Information on 279 Single Nucleotide Polymorphisms (SNPs) from the genome-wide association study on lung function and chronic obstructive lung disease by Shrine et al. (2019; n = 400,102). 278 SNPs were included in our calculation of the polygenic risk score. Marked SNPs (*) were also included in our polygenic risk score of sentinel SNPs belonging to causal genes (see Additional file 2: Table S10). Annotation data of the 278 SNPs included in our calculation of the polygenic risk score such as CHROM chromosome, rsID reference SNP cluster ID, POS reference position, REF reference allele, ALT alternative non-reference allele, SNP CHROM:POS:REF:ALT, MAF minor allele frequency in the specific cohort as the second most common allele count from the number of alleles in called genotypes in the specific cohort, TYPED indicates if the variant was genotyped or imputed, R2 imputation quality as the estimated value of the squared correlation between imputed genotypes and true/unobserved genotypes, ER2 empirical R2 for genotyped variants (not calculated for imputed variants), * = SNPs included in our polygenic risk score of sentinel SNPs belonging to causal genes. Table S2. Descriptive statistics on each study and model sample, airway inflammatory biomarker levels and air pollution exposures in the SALIA cohort. Descriptive statistics on each study and model sample using arithmetic and geometric mean and standard deviation of airway inflammatory biomarkers (tumor necrosis factor-α, leukotriene B4, and the sum of eosinophils, macrophages, neutrophils and epithelial cells in induced sputum), study characteristics including mean age, body mass index, education, smoking, indoor air pollution, and chronic inflammatory respiratory condition defined as any condition of asthma, chronic bronchitis, hay fever, cough, cough with sputum or chronic obstructive pulmonary disease, and median and interquartile ranges of chronic air pollution exposure of nitrogen dioxide, nitrogen oxides, particulate matter with aerodynamic diameters ≤ 2.5/ ≤ 10/2.5–10 µm, reflectance of PM2.5 filters calculated as the mean of the annual average concentrations from baseline and first follow-up examinations within a time window of 15 years before the biomarker assessments, statistically centred across the participants. Table S3. Environmental main effects of chronic air pollution exposure and the polygenic main effects on natural log-transformed airway inflammatory biomarker level in elderly women using adjusted linear regression models in test dataset. Environmental main effects as the effect of chronic air pollution exposure (nitrogen dioxide, nitrogen oxides, particulate matter with aerodynamic diameters of ≤ 2.5/ ≤ 10/2.5–10 µm, reflectance of PM2.5 filters calculated as the mean of the annual average concentrations from baseline and first follow-up examinations within a time window of 15 years prior to the biomarker assessments, statistically centred across the participants, standardized using interquartile ranges) on airway inflammatory biomarkers (tumor necrosis factor-α, leukotriene B4, and the sum of eosinophils, macrophages, neutrophils and epithelial cells in induced sputum) adjusted for: age, body mass index (BMI in kg/m2), highest education of the participant or her spouse (low < 10 years, medium = 10 years, high > 10 years of education), ever-/never-smoking, and second-hand smoking using adjusted linear regression models in test dataset: beta estimate with 95% confidence intervals on natural log-transformed airway inflammatory biomarker level and percentage change with 95% confidence interval in airway inflammatory biomarker level. P-values < 0.05 are highlighted bold and p-values < 0.1 cursive. Polygenic main effects as the effect of polygenic risk score (normally distributed (Shapiro–Wilk normality test: n = 194, p-value = 0.002) on airway inflammatory biomarkers (tumor necrosis factor-α, leukotriene B4, and the sum of eosinophils, macrophages, neutrophils and epithelial cells in induced sputum) adjusted for: age, body mass index (BMI in kg/m2), highest education of the participant or her spouse (low < 10 years, medium = 10 years, high > 10 years of education), ever-/never-smoking, and second-hand smoking using adjusted linear regression models in test dataset: beta estimate with 95% confidence intervals on natural log-transformed airway inflammatory biomarker level and percentage change with 95% confidence interval in airway inflammatory biomarker level. P-values < 0.05 are highlighted bold and p-values < 0.1 cursive. The polygenic weights are gained from the interaction terms between each SNP and the air pollution exposure using elastic net regression models, hence it results one polygenic main effect with each air pollutant per airway inflammatory biomarker. Table S4. The environmental main effects of chronic air pollution exposure and the polygenic main effects on natural log-transformed airway inflammatory biomarker level in elderly women using adjusted linear regression models (without adjustment for body mass index) in test dataset. Environmental main effects as the effect of chronic air pollution exposure (nitrogen dioxide, nitrogen oxides, particulate matter with aerodynamic diameters of ≤ 2.5/ ≤ 10/2.5–10 µm, reflectance of PM2.5 filters calculated as the mean of the annual average concentrations from baseline and first follow-up examinations within a time window of 15 years prior to the biomarker assessments, statistically centred across the participants, standardized using interquartile ranges) on airway inflammatory biomarkers (tumor necrosis factor-α, leukotriene B4, and the sum of eosinophils, macrophages, neutrophils and epithelial cells in induced sputum) adjusted for: age, highest education of the participant or her spouse (low < 10 years, medium = 10 years, high > 10 years of education), ever-/never-smoking, and second-hand smoking using adjusted linear regression models in test dataset: beta estimate with 95% confidence intervals on natural log-transformed airway inflammatory biomarker level and percentage change with 95% confidence interval in airway inflammatory biomarker level. P-values < 0.05 are highlighted bold and p-values < 0.1 cursive. Polygenic main effects as the effect of polygenic risk score (normally distributed (Shapiro–Wilk normality test: n = 194, p-value = 0.002) on airway inflammatory biomarkers (tumor necrosis factor-α, leukotriene B4, and the sum of eosinophils, macrophages, neutrophils and epithelial cells in induced sputum) adjusted for: age, highest education of the participant or her spouse (low < 10 years, medium = 10 years, high > 10 years of education), ever-/never-smoking, and second-hand smoking using adjusted linear regression models in test dataset: beta estimate with 95% confidence intervals on natural log-transformed airway inflammatory biomarker level and percentage change with 95% confidence interval in airway inflammatory biomarker level. P-values < 0.05 are highlighted bold and p-values < 0.1 cursive. The polygenic weights are gained from the interaction terms between each SNP and the air pollution exposure using elastic net regression models, hence it results one polygenic main effect with each air pollutant per airway inflammatory biomarker. Table S5. Gene-environment interaction effects between the weighted polygenic risk score and chronic air pollution exposure on natural log-transformed airway inflammatory biomarker levels in elderly women using adjusted linear regression models in test dataset. Gene-environment interaction effects between the weighted polygenic risk score (derived by the interaction-training approach, standardized using interquartile ranges) and chronic air pollution exposure (nitrogen dioxide, nitrogen oxides, particulate matter with aerodynamic diameters of ≤ 2.5/ ≤ 10/2.5–10 µm, reflectance of PM2.5 filters calculated as the mean of the annual average concentrations from baseline and first follow-up examinations within a time window of 15 years prior to the biomarker assessments, statistically centred across the participants, standardized using interquartile ranges) on natural log-transformed airway inflammatory biomarker levels (tumor necrosis factor-α, leukotriene B4, and the sum of eosinophils, macrophages, neutrophils and epithelial cells in induced sputum) adjusted for: age, body mass index (BMI in kg/m2), highest education of the participant or her spouse (low < 10 years, medium = 10 years, high > 10 years of education), ever-/never-smoking, and second-hand smoking using adjusted linear regression models in test dataset: beta estimate with 95% confidence intervals on natural log-transformed airway inflammatory biomarker level and percentage change with 95% confidence interval in airway inflammatory biomarker level. P-values < 0.05 are highlighted bold and p-values < 0.1 cursive. Table S6. Gene-environment interaction effects between the weighted binary polygenic risk score (genetic low-risk vs. high-risk group) and chronic air pollution exposure on natural log-transformed airway inflammatory biomarker levels in elderly women using adjusted linear regression models in test dataset. Gene-environment interaction effects between the weighted binary polygenic risk score (derived by the interaction-training approach, dichotomized using the median of weighted polygenic risk score) and chronic air pollution exposure (nitrogen dioxide, nitrogen oxides, particulate matter with aerodynamic diameters of ≤ 2.5/ ≤ 10/2.5–10 µm, reflectance of PM2.5 filters calculated as the mean of the annual average concentrations from baseline and first follow-up examinations within a time window of 15 years prior to the biomarker assessments, statistically centred across the participants, standardized using interquartile ranges) on natural log-transformed airway inflammatory biomarker levels (tumor necrosis factor-α, leukotriene B4, and the sum of eosinophils, macrophages, neutrophils and epithelial cells in induced sputum) adjusted for: age, body mass index (BMI in kg/m2), highest education of the participant or her spouse (low < 10 years, medium = 10 years, high > 10 years of education), ever-/never-smoking, and second-hand smoking using adjusted linear regression models in test dataset: beta estimate with 95% confidence intervals on natural log-transformed airway inflammatory biomarker level and percentage change with 95% confidence interval in airway inflammatory biomarker level. P-values < 0.05 are highlighted bold and p-values < 0.1 cursive. Table S7. Gene-environment interaction effects between the weighted polygenic risk score and chronic air pollution exposure on natural log-transformed airway inflammatory biomarker levels in elderly women using linear regression models in test dataset with additional adjustment according to indoor air pollution (exposure to mould), and heating with fossil fuels. Gene-environment interaction effects between the weighted polygenic risk score (derived by the interaction-training approach, standardized using interquartile ranges) and chronic air pollution exposure (nitrogen dioxide, nitrogen oxides, particulate matter with aerodynamic diameters of ≤ 2.5/ ≤ 10/2.5–10 µm, reflectance of PM2.5 filters calculated as the mean of the annual average concentrations from baseline and first follow-up examinations within a time window of 15 years prior to the biomarker assessments, statistically centred across the participants, standardized using interquartile ranges) on natural log-transformed airway inflammatory biomarker levels (tumor necrosis factor-α, leukotriene B4, and the sum of eosinophils, macrophages, neutrophils and epithelial cells in induced sputum) adjusted for: age, body mass index (BMI in kg/m2), highest education of the participant or her spouse (low < 10 years, medium = 10 years, high > 10 years of education), ever-/never-smoking, second-hand smoking, indoor air pollution (exposure to mould), and heating with fossil fuels using adjusted linear regression models in test dataset: beta estimate with 95% confidence intervals on natural log-transformed airway inflammatory biomarker level and percentage change with 95% confidence interval in airway inflammatory biomarker level. P-values < 0.05 are highlighted bold and p-values < 0.1 cursive. Table S8. Gene-environment interaction effects between the weighted polygenic risk score and chronic air pollution exposure on natural log-transformed airway inflammatory biomarker levels in elderly women using linear regression models in test dataset with no adjustment according to body mass index. Gene-environment interaction effects between the weighted polygenic risk score (derived by the interaction-training approach, standardized using interquartile ranges) and chronic air pollution exposure (nitrogen dioxide, nitrogen oxides, particulate matter with aerodynamic diameters of ≤ 2.5/ ≤ 10/2.5–10 µm, reflectance of PM2.5 filters calculated as the mean of the annual average concentrations from baseline and first follow-up examinations within a time window of 15 years prior to the biomarker assessments, statistically centred across the participants, standardized using interquartile ranges) on natural log-transformed airway inflammatory biomarker levels (tumor necrosis factor-α, leukotriene B4, and the sum of eosinophils, macrophages, neutrophils and epithelial cells in induced sputum) adjusted for: age, highest education of the participant or her spouse (low < 10 years, medium = 10 years, high > 10 years of education), ever-/never-smoking, and second-hand smoking using adjusted linear regression models in test dataset: beta estimate with 95% confidence intervals on natural log-transformed airway inflammatory biomarker level and percentage change with 95% confidence interval in airway inflammatory biomarker level. P-values < 0.05 are highlighted bold and p-values < 0.1 cursive. Table S9. Gene-environment interaction effects between the weighted polygenic risk score and chronic air pollution exposure on natural log-transformed airway inflammatory biomarker levels in elderly women using adjusted linear regression models in test dataset excluding women with any chronic inflammatory respiratory condition. Gene-environment interaction effects between the weighted polygenic risk score (derived by the interaction-training approach, standardized using interquartile ranges) and chronic air pollution exposure (nitrogen dioxide, nitrogen oxides, particulate matter with aerodynamic diameters of ≤ 2.5/ ≤ 10/2.5–10 µm, reflectance of PM2.5 filters calculated as the mean of the annual average concentrations from baseline and first follow-up examinations within a time window of 15 years prior to the biomarker assessments, statistically centred across the participants, standardized using interquartile ranges) on natural log-transformed airway inflammatory biomarker levels (tumor necrosis factor-α, leukotriene B4, and the sum of eosinophils, macrophages, neutrophils and epithelial cells in induced sputum) adjusted for: age, body mass index (BMI in kg/m2), highest education of the participant or her spouse (low < 10 years, medium = 10 years, high > 10 years of education), ever-/never-smoking, and second-hand smoking in women without any condition of asthma, chronic bronchitis, hay fever, cough, cough with sputum or chronic obstructive pulmonary disease using adjusted linear regression models in test dataset: beta estimate with 95% confidence intervals on natural log-transformed airway inflammatory biomarker level and percentage change with 95% confidence interval in airway inflammatory biomarker level. P-values < 0.05 are highlighted bold and p-values < 0.1 cursive. Table S10. Gene-environment interaction effects between the weighted polygenic risk score including only the Sentinel SNPs belonging to causal genes and chronic air pollution exposure on natural log-transformed airway inflammatory biomarker levels in elderly women using adjusted linear regression models in test dataset. Gene-environment interaction effects between the weighted polygenic risk score (derived by the interaction-training approach, standardized using interquartile ranges, including only the Sentinel SNPs belonging to causal genes) and chronic air pollution exposure (nitrogen dioxide, nitrogen oxides, particulate matter with aerodynamic diameters of ≤ 2.5/ ≤ 10/ 2.5–10 µm, reflectance of PM2.5 filters calculated as the mean of the annual average concentrations from baseline and first follow-up examinations within a time window of 15 years prior to the biomarker assessments, statistically centred across the participants, standardized using interquartile ranges) on natural log-transformed airway inflammatory biomarker levels (tumor necrosis factor-α, leukotriene B4, and the sum of eosinophils, macrophages, neutrophils and epithelial cells in induced sputum) adjusted for: age, body mass index (BMI in kg/m2), highest education of the participant or her spouse (low < 10 years, medium = 10 years, high > 10 years of education), ever-/never-smoking, and second-hand smoking using adjusted linear regression models in test dataset: beta estimate with 95% confidence intervals on natural log-transformed airway inflammatory biomarker level and percentage change with 95% confidence interval in airway inflammatory biomarker level. P-values < 0.05 are highlighted bold and p-values < 0.1 cursive.