Reply to Ward et al.: Air Pollution as a Risk Factor for Lung Cancer: Potential Mechanisms

From the Authors:

We read the response letter by Ward and colleagues to our manuscript (1), and the accompanying editorial by Christiani (2), with great interest and value the insightful suggestions they raise. We agreed with that there is a mechanistically plausible etiological link between chronic obstructive pulmonary disease (COPD) and lung cancer development. Previous studies have also shown that personal medical history of COPD was an independent risk factor of lung cancer (3).

As suggested, we further integrated COPD at baseline into our analysis to dissect the complex relations of air pollution, COPD, and incident lung cancer events in the UK Biobank. We defined participants with hospital admission records of COPD before the date of baseline assessment, self-reported COPD, or those with mild to moderate airflow obstruction (defined as post-bronchodilator FEV₁/FVC <0.70) at the baseline assessment as prevalent COPD (4). We also defined particulate matter ⩽2.5 μm in aerodynamic diameter (PM_2.5), particulate matter ⩽10 μm in aerodynamic diameter (PM₁₀), and nitrogen dioxide (NO₂) as high- and low-exposure category according to World Health Organization guidelines, except for nitrogen oxides (NO_x), of which the median level was used. Cross-sectional analyses showed that higher exposures to PM_2.5, PM₁₀, NO₂, and NO_x were significantly associated with higher risk of prevalent COPD at baseline after adjusting for age, sex, body mass index, household income, education level, smoking status, and pack-years of smoking (PM_2.5 [odds ratio (OR), 1.47; 95% confidence interval (CI), 1.40–1.54, per 5 μg/m³]; PM₁₀ [OR, 1.13; 95% CI, 1.07–1.20, per 10 μg/m³]; NO₂ [OR, 1.08; 95% CI, 1.06–1.09, per 10 μg/m³]; and NO_x [OR, 1.10; 95% CI, 1.08–1.11, per 20 μg/m³]). In addition, we observed that participants with prevalent COPD at baseline had significantly higher risk of incident lung cancer during the follow-up, with a hazard ratio (HR) of 2.68 (95% CI, 2.37–3.02) after adjusting the above covariates. These results indicated that COPD might be a mediator between air pollution and incident lung cancer.

We then performed a mediation analysis using natural effect models within the R package “medflex” (5). As expected, we observed significant mediation effects by prevalent COPD in the apparent air pollution effect on lung cancer, with a mediation proportion of 17.56% for PM_2.5, 7.91% for PM₁₀, 19.99% for NO₂, and 19.43% for NO_x (Table 1). Compared with participants without COPD at baseline and with low exposure to air pollution, those participants with COPD and with high exposure to air pollution had significantly increased risk of incident lung cancer (PM_2.5 [HR, 3.01; 95% CI, 2.55–3.55]; PM₁₀ [HR, 3.02; 95% CI, 2.53–3.60]; NO₂ [HR, 2.64; 95% CI, 2.01–3.45]; and NO_x [HR, 3.07; 95% CI, 2.62–3.58]). Furthermore, we also observed positive additive interactions for air pollutants and prevalent COPD in the development of lung cancer (Table 2). The relative excess risk because of the interactions were estimated to be 0.36 (95% CI, 0.14–0.58) for PM_2.5, 0.53 (95% CI, 0.29–0.76) for PM₁₀, and 0.63 (95% CI, 0.42–0.84) for NO_x, which indicated that there were 12%, 17%, and 20% of the increased risk attributable to the additive interactions. However, we did not observe significant additive interactions for NO₂ exposure and prevalent COPD.

Table 1.

Mediation Analysis of COPD on the Associations between Air Pollution and Lung Cancer Risk

Pollution	Total Effect [β (95% CI)]	Natural Direct Effect [β (95% CI)]	Natural Indirect Effect [β (95% CI)]	Mediation Proportion [% (95% CI)]
PM2.5	0.454 (0.198–0.691)	0.374 (0.117–0.615)	0.080 (0.066–0.095)	17.56 (11.13–41.43)
PM10	0.336 (0.048–0.618)	0.309 (0.024–0.592)	0.027 (0.015–0.039)	7.91 (2.97–34.27)
NO2	0.076 (0.011–0.134)	0.061 (–0.004–0.119)	0.015 (0.012–0.018)	19.99 (10.15–95.45)
NOx	0.098 (0.034–0.151)	0.079 (0.017–0.133)	0.019 (0.016–0.023)	19.43 (11.85–50.97)

Definition of abbreviations: CI = confidence interval; COPD = chronic obstructive pulmonary disease; NO₂ = nitrogen dioxide; NO_x = nitrogen oxides; PM_2.5 = particulate matter ⩽2.5 μm in aerodynamic diameter; PM₁₀ = particulate matter ⩽10 μm in aerodynamic diameter.

Adjusted for age, sex, body mass index, household income, education level, smoking status, and pack-years of smoking.

Table 2.

Joint Effect and Additive Interaction between Air Pollution and COPD

Air Pollutant	Pollution Category	COPD	HR (95% CI)	P Value	RERI* (95% CI)	AP* (95% CI)
PM2.5†	Low	Without	Ref.
	High	Without	1.09 (0.93 to 1.29)	0.290
	Low	With	2.56 (2.15 to 3.04)	<2 × 10−16
	High	With	3.01 (2.55 to 3.55)	<2 × 10−16	0.36 (0.14 to 0.58)	0.12 (0.05 to 0.19)
PM10‡	Low	Without	Ref.
	High	Without	0.99 (0.83 to 1.18)	0.898
	Low	With	2.51 (2.16 to 2.90)	<2 × 10−16
	High	With	3.02 (2.53 to 3.60)	<2 × 10−16	0.53 (0.29 to 0.76)	0.17 (0.10 to 0.24)
NO2§	Low	Without	Ref.
	High	Without	1.03 (0.78 to 1.37)	0.815
	Low	With	2.69 (2.38 to 3.04)	<2 × 10−16
	High	With	2.64 (2.01 to 3.45)	<2 × 10−16	−0.09 (−0.43 to 0.27)	−0.03 (−0.19 to 0.09)
NOxǁ	Low	Without	Ref.
	High	Without	1.06 (0.91 to 1.24)	0.443
	Low	With	2.38 (2.00 to 2.83)	<2 × 10−16
	High	With	3.07 (2.62 to 3.58)	<2 × 10−16	0.63 (0.42 to 0.84)	0.20 (0.14 to 0.27)

Definition of abbreviations: AP = attributable proportion due to the interaction; CI = confidence interval; COPD = chronic obstructive pulmonary disease; HR = hazard ratio; NO₂ = nitrogen dioxide; NO_x = nitrogen oxides; PM_2.5 = particulate matter ⩽2.5 μm in aerodynamic diameter; PM₁₀ = particulate matter ⩽10 μm in aerodynamic diameter; Ref. = reference; RERI = relative excess risk due to the interaction; WHO = World Health Organization.

Adjusted for age, sex, body mass index, household income, education level, smoking status, and pack-years of smoking.

^*To estimate RERI and AP, the low-pollution category and the without COPD groups were the reference categories.

^† Defined by WHO guideline value of PM_2.5: low (<10 μg/m³) and high (⩾10 μg/m³).

^‡ Defined by WHO guideline value of PM₁₀: low (<20 μg/m³) and high (⩾20 μg/m³).

^§ Defined by WHO guideline value of NO₂: low (<40 μg/m³) and high (⩾40 μg/m³).

^ǁ Defined by median of NO_x: low (<41.75 μg/m³) and high (⩾41.75 μg/m³).

Taken together, our analyses supported the core relationships between environmental air pollution exposures, airway pathology, and the etiology of lung cancer in the globally relevant context of air pollution. Chronic inflammation associated with COPD may result in repeated airway epithelial injury and accompanying high cell turnover rates and propagation of DNA errors resulting in amplification of the carcinogenic effects of air pollution exposures (6). However, the hypothesized pathway was only one of the potential carcinogenic mechanisms of air pollution, as less than one-fifth of the total effects were mediated by COPD (7). In addition, it should be noted that it is difficult to disentangle the effects of single components from the complex mixture of air pollution; thus, further exposure pattern analysis may provide deeper insights.