory cytokines and the risk to develop type 2diabetes: results of the prospective population-based European Prospective Investigation intoCancer and Nutrition (EPIC)-Potsdam Study. Di-abetes 2003;52:812–81718. Thorand B, Kolb H, Baumert J, et al. Ele-vated levels of interleukin-18 predict the devel-opment of type 2 diabetes: results from theMONICA/KORA Augsburg Study, 1984-2002. Di-abetes 2005;54:2932–293819. Pickup JC. Inflammation and activated in-nate immunity in the pathogenesis of type 2diabetes. Diabetes Care 2004;27:813–82320. American Diabetes Association. Diagnosisand classification of diabetes mellitus. DiabetesCare 2014;37(Suppl. 1):S81–S9021. Kengne AP, Batty GD, Hamer M, StamatakisE, Czernichow S. Association of C-reactive pro-tein with cardiovascular disease mortality ac-cording to diabetes status: pooled analyses of25,979 participants from four U.K. prospectivecohort studies. Diabetes Care 2012;35:396–40322. Schottker B, Herder C, Rothenbacher D,et al. Proinflammatory cytokines, adiponectin,and increased risk of primary cardiovascularevents in diabetic patients with or without renaldysfunction: results from the ESTHER study. Di-abetes Care 2013;36:1703–171123. Dalla Vestra M, Mussap M, Gallina P, et al.Acute-phase markers of inflammation and glo-merular structure in patients with type 2diabetes. J Am Soc Nephrol 2005;16(Suppl. 1):S78–S8224. Vozarova B, Weyer C, Lindsay RS, PratleyRE, Bogardus C, Tataranni PA. High white bloodcell count is associated with a worsening of in-sulin sensitivity and predicts the developmentof type 2 diabetes. Diabetes 2002;51:455–46125. Herder C, Carstensen M, Ouwens DM. Anti-inflammatory cytokines and risk of type 2 dia-betes. Diabetes Obes Metab 2013;15(Suppl. 3):39–5026. Wild PS, Zeller T, Beutel M, et al. The Gu-tenberg Health Study. BundesgesundheitsblattGesundheitsforschung Gesundheitsschutz2012;55:824–829 [in German]27. Report of the Expert Committee on the Di-agnosis and Classification of Diabetes Mellitus.Diabetes Care 1997;20:1183–119728. ArkanMC, Hevener AL, Greten FR, et al. IKK-beta links inflammation to obesity-induced in-sulin resistance. Nat Med 2005;11:191–19829. Carstensen M, Herder C, Kivimaki M, et al.Accelerated increase in serum interleukin-1 re-ceptor antagonist starts 6 years before diagno-sis of type 2 diabetes: Whitehall II prospectivecohort study. Diabetes 2010;59:1222–122730. Salomaa V, Havulinna A, Saarela O, et al.Thirty-one novel biomarkers as predictors for

care.diabetesjournals.org Grossmann and Associates 1363

clinically incident diabetes. PLoS ONE 2010;5:e1010031. Schennach H, Murr C, Gachter E,Mayersbach P, Schonitzer D, Fuchs D. Factorsinfluencing serum neopterin concentrationsin a population of blood donors. Clin Chem2002;48:643–64532. Talukdar S, Oh Y, Bandyopadhyay G, et al.Neutrophils mediate insulin resistance in micefed a high-fat diet through secreted elastase.Nat Med 2012;18:1407–141233. Elhadd TA, Bancroft A, McLarenM, NewtonRW, Belch JJ. Increased granulocyte aggrega-tion in vitro in diabetes mellitus. QJM 1997;90:461–464

34. Pickup JC, Mattock MB, Chusney GD, BurtD. NIDDM as a disease of the innate immunesystem: association of acute-phase reactantsand interleukin-6 with metabolic syndrome X.Diabetologia 1997;40:1286–129235. Mortensen RF. Macrophages and acute-phase proteins. Immunol Ser 1994;60:143–15836. Steel DM, Whitehead AS. The major acutephase reactants: C-reactive protein, serum am-yloid P component and serum amyloid A pro-tein. Immunol Today 1994;15:81–8837. Danesh J, Collins R, Peto R, Lowe GD. Hae-matocrit, viscosity, erythrocyte sedimentationrate: meta-analyses of prospective studies of

coronary heart disease. Eur Heart J 2000;21:515–52038. Tracy RP. Thrombin, inflammation, and car-diovascular disease: an epidemiologic perspec-tive. Chest 2003;124(Suppl.):49S–57S39. Bulum T, Prkacin I, Blaslov K, Zibar K,Duvnjak L. Association between red blood cellcount and renal function exist in type 1 diabeticpatients in the absence of nephropathy. CollAntropol 2013;37:777–78240. Robles NR, Ferreira F, Martinez-Gallardo R,et al. Hematocrit, urea and gender: the Hemat-ocrit, Urea and GEnder formula for prognosingprogressive renal failure in diabetic nephropa-thy. Eur J Intern Med 2012;23:283–286

1364 Immune and Inflammatory Response in Diabetes Diabetes Care Volume 38, July 2015