臺灣博碩士論文加值系統

經皮吸收半致死劑量(dermal LD50)為常用於評估化學物皮膚暴露急性效應之毒性數據，主要透過活體內動物實驗產生。然而隨著動物福祉成為關注焦點，利用傳統方法評估皮膚暴露毒害效應日漸困難。為滿足毒性評估資料之需求，歐美各國逐步開始發展包含量化結構−活性關係(QSAR)等可用以替代動物實驗之評估方法，以輔助化學物管理所需之危害辨識。本研究針對美國國家醫學圖書館危害物質職業暴露資料庫(Haz-Map®)收錄之所有化學物，結合美國政府工業衛生師協會(ACGIH)授有閾限值(TLV®)、化學品全球分類及標示調和制度(GHS)危害分級架構、及Haz-Map®分類標準等條件篩選標的化學物，建立相應之dermal LD50 QSAR模式，並比較不同模式間所涵納之分子描述符，探討影響化學物經皮吸收毒害潛能之分子特徵。研究結果顯示：本研究以Haz-Map®資料庫中所有化學物建立之QSAR模式預測效能不佳，或與Haz-Map®資料庫整體數據中影響不同類型化學物經皮吸收毒性潛能之分子特徵較為分歧有關。當以GHS危害分級系統第一級毒性物質、ACGIH授有TLV®、及Haz-Map®分類為農藥類、有機溶劑類、與含氮化合物類之化學物分別發展QSAR模式時判斷係數均顯著提昇，其中又以GHS危害分級系統第一級毒性物質及有機溶劑類化學物所發展之QSAR模式預測效能表現最佳。顯見當標的化學物群組之分子結構特性或致毒機轉接近時，影響其經皮暴露毒害效應之分子特徵亦較相似。透過進一步規範及篩選標的化學物，可有效降低急性毒害效應形成機轉間之分歧，加強模式之預測效能。本研究所發展QSAR模式中重覆出現與強陰電性原子鍵結組合相關之分子描述符，顯示分子結構中電荷重分配及分佈不均之現象或為影響化學物經皮吸收毒害效應之重要因子。此外模式亦出現影響化學物皮膚滲透性之辛醇−水分配係數量化指標，顯示化學物皮膚滲透性為影響化學物經皮吸收毒害效應之先決因子。

The dermal lethal dose 50% (dermal LD50) is among the toxicological data most frequently used in hazard identification for evaluating acute toxicity of chemicals resulting from dermal absorption. Traditionally dermal LD50 is determined from in vivo animal tests. However, considering animal welfare in vivo testing has become less of a choice in furnishing the dermal LD50 data required of for hazard characterization. To meet the increasing regulatory demand of toxicological data, the methods alternative to animal testing including the development of quantitative structure-activity relationships (QSARs) have been attempted and encouraged by regulatory agencies in the US and Europe to facilitate proper hazard identification in the management of chemical substances. This study developed molecular descriptor-based QSARs for estimating dermal LD50 using different groupings of chemical substances, including all of the chemicals reported in the National Library of Medicine Occupational Exposure to Hazardous Agents Database (Haz-Map®) database, chemicals awarded a Threshold Limit Value (TLV®) by the American Conference of Governmental Industrial Hygienists (ACGIH), chemicals ranked as Hazard Category 1 hazards in the Globally Harmonized System of Classification and Labelling of Chemicals (GHS) promulgated by the United Nations, and those grouped as “Pesticides”, “Solvents”, and “Nitrogen Compounds” in the Haz-Map®. The molecular descriptors described in each QSAR were cross-analyzed to explore the molecular characteristics involved in transdermal permeation of and subsequent toxicity by chemicals of varying toxicological potency and physicochemical properties. As the results show, the QSAR built using all chemical in the Haz-Map® performed poorly in terms of predictability and reliability, possibly being a result of diverse molecular characteristics and toxicological modes of action among these target chemicals, hampering the identification of the most relevant descriptors in the QSAR. A significant improvent in the coefflicients of determination in the model was observed when the training and validating sets were re-defined and limited to only chemicals that were ranked as the GHS Hazard Catetgory 1 hazards, awarded the TLV®, and grouped as the Haz-Map® “Pesticides”, “Solvents”, or “Nitrogen Compounds”. Among which the best performance was observed in the QSARs for GHS Hazard Catetgory 1 chemicals and for the Haz-Map® “Solvents”. These observations supported the assumption that defining of application domains for candidate compounds effectively highlighted the molecular characteristics or toxicological mechanisms underlying dermal acute toxicity of candidate compounds and reduced the diversity encountered among chemicals in the QSAR development. The repeated appearance of molecular descriptors reporting loations and bonding for atoms of significant electronegativity in the QSARs indicated that the density and re-distribution of electrons in the structure were highly associated with the transdermal penetration and acute toxicity of chemicals. The appearance of octanol-water partition coeffieient further attested to that the skin permeability of chemicals was a key factor controlling the dermal absorption-related toxicity.

誌謝 i
摘要 iii
ABSTRACT v
目錄 vii
表目錄 xi
圖目錄 xvii
第一章 緒論 1
第一節 研究緣起 1
第二節 研究之重要性 2
第三節 研究目的 3
第四節 研究假設 4
第五節 名詞界定 5
第二章 文獻探討 6
第一節 化學品之現行危害分類判斷機制與方法 6
第二節 替代性評估工具 9
第三節 量化結構−活性關係模式 10
第四節 分子特性及結構資料庫 13
第五節 文獻回顧總結 15
第三章 研究方法 16
第一節 研究設計 16
3.1.1　Dermal LD50 QSAR預測模式發展與驗證所需之標的化學物選定 16
3.1.2　Dermal LD50 QSAR預測模式建立所需之參數資料庫建立 17
3.1.3　Dermal LD50 QSAR預測模式之發展與驗證 18
3.1.4　對應Haz-Map®、TLV®、GHS危害等級、及不同化學物類別之dermal LD50 QSAR預測模式建立 25
第四章 研究結果與討論 26
第一節　樣本敘述 26
第二節　以美國國家醫學圖書館危害物質職業暴露資料庫中具經皮吸收急性毒性之化學物為基礎所發展之經皮吸收半致死劑量預測模式 31
第三節　以化學品全球分類及標示調和制度危害分級系統第一級毒性物質之化學物為基礎發展之經皮吸收半致死劑量預測模式 48
第四節　以美國政府工業衛生師協會授有閾限值之化學物為基礎發展之經皮吸收半致死劑量預測模式 66
第五節　以美國國家醫學圖書館危害物質職業暴露資料庫中屬農藥類之化學物為基礎發展之經皮吸收半致死劑量預測模式 82
第六節　以美國國家醫學圖書館危害物質職業暴露資料庫中屬有機溶劑類之化學物為基礎發展之經皮吸收半致死劑量預測模式 97
第七節　以美國國家醫學圖書館危害物質職業暴露資料庫中屬含氮化合物類之化學物為基礎發展之經皮吸收半致死劑量預測模式 113
第五章　結論與建議 130
第一節　結論 130
第二節　研究限制 134
第三節　應用與建議 134
參考文獻 137
附錄一 本研究發展經皮吸收半致死劑量量化結構−活性關係預測模式所使用收錄於美國國家醫學圖書館危害物質職業暴露資料庫之標的化學物與相關化學特性表列 142
附錄二 本研究所建立不同經皮吸收半致死劑量量化結構−活性關係預測模式於發展階段與驗證階段選用之標的化學物表列 176
附錄三 本研究發展經皮吸收半致死劑量量化結構−活性關係預測模式所使用具美國政府工業衛生師協會閾限值之標的化學物與相關化學特性表列 243

1. 勞動部：職業安全衛生法。臺北市，臺灣：行政院勞動部；2013。(民國102年7月3日修正條文)。
2. 勞動部：危害性化學品標示與通識規則。臺北市，臺灣：行政院勞動部；2014。(民國103年6月27日修正條文)。
3. 經濟部標準檢驗局：化學品分類及標示。CNS 15030。臺北市，臺灣：行政院經濟部標準檢驗局；2008。(民國97年12月29日修訂公布)。
4. United Nations Economic Commission for Europe (UNECE). Globally Harmonized System of Classification and Labelling of Chemicals (GHS). ST/SG/AC.10/30/Rev.6. New York, NY, USA and Geneva, Switzerland: UNECE; 2015.
5. Dotson GS, Chen CP, Gadagbui B, Maier A, Ahlers HW, Lentz TJ. The evolution of skin notations for occupational risk assessment: A new NIOSH strategy. Regulatory Toxicology and Pharmacology 2011; 61:53-62.
6. National Library of Medicine (NLM). Haz-Map??: Occupational Exposure to Hazardous Agent Database, http://hazmap.nlm.nih.gov/. Bethesda, MD, USA: NLM; 2013.
7. Doke SK, Dhawale SC. Alternatives to animal testing: A review. Saudi Pharmaceutical Journal 2015; 23:223-229.
8. National Institute for Occupational Safety and Health (NIOSH). NIOSH Current Intelligence Bulletin 61: A Strategy for Improvement of Skin Notations. DHHS (NIOSH) Publication No. 2009-147. Cincinnati, OH, USA: NIOSH; 2009.
9. Chen CP, Ahlers HW, Dotson GS, Lin YC, Chang WC, Maier A, Gadagbui B. Efficacy of predictive modeling as s scientific criterion in dermal hazard identification for assignment of skin notations. Regulatory Toxicology and Pharmacology 2011; 61:63-72.
10. The European Parliament and the Council of the European Union. Regulation (EC) No 1907/2006 of the European Parliament and of the Council of 18 December 2006 concerning the Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH). Official Journal of European Union 2006; L396 (30.12.2006):1-849.
11. 張晏青：皮膚暴露危害評估模式中皮膚滲透性及系統吸收閾值劑量之預測。臺中市，臺灣：中國醫藥大學；2013。
12. Maran U, Sild S. QSAR modeling of genotoxicity on non-congeneric sets of organic compounds. Artificial Intelligence Review 2003; 20:13-28.
13. Gramatica P. Priciples of QSAR models validation: internal and external. QSAR and Combimatorial Science 2007; 26:694-701.
14. Roy K, Kar S, Das RN. A primer of QSAR/OSPR modeling: foundamental concepts. Cham, Switzerland: Springer International Publishing AG; 2015.
15. Organisation for Economic Co-operation and Development (OECD). Guidance Document on the Validation of (Quantitative) Structure-Activity Relationships [(Q)SAR] Models. ENV/JM/MONO(2007)2. Paris, France: OECD; 2007.
16. Talete s.r.l. DRAGON?? User Manual. DRAGON?? for Windows?? and Linux 2007, Version 6, http://www.talete.mi.it/. Milano, Italy: Talete s.r.l.; 2013.
17. Chang YC, Chen CP, Chen CC. Predicting skin permeability of chemical substances using a quantitative structure-activity relationship. Procedia Engineering 2012; 45:875-879.
18. National Institute for Occupational Safety and Health (NIOSH). Features of the RTECS?? Database, http://www.cdc.gov/niosh/rtecs/ rtecsfeatures.html. Cincinnati, OH, USA: NIOSH; 2011.
19. Linstroom PJ, Malard WG. The NIST Chemistry WebBook: A Chemical Data Resource on the Internet. Journal of Chemical and Engineering Data 2001; 46:1059-1063.
20. National Library Medicine (NLM). TOXNET??: Toxicology Data Network Fact Sheet, https://www.nlm.nih.gov/pubs/factsheets/
toxnetfs.html. Bethesda, MD, USA: NLM; 2016.
21. National Library Medicine (NLM). TOXNET?? & Beyond: Using the National Library of Medicine?? Environmental Health and Toxicology Protal. Bethesda, MD, USA: NLM; 2014.
22. American Conference of Governmental Industrial Hygienists (ACGIH). 2013 TLVs?? and BEIs?? Based on the Documentation of the Threshold Limit Values for Chemical Substances and Physical Agents & Biological Exposure Indices. Cincinnati, OH, USA: ACGIH; 2013.
23. National Institute for Occupational Safety and Health (NIOSH). Registry of Toxic Effect of Chemical Substances (RTECS??) Database. CD-ROM. Cincinnati, OH, USA: NIOSH; 2001.
24. Sweet DV. Registry of Toxic Effects of Chemical Substances (RTECS??): Comprehensive Guide to the RTECS??. DHHS (NIOSH) Publication No. 97-119. Cincinnati, OH, USA: NIOSH; 1997.
25. Organisation for Economic Co-operation and Development (OECD). Test No. 402: Acute Dermal Toxicity. Paris, France: OECD; 2007.
26. National Institute of Standards and Technology (NIST). Chemistry WebBook, http://webbook.nist.gov/chemistry/. Gaithersburg, MD, USA: NIST; 2011.
27. National Library of Medicine (NLM). Toxicology Data Network (TOXNET), http://chem.sis.nlm.nih.gov/. Bethesda, MD, USA: NLM; 2005.
28. Hypercube Inc. Reference Manual. HyperChem?? Release 8.0 for Windows?? Molecular Modeling System. Gainesville, FL, USA: Hypercube Inc.; 2008.
29. The MathWorks Inc. MATLAB?? Statistical Toolbox v7.7 — User Guide. Natick, MA, USA: The MathWorks Inc.; 2009.
30. Chan CC, Liaw HJ, Tsai YJ. Prediction of flash point of organosilicon compounds using quantitative structure property relationship approach. Industrial and Engineering Chemistry Research 2010; 49:12702-12708.
31. Martin TM, Harten P, Young DM, Muratov EN, Golbraikh A, Zhu H, Tropsha A. Dose Rational selection of training and test sets improve the outcome of QSAR modeling? Journal of Chemical Information and Modeling 2012; 52:2570-2578.
32. Topliss JG, Costello RJ. Chance correlations in structure activity studies using multiple regression analysis. Journal of Medicinal Chemistry 1979; 15:1066-1068.
33. Golbraikh A, Tropsha A. Beware of q2! Journal of Molecular Graphics and Modelling 2002; 20:269-276.
34. Sahigara F, Ballabio D, Todeschini R, Consonni V. Assessing the validity of QSARs for ready biodegradability of chemicals: an Applicability Domain perspective. Current Computer-Aided Drug Design 2014; 10:137-147.
35. Sahigara F, Mansouri K, Ballabio D, Mauri A, Consonni V, Todeschini R. Comparison of different approaches to define the Applicability Domain of QSAR models. Molecules 2012; 17:4791-4810.
36. Lowell H. Hall, Mohney B. The electrotopological state: an atom index for QSAR. Quantitative Structure-Activity Relationships 1991; 10:43-51.
37. Dvoř?鴕 Z, Lidický B, Škrekovski R. Randić index and the diameter of a graph. European Journal of Combinatorics 2011; 32:434-442.
38. Randić M. On characterization of molecular branching. Journal of the American Chemical Society 1975; 97:6609-6615.