A common functional variant on the pro-inflammatory Interleukin-6 gene may modify the association between long-term PM10 exposure and diabetes

Table 3 Associations and interactions between PM₁₀ and candidate SNPs on odds of diabetes, applying inverse probability weighting (IPW) to account for potential selection bias

Genotype	Genotype-specific PM₁₀ and diabetes association OR (95 % CI)	P-value^*	P-value of interaction^**	P-value of interaction^***	P-value of interaction^****
IL6 -572G > C
Adjusted model without IPW
GG	1.53 (1.22, 1.92)	<0.001	0.031	n.d.	0.058
GC + CC	0.87 (0.51, 1.49)	0.618
Adjusted model applying IPW
GG	1.53 (1.23, 1.91)	<0.001	0.003	n.d.	0.006
GC + CC	0.74 (0.46, 1.20)	0.225
IL6 -174G > C
Adjusted model without IPW
GG	1.49 (1.09, 2.04)	0.012	0.763	0.966	0.645
GC	1.35 (1.01, 1.80)	0.046
CC	1.43 (0.80, 2.54)	0.226
Adjusted model applying IPW
GG	1.44 (1.07, 1.93)	0.016	0.847	0.955	0.749
GC	1.34 (1.01, 1.78)	0.044
CC	1.41 (0.76, 2.61)	0.226

Adjusted models include age, sex, educational attainment, neighborhood-level socio-economic status, smoking status, pack-years of smoking, exposure to passive smoke and occupational dusts and fumes, dietary fibre intake, alcohol consumption, physical activity, body mass index (BMI), PM₁₀. Study area was treated as random effects in all models. OR: odds ratio; CI: confidence intervals; OR values represent percent increase in odds of diabetes per 10 μg/m³ increase in PM₁₀ exposure. PM₁₀: particulate matter <10 μm in diameter.^* P-value of genotype specific association between PM₁₀ and diabetes.^** additive model (per G allele); ^***dominant model (GG + GC vs. CC); ^****recessive model (GG vs. GC + CC); n.d.: not done due to the few number of CC allele carriers

ISSN: 1476-069X