Introduction: Epidemiological and toxicological evidence has indicated that PM10 and PM2.5 exposure increase the incidence and mortality of lung cancer. It is possible that PM10 and PM2.5 exposure also reduce length of lung cancer survival from the time of diagnosis. This thesis aims to examine the survival differences among lung cancer patients exposed to different levels of particulate matter.

Methods: This thesis conducted a secondary analysis of the association between PM2.5 and PM10 exposure and survival from lung cancer. Exposure data were obtained from the Ministry of Health and the Land Air Water Aotearoa. Participants consisted of Christchurch residents initially diagnosed with lung cancer in 2014. The Kaplan-Meier method was used for survival analysis and the log-rank test was conducted to compare whether survival probabilities differed between groups. Temperature, wind speed, relative humidity, O3, NO, NO2, SO2, CO, PM10 and PM2.5 (p<0.05 in univariate analysis) were included in the multivariable Cox proportional hazards model. Age and socioeconomic status were also included in the multivariable Cox proportional hazards model due to their important effect on lung cancer survival. In the competing risk model, death from lung cancer was regarded as the major event, and death from causes other than lung cancer was deemed the competing risk event. Adjusted survival curves for PM10 were drawn in accordance with the results of the Cox proportional hazards model.

Results: A total of 176 lung cancer patients were eligible for the analysis. The lung cancer-specific mortality was 5.24 (95% CI: 2.88, 9.53) times higher in the high PM10 exposure group (PM10>19.82μg/m3) than in the low PM10 exposure group (PM10≤19.82μg/m3) in multivariable Cox proportional hazards model. The HR decreased to 1.87 (95% CI: 1.16, 2.98) in the competing risk model. The association between PM2.5 exposure and lung cancer survival was not statistically significant. For PM10, the high exposure group (150 days) suffered from a 193-day reduction in median survival time compared with the low exposure group (343 days) after adjusting age, socioeconomic status, temperature, wind speed, relative humidity, O3, NO, NO2, SO2, CO and PM2.5.

Discussion: The correlation coefficient between PM and lung cancer mortality increases with the lagged years, reaching 0.9 in the fourth year and the highest in the seventh year. Therefore higher HR values for PM10 observed in this thesis can result from considering the latency effect of PM exposure. The association between PM2.5 exposure and lung cancer survival was not statistically significant. Highly correlated variables in this thesis may cause this result. Based on a shrunken survival time (a 193-day decrease in median survival) in the high PM10 exposure group (PM10>19.82μg/m3), this thesis proposes that PM10 exposure accelerates lung cancer progression in lung cancer patients.

Conclusion: The finding that high PM10 exposure shortens lung cancer survival adds support for lung cancer patients to take action to reduce PM10 exposure, such as using an air purifier and avoiding outdoor activities when PM10 concentration is high. Future work can use the satellite-based measurement for accurate PM exposure estimates, increase the sample size, utilise a Cox proportional hazards model with time-dependent covariates or a non-proportional hazard model to deal with the inconstant hazard between high and low wind exposure groups, and account for additional variables, e.g., smoking, marital status and such.