長期PM2.5暴露與肺腺癌發生率、死亡率及存活率風險之相關研究

Abstract

研究背景： 肺癌多年蟬聯臺灣癌症發生第一名，並在2023年成為臺灣癌症死亡的首要原因。由於缺乏長期歷史空氣污染資料，過去30年以上細懸浮微粒(PM2.5) 暴露對肺腺癌的長期影響尚未被探討。本研究目的為結合自1968年以來的空氣污染數據與自1997年以來的全國性臺灣癌症登記資料，以評估30年PM2.5暴露對肺腺癌的相對風險（以非肺腺癌作為對照），此外，本研究也將量化PM2.5暴露減少對非小細胞肺癌（NSCLC）患者的生存效益。最後，本研究也將運用生態學方法，評估二氧化碳排放（包括煤炭和石油產業的排放）與肺腺癌發病率和死亡率之間的關聯。 研究方法：本研究包含三個部分。第一部分為病例對照研究，利用自1997年起的臺灣全國癌症登記資料與自1968年起的空氣污染資料，評估自診斷日前30年PM2.5暴露對肺腺癌風險的影響。病例為所有肺腺癌個案，對照則為所有非肺腺癌個案。每個病例的30年PM2.5暴露量依照其診斷日前30年之居住地PM2.5濃度進行估算。此研究使用多元邏輯斯迴歸模型評估PM2.5暴露與肺腺癌的發生比(IRR)，調整變項包含性別、年齡、吸菸狀態、癌症分期與肺癌表皮生長因子受體 (EGFR)突變狀態。 第二部分為生存分析，研究對象為1997至2017年間診斷之非小細胞肺癌 (NSCLC) 患者，並追蹤至死亡或追蹤結束之 2022 年 12 月 31 日。以 Cox 比例風險模型評估患者 PM2.5 暴露與其癌症存活之間的關聯，並控制年齡、性別、體重指數、吸菸狀況、癌症分期、EGFR 突變、治療以及 PM2.5 暴露與治療效果之間的交互作用。 第三部分為生態研究，利用線性迴歸模型評估空氣污染排放源與肺腺癌年齡調整發生率與死亡率之關聯。空氣汙染暴露指標包含30年移動平均的二氧化碳 (CO2) 年排放量、煤、石油產業CO2年排放量，以及全臺灣機動車輛、汽車、機車之車輛數量。 研究結果：自1968年以來環境中年平均PM2.5濃度上升與自1997年起年齡調整後肺腺癌發生率增加相關。肺腺癌發生率在女性、非吸菸者、65歲以上長者、ⅢB期至第Ⅳ期個案及EGFR突變者中較高。研究對象的30年平均PM2.5暴露濃度為33.7 ± 7.4 μg/m3，平均高污染日數為162 ± 130天。多元邏輯斯模型結果顯示，每增加10 μg/m3 PM2.5暴露量，2011–2020年間肺腺癌與非肺腺癌的發生比IRR顯著為1.044。 生存分析中，納入了1997年至2017年間共67,967例非小細胞肺患者。觀察到在2011年至2017年間，5年存活率提高了18%至36%，而5年移動平均PM2.5濃度從2011年的28.5 µg/m3下降至2017年的18.6 µg/m3。交互作用分析顯示，PM2.5濃度的降低為接受藥物治療的NSCLC患者帶來了額外的生存獲益，自診斷以來，PM2.5濃度每降低8.0 µg/m3，肺癌特異性死亡的風險比為0.75（95% CI：0.74–0.76）；自2006年以來，每降低7.1 µg/m3，肺癌特異性死亡的風險比為0.83（95% CI：0.82–0.84）。未接受治療的NSCLC患者的存活獲益較小。 整體而言，每百萬公噸CO2排放量增加與肺腺癌發生率上升0.13有關，其中來自煤與石油產業的排放分別與肺腺癌發生率上升0.21與0.62相關。類似的關聯亦見於30年移動平均機動車輛數，每增加一百萬輛車，肺腺癌發生率增加1.62。 結論： 本研究證實30年長期PM2.5暴露與肺腺癌相較於非肺腺癌的風險增加有關，且無論性別、年齡、吸菸與否、癌症分期或EGFR突變狀態皆具一致性。同時也發現PM2.5濃度下降使NSCLC患者在癌症治療之外獲得額外的生存益處。此外，本研究亦提供CO2排放量、車輛數與肺腺癌發生率關聯的佐證。

Background: Lung cancer remained the leading cause of cancer-related death in Taiwan in 2023. The long-term effects of over 30 years of PM2.5 exposure on adenocarcinoma lung cancer (AdLC) have not been previously studied, mainly due to the lack of long-term historical air pollution data. By combining air pollution data from 1968 with nationwide Taiwan Cancer Registry (TCR) data since 1997 in Taiwan, this study aims to estimate the relative risks of 30-year PM2.5 exposure on AdLC using non-AdLC cases as referents. Also to quantify the potential survival gains in non-small cel lung cancer (NSCLC) associated with reductions in PM2.5 exposure. And will further apply an ecological study to estimate the associations between CO2 emissions and AdLC incidence and mortality, including emissions from coal and oil industries. Material and Methods: This study comprises three components. The first is a case-referent study using nationwide TCR data from 1997 onward and air pollution data since 1968 in Taiwan to estimate the risk of 30-year PM2.5 exposure on AdLC. Cases were all AdLC patients, while referents were all non-AdLC patients. Individual 30-year PM2.5 exposures were estimated based on PM2.5 levels at their residential locations during the 30 years prior to their diagnosis dates. Multiple logistic regression models were applied to estimate the incidence rate ratios (IRRs) associated with PM2.5 exposure, adjusting for sex, age, smoking status, cancer stage, and epidermal growth factor receptor (EGFR) mutation. The second component is a survival analysis of NSCLC patient diagnosed between 1997 and 2017 and followed until death or until their death or December 31, 2022. Cox proportional hazard models are used to assess associations between patients’ PM2.5 exposure and their cancer survivals, controlling age, sex, body mass index, smoking status, cancer stage, EGFR mutation, treatment, and interaction between PM2.5 exposure and treatment effect. The third component is an ecological study employing linear regression models to evaluate the associations between air pollution emission sources and the age-adjusted annual incidence and mortality rates of AdLC. The 30-year moving averages of annual carbon dioxide (CO2) emissions, emissions from coal and oil industries, and the number of vehicles, including both cars and motorcycles in Taiwan were used as exposure indicators. Results: Elevation in annual ambient PM2.5 concentrations since 1968 were associated with increase in annual age-adjusted AdLC incidence since 1997. AdLC incidence were higher among females, nonsmokers, the elderly aged above 65, cases of stages IIIB to IV, and those with EGFR mutation. Study subjects’ PM2.5 exposures averaged at 33.7 ± 7.4 μg/m3 with 162 ± 130 high PM2.5 pollution days over 30 years. Multiple logistic models showed an increase in 10 μg/m3 of PM2.5 exposures was significantly associated with 1.044 of IRR between AdLC and non-AdLC cases during 2011–2020. For survival analysis, a total of 67,967 NSCLC patients were registered in Taiwan from 1997 to 2017. We observed 5-year survival rates from 2011 to 2017 increased by 18% to 36%, and 5-year moving average PM2.5 concentrations decreased from 28.5 µg/m3 since 2011 to 18.6 μg/m3 since 2017. Interaction analyses showed that PM2.5 reduction yielded additional survival benefits for NSCLC patients with medical treatments, with hazard ratios for lung cancer-specific mortality of 0.75 (95% CI: 0.74-0.76) per 8.0 µg/m3 reduction since diagnosis and 0.83 (95% CI: 0.82-0.84) per 7.1 µg/m3 reduction since 2006. Survival improvements were smaller among untreated NSCLC patients. Overall, per million tonnes of CO₂ emissions was associated with a 0.13 increase in AdLC incidence, across both emission from coal and oil industries for 0.21 and 0.62 increase. Similar patterns were observed in 30-year moving averages of vehicle numbers, with each increase of one million vehicles associated with a 1.62 increase in AdLC incidence. Conclusion: This research demonstrates that long-term exposures to PM2.5 over 30 years were associated with elevated risks of AdLC against non-AdLC, regardless of sex, age, smoking status, cancer stage, or EGFR mutation. It further shows that reducing PM2.5 exposure yielded additional survival benefits beyond cancer treatments among NSCLC patients. Additionally, this study provides supporting evidence for the association between CO2 emissions, vehicle numbers, and AdLC incidence.