Insulin resistance (IR) is present long before the onset of type 2 diabetes and results not only from inherited and lifestyle factors but also likely from environmental conditions. We investigated the association between modeled long-term exposure to air pollution at residence and biomarkers related to IR, subclinical inflammation, and adipokines. Data were based on 2,944 participants of the KORA (Cooperative Health Research in the Region Augsburg) F4 study conducted in southern Germany (2006–2008). We analyzed associations between individual air pollution concentration estimated by land use regression and HOMA-IR, glucose, insulin, HbA1c, leptin, and high-sensitivity C-reactive protein levels from fasting samples using multivariable linear regression models. Effect estimates were calculated for the whole study population and subgroups of individuals who did not have diabetes, had prediabetes, or had diabetes. Among all participants, a 7.9 μg/m3 increment in particulate matter of <10 μm was associated with higher HOMA-IR (15.6% [95% CI 4.0; 28.6]) and insulin (14.5% [3.6; 26.5]). Nitrogen dioxide was associated with HOMA-IR, glucose, insulin, and leptin. Effect estimates for individuals with prediabetes were much larger and highly statistically significant, whereas individuals who did not have diabetes or had diabetes showed rather weak associations. No association was seen for HbA1c level. Our results suggested an association between long-term exposure to air pollution and IR in the general population that was attributable mainly to individuals with prediabetes.

Insulin resistance (IR) is a condition that is characterized by decreased tissue sensitivity to the action of insulin. After an initial compensatory stage when increased insulin secretion compensates for the low insulin action, the following stage indicates rapidly rising glucose levels. At this stage, IR can be diagnosed as isolated impaired fasting glucose (i-IFG), isolated impaired glucose tolerance (i-IGT), or both IFG and IGT. IR has been shown to be associated with a high risk of the development of type 2 diabetes in later life (1,2) and is considered to be an independent predictor for type 2 diabetes and therefore is often referred to as a prediabetic state (35). However, diabetes will not develop in all insulin-resistant people.

In 2013, the International Diabetes Federation estimated a diabetes prevalence of ∼8.3% (382 million people) worldwide, with type 2 diabetes accounting for ∼85–95% of all cases of diabetes (6). Obesity (7,8) and certain gene variants (9) were associated with diabetes and may cause the disease. Further critical factors determining the susceptibility for diabetes may be poor nutrition and sedentary lifestyle (10). In recent years, air pollution has also been discussed as a potential risk factor for the onset of type 2 diabetes (1115). Several reviews and a meta-analysis combined the epidemiological findings and quantified the risk increases for type 2 diabetes per 10 μg/m3 increase in exposure between 5% and 27% for particulate matter (PM) ≤2.5μm in diameter (PM2.5), 1–15% for PM ≤10μm in diameter (PM10), and 1–11% for nitrogen dioxide (NO2) (14,1620), depending on the studies included. Still, the underlying mechanisms are not fully understood, although a number of plausible pathways have been suggested, including systemic inflammation, oxidative stress, and neuronal mechanisms (12,14,15,21). Recent studies (10,2227) investigating air pollution and IR as a precursor state of type 2 diabetes showed positive associations. However, these studies were experimental, investigated short-term effects, or were focused on children.

With this cross-sectional analysis, we aimed to assess the associations between long-term exposure to air pollution at residence and biomarkers related to IR, subclinical inflammation, and adipokines 1) in the general population and 2) in individuals without diabetes, those with prediabetes, and individuals with diabetes. For all markers investigated, we hypothesized incremented levels in association with incremented air pollutant concentrations. We evaluated HOMA-IR, fasting glucose, fasting insulin, hemoglobin A1c (HbA1c) levels, as well as high-sensitivity C-reactive protein (hs-CRP) levels as established markers of inflammation. In addition, leptin was examined as adipokine, which has been suggested to be associated with IR (28). In terms of environmental stressors, we evaluated modeled long-term exposures to PM (PM10, PM between 2.5 and 10 μm in diameter [PMcoarse], PM2.5, and PM2.5 absorbance), and nitrogen oxides based on measurements conducted in 2008/2009 as well as two traffic indicators.

Study Population

The current analysis is based on data collected within the Cooperative Health Research in the Region of Augsburg (KORA) F4 study conducted in the city of Augsburg and two adjacent rural counties (southern Germany) during 2006–2008.

Altogether, 3,080 participants of the KORA F4 study were invited to the study center in Augsburg, where they answered a computer-assisted personal interview and completed a self-administered questionnaire. All individuals were physically examined, and fasting blood samples were taken. Individuals without diabetes underwent an oral glucose tolerance test (OGTT) (29).

In the KORA F4 study, data on glucose metabolism were gathered as follows: previously diagnosed diabetes was defined as a validated physician diagnosis or current use of glucose-lowering agents; and newly diagnosed diabetes, i-IGT, i-IFG, and normal glucose tolerance were defined according to World Health Organization (WHO) 1999 diagnostic criteria (30) based on fasting and post-OGTT values. For definition, diagnosis, and classification of glucose metabolism, see Supplementary Table 1.

Because of missing address information (n = 20), missing information on glucose metabolism (n = 94), the nonfasting status of some individuals (n = 10), or missing information on the main confounders (n = 12), 136 participants had to be excluded. For subgroup analysis, we stratified the remaining number of 2,944 study participants into the following subgroups: 1) people having no diabetes (nondiabetes group n = 2,125) with normal glucose tolerance; 2) people representing a group without diabetes with conditions that are associated with IR (prediabetes group n = 496) with i-IFG, i-IGT, or IFG and IGT; and 3) people who already have type 2 diabetes (diabetes group n = 323). Although diabetes will not develop in all individuals in whom IR has been diagnosed, for reasons of brevity and simplicity we named all individuals with diagnosable IR as the prediabetes group. More details on study design, sampling method, and data collection are provided elsewhere (31).

Outcome Definition

Blood was collected with minimal stasis, refrigerated at 4–8°C, and shipped in refrigerant packaging within 2–4 h to the laboratory of Augsburg Central Hospital. Fasting venous blood glucose was sampled in the morning (7:00 a.m. to 11:00 a.m.). All participants without diabetes were given a 75-g dose of anhydrous glucose (Dextro OGT; Boehringer Mannheim, Ingelheim, Germany), and another blood sample was collected after 2 h. Serum glucose was measured using a hexokinase method (GLU Flex; Dade Behring Marburg, Marburg, Germany). Insulin was determined using ELISA kits from Invitrogen (Camarillo, CA). As a surrogate of IR, the homeostatic model assessment was used and defined as HOMA-IR = (fasting insulin [μU/mL]) × (fasting glucose [mmol/L])/22.5 (32).

HbA1c was measured with a reverse-phase, cation-exchange, high-performance liquid chromotography method (analyzer HA 8160; Menarini Group). Leptin concentrations were assessed using ELISA kits from Mercodia (Uppsala, Sweden). The measurement of hs-CRP was in anticoagulated plasma samples using a high-sensitivity latex-enhanced nephelometric assay on a BN II analyzer (Dade Behring), with intra-assay and interassay coefficients of variation of 2.7% and 6.3%, respectively.

Air Pollution Exposure

Residential exposure to ambient air pollution assessed as mean annual levels was estimated within the ESCAPE study (European Study of Cohorts for Air Pollution Effects [www.escapeproject.eu]). Air pollution measurements of PM10, PM2.5, NO2, and the sum of NO2 and nitrogen monoxides (NOx) were collected at 20 (PM) and 40 (NOx) monitoring sites for three periods of 2 weeks in the cold, warm, and one intermediate season during the period from October 2008 to July 2009. Land use regression (LUR) models were developed on the basis of annual average measurements and predictor variables like traffic, land use, industry, and population density derived from geographic information systems (33,34). These regression models were then applied to the residence addresses of study participants to assess individual long-term concentrations.

In addition to the modeled air pollution concentration, we considered two traffic indicators: 1) traffic intensity on the nearest road (number of vehicles/day) and 2) traffic load on major roads within 100 m of the residence (number of vehicles*meter/day), defined as the sum of traffic intensity on roads with >5,000 vehicles/day multiplied by the length of those roads in a 100 m buffer around the home addresses.

Covariates

As potential confounding factors might affect the different outcomes in different ways, we hierarchically optimized our confounder models for each outcome separately. First, we specified a minimum set of a priori defined covariates for all outcome variables, including age, sex, smoking status, BMI, waist-to-hip ratio (Spearman correlation coefficient with BMI was 0.52), and month of blood withdrawal. Second, we selected from several socioeconomic variables, as follows: occupational status, years of education, per capita income, and socioeconomic status (categorical variable combining education and income). In a third step, we offered further lifestyle-related variables: years and pack-years of smoking, physical activity, and alcohol intake. The selection in steps 2 and 3 was based on minimizing the Bayesian information criterion as it deals with the trade-off between the goodness-of-fit and the complexity of the model. For a detailed description of the covariates and the final confounder models, see Table 1 and Supplementary Table 2.

Table 1

Description of the study population: all participants and stratified by state of glucose metabolism

VariableMissing values N (%)All (N = 2,944)
Nondiabetes (N = 2,125)
Prediabetes (N = 496)
Diabetes (N = 323)
P value*
Arithmetic mean(SD)Arithmetic mean(SD)Arithmetic mean(SD)Arithmetic mean(SD)
Age (years)  56.2 (13.1) 52.9 (12.5) 63.2 (11.0) 66.9 (9.4) <0.001 
BMI (kg/m2 27.6 (4.8) 26.6 (4.3) 29.7 (4.8) 31.3 (5.0) <0.001 
Waist-to-hip ratio  0.88 (0.09) 0.86 (0.08) 0.92 (0.08) 0.95 (0.07) <0.001 
Education (years) 5 (0.2) 11.8 (2.6) 12.0 (2.7) 11.3 (2.5) 11.0 (2.4) <0.001 
Per capita income (€) 109 (3.7) 1,120 (576) 1,133 (593) 1,105 (593) 1,054 (562) 0.058 
Smoking (years) 8 (0.3) 13.1 (15.3) 12.6 (14.6) 13.2 (15.7) 16.9 (18.4) 0.003 
Pack-years of smoking (total pack-years smoked) 81 (2.8) 11.6 (19.2) 10.2 (16.1) 12.1 (19.6) 19.9 (31.7) 0.002 
Alcohol intake (g/day)  14.4 (19.6) 14.1 (18.9) 15.4 (19.8) 15.0 (23.6) 0.149 
  N (%) N (%) N (%) N (%) P value* 
Male sex  1,424 (48.4) 969 (45.6) 262 (52.8) 193 (59.8) <0.001 
Smoking status           
 Current  522 (17.7) 437 (20.6) 46 (9.3) 39 (12.1) <0.001 
 Ex  1,195 (40.6) 811 (38.2) 222 (44.8) 162 (50.2) <0.001 
 Never  1,227 (41.7) 877 (41.3) 228 (46.0) 122 (37.8)  
Occupational status           
 Employed/self-employed  1,483 (50.4) 1,265 (59.5) 165 (33.3) 53 (16.4) <0.001 
 Unemployed  60 (2.0) 43 (2.0) (1.4) 10 (3.1)  
 Homemaker  304 (10.3) 242 (11.4) 42 (8.5) 20 (6.2)  
 Retired  1,097 (37.3) 575 (27.1) 282 (56.9) 240 (74.3)  
Socioeconomic status           
 Low 114 (3.9) 915 (32.3) 658 (32.2) 149 (31.5) 108 (34.6) 0.002 
 Medium  853 (30.1) 584 (28.6) 153 (32.3) 116 (37.2)  
 High  1,062 (37.5) 803 (39.3) 171 (36.2) 88 (27.2)  
Physical activity           
 Low  945 (32.1) 597 (28.1) 193 (38.9) 155 (48.0) <0.001 
 Medium  1,277 (43.4) 958 (45.1) 204 (41.1) 115 (35.6)  
 High  722 (24.5) 570 (26.8) 99 (20.0) 53 (16.4)  
History of           
 Hypertension 4 (0.1) 1,124 (38.2) 584 (27.5) 281 (56.7) 259 (80.2) <0.001 
 Myocardial infarction  76 (2.6) 26 (1.2) 19 (3.8) 31 (9.6) <0.001 
 Stroke  58 (2.0) 30 (1.4) 11 (2.2) 17 (5.3) <0.001 
Medication intake           
 Antidiabetics 2 (0.1) 155 (5.3)     155 (48.0)  
 Insulin 2 (0.1) 41 (1.4)     41 (12.7)  
 Oral antidiabetic agents 2 (0.1) 139 (4.7)     139 (43.0)  
 Statins 2 (0.1) 352 (12.0) 176 (8.3) 74 (14.9) 102 (31.6) <0.001 
 Antihypertensive agents  899 (30.5) 430 (20.2) 232 (46.8) 237 (73.4) <0.001 
 Diuretics  517 (17.6) 228 (10.7) 139 (28.0) 150 (46.4) <0.001 
 β-Blocker  551 (18.7) 253 (11.9) 152 (30.6) 146 (45.2) <0.001 
 Calcium channel blocker 1 (0.0) 231 (7.9) 100 (4.7) 63 (12.7) 68 (21.1) <0.001 
 ACE inhibitor 1 (0.0) 388 (13.2) 162 (7.6) 99 (20.0) 127 (39.3) <0.001 
 Angiotensin receptor blocker 2 (0.1) 219 (7.4) 121 (5.7) 50 (10.1) 48 (14.9) <0.001 
 Cortisone 2 (0.1) 41 (1.4) 27 (1.3) (1.2) (2.5) 0.208 
VariableMissing values N (%)All (N = 2,944)
Nondiabetes (N = 2,125)
Prediabetes (N = 496)
Diabetes (N = 323)
P value*
Arithmetic mean(SD)Arithmetic mean(SD)Arithmetic mean(SD)Arithmetic mean(SD)
Age (years)  56.2 (13.1) 52.9 (12.5) 63.2 (11.0) 66.9 (9.4) <0.001 
BMI (kg/m2 27.6 (4.8) 26.6 (4.3) 29.7 (4.8) 31.3 (5.0) <0.001 
Waist-to-hip ratio  0.88 (0.09) 0.86 (0.08) 0.92 (0.08) 0.95 (0.07) <0.001 
Education (years) 5 (0.2) 11.8 (2.6) 12.0 (2.7) 11.3 (2.5) 11.0 (2.4) <0.001 
Per capita income (€) 109 (3.7) 1,120 (576) 1,133 (593) 1,105 (593) 1,054 (562) 0.058 
Smoking (years) 8 (0.3) 13.1 (15.3) 12.6 (14.6) 13.2 (15.7) 16.9 (18.4) 0.003 
Pack-years of smoking (total pack-years smoked) 81 (2.8) 11.6 (19.2) 10.2 (16.1) 12.1 (19.6) 19.9 (31.7) 0.002 
Alcohol intake (g/day)  14.4 (19.6) 14.1 (18.9) 15.4 (19.8) 15.0 (23.6) 0.149 
  N (%) N (%) N (%) N (%) P value* 
Male sex  1,424 (48.4) 969 (45.6) 262 (52.8) 193 (59.8) <0.001 
Smoking status           
 Current  522 (17.7) 437 (20.6) 46 (9.3) 39 (12.1) <0.001 
 Ex  1,195 (40.6) 811 (38.2) 222 (44.8) 162 (50.2) <0.001 
 Never  1,227 (41.7) 877 (41.3) 228 (46.0) 122 (37.8)  
Occupational status           
 Employed/self-employed  1,483 (50.4) 1,265 (59.5) 165 (33.3) 53 (16.4) <0.001 
 Unemployed  60 (2.0) 43 (2.0) (1.4) 10 (3.1)  
 Homemaker  304 (10.3) 242 (11.4) 42 (8.5) 20 (6.2)  
 Retired  1,097 (37.3) 575 (27.1) 282 (56.9) 240 (74.3)  
Socioeconomic status           
 Low 114 (3.9) 915 (32.3) 658 (32.2) 149 (31.5) 108 (34.6) 0.002 
 Medium  853 (30.1) 584 (28.6) 153 (32.3) 116 (37.2)  
 High  1,062 (37.5) 803 (39.3) 171 (36.2) 88 (27.2)  
Physical activity           
 Low  945 (32.1) 597 (28.1) 193 (38.9) 155 (48.0) <0.001 
 Medium  1,277 (43.4) 958 (45.1) 204 (41.1) 115 (35.6)  
 High  722 (24.5) 570 (26.8) 99 (20.0) 53 (16.4)  
History of           
 Hypertension 4 (0.1) 1,124 (38.2) 584 (27.5) 281 (56.7) 259 (80.2) <0.001 
 Myocardial infarction  76 (2.6) 26 (1.2) 19 (3.8) 31 (9.6) <0.001 
 Stroke  58 (2.0) 30 (1.4) 11 (2.2) 17 (5.3) <0.001 
Medication intake           
 Antidiabetics 2 (0.1) 155 (5.3)     155 (48.0)  
 Insulin 2 (0.1) 41 (1.4)     41 (12.7)  
 Oral antidiabetic agents 2 (0.1) 139 (4.7)     139 (43.0)  
 Statins 2 (0.1) 352 (12.0) 176 (8.3) 74 (14.9) 102 (31.6) <0.001 
 Antihypertensive agents  899 (30.5) 430 (20.2) 232 (46.8) 237 (73.4) <0.001 
 Diuretics  517 (17.6) 228 (10.7) 139 (28.0) 150 (46.4) <0.001 
 β-Blocker  551 (18.7) 253 (11.9) 152 (30.6) 146 (45.2) <0.001 
 Calcium channel blocker 1 (0.0) 231 (7.9) 100 (4.7) 63 (12.7) 68 (21.1) <0.001 
 ACE inhibitor 1 (0.0) 388 (13.2) 162 (7.6) 99 (20.0) 127 (39.3) <0.001 
 Angiotensin receptor blocker 2 (0.1) 219 (7.4) 121 (5.7) 50 (10.1) 48 (14.9) <0.001 
 Cortisone 2 (0.1) 41 (1.4) 27 (1.3) (1.2) (2.5) 0.208 

* Kruskal-Wallis rank sum test (for continuous variables) or Pearson χ2 test (for categorical variables) to test for differences between the subgroups.

†Sum of “education in categories” (1, <10 years of education; 2, ≥10 years of education but not university degree; 3, university degree) and “per capita income in tertiles” (1, lower; 2, medium; 3, upper): low (sum = 2 or 3), medium (sum = 4); high: (sum = 5 or 6).

Statistical Analyses

We performed Pearson x2 tests for categorical variables and Kruskal-Wallis tests for continuous variables to test for differences between the subgroups. Correlations between air pollutants and residential proximity to traffic were examined using Spearman correlation coefficients.

To assess the association between long-term residential exposure to air pollution and the biomarkers, we performed multivariable linear regression analyses. All outcomes were log transformed since residuals deviated from normality. We included the annual mean concentration of each air pollutant separately as a linear term in addition to the chosen covariates. Traffic variables were additionally adjusted for background NO2 levels to investigate traffic effects independent of the background air pollution concentrations. To investigate potential effect modification by sex, we included an interaction term in the model. Results are presented as the percentage change of the geometric mean value per 5th − 95th percentile difference of the exposure concentrations and corresponding 95% CIs. Effect estimates were calculated separately for all participants, and for the nondiabetes, prediabetes, and diabetes subgroups. Tests of interaction were calculated for differences between subgroups (35).

As a sensitivity analysis, we applied a very basic confounder model including only sex, age, and BMI to verify whether the level of covariate adjustment was probably too high to detect an association. We used additive models incorporating separately each exposure variable as a cubic regression spline with 3 df to check the linearity of the dose-response function. In addition, we categorized the exposure variables into quartiles and alternatively ran quantile regression. Since an underlying systemic inflammation due to an acute infection might change markers of IR, we excluded study participants with an hs-CRP value >10 mg/L. To investigate the influence of glucose- and lipid-lowering medication, we excluded persons taking antidiabetic medications or statins. We also excluded persons who reported an intake of diuretics and/or β-blockers, as this type of medication may promote susceptibility for IR (36,37). To assess the sensitivity of our results to the influence of the degree of impaired glucose regulation, we stratified the prediabetes group into i-IFG (n = 113), i-IGT (n = 307), and IFG-IGT (n = 76) groups. Moreover, we stratified the diabetes group into persons with newly diagnosed diabetes (n = 113) and individuals with known type 2 diabetes (n = 210).

All statistical analyses were performed with SAS version 9.3 (SAS Institute, Cary, NC) and R version 3.1.0 (The R Foundation for Statistical Computing, Vienna, Austria).

Study Population

Table 1 shows the baseline characteristics of the 2,944 study participants. The mean age was 56.2 years, the mean BMI was 27.6 kg/m2, and marginally more women participated. Participants with prediabetes and participants with diabetes, on average, were older, BMI was higher, and more usually male than participants without diabetes. In addition, they showed a lower prevalence of current smokers, but a higher prevalence of ex-smokers. Also, socioeconomic status and physical activity were in general lower for individuals with prediabetes and individuals with diabetes. The prevalence of hypertension, myocardial infarction, stroke, and medication intake were higher with worsened insulin sensitivity.

Plasma concentrations of the six selected biomarkers are described in Table 2 by the arithmetic mean and SD as well as the geometric mean. Significant differences were found for all markers between study groups. All considered blood markers showed higher concentrations with deteriorating glucose metabolism.

Table 2

Plasma concentrations of blood markers for all participants and stratified by state of glucose metabolism

Blood markersAll (N = 2,944)
Nondiabetes (N = 2,125)
Prediabetes (N = 496)
Diabetes (N = 323)
P value*
NArithmetic mean(SD)Geometric meanNArithmetic mean(SD)Geometric meanNArithmetic mean(SD)Geometric meanNArithmetic mean(SD)Geometric mean
HOMA-IR 2,928 2.3 (8.7) 1.1 2,114 1.6 (6.0) 0.9 492 3.6 (14.8) 1.8 322 5.1 (10.2) 2.8 <0.001 
Glucose (mg/dL) 2,944 98.3 (19.1) 96.9 2,125 91.7 (7.6) 91.4 496 103.3 (11.1) 102.7 323 133.7 (34.4) 130.1 <0.001 
Insulin (μIU/mL) 2,928 9.0 (34.6) 4.7 2,114 6.9 (26.0) 3.9 492 13.9 (59.2) 7.1 322 15.7 (31.9) 8.6 <0.001 
HbA1c (%) 2,943 5.6 (0.6) 5.5 2,125 5.4 (0.3) 5.4 496 5.7 (0.3) 5.6 322 6.6 (1.1) 6.5 <0.001 
HbA1c (mmol/mol)‡ 2,943 37.1 (6.6) 36.6 2,125 35.1 (3.3) 34.9 496 38.2 (3.7) 38.1 322 48.7 (11.8) 47.6 <0.001 
Leptin (ng/mL) 2,929 18.9 (20.4) 11.1 2,116 16.4 (18.4) 9.5 493 23.3 (21.2) 15.2 320 28.8 (27.1) 18.9 <0.001 
hs-CRP (mg/L) 2,929 2.5 (5.4) 1.2 2,116 2.0 (3.4) 1.0 491 3.2 (4.6) 1.8 322 4.9 (12.1) 2.3 <0.001 
Blood markersAll (N = 2,944)
Nondiabetes (N = 2,125)
Prediabetes (N = 496)
Diabetes (N = 323)
P value*
NArithmetic mean(SD)Geometric meanNArithmetic mean(SD)Geometric meanNArithmetic mean(SD)Geometric meanNArithmetic mean(SD)Geometric mean
HOMA-IR 2,928 2.3 (8.7) 1.1 2,114 1.6 (6.0) 0.9 492 3.6 (14.8) 1.8 322 5.1 (10.2) 2.8 <0.001 
Glucose (mg/dL) 2,944 98.3 (19.1) 96.9 2,125 91.7 (7.6) 91.4 496 103.3 (11.1) 102.7 323 133.7 (34.4) 130.1 <0.001 
Insulin (μIU/mL) 2,928 9.0 (34.6) 4.7 2,114 6.9 (26.0) 3.9 492 13.9 (59.2) 7.1 322 15.7 (31.9) 8.6 <0.001 
HbA1c (%) 2,943 5.6 (0.6) 5.5 2,125 5.4 (0.3) 5.4 496 5.7 (0.3) 5.6 322 6.6 (1.1) 6.5 <0.001 
HbA1c (mmol/mol)‡ 2,943 37.1 (6.6) 36.6 2,125 35.1 (3.3) 34.9 496 38.2 (3.7) 38.1 322 48.7 (11.8) 47.6 <0.001 
Leptin (ng/mL) 2,929 18.9 (20.4) 11.1 2,116 16.4 (18.4) 9.5 493 23.3 (21.2) 15.2 320 28.8 (27.1) 18.9 <0.001 
hs-CRP (mg/L) 2,929 2.5 (5.4) 1.2 2,116 2.0 (3.4) 1.0 491 3.2 (4.6) 1.8 322 4.9 (12.1) 2.3 <0.001 

*Kruskal-Wallis rank sum test to test for differences between the subgroups.

†Expressed as % according to National Glycohemoglobin Standardization Program/Diabetes Control and Complications Trial (NGSP/DCCT): HbA1c (%) = (HbA1c/Hb) × 91.48 + 2.15.

§For mmol/mol according to International Federation of Clinical Chemistry and Laboratory Medicine (IFCC): HbA1c (mmol/mol) = (HbA1c/Hb) × 1,000.

Long-term Air Pollution

The distribution of modeled annual average concentrations of air pollutants and traffic indicators at participants’ residences can be found in Table 3. Air pollution concentrations were below European Union limits (European Union Directive 2008/50/EC) but exceeded WHO recommendations (38). Correlations between air pollution and traffic indicators were only low to moderate. There were no significant differences in the exposure levels between the three subgroups (Supplementary Table 3).

Table 3

Annual average concentrations of air pollutants and traffic indicators and corresponding Spearman correlation coefficients (N = 2,944)

ExposureDescriptives
Spearman correlation coefficients
MeanSD5%Median95%PM10PMcoarsePM2.5PM2.5absNOxNO2
PM10 (μg/m320.4 (2.4) 16.5 20.5 24.4      
PMcoarse (μg/m36.2 (1.1) 4.9 6.1 8.4 0.76     
PM2.5 (μg/m3)  13.6 (0.9) 12.5 13.4 15.3 0.43 0.32    
PM2.5abs (10−5/m) 1.7 (0.2) 1.5 1.7 2.0 0.67 0.84 0.48   
NOx (μg/m332.7 (7.2) 23.9 31.4 46.7 0.69 0.85 0.48 0.76  
NO2 (μg/m318.8 (3.8) 13.8 18.3 25.6 0.67 0.79 0.45 0.66 0.92 
Traffic intensity on the nearest road (vehicles/day), per 1,000 1.6 (3.2) 0.5 0.5 8.1 0.13 0.19 0.22 0.20 0.26 0.26 
Traffic load within 100 m on major roads (vehicles*meter/day), per 10,000 40.7 (102.3) 0.0 0.0 243.6 0.27 0.33 0.30 0.42 0.39 0.37 
ExposureDescriptives
Spearman correlation coefficients
MeanSD5%Median95%PM10PMcoarsePM2.5PM2.5absNOxNO2
PM10 (μg/m320.4 (2.4) 16.5 20.5 24.4      
PMcoarse (μg/m36.2 (1.1) 4.9 6.1 8.4 0.76     
PM2.5 (μg/m3)  13.6 (0.9) 12.5 13.4 15.3 0.43 0.32    
PM2.5abs (10−5/m) 1.7 (0.2) 1.5 1.7 2.0 0.67 0.84 0.48   
NOx (μg/m332.7 (7.2) 23.9 31.4 46.7 0.69 0.85 0.48 0.76  
NO2 (μg/m318.8 (3.8) 13.8 18.3 25.6 0.67 0.79 0.45 0.66 0.92 
Traffic intensity on the nearest road (vehicles/day), per 1,000 1.6 (3.2) 0.5 0.5 8.1 0.13 0.19 0.22 0.20 0.26 0.26 
Traffic load within 100 m on major roads (vehicles*meter/day), per 10,000 40.7 (102.3) 0.0 0.0 243.6 0.27 0.33 0.30 0.42 0.39 0.37 

Current European air quality standards (1-year average): 40 µg/m3 (PM10); 25 µg/m3 (PM2.5); 40 µg/m3 (NO2). WHO recommendations (1-year average): 20 µg/m3 (PM10); 10 µg/m3 (PM2.5); 40 µg/m3 (NO2). PM2.5abs, soot content (absorbance) of PM2.5.

Association Between Long-term Air Pollution and Biomarkers Related to IR

Tables 4 and 5 show the associations between long-term residential exposure to air pollutants, traffic indicators, and biomarkers related to IR. Among all study participants, exposure to NO2 was significantly positively associated with HOMA-IR, glucose, insulin, and leptin levels. Also, PM10, PMcoarse, PM2.5 absorbance, and NOx showed a positive association with HOMA-IR and insulin, whereas PM2.5 was borderline significant for glucose. Both traffic indicators were not significantly associated with any of the blood markers.

Table 4

Association between long-term air pollutants, traffic indicators, and biomarkers

Pollutant ExposureIncrement
All
Nondiabetes
Prediabetes
Diabetes
% Change (95% CI)% Change (95% CI)% Change (95% CI)P value% Change (95% CI)P value§
HOMA-IR        
 PM10 (μg/m37.9 15.6 (4.0; 28.6) 10.4 (−1.8; 24.1) 45.9 (12.8; 88.8) 0.027 3.5 (−29.2; 51.1) 0.374 
 PMcoarse (μg/m33.6 18.8 (6.6; 32.4) 11.1 (−1.5; 25.3) 68.7 (30.8; 117.7) 0.002 13.5 (−22.2; 65.5) 0.458 
 PM2.5 (μg/m32.8 9.7 (−1.3; 21.9) 0.6 (−10.7; 13.2) 26.8 (−0.8; 62.1) 0.048 38.6 (−2.1; 96.4) 0.044 
 PM2.5abs (10−5/m) 0.5 13.2 (2.9; 24.6) 7.6 (−3.1; 19.5) 61.3 (28.1; 103.0) 0.001 0.1 (−28.6; 40.4) 0.345 
 NOx (μg/m322.9 21.2 (9.4; 34.3) 12.4 (0.4; 25.8) 78.1 (39.6; 127.3) 0.000 23.2 (−14.5; 77.5) 0.319 
 NO2 (μg/m311.9 19.2 (7.7; 31.8) 9.2 (−2.3; 22.1) 73.1 (36.1; 120.1) 0.000 24.9 (−12.6; 78.5) 0.241 
 Traffic intensity on the nearest road (vehicles/day), per 1,000 7.5 −5.6 (−13.0; 2.4) −6.8 (−14.7; 1.7) 16.3 (−5.4; 43.0) 0.026 −9.1 (−34; 25.3) 0.443 
 Traffic load within 100 m on major roads (vehicles*meter/day), per 10,000 243.6 6.9 (−1.2; 15.6) 0.6 (−7.4; 9.4) 50.6 (22.4; 85.5) 0.000 28.0 (−6.4; 75) 0.072 
Glucose     
 PM10 (μg/m37.9 1.0 (−0.7; 2.6) 0.6 (−0.4; 1.7) 0.2 (−2.9; 3.4) 0.389 0.3 (−7.7; 9.0) 0.468 
 PMcoarse (μg/m33.6 1.4 (−0.3; 3.1) 0.9 (−0.2; 1.9) 1.3 (−1.9; 4.5) 0.405 4.4 (−3.9; 13.5) 0.207 
 PM2.5 (μg/m32.8 1.6 (0.0; 3.3) 0.8 (−0.3; 1.8) 1.9 (−1.1; 5.0) 0.248 0.4 (−7.0; 8.5) 0.468 
 PM2.5abs (10−5/m) 0.5 0.6 (−0.9; 2.1) 0.8 (−0.2; 1.7) 0.8 (−2.1; 3.7) 0.497 0.9 (−6.4; 8.6) 0.489 
 NOx (μg/m322.9 1.3 (−0.3; 2.9) 1.0 (0.0; 2.0) 2.2 (−0.8; 5.3) 0.239 2.6 (−5.3; 11.2) 0.353 
 NO2 (μg/m311.9 1.7 (0.1; 3.3) 0.9 (−0.1; 1.9) 2.4 (−0.6; 5.5) 0.184 3.5 (−4.3; 11.9) 0.267 
 Traffic intensity on the nearest road (vehicles/day), per 1,000 7.5 −0.2 (−1.5; 1.0) 0.7 (−0.1; 1.4) 1.0 (−1.5; 3.6) 0.408 −0.7 (−7.5; 6.6) 0.357 
 Traffic load within 100 m on major roads (vehicles*meter/day), per 10,000 243.6 0.3 (−0.9; 1.5) −0.4 (−1.1; 0.3) 3.3 (0.7; 6.0) 0.004 8.9 (1.6; 16.6) 0.006 
Insulin     
 PM10 (μg/m37.9 14.5 (3.6; 26.5) 10.3 (−1.4; 23.4) 44.5 (13.0; 84.7) 0.025 0.0 (−29.8; 42.4) 0.302 
 PMcoarse (μg/m33.6 17.8 (6.4; 30.5) 11.0 (−1.0; 24.6) 65.8 (30.1; 111.4) 0.002 6.7 (−25.3; 52.4) 0.417 
 PM2.5 (μg/m32.8 8.3 (−2.0; 19.6) 0.4 (−10.4; 12.6) 24.2 (−1.9; 57.1) 0.056 35.6 (−2.9; 89.4) 0.047 
 PM2.5abs (10−5/m) 0.5 13.2 (3.4; 23.8) 7.9 (−2.4; 19.4) 59.1 (27.6; 98.4) 0.001 −2.5 (−29.3; 34.5) 0.278 
 NOx (μg/m322.9 20.0 (8.9; 32.2) 11.9 (0.4; 24.7) 72.0 (36.4; 117.1) 0.001 17.8 (−16.7; 66.5) 0.392 
 NO2 (μg/m311.9 17.2 (6.6; 29.0) 8.5 (−2.5; 20.8) 67.2 (32.8; 110.4) 0.000 19.6 (−14.7; 67.8) 0.295 
 Traffic intensity on the nearest road (vehicles/day), per 1,000 7.5 −5.6 (−12.6; 2.1) −7.7 (−15.2; 0.5) 14.7 (−5.9; 39.8) 0.024 −8.4 (−32.7; 24.7) 0.481 
 Traffic load within 100 m on major roads (vehicles*meter/day), per 10,000 243.6 6.4 (−1.3; 14.6) 1.0 (−6.8; 9.5) 44.9 (18.5; 77.2) 0.001 16.6 (−13.7; 57.5) 0.183 
Pollutant ExposureIncrement
All
Nondiabetes
Prediabetes
Diabetes
% Change (95% CI)% Change (95% CI)% Change (95% CI)P value% Change (95% CI)P value§
HOMA-IR        
 PM10 (μg/m37.9 15.6 (4.0; 28.6) 10.4 (−1.8; 24.1) 45.9 (12.8; 88.8) 0.027 3.5 (−29.2; 51.1) 0.374 
 PMcoarse (μg/m33.6 18.8 (6.6; 32.4) 11.1 (−1.5; 25.3) 68.7 (30.8; 117.7) 0.002 13.5 (−22.2; 65.5) 0.458 
 PM2.5 (μg/m32.8 9.7 (−1.3; 21.9) 0.6 (−10.7; 13.2) 26.8 (−0.8; 62.1) 0.048 38.6 (−2.1; 96.4) 0.044 
 PM2.5abs (10−5/m) 0.5 13.2 (2.9; 24.6) 7.6 (−3.1; 19.5) 61.3 (28.1; 103.0) 0.001 0.1 (−28.6; 40.4) 0.345 
 NOx (μg/m322.9 21.2 (9.4; 34.3) 12.4 (0.4; 25.8) 78.1 (39.6; 127.3) 0.000 23.2 (−14.5; 77.5) 0.319 
 NO2 (μg/m311.9 19.2 (7.7; 31.8) 9.2 (−2.3; 22.1) 73.1 (36.1; 120.1) 0.000 24.9 (−12.6; 78.5) 0.241 
 Traffic intensity on the nearest road (vehicles/day), per 1,000 7.5 −5.6 (−13.0; 2.4) −6.8 (−14.7; 1.7) 16.3 (−5.4; 43.0) 0.026 −9.1 (−34; 25.3) 0.443 
 Traffic load within 100 m on major roads (vehicles*meter/day), per 10,000 243.6 6.9 (−1.2; 15.6) 0.6 (−7.4; 9.4) 50.6 (22.4; 85.5) 0.000 28.0 (−6.4; 75) 0.072 
Glucose     
 PM10 (μg/m37.9 1.0 (−0.7; 2.6) 0.6 (−0.4; 1.7) 0.2 (−2.9; 3.4) 0.389 0.3 (−7.7; 9.0) 0.468 
 PMcoarse (μg/m33.6 1.4 (−0.3; 3.1) 0.9 (−0.2; 1.9) 1.3 (−1.9; 4.5) 0.405 4.4 (−3.9; 13.5) 0.207 
 PM2.5 (μg/m32.8 1.6 (0.0; 3.3) 0.8 (−0.3; 1.8) 1.9 (−1.1; 5.0) 0.248 0.4 (−7.0; 8.5) 0.468 
 PM2.5abs (10−5/m) 0.5 0.6 (−0.9; 2.1) 0.8 (−0.2; 1.7) 0.8 (−2.1; 3.7) 0.497 0.9 (−6.4; 8.6) 0.489 
 NOx (μg/m322.9 1.3 (−0.3; 2.9) 1.0 (0.0; 2.0) 2.2 (−0.8; 5.3) 0.239 2.6 (−5.3; 11.2) 0.353 
 NO2 (μg/m311.9 1.7 (0.1; 3.3) 0.9 (−0.1; 1.9) 2.4 (−0.6; 5.5) 0.184 3.5 (−4.3; 11.9) 0.267 
 Traffic intensity on the nearest road (vehicles/day), per 1,000 7.5 −0.2 (−1.5; 1.0) 0.7 (−0.1; 1.4) 1.0 (−1.5; 3.6) 0.408 −0.7 (−7.5; 6.6) 0.357 
 Traffic load within 100 m on major roads (vehicles*meter/day), per 10,000 243.6 0.3 (−0.9; 1.5) −0.4 (−1.1; 0.3) 3.3 (0.7; 6.0) 0.004 8.9 (1.6; 16.6) 0.006 
Insulin     
 PM10 (μg/m37.9 14.5 (3.6; 26.5) 10.3 (−1.4; 23.4) 44.5 (13.0; 84.7) 0.025 0.0 (−29.8; 42.4) 0.302 
 PMcoarse (μg/m33.6 17.8 (6.4; 30.5) 11.0 (−1.0; 24.6) 65.8 (30.1; 111.4) 0.002 6.7 (−25.3; 52.4) 0.417 
 PM2.5 (μg/m32.8 8.3 (−2.0; 19.6) 0.4 (−10.4; 12.6) 24.2 (−1.9; 57.1) 0.056 35.6 (−2.9; 89.4) 0.047 
 PM2.5abs (10−5/m) 0.5 13.2 (3.4; 23.8) 7.9 (−2.4; 19.4) 59.1 (27.6; 98.4) 0.001 −2.5 (−29.3; 34.5) 0.278 
 NOx (μg/m322.9 20.0 (8.9; 32.2) 11.9 (0.4; 24.7) 72.0 (36.4; 117.1) 0.001 17.8 (−16.7; 66.5) 0.392 
 NO2 (μg/m311.9 17.2 (6.6; 29.0) 8.5 (−2.5; 20.8) 67.2 (32.8; 110.4) 0.000 19.6 (−14.7; 67.8) 0.295 
 Traffic intensity on the nearest road (vehicles/day), per 1,000 7.5 −5.6 (−12.6; 2.1) −7.7 (−15.2; 0.5) 14.7 (−5.9; 39.8) 0.024 −8.4 (−32.7; 24.7) 0.481 
 Traffic load within 100 m on major roads (vehicles*meter/day), per 10,000 243.6 6.4 (−1.3; 14.6) 1.0 (−6.8; 9.5) 44.9 (18.5; 77.2) 0.001 16.6 (−13.7; 57.5) 0.183 

Data presented as the percentage change (95% CI) from geometric mean per 5th–95th percentile difference increment in air pollutants adjusted for age, sex, smoking, BMI, waist-to-hip ratio, month of blood withdrawal, and selected socioeconomic and lifestyle variables (see Supplementary Table 2). PM2.5abs, soot content (absorbance) of PM2.5.

Boldface type indicates effect estimates were statistically significant, P < 0.05.

†Corresponds to the difference between the 5th and 95th percentile of the corresponding exposure.

‡Test of interaction to test for differences between participants with prediabetes and participants without diabetes.

§Test of interaction to test for differences between participants with and without diabetes.

Table 5

Association between long-term air pollutants, traffic indicators, and biomarkers

Pollutant exposureIncrement
All
Nondiabetes
Prediabetes
Diabetes
% Change (95% CI)% Change (95% CI)% Change (95% CI)P value% Change (95% CI)P value§
HbA1c        
 PM10 (μg/m37.9 0.5 (−0.5; 1.5) 0.3 (−0.4; 1.0) −0.4 (−2.0; 1.2) 0.200 2.7 (−2.6; 8.4) 0.193 
 PMcoarse (μg/m33.6 0.1 (−0.9; 1.1) 0.1 (−0.6; 0.8) −0.2 (−1.8; 1.4) 0.368 1.1 (−4.3; 6.7) 0.365 
 PM2.5 (μg/m32.8 0.9 (−0.1; 1.9) 0.4 (−0.2; 1.1) 0.7 (−0.8; 2.3) 0.360 0.7 (−4.2; 5.8) 0.465 
 PM2.5abs (10−5/m) 0.5 −0.1 (−0.9; 0.8) 0.1 (−0.5; 0.7) −0.4 (−1.9; 1.0) 0.248 0.1 (−4.6; 5.0) 0.494 
 NOx (μg/m322.9 0.0 (−0.9; 1.0) 0.1 (−0.5; 0.8) 0.1 (−1.4; 1.7) 0.498 −0.5 (−5.6; 4.9) 0.413 
 NO2 (μg/m311.9 0.2 (−0.8; 1.1) 0.2 (−0.4; 0.9) 0.3 (−1.3; 1.8) 0.483 −1.3 (−6.3; 3.9) 0.274 
 Traffic intensity on the nearest road (vehicles/day), per 1,000 7.5 −0.3 (−1.1; 0.5) 0.0 (−0.5; 0.5) 0.7 (−0.6; 2.0) 0.161 1.1 (−3.5; 5.9) 0.332 
 Traffic load within 100 m on major roads (vehicles*meter/day), per 10,000 243.6 −0.2 (−0.9; 0.5) −0.3 (−0.7; 0.2) 0.7 (−0.6; 2.0) 0.090 2.6 (−1.9; 7.3) 0.107 
Leptin     
 PM10 (μg/m37.9 7.9 (−0.4; 17.0) 6.6 (−3.1; 17.3) 20.2 (0.7; 43.5) 0.122 −1.8 (−21.7; 23.0) 0.254 
 PMcoarse (μg/m33.6 6.5 (−1.9; 15.7) 3.6 (−6.1; 14.3) 19.9 (0.4; 43.1) 0.079 5.4 (−16.0; 32.2) 0.446 
 PM2.5 (μg/m32.8 5.8 (−2.4; 14.6) 2.3 (−7.2; 12.7) 18.8 (0.3; 40.7) 0.066 9.5 (−11.5; 35.5) 0.282 
 PM2.5abs (10−5/m) 0.5 1.4 (−5.7; 9.1) −1.2 (−9.3; 7.7) 25.7 (7.1; 47.6) 0.005 −8.4 (−25.4; 12.4) 0.251 
 NOx (μg/m322.9 12.6 (4.2; 21.7) 8.1 (−1.4; 18.5) 34.5 (13.6; 59.3) 0.013 17.9 (−5.3; 46.7) 0.238 
 NO2 (μg/m311.9 15.3 (6.8; 24.5) 10.9 (1.3; 21.5) 29.3 (9.3; 52.9) 0.058 23.7 (−0.1; 53.2) 0.180 
 Traffic intensity on the nearest road (vehicles/day), per 1,000 7.5 −3.8 (−9.6; 2.4) −4.4 (−11.1; 2.7) 4.3 (−9.5; 20.3) 0.141 0.4 (−17.3; 22.0) 0.321 
 Traffic load within 100 m on major roads (vehicles*meter/day), per 10,000 243.6 −2.8 (−8.6; 3.3) −3.0 (−9.6; 4.1) 3.3 (−10.8; 19.6) 0.225 5.0 (−13.2; 27.1) 0.222 
hs-CRP     
 PM10 (μg/m37.9 8.5 (−3.5; 22.0) 3.1 (−9.8; 17.9) 35.7 (2.6; 79.5) 0.041 3.3 (−31.3; 55.3) 0.497 
 PMcoarse (μg/m33.6 2.0 (−9.5; 15.0) −3.9 (−16.3; 10.3) 32.9 (0.5; 75.8) 0.021 −0.5 (−33.9; 49.7) 0.437 
 PM2.5 (μg/m32.8 8.4 (−3.6; 21.8) 5.4 (−8.0; 20.8) 9.7 (−16.1; 43.4) 0.396 21.3 (−17.4; 78.0) 0.250 
 PM2.5abs (10−5/m) 0.5 4.6 (−5.9; 16.3) −0.3 (−11.6; 12.5) 21.7 (−5.6; 57.0) 0.083 15.7 (−19.9; 67.1) 0.226 
 NOx (μg/m322.9 10.8 (−1.1; 24.1) 4.0 (−8.6; 18.4) 35.2 (3.4; 76.8) 0.042 24.2 (−16.4; 84.5) 0.202 
 NO2 (μg/m311.9 11.1 (−0.6; 24.3) 1.1 (−11.1; 14.9) 33.9 (2.7; 74.6) 0.031 44.6 (−1.6; 112.6) 0.042 
 Traffic intensity on the nearest road (vehicles/day), per 1,000 7.5 3.8 (−5.2; 13.7) 0.5 (−9.2; 11.3) 8.3 (−13.8; 36.0) 0.279 27.5 (−10.3; 81.1) 0.101 
 Traffic load within 100 m on major roads (vehicles*meter/day), per 10,000 243.6 −1.2 (−9.4; 7.9) −1.9 (−10.9; 8.1) 6.7 (−15.5; 34.7) 0.257 −4.0 (−32.1; 35.6) 0.452 
Pollutant exposureIncrement
All
Nondiabetes
Prediabetes
Diabetes
% Change (95% CI)% Change (95% CI)% Change (95% CI)P value% Change (95% CI)P value§
HbA1c        
 PM10 (μg/m37.9 0.5 (−0.5; 1.5) 0.3 (−0.4; 1.0) −0.4 (−2.0; 1.2) 0.200 2.7 (−2.6; 8.4) 0.193 
 PMcoarse (μg/m33.6 0.1 (−0.9; 1.1) 0.1 (−0.6; 0.8) −0.2 (−1.8; 1.4) 0.368 1.1 (−4.3; 6.7) 0.365 
 PM2.5 (μg/m32.8 0.9 (−0.1; 1.9) 0.4 (−0.2; 1.1) 0.7 (−0.8; 2.3) 0.360 0.7 (−4.2; 5.8) 0.465 
 PM2.5abs (10−5/m) 0.5 −0.1 (−0.9; 0.8) 0.1 (−0.5; 0.7) −0.4 (−1.9; 1.0) 0.248 0.1 (−4.6; 5.0) 0.494 
 NOx (μg/m322.9 0.0 (−0.9; 1.0) 0.1 (−0.5; 0.8) 0.1 (−1.4; 1.7) 0.498 −0.5 (−5.6; 4.9) 0.413 
 NO2 (μg/m311.9 0.2 (−0.8; 1.1) 0.2 (−0.4; 0.9) 0.3 (−1.3; 1.8) 0.483 −1.3 (−6.3; 3.9) 0.274 
 Traffic intensity on the nearest road (vehicles/day), per 1,000 7.5 −0.3 (−1.1; 0.5) 0.0 (−0.5; 0.5) 0.7 (−0.6; 2.0) 0.161 1.1 (−3.5; 5.9) 0.332 
 Traffic load within 100 m on major roads (vehicles*meter/day), per 10,000 243.6 −0.2 (−0.9; 0.5) −0.3 (−0.7; 0.2) 0.7 (−0.6; 2.0) 0.090 2.6 (−1.9; 7.3) 0.107 
Leptin     
 PM10 (μg/m37.9 7.9 (−0.4; 17.0) 6.6 (−3.1; 17.3) 20.2 (0.7; 43.5) 0.122 −1.8 (−21.7; 23.0) 0.254 
 PMcoarse (μg/m33.6 6.5 (−1.9; 15.7) 3.6 (−6.1; 14.3) 19.9 (0.4; 43.1) 0.079 5.4 (−16.0; 32.2) 0.446 
 PM2.5 (μg/m32.8 5.8 (−2.4; 14.6) 2.3 (−7.2; 12.7) 18.8 (0.3; 40.7) 0.066 9.5 (−11.5; 35.5) 0.282 
 PM2.5abs (10−5/m) 0.5 1.4 (−5.7; 9.1) −1.2 (−9.3; 7.7) 25.7 (7.1; 47.6) 0.005 −8.4 (−25.4; 12.4) 0.251 
 NOx (μg/m322.9 12.6 (4.2; 21.7) 8.1 (−1.4; 18.5) 34.5 (13.6; 59.3) 0.013 17.9 (−5.3; 46.7) 0.238 
 NO2 (μg/m311.9 15.3 (6.8; 24.5) 10.9 (1.3; 21.5) 29.3 (9.3; 52.9) 0.058 23.7 (−0.1; 53.2) 0.180 
 Traffic intensity on the nearest road (vehicles/day), per 1,000 7.5 −3.8 (−9.6; 2.4) −4.4 (−11.1; 2.7) 4.3 (−9.5; 20.3) 0.141 0.4 (−17.3; 22.0) 0.321 
 Traffic load within 100 m on major roads (vehicles*meter/day), per 10,000 243.6 −2.8 (−8.6; 3.3) −3.0 (−9.6; 4.1) 3.3 (−10.8; 19.6) 0.225 5.0 (−13.2; 27.1) 0.222 
hs-CRP     
 PM10 (μg/m37.9 8.5 (−3.5; 22.0) 3.1 (−9.8; 17.9) 35.7 (2.6; 79.5) 0.041 3.3 (−31.3; 55.3) 0.497 
 PMcoarse (μg/m33.6 2.0 (−9.5; 15.0) −3.9 (−16.3; 10.3) 32.9 (0.5; 75.8) 0.021 −0.5 (−33.9; 49.7) 0.437 
 PM2.5 (μg/m32.8 8.4 (−3.6; 21.8) 5.4 (−8.0; 20.8) 9.7 (−16.1; 43.4) 0.396 21.3 (−17.4; 78.0) 0.250 
 PM2.5abs (10−5/m) 0.5 4.6 (−5.9; 16.3) −0.3 (−11.6; 12.5) 21.7 (−5.6; 57.0) 0.083 15.7 (−19.9; 67.1) 0.226 
 NOx (μg/m322.9 10.8 (−1.1; 24.1) 4.0 (−8.6; 18.4) 35.2 (3.4; 76.8) 0.042 24.2 (−16.4; 84.5) 0.202 
 NO2 (μg/m311.9 11.1 (−0.6; 24.3) 1.1 (−11.1; 14.9) 33.9 (2.7; 74.6) 0.031 44.6 (−1.6; 112.6) 0.042 
 Traffic intensity on the nearest road (vehicles/day), per 1,000 7.5 3.8 (−5.2; 13.7) 0.5 (−9.2; 11.3) 8.3 (−13.8; 36.0) 0.279 27.5 (−10.3; 81.1) 0.101 
 Traffic load within 100 m on major roads (vehicles*meter/day), per 10,000 243.6 −1.2 (−9.4; 7.9) −1.9 (−10.9; 8.1) 6.7 (−15.5; 34.7) 0.257 −4.0 (−32.1; 35.6) 0.452 

Data are reported as the presented as percentage change (95% CI) from geometric mean per 5th–95th percentile difference increment in air pollutants adjusted for age, sex, smoking, BMI, waist-to-hip ratio, month of blood withdrawal, and selected socioeconomic and lifestyle variables (see Supplementary Table 2). PM2.5abs, soot content (absorbance) of PM2.5.

Boldface type indicates effect estimates were statistically significant, P < 0.05.

†Corresponds to the difference between the 5th and 95th percentiles of the corresponding exposure.

‡Test of interaction to test for differences between participants with prediabetes and those without diabetes.

§Test of interaction to test for differences between participant with and without diabetes.

For individuals without diabetes, exposure to NOx was associated with HOMA-IR, glucose, and insulin, whereas NO2 was associated with leptin. Among people with diabetes, the only significant association was seen between traffic load on major roads within 100 m from the residence and glucose levels.

The group of individuals with prediabetes yielded the strongest association with highest effect estimates. In this group, HOMA-IR was associated with all air pollutants except PM2.5 and traffic load on major roads within 100 m. With HOMA-IR being the product of fasting glucose and insulin, regression results between air pollutants and insulin mainly replicated the associations found for HOMA-IR, whereas glucose was associated only with traffic load on major roads. Further associations were seen for leptin with all air pollutants (Table 5). Results in the prediabetes subgroup may indicate some underlying systemic inflammatory processes. Our effect estimates for hs-CRP pointed in this direction because most air pollutants were significantly associated with higher percentage changes in hs-CRP in individuals with prediabetes (Table 5).

The investigation of sex as effect modifier did not reveal any significant differences between men and women (data not shown).

Sensitivity Analyses

The reduction of covariate adjustment led to similar results (Supplementary Tables 4 and 5). We checked the linearity of the dose-response function by including the air pollutants as cubic regression splines exemplarily for HOMA-IR in the prediabetes group. All air pollutants indicated no clear deviation from linearity when incorporated as a smooth function in the model (Supplementary Fig. 1). Also, the categorical analyses comparing the quartiles of exposure generally confirmed the linear trend except for PMcoarse and PM2.5 in the prediabetes group, and the diabetes group that showed no clear pattern, potentially due to the reduced power (Supplementary Fig. 2). The alternative quantile regression indicated no clear heterogeneity across the deciles (Supplementary Fig. 3A–D). When excluding participants with hs-CRP >10 mg/L (n = 97), results were robust for all participants (data not shown). In the prediabetes group, effect estimates were slightly attenuated for HOMA-IR, insulin, and leptin. For the latter, the association with PM10, PMcoarse, and PM2.5 was not significant anymore. For HOMA-IR, the PM2.5 estimate was slightly higher and significant for individuals with prediabetes and individuals with diabetes (data not shown).

The exclusion of 155 participants who reported an intake of antidiabetic medication showed quite robust estimates for all participants (Fig. 1, exemplarily shown for NO2). Estimates for the diabetes group changed considerably because the exclusion affected almost half of the participants, resulting in lower estimates for glucose and HbA1c levels, but higher estimates for HOMA-IR, insulin, leptin, and hs-CRP levels. The exclusion of 352 participants taking statins slightly attenuated the estimates for HOMA-IR, glucose, and insulin levels (Supplementary Fig. 4, exemplarily shown for NO2). Leptin and hs-CRP estimates were generally robust but were higher for the diabetes group, resulting in significant estimates for NO2.

Figure 1

Association between NO2 and biomarkers presented as the percentage change (%-change) (95% CI) from geometric mean per 11.9 µg/m3 increment in NO2 for all participants (top) and individuals with diabetes (bottom). Squares, NO2 estimates for the whole group; circles, NO2 estimates for participants without antidiabetic medication intake. Models were adjusted for age, sex, smoking, BMI, waist-to-hip ratio, month of blood withdrawal, and selected socioeconomic and lifestyle variables (see Supplementary Table 2). T2D, type 2 diabetes.

Figure 1

Association between NO2 and biomarkers presented as the percentage change (%-change) (95% CI) from geometric mean per 11.9 µg/m3 increment in NO2 for all participants (top) and individuals with diabetes (bottom). Squares, NO2 estimates for the whole group; circles, NO2 estimates for participants without antidiabetic medication intake. Models were adjusted for age, sex, smoking, BMI, waist-to-hip ratio, month of blood withdrawal, and selected socioeconomic and lifestyle variables (see Supplementary Table 2). T2D, type 2 diabetes.

The exclusion of 762 study participants taking diuretics and/or β-blockers (with 306 who reported an intake of both) showed robust effect estimates for all participants, but showed attenuated estimates for glucose as well as for glucose and insulin in individuals without diabetes. In the prediabetes group, effect estimates for leptin and hs-CRP were not significant anymore (data not shown). The stratification of the prediabetes subgroup into individuals with i-IFG, i-IGT, or IFG-IGT showed in general higher percentage changes for persons with i-IFG and i-IGT, especially for HOMA-IR, insulin, and hs-CRP (Fig. 2, exemplarily shown for HOMA-IR). For leptin, only participants with i-IGT showed an association (data not shown). The stratification of the diabetes subgroup into participants with newly diagnosed diabetes and individuals with known type 2 diabetes for HOMA-IR showed significant associations for almost all exposures for the first strata but no association for the second strata (Fig. 3).

Figure 2

Associations between air pollutants, traffic indicators, and HOMA-IR presented as the percentage change (%-change) (95% CI) from geometric mean per 5th–95th percentile difference in air pollutants among participants with prediabetes stratified by subgroup of i-IFG (squares, N = 110), i-IGT (circles, N = 298), and IFG-IGT (triangles, N = 73). N represents the number of observation finally used for the analysis with HOMA-IR. Models were adjusted for age, sex, smoking, BMI, waist-to-hip ratio, month of blood withdrawal, and pack-years smoked. PM2.5abs, soot content (absorbance) of PM2.5; traffic major, traffic load within 100 m on major roads; traffic near, traffic intensity on the nearest road.

Figure 2

Associations between air pollutants, traffic indicators, and HOMA-IR presented as the percentage change (%-change) (95% CI) from geometric mean per 5th–95th percentile difference in air pollutants among participants with prediabetes stratified by subgroup of i-IFG (squares, N = 110), i-IGT (circles, N = 298), and IFG-IGT (triangles, N = 73). N represents the number of observation finally used for the analysis with HOMA-IR. Models were adjusted for age, sex, smoking, BMI, waist-to-hip ratio, month of blood withdrawal, and pack-years smoked. PM2.5abs, soot content (absorbance) of PM2.5; traffic major, traffic load within 100 m on major roads; traffic near, traffic intensity on the nearest road.

Figure 3

Association between air pollutants, traffic indicators, and HOMA-IR presented as the percentage change (%-change) (95% CI) from geometric mean per 5th–95th percentile difference in air pollutants among participants with diabetes stratified by subgroup of participants with newly diagnosed type 2 diabetes (squares, N = 109) and known type 2 diabetes (circles, N = 205). N represents the number of observations finally used for the analysis with HOMA-IR. Models were adjusted for age, sex, smoking, BMI, waist-to-hip ratio, month of blood withdrawal, and pack-years smoked. PM2.5abs, soot content (absorbance) of PM2.5; T2D, type 2 diabetes; traffic major, traffic load within 100 m on major roads; traffic near, traffic intensity on the nearest road.

Figure 3

Association between air pollutants, traffic indicators, and HOMA-IR presented as the percentage change (%-change) (95% CI) from geometric mean per 5th–95th percentile difference in air pollutants among participants with diabetes stratified by subgroup of participants with newly diagnosed type 2 diabetes (squares, N = 109) and known type 2 diabetes (circles, N = 205). N represents the number of observations finally used for the analysis with HOMA-IR. Models were adjusted for age, sex, smoking, BMI, waist-to-hip ratio, month of blood withdrawal, and pack-years smoked. PM2.5abs, soot content (absorbance) of PM2.5; T2D, type 2 diabetes; traffic major, traffic load within 100 m on major roads; traffic near, traffic intensity on the nearest road.

We examined the association between residential long-term exposure to air pollutants and traffic indicators on biomarkers related to IR, subclinical inflammation, and adipokines in a cross-sectional study conducted in the region of Augsburg, in southern Germany. Among all study participants, we found a positive association between PM10, PMcoarse, PM2.5 absorbance, NOx, and NO2, and HOMA-IR and insulin levels. Furthermore, NO2 was significantly associated with glucose and leptin. When stratifying by glucose tolerance, most pollutants were statistically significant in association with HOMA-IR, insulin, leptin, and hs-CRP in the prediabetes subgroup. Individuals with or without diabetes showed, rather, no or only weak associations between air pollution and blood markers.

Sensitivity analyses suggested in general robust results for all participants and for most of the subgroup analyses. However, medication intake seemed to play a complex role, especially for the diabetes subgroup. Thus, the exclusion of participants taking glucose-lowering medication (diabetes group only) led to higher air pollution estimates for HOMA-IR, insulin, leptin, and hs-CRP levels, but lower estimates for glucose and HbA1c levels. Estimates in this group were also higher for leptin and hs-CRP levels when excluding persons taking statins, whereas results were robust for the group without diabetes and the prediabetes group. This might indicate a mitigating or inhibiting role of this medication type with regard to inflammatory effects of air pollution in patients with diabetes, whereas healthy individuals may not be as susceptible to inflammation as individuals with metabolic disorders. Also, the stratification into participants with newly diagnosed diabetes and those with known type 2 diabetes pointed in this direction, suggesting an increased susceptibility of the first subgroup to air pollution exposure that has not been properly medicated. The exclusion of individuals with diuretic and/or β-blocker intake mainly affected the results of the prediabetes group, leading to nonsignificant estimates for all air pollutants in association with leptin and hs-CRP levels.

The first studies on adverse the health effects of ambient air pollution mainly looked at respiratory outcomes and somewhat later on cardiovascular outcomes (39,40). Recent research also suggested a link between air pollution and type 2 diabetes involving multiple pathophysiological pathways (11,12,15). Several reviews and meta-analyses (1420) have been published since, mainly referring to the same studies. However, the pooled effect estimates varied to some extent, depending on the inclusion and exclusion criteria. As the number of eligible studies is quite sparse, the meta-analyses usually combined prevalent and incident diabetes and could not distinguish between type 1 and type 2 diabetes. Thus, clear evidence is still limited because of differences in outcome definition, exposure metrics, population characteristics, and the covariates that were considered (14). As IR is a powerful predictor of the future development of type 2 diabetes, it came into focus in several recent epidemiological studies on air pollution (10,2427,41). However, these studies either investigated short-term effects or focused on children, and, thus, are not directly comparable to our study. Short- and long-term effects of air pollution are hypothesized to arise partly from different biological pathways, and recent epidemiological evidence showed that the adverse health effects of long-term exposure are generally larger than those observed for short-term exposure (42). In addition, studies among children might be more pronounced due to the children being more vulnerable to environmental stressors. To the best of our knowledge, this is the first study investigating the long-term effects of air pollution in association with biomarkers of IR in the general population.

A German study by Teichert et al. (41) used data that were partly comparable to ours; however, they applied a different approach to assess the association between long-term exposure to air pollution, subclinical inflammation, and impaired glucose metabolism in 363 women. The authors stratified the women by impaired (defined as i-IFG or previous diagnosis of type 2 diabetes by a physician) versus normal glucose metabolism and compared the risk differences in association with air pollution, 14 proinflammatory and anti-inflammatory immune mediators, and fasting glucose and insulin levels. The authors reported higher odds ratios for NO2 and NOx, but not for the PM fractions. Among all exposures investigated in our study, NO2 and NOx effect estimates were the most consistent and most pronounced, although we also observed an association for the coarse PM fraction. Similar to our analysis, the study by Thiering et al. (26) was based on long-term exposure estimates from the ESCAPE project. The authors looked at HOMA-IR in 397 10-year-old children in two prospective German birth cohort studies and observed increments of 17.0% (95% CI 5.0; 30.3) and 18.7% (95% CI 2.9; 36.9) for increments of 10.6 μg/m3 in ambient NO2 and 6.0 μg/m3 in PM10. Calderón-Garcidueñas et al. (27) matched 54 children living in the metropolitan area of Mexico City, who were thus chronically exposed to PM2.5 and O3 concentrations above the standards, with 26 control children and found significantly higher leptin and glucose levels in the first group, but no differences for insulin and HOMA-IR. Three studies on the short-term effects of air pollution conducted in highly selected populations reported higher levels of HOMA-IR in association with NO2, PM10, and PM2.5 (10,24,25).

Regarding long-term air pollution and the incidence of type 2 diabetes, only five articles have been published so far (4347). Although two studies from the U.S. (43,44) did not observe an association between diabetes incidence and long-term air pollution, Coogan et al. (47) reported a significant risk increase in association with NOx, but not with PM2.5, in a cohort of black women living in Los Angeles. A similar pattern was seen in our study, with rather significant estimates for NOx, but rarely for PM2.5. Also, a prospective study among women from the highly industrialized Ruhr district (western Germany) observed stronger associations with NO2 than with PM10 (45). In addition, significant associations were seen for PM2.5 absorbance and proximity to major roads, which were also associated with elevated levels of markers related to IR in our analysis. A further study from the Ruhr area (46) reported higher effect estimates for PM10 compared with PM2.5, which was similar to our results.

Biological Mechanism

Potential pathways between air pollution and adverse cardiometabolic changes may occur through a multitude of mechanisms. Liu et al. (11) compiled a comprehensive evaluation of potentially underlying biological mechanisms linking air pollution and IR/type 2 diabetes, including pulmonary and systemic inflammation, endoplasmic reticulum stress, alterations in glucose and lipid metabolism, and activation of the central nervous system, among others. Alveolar macrophages and bronchial epithelial cells are initial cellular sensors of PM. These sensors do not react on PM per se but more on biological components intrinsic to PM, such as lipopolysaccharide (LPS), which has a lower concentration in PM2.5 than in PM10 (11). LPS-binding protein, a soluble acute-phase protein that binds to bacterial LPS, has been found to be associated with obesity, metabolic syndrome, and type 2 diabetes (48). This might be a possible explanation for the higher effect estimates we and others observed for PM10 and PMcoarse compared with PM2.5.

Strengths and Limitations

The major strengths of this study are the well-characterized nature of the KORA F4 cohort, the standardized and comprehensive estimation of residential air pollution exposure, and the availability of OGTT measurements to allow for stratification by impaired glucose regulation. Thus, the study delivered a high degree of representativeness in terms of a large number of study participants to conduct subgroup analyses and a large scale of information on patient characteristics for the examination of potential confounding.

The cross-sectional study design limits our study findings in a way that we have one-time measurements giving no indication of the sequence of events. The observed elevation of IR-related biomarkers at one time point may have occurred before the onset of adverse health effects due to air pollution. Since biomarkers were determined up to 3 years before the air pollution measurements, it is not possible to infer causation based on our associations. However, we are investigating the spatial contrasts of air pollution. Several studies could show that spatial contrasts remained stable for periods of up to 10 years and longer, even with decreases in concentrations over time (49,50). Thus, we believe that our LUR models based on measurements from the years 2008/2009 are not necessarily restricted to this period but may be also valid predictors of the historic spatial contrasts. With the application of LUR models to estimate the residential long-term concentrations, we cannot rule out the possibility of exposure misclassification. However, we assume that the measurement error biases our effect estimates toward the null.

Furthermore, disparities in the inferences that can be drawn from IR measures in people with or without β-cell failure might have limited the comparability of the results.

Conclusion

In conclusion, our results point to an association between traffic-related air pollution and biomarkers related to IR, subclinical inflammation, and adipokines in the general population. The effect estimates were remarkably high for individuals with i-IFG, i-IGT, or both, suggesting that this subgroup is particularly susceptible to adverse health effects due to air pollution exposure.

Appendix

Members of the KORA Study Group. A. Peters (speaker), J. Heinrich, R. Holle, R. Leidl, C. Meisinger, K. Strauch, and their coworkers, who are responsible for the design and conduct of the KORA studies.

Funding. This research has been supported by funding from the European Community Seventh Framework Program (FP7/2007‐2011) under grant agreement 211250. The KORA research platform is financed by the Helmholtz Zentrum München, German Research Center for Environmental Health (GmbH), which is funded by the German Federal Ministry of Education, Science, Research and Technology and by the State of Bavaria. Additional support was obtained from the Ministry of Science and Research of the State of North Rhine-Westphalia and by the German Federal Ministry of Health. Further support for this study was obtained by a grant from the German Federal Ministry of Education and Research to the German Center for Diabetes Research.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. K.W. and A.Po. performed the analyses and wrote the manuscript. A.S., S.B., R.H., W.R., C.H., M.R., W.K., and C.M. contributed to the design of the study and reviewed and edited the manuscript. A.Pe. researched the data, conceived the research, provided overall supervision, and reviewed and edited the manuscript. K.W. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

1.
Cameron
AJ
,
Magliano
DJ
,
Zimmet
PZ
, et al
.
The metabolic syndrome as a tool for predicting future diabetes: the AusDiab study
.
J Intern Med
2008
;
264
:
177
186
[PubMed]
2.
Morris
DH
,
Khunti
K
,
Achana
F
, et al
.
Progression rates from HbA1c 6.0-6.4% and other prediabetes definitions to type 2 diabetes: a meta-analysis
.
Diabetologia
2013
;
56
:
1489
1493
[PubMed]
3.
Weyer
C
,
Tataranni
PA
,
Bogardus
C
,
Pratley
RE
.
Insulin resistance and insulin secretory dysfunction are independent predictors of worsening of glucose tolerance during each stage of type 2 diabetes development
.
Diabetes Care
2001
;
24
:
89
94
[PubMed]
4.
Meigs
JB
,
Rutter
MK
,
Sullivan
LM
,
Fox
CS
,
D’Agostino
RB
 Sr
,
Wilson
PW
.
Impact of insulin resistance on risk of type 2 diabetes and cardiovascular disease in people with metabolic syndrome
.
Diabetes Care
2007
;
30
:
1219
1225
[PubMed]
5.
Tabák
AG
,
Herder
C
,
Rathmann
W
,
Brunner
EJ
,
Kivimäki
M
.
Prediabetes: a high-risk state for diabetes development
.
Lancet
2012
;
379
:
2279
2290
[PubMed]
6.
International Diabetes Federation
.
IDF Diabetes Atlas
. 6th ed.
Brussels, Belgium
,
International Diabetes Federation
,
2013
7.
Kahn
SE
,
Hull
RL
,
Utzschneider
KM
.
Mechanisms linking obesity to insulin resistance and type 2 diabetes
.
Nature
2006
;
444
:
840
846
[PubMed]
8.
Olefsky
JM
,
Glass
CK
.
Macrophages, inflammation, and insulin resistance
.
Annu Rev Physiol
2010
;
72
:
219
246
[PubMed]
9.
Herder
C
,
Roden
M
.
Genetics of type 2 diabetes: pathophysiologic and clinical relevance
.
Eur J Clin Invest
2011
;
41
:
679
692
[PubMed]
10.
Kelishadi
R
,
Mirghaffari
N
,
Poursafa
P
,
Gidding
SS
.
Lifestyle and environmental factors associated with inflammation, oxidative stress and insulin resistance in children
.
Atherosclerosis
2009
;
203
:
311
319
[PubMed]
11.
Liu
C
,
Ying
Z
,
Harkema
J
,
Sun
Q
,
Rajagopalan
S
.
Epidemiological and experimental links between air pollution and type 2 diabetes
.
Toxicol Pathol
2013
;
41
:
361
373
[PubMed]
12.
Rajagopalan
S
,
Brook
RD
.
Air pollution and type 2 diabetes: mechanistic insights
.
Diabetes
2012
;
61
:
3037
3045
[PubMed]
13.
Peters
A
.
Epidemiology: air pollution and mortality from diabetes mellitus
.
Nat Rev Endocrinol
2012
;
8
:
706
707
[PubMed]
14.
Esposito
K
,
Petrizzo
M
,
Maiorino
MI
,
Bellastella
G
,
Giugliano
D
.
Particulate matter pollutants and risk of type 2 diabetes: a time for concern?
Endocrine
2016
;
51
:
32
37
[PubMed]
15.
Rao
X
,
Montresor-Lopez
J
,
Puett
R
,
Rajagopalan
S
,
Brook
RD
.
Ambient air pollution: an emerging risk factor for diabetes mellitus
.
Curr Diab Rep
2015
;
15
:
603
[PubMed]
16.
Balti
EV
,
Echouffo-Tcheugui
JB
,
Yako
YY
,
Kengne
AP
.
Air pollution and risk of type 2 diabetes mellitus: a systematic review and meta-analysis
.
Diabetes Res Clin Pract
2014
;
106
:
161
172
[PubMed]
17.
Wang
B
,
Xu
D
,
Jing
Z
,
Liu
D
,
Yan
S
,
Wang
Y
.
Effect of long-term exposure to air pollution on type 2 diabetes mellitus risk: a systemic review and meta-analysis of cohort studies
.
Eur J Endocrinol
2014
;
171
:
R173
R182
.
18.
Li
C
,
Fang
D
,
Xu
D
,
Wang
B
,
Zhao
S
,
Yan
S
,
Wang
Y
.
Main air pollutants and diabetes-associated mortality: a systematic review and meta-analysis
.
Eur J Endocrinol
2014
;
171
:
R183
R190
.
19.
Janghorbani
M
,
Momeni
F
,
Mansourian
M
.
Systematic review and metaanalysis of air pollution exposure and risk of diabetes
.
Eur J Epidemiol
2014
;
29
:
231
242
[PubMed]
20.
Eze
IC
,
Hemkens
LG
,
Bucher
HC
, et al
.
Association between ambient air pollution and diabetes mellitus in Europe and North America: systematic review and meta-analysis
.
Environ Health Perspect
2015
;
123
:
381
389
[PubMed]
21.
Kodavanti
UP
.
Air pollution and insulin resistance: do all roads lead to Rome?
Diabetes
2015
;
64
:
712
714
[PubMed]
22.
Sun
Q
,
Yue
P
,
Deiuliis
JA
, et al
.
Ambient air pollution exaggerates adipose inflammation and insulin resistance in a mouse model of diet-induced obesity
.
Circulation
2009
;
119
:
538
546
[PubMed]
23.
Xu
X
,
Liu
C
,
Xu
Z
, et al
.
Long-term exposure to ambient fine particulate pollution induces insulin resistance and mitochondrial alteration in adipose tissue
.
Toxicol Sci
2011
;
124
:
88
98
[PubMed]
24.
Kim
JH
,
Hong
YC
.
GSTM1, GSTT1, and GSTP1 polymorphisms and associations between air pollutants and markers of insulin resistance in elderly Koreans
.
Environ Health Perspect
2012
;
120
:
1378
1384
[PubMed]
25.
Brook
RD
,
Xu
X
,
Bard
RL
, et al
.
Reduced metabolic insulin sensitivity following sub-acute exposures to low levels of ambient fine particulate matter air pollution
.
Sci Total Environ
2013
;
448
:
66
71
[PubMed]
26.
Thiering
E
,
Cyrys
J
,
Kratzsch
J
, et al
.
Long-term exposure to traffic-related air pollution and insulin resistance in children: results from the GINIplus and LISAplus birth cohorts
.
Diabetologia
2013
;
56
:
1696
1704
[PubMed]
27.
Calderón-Garcidueñas
L
,
Franco-Lira
M
,
D’Angiulli
A
, et al
.
Mexico City normal weight children exposed to high concentrations of ambient PM2.5 show high blood leptin and endothelin-1, vitamin D deficiency, and food reward hormone dysregulation versus low pollution controls. Relevance for obesity and Alzheimer disease
.
Environ Res
2015
;
140
:
579
592
[PubMed]
28.
Blüher
M
.
Clinical relevance of adipokines
.
Diabetes Metab J
2012
;
36
:
317
327
[PubMed]
29.
Meisinger
C
,
Strassburger
K
,
Heier
M
, et al
.
Prevalence of undiagnosed diabetes and impaired glucose regulation in 35-59-year-old individuals in Southern Germany: the KORA F4 study
.
Diabet Med
2010
;
27
:
360
362
.
30.
World Health Organisation (WHO)
.
Definition, Diagnosis and Classification of Diabetes Mellitus and its Complications. Part 1: Diagnosis and Classification of Diabetes Mellitus
.
Geneva, Switzerland
,
World Health Organization
,
1999
31.
Rathmann
W
,
Haastert
B
,
Icks
A
, et al
.
High prevalence of undiagnosed diabetes mellitus in Southern Germany: target populations for efficient screening. The KORA survey 2000
.
Diabetologia
2003
;
46
:
182
189
[PubMed]
32.
Matthews
DR
,
Hosker
JP
,
Rudenski
AS
,
Naylor
BA
,
Treacher
DF
,
Turner
RC
.
Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man
.
Diabetologia
1985
;
28
:
412
419
[PubMed]
33.
Eeftens
M
,
Beelen
R
,
de Hoogh
K
, et al
.
Development of land use regression models for PM(2.5), PM(2.5) absorbance, PM(10) and PM(coarse) in 20 European study areas; results of the ESCAPE project
.
Environ Sci Technol
2012
;
46
:
11195
11205
[PubMed]
34.
Beelen
R
,
Hoek
G
,
Vienneau
D
, et al
.
Development of NO2 and NOx land use regression models for estimating air pollution exposure in 36 study areas in Europe—the ESCAPE project
.
Atmos Environ
2013
;
72
:
10
23
35.
Altman
DG
,
Bland
JM
.
Interaction revisited: the difference between two estimates
.
BMJ
2003
;
326
:
219
[PubMed]
36.
Dronavalli
S
,
Bakris
GL
.
Mechanistic insights into diuretic-induced insulin resistance
.
Hypertension
2008
;
52
:
1009
1011
[PubMed]
37.
Jacob
S
,
Balletshofer
B
,
Henriksen
EJ
, et al
.
Beta-blocking agents in patients with insulin resistance: effects of vasodilating beta-blockers
.
Blood Press
1999
;
8
:
261
268
[PubMed]
38.
World Health Organization. Air quality guidelines—global update 2005 [article online], 2005. Available from http://www.who.int/phe/health_topics/outdoorair/outdoorair_aqg/en/index.html. Accessed 4 March 2014
39.
Rückerl
R
,
Schneider
A
,
Breitner
S
,
Cyrys
J
,
Peters
A
.
Health effects of particulate air pollution: a review of epidemiological evidence
.
Inhal Toxicol
2011
;
23
:
555
592
[PubMed]
40.
Pope
CA
 3rd
,
Dockery
DW
.
Health effects of fine particulate air pollution: lines that connect
.
J Air Waste Manag Assoc
2006
;
56
:
709
742
[PubMed]
41.
Teichert
T
,
Vossoughi
M
,
Vierkötter
A
, et al
.
Association between traffic-related air pollution, subclinical inflammation and impaired glucose metabolism: results from the SALIA study
.
PLoS One
2013
;
8
:
e83042
[PubMed]
42.
World Health Organization, Regional Office for Europe. Review of evidence on health aspects of air pollution—REVIHAAP project: final technical report [article online], 2013. Available from http://www.euro.who.int/en/health-topics/environment-and-health/air-quality/publications/2013/review-of-evidence-on-health-aspects-of-air-pollution-revihaap-project-final-technical-report. Accessed 22 September 2016
43.
Puett
RC
,
Hart
JE
,
Schwartz
J
,
Hu
FB
,
Liese
AD
,
Laden
F
.
Are particulate matter exposures associated with risk of type 2 diabetes?
Environ Health Perspect
2011
;
119
:
384
389
[PubMed]
44.
Park
SK
,
Adar
SD
,
O’Neill
MS
, et al
.
Long-term exposure to air pollution and type 2 diabetes mellitus in a multiethnic cohort
.
Am J Epidemiol
2015
;
181
:
327
336
[PubMed]
45.
Krämer
U
,
Herder
C
,
Sugiri
D
, et al
.
Traffic-related air pollution and incident type 2 diabetes: results from the SALIA cohort study
.
Environ Health Perspect
2010
;
118
:
1273
1279
[PubMed]
46.
Weinmayr
G
,
Hennig
F
,
Fuks
K
, et al.;
Heinz Nixdorf Recall Investigator Group
.
Long-term exposure to fine particulate matter and incidence of type 2 diabetes mellitus in a cohort study: effects of total and traffic-specific air pollution
.
Environ Health
2015
;
14
:
53
[PubMed]
47.
Coogan
PF
,
White
LF
,
Jerrett
M
, et al
.
Air pollution and incidence of hypertension and diabetes mellitus in black women living in Los Angeles
.
Circulation
2012
;
125
:
767
772
[PubMed]
48.
Sun
L
,
Yu
Z
,
Ye
X
, et al
.
A marker of endotoxemia is associated with obesity and related metabolic disorders in apparently healthy Chinese
.
Diabetes Care
2010
;
33
:
1925
1932
[PubMed]
49.
Eeftens
M
,
Beelen
R
,
Fischer
P
,
Brunekreef
B
,
Meliefste
K
,
Hoek
G
.
Stability of measured and modelled spatial contrasts in NO(2) over time
.
Occup Environ Med
2011
;
68
:
765
770
[PubMed]
50.
Wang
R
,
Henderson
SB
,
Sbihi
H
,
Allen
RW
,
Brauer
M
.
Temporal stability of land use regression models for traffic-related air pollution
.
Atmos Environ
2013
;
64
:
312
319
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at http://www.diabetesjournals.org/content/license.

Supplementary data