大氣中致癌揮發性有機化合物的特性及健康風險評估

Abstract

揮發性有機化合物 (Volatile Organic Compounds, VOCs) 是日常生活環境中常會接觸到的空氣污染物，常溫常壓下多以氣態形式存在，主要透過吸入途徑進入人體。其主要排放源包括交通運輸、石化製程及各類工業活動。其中有些VOCs會參與光化學反應並促成臭氧生成，因此臺灣環保署 (現改為環境部) 自 2003 年起，於臺灣西部地區陸續設置光化學評估監測站 (Photochemical Assessment Monitoring Stations, PAMS)，目前全臺共有11座光化測站持續運行中。測站長期逐時監測54種 VOCs的濃度，為臭氧生成分析與空氣品質研究提供重要資料庫。 長期暴露於部分 VOCs 可能會對人體健康造成不良影響，包括損害神經、呼吸及免疫系統，甚至具致癌性。根據國際癌症研究機構 (International Agency for Research on Cancer, IARC) 分類，在PAMS所監測的54種VOCs中，苯、苯乙烯、異戊二烯、異丙基苯與乙苯分別被列為已知、極可能或可能的人類致癌物，顯示長期暴露於VOCs對人體具有潛在健康影響，其健康風險值得進一步評估。 儘管臺灣已累積多年VOCs監測資料，針對其長期暴露所造成的健康風險，特別是致癌風險評估研究仍然非常有限。過去國內相關評估多引用美國環保署風險資訊系統 (Integrated Risk Information System, IRIS) 所提供的致癌斜率係數 (Cancer slope factor, CSF)或吸入單位風險值 (Inhalation Unit Risk, IUR)，然而目前僅苯有提供相關IUR值，苯乙烯、異戊二烯、異丙基苯與乙苯，則因尚未完成評估或未曾評估，使得IRIS目前仍無提供這些致癌物質的CSF或IUR值，這也解釋了國內未曾針對這些物質進行致癌風險評估的原因。 因此，本研究整合長期 PAMS 監測數據與 VOCs 毒理學文獻，針對被 IARC 歸類為 2B 級以上、且具基因毒性的VOCs，進行CSF與IUR的推估與健康風險評估。研究篩選出苯、苯乙烯、異戊二烯、異丙基苯與乙苯等五種目標物質，並依循風險評估四大步驟進行分析。在劑量效應評估方面，採用美國國家毒理計畫(National Toxicology Program, NTP) 與其他長期動物吸入試驗數據，並使用美國環保署開發的貝氏統計基準劑量軟體 (Bayesian Benchmark Dose Software, BSBMDS)進行劑量效應關係擬合，推估各物質之基準劑量下限 (Benchmark Dose Lower Confidence Limit, BMDL)，以進一步計算CSF及IUR。在暴露評估部分，彙整 2006至2023 年間全臺 11 座 PAMS 測站之逐時 VOCs 監測資料，針對未檢出值 (Not Detected, ND) 以 0 或 1/2 方法偵測極限進行補值，計算年平均濃度。並整理各測站自設立起至 2023 年間的年平均濃度數據，取得平均值、最大值、最小值與標準差，以代表民眾長期暴露情況。並透過 Mann-Kendall 檢定分析濃度變化趨勢。最後，結合暴露濃度分布與 IUR，假設濃度符合截斷常態分布 (Truncated normal distribution)，使用 Crystal Ball 軟體進行 10,000 次蒙地卡羅模擬，推估不同地區居民之終身吸入致癌風險。 研究結果顯示，多數監測站的總致癌風險已超過 1 × 10-6，部分地區甚至達到 10-4，顯示特定區域居民長期暴露於較高風險環境中，尤以工業與交通活動密集的高雄、臺南與臺北萬華最為顯著。相較之下，雲林臺西與嘉義朴子等農業區風險則較低。而苯乙烯與異丙基苯因 IUR 較高，是主要風險貢獻物質。 時段分析指出，苯、異丙基苯與乙苯在通勤尖峰時段濃度上升，可能與交通排放相關；異戊二烯則於清晨六時濃度大幅升高，推測可能主要來自移動污染源排放。時間趨勢分析則發現VOCs 濃度多呈下降趨勢，反映污染控制政策的成效。此外，若僅以近年濃度資料進行評估，將低估長期暴露風險約 1 至 3 倍，顯示納入長期監測數據的重要性。 本研究亦存在若干限制與不確定性，需在解讀結果時加以審慎考量。例如，動物實驗顯示雌性老鼠對苯乙烯與異戊二烯所引發的腫瘤反應較為敏感，惟因可能存在代謝飽和效應，導致雌鼠資料擬合不佳，故本研究改採雄性老鼠之腫瘤資料進行劑量效應擬合，可能低估女性或其他易感族群之健康風險。此外，部分常見 VOCs (如1,3-丁二烯) 因僅於特殊工業區進行監測，未能納入本研究的致癌風險總量分析，亦為一潛在限制。 總結而言，本研究針對臺灣地區的 VOCs 致癌風險進行量化分析，首次補足苯乙烯、異戊二烯、異丙基苯與乙苯等物質在IUR的知識缺口，並辨識出潛在高風險區域與主要風險貢獻物質。研究成果可作為未來風險溝通、污染源管理與環境政策擬定的科學依據，亦強調持續更新監測資料與精進風險評估方法的必要性。

Volatile organic compounds (VOCs) are common air pollutants in our environment, originating mainly from vehicular emissions, petrochemical processes, and industrial activities. The primary exposure route for humans is inhalation. VOCs are precursors for the formation of ozone through photochemical reactions. Therefore, the former Taiwan Protection Administration (current the Taiwan Ministry of Environmental Protection) gradually established 11 photochemical assessment monitoring stations (PAMSs) in western Taiwan since 2003. Hourly samples have been collected and analyzed for 54 VOCs at each PAMS. Among the 54 VOCs monitored, some of them have been associated with adverse health effects, including neurological, respiratory, and immune toxicity, as well as carcinogenicity. The International Agency for Research on Cancer (IARC) classified benzene, styrene, isoprene, isopropylbenzene, and ethylbenzene as known, possible, or probable human carcinogens, highlighting potential health effects due to long-term exposures to these VOCs and the need for further health risk assessment. Although the potential health impacts caused by long-term VOC exposures have been of great concern, cancer risk has remained limited. One of the reasons was the lack of cancer slope factors (CSFs) and/or inhalation unit risks (IURs) for many VOCs in the U.S. EPA’s IRIS database, except for benzene, given that the chronological PAMS VOC dataset is valuable for the study of ozone formation and the assessment of VOC exposures and risk to address the potential health impacts to the residents near the PAMSs. This study aims to integrate 11 PAMS monitoring data (2006–2023) with toxicological data to derive CSFs and IURs for four VOCs: styrene, isoprene, isopropylbenzene, and ethylbenzene, and to assess cancer risk for the 4 VOCs and benzene. Benchmark Dose Lower Confidence Limits (BMDLs) were estimated by fitting data from long-term animal studies using the Bayesian statistics Benchmark Dose Software (BSBMDS). The BMDLs were used to derive CSFs and IURs of the four VOCs. Exposures were assessed using hourly VOC data, with non-detectable samples assigned to half the method detection limit. The annual average concentrations across the 11 stations were analyzed for temporal trends with the Mann-Kendall test and assumed to belong to truncated normal distributions. Monte Carlo simulations were applied to estimate lifetime cancer risks with 10000 iterations using the Crystal Ball software. Results showed elevated risk in industrial and high-traffic areas such as Kaohsiung (Xiaogang, Qiaotou), Tainan, and Wanhua stations. In these regions, styrene and isopropylbenzene—both with high IURs—were major contributors. Conversely, agricultural areas associated with a petrochemical industrial area, such as Taixi and Puzi stations, revealed notably lower concentrations and cancer risks. Temporal trends indicated a general decline in VOC levels, reflecting the effectiveness of air pollution control policies. Hourly concentration patterns further suggested that traffic emissions were responsible for certain VOCs (e.g., benzene, isopropylbenzene, ethylbenzene). Most of the 11 PAMSs reported total cancer risks exceeding the negligible risk of 1 × 10-6, with some approaching 1 × 10-4, underscoring potential chronic health concerns. Notably, previous assessments that relied on recent year data (e.g., 2023) could underestimate long-term risk by up to 1–3 times. This study acknowledges limitations, such as potential underestimation of risks for sensitive populations (e.g., females). The female rat rodents showed high tumor incidences, compared with the males, and lacked a dose-dependent increase, which was a poor fit for BSBMDS. Therefore, the IURs and cancer risks of isoprene and styrene may have been underestimated. Additionally, certain carcinogenic VOCs like 1,3-butadiene and acrylonitrile were not included because they were not monitored by the PAMSs. Overall, this assessment, based on the best toxicological and VOC data of the PAMS dataset, provides critical updates to VOC risk parameters and identifies priority pollutants and regions, offering a scientific foundation for future risk communication, regulatory actions, and environmental policy development in Taiwan to improve public protection.