前言
很荣幸有机会再次向广大读者介绍这本毒理学百科全书。第二版百科全书在第一版
相关内容的基础上，进行了扩展和精练，以便能够很好地服务于毒理学工作者。特别是在
目前科学分类日益精细，导致个人关注面相对狭窄的情况下，此书有助于大家对毒理学学
科的整体研究范畴和功能作用作一个更全面的理解。
自第一版百科全书出版以后，毒理学的发展又发生了显著的变化，而且现在这种变化
有加速的趋势。经过四五十年的发展，毒理学从主要以体内毒性为基础的描述性科学，发
展为包含现代生物学和化学各方面的综合性科学，从分子生物学到尖端仪器分析都有涉
及。其理论基础已经从原来的主要基于病原检测和体内毒性分析的常规危害分析检测，
转变为在器官、细胞和分子水平上强调毒性机制。这些变化也带来了大量毒理学著作的
发行。
从那时起，分子生物学技术在毒理学的各个方面发挥了越来越重要的作用，如毒性机
制的阐明、外源性化学物质代谢研究、更安全有效的药物和化学物开发、毒物暴露和效应
生物标志物的研究等等，这些重要的毒理学研究都受分子生物学技术影响。此外，分析化
学也继续被应用于微量外源性化学物质的检测，其检测灵敏度非常高，甚至在毒物效应短
期内还未表现时就能将其检测到。目前，因为危害评价很大程度上还是依靠数学模型而
不是毒理学科学，这些新的科学在危害或风险性评价的应用方面还有很多问题，但是人类
的健康危害评价和环境风险评估的发展还是很显著的。
始终不变的是毒理学科技文献服务于广大专家的需求。考虑到毒理学方法基础和实
践需要的双方面特征，现在更需要有综合性的著作。毒理学百科全书在其中可以起到很
好的媒介作用，它比字典更详细，更便于从事危害评价、法规、教育和咨询等工作的普通工
作者以及专家们获得所需要的信息，这些信息往往是他们的专长之外的知识。它还为非
毒理学家了解毒理学的科学体系、原理、方法和应用这门科学提供了方便。
总之，此书是对毒理学科学的巨大贡献，是热衷于毒理学的科学家和服务于毒理学家
的图书馆的必备之物。
Ernest Hodgson
William Neal Reynolds Professor
Environmental and Molecular Toxicology
North Carolina State University
(周平坤 译)
第二版序
时光流逝，人们对毒理学的理解和认识的追求仍在持续。我们一直希望能够结束贫
穷、无知、饥饿和有害化学物暴露等问题，也一直在向这些目标努力，但困难巨大，目标依
然遥不可及。化学物和由化学物而来的终产物在我们的生活中仍然扮演着重要的角色。
尽管现在还不确定化学物的好处一定能超过它们带来的危害，但人们很少怀疑的是：为数
众多的化学物和药物可以延长人类寿命，提高人们的生活质量。与此同时，某些化学物也
会在特定的情况下对特定的人群产生危害作用。毒理学家的研究成果中就包括提供信
息，告诉人们如何去更好地消除、减少和预防这些危害。
第一版毒理学百科全书发表后的7年中，毒理学学科的发展迈出了很大的一步。
目前，分子毒理学知识在持续快速地发展。的确，通过研究取得数据结果的过程相对
容易，而花时间对资料进行充分的分析和评估则要困难得多。基因组学、蛋白质组学
和其他组学技术的发展，使我们能更有效地揭示环境化学物暴露和疾病易感性之间的
复杂联系。美国国家毒物基因组中心(The US National Center for Toxicogenomics)建立
于2000年，致力于信息学和计算机毒理学的研究。该机构的研究成果与其他团体的
研究一道，使我们能用更有效的方法来确定化学物的安全性，研究结构一活性关系。另
外，随着分析工具越来越精细和敏感，现在可以对生物系统和环境中更微量的污染物
进行检测和定量。
现在越来越多的消费者(特别是在西方国家)更倾向于接受药补和替代医学疗法，人
们会比以前接触更大量的草药和其他植物类药物。尽管毒理学家一直认为“天然的”不等
于是“安全的”，但对草药添加物以及它们与其他化学物相互作用的危害性评价，所做的工
作并不多。然而，这种状况正在发生变化。
化学战、生物战和核战争历来就是社会关注的热点，它们有时是作为现实事件，更多
的是作为学术问题引起关注。在2001年9．11事件发生后，人们紧迫地意识到：需要了解
非常规战争及其战剂的知识，如它们如何发挥作用、如何使人类自己免受危害等。毒理学
的研究范畴也扩展到防范这种武器复活的危害。
随着现代科技的发展，制造业、加工过程以及化学物和其他产品的使用使我们获益，
但也加速了有害废弃染物污染的扩散。在美国，每年有两百万吨的电子产品废弃待处理，
包括5 000万台电脑和13 000万部手机。据国际电子产品回收协会统计，到2010年这种
电子废品的数量将增加3倍以上。如此巨量的废物埋藏在山地或水域中，将不可避免地
对我们的空气和水造成污染，其潜在的污染物包括铅、镉和铍。
动物实验的替代实验正在逐渐渗透到毒理学研究领域中。虽然整体动物实验还不会
很快消失，但其他一些确定危害性和安全性的方法，现在越来越受到毒理学界的追捧，并
且正在成为化学物评价的主流方法的一部分。“体外实验”方法(如细胞培养和皮肤刺激
试验)和计算机分析方法提供的毒性信息正在日益增加。其中计算机分析方法基于已有
的数据资料，利用计算机程序来评估类似化学物的毒性特征，有无补充的化学和物理的特
征数据均可。
现在市场上纳米技术产品已经越来越多，纳米技术的研究和发展正处于上升势头。
美国于2001年建立了国家纳米技术中心，联邦机构和大学也开始研究纳米材料对环境和
人类健康的影响。
对化学物暴露(包括实际的和预计的)的深入研究，有助于我们更好地了解化学物对
环境和人类健康的危害性。毒理学家与危害评估管理者问不断增加的合作非常必要，因
为密切的合作可以使危害评估有确实的科学依据，为评估者和管理者们提供可信的材料，
从而加强化学物危害的控制。
在全球范围内，对化学物的控制和管理已经取得了很大的发展。里约热内卢地球峰
会后10年，世界可持续发展峰会(WSSD)于2002年在南非约翰内斯堡举行。会议设立
的目标之一是，到2020年，以不对人体健康和环境产生明显副作用的方式使用和生产化
学物。
2004年5月17日，保护人类健康和环境免受持续性有机污染物(POPs)危害的斯德
哥尔摩公约正式生效。POPs是有毒性、持续存在、具有累积性、并且能在环境中长距离
传播的有机污染物。此公约的目的是消除和限制这类化学物的生产和使用。而京都议定
书是用来限制温室气体排放的，现在已经成为国际公认的法律，只有极少数国家对它进行
抵制。
美国拥有活跃的、不断壮大的毒理学专业队伍，他们进行着开创性的毒理学研究工
作，其他国家的科学家也是如此。通过因特网及其优化技术可用性的不断增强，各国科学
家可以方便的进行合作和信息共享。有些重要的工作在多国共同体的赞助支持下积极开
展，如经济合作与发展组织、欧盟委员会、国际化学安全署等。
全球范围的毒理学资料和信息的协调和联络工作也在积极地开展。用于化学物分类
和标记的全球调和制度(GHS)已经被采纳，并且即将付诸执行。这将提供统一、连续的
途径对有害化学物进行识别，并且能提供有害物的信息和针对暴露人群的防护措施。同
时，在欧盟，一个叫做《化学品的注册、评估和授权》(REACH)的法规框架已建立，其提出
了对于年产量或进口量超过1吨的化学品进行登记的制度。
最后，毒物在个人和政治谋杀或仇杀中的作用：它可能在博尔吉亚家族(Borgias)统
治时代达到顶峰，但从那以后并没有终止。特别提出的一次事件发生在2004年乌克兰总
统竞选期间。经过激烈的竞选，亲西方反对派领袖尤先科获胜，并且于2005年1月举行
就职典礼。这个民主的快乐节日却被中毒事件毁坏。医生称尤先科因为大剂量二嗯英中
毒导致脸部毁容和其他疾病。据说二囗英是被混合于他食用的汤中而导致其中毒的。尽
管整个事件还没有水落石出，但怀疑其中有政治动机。
第二版百科全书汇集了392作者提供的1057条目录，而第一版只包括200名作者提
供的749条目录。实际上，第一版中所有条目的内容在第二版中都得到了充实，而某些条
目则完全进行了更新。在第二版中新出现的308个条目中，包括鸟类生态毒理学、基准剂
量、杀虫剂、计算机毒理学、致癌效力因素、代谢组学、化学事件、蒙地卡罗分析、非致死性
化学武器、无脊椎动物生态毒理学、药物滥用、癌症化疗因子和消费产品等等。许多针对
特殊化学物的条目都是全新的，而书中搜录的国际组织也得到了很大的扩展。某些条目
罗列了许多著名的毒物中毒相关事件：如美国拉夫运河事件、泰晤士海滩事件、切尔诺贝
利事件和三里岛事件等。另外许多条目涉及毒理学知识的社会应用，如文化毒理学、环境
<节选内容>=犯罪、著名投毒者和中毒事件、古代化学生物战以及美国环境运动历史等。因此，新版的
毒理学百科全书在长度、广度、以及深度上都对毒理学的许多方面进行了更为广泛和深入
的概述。
Philip Wexler
(周平坤 译)
第一版序
现在有许多关于毒理学方面的普通或专业的单行本著作，其中大多数都针对于毒理
学家和毒理专业的学生，少部分适用于非专业人员。这本毒理学百科全书的出版不是为
了取代这些著作，而是希望能满足更广大读者和新的读者群对毒理学信息和知识的需求。
所以在编写上采取了一种新的组织形式，减少了入门级知识和专业论文的数据资料元素，
使本书定格于中级水平。
本百科全书尽管不求详尽无遗，但其内容仍然非常宽广。我们的想法是应考虑到
毒理学方面基础的、关键的甚至有争议的要素，因为这些内容要素是理解毒理学学科
基础及派生的社会影响所必需的。因此，这本百科全书不仅必须包括一些重要概念，
如剂量效应、作用机制、试验程序、终点反应、靶位等，还应包含单个的化学物及其分
类。尽管本书的重点在化学物上，我们也搜集了其他一些概念如辐射和噪音，并介绍
了毒理学相关的历史、法律、条例、教育、组织和数据库。本书还需考虑到环境和生态
毒理学，以便在某种程度上平衡那些侧重于实验动物和人类的知识点，因为最终需要
将所有的联系深入。
在化学物选择方面，编者们根据所掌握的知识，选取了那些具有相对较高的毒性、暴
露量、产量、争议性、受关注度或在其他方面引起人们兴趣的化学物。这些化学物并不代
表一个化学物条规清单或数据库组合，而是我们认为毒理学必须关注的化学物质。本书
没打算成为一本大规模的有毒化学物的一览表(这种一览表现在已经有好几本)。
依照许多传统的科学或其他百科全书的标准，本书完全按字母顺序排列。除了一些
有用的小字典外，此编排方式在毒理学上还没采用过。这种编排，再加上详细的索引和广
泛的交叉参照，应该使读者能够快速地查阅到所需要的信息。
尽管本书对毒理学家会相当有用，但更适用于其他人员，包括普通科学工作人员、医
生、法律和条例专业人员和有科学基础的非专业人员。在他们工作、学习的过程中，或者
有兴趣的时候，如果需要了解毒理学方面的知识，就可在此全书中获取。毒理学家在需要
温习或者快速浏览一下非本专业的内容时，也可以从中查阅；但要寻求深入的处理方案
时，就有必要去查阅专题论著或者专业期刊文献。
本百科全书的目的在于对那些有时很复杂的项目进行简要的概述。书中没有正式的
参考文献和脚注，因为这些与百科全书的目的相关性不大。它只有简单的目录引导读者
到相关的特别条目下去阅读更详细的信息。如信息资源的条目可以引导读者了解毒理学
方面的电子信息。
首先，我要感谢副主编们和所有撰稿人，他们的努力成就了本百科全书。感谢原策划
编辑Yale Altman和Linda Marshall，他们所做的铺垫性工作对本书的编辑出版有很大
的帮助；感谢来自Academic Press的现策划编辑Tari Paschall和高级制作编辑Monique
Larson，他们的学识、专长和高效率工作使本书得以顺利完成；原稿的组织和格式审定工
作由Mary Hall、Christen Bosh和Jennifer Brewster完成，他们进行了非常技巧性的、耐
心和细致的工作。
我以个人的名义而不是政府职员的身份参加本百科全书的工作。书中的观点仅代表
我的个人观点，不代表美国国家医学图书馆或美国联邦其他机构的意见。
Philip Wexler
(周平坤译)
LD5o/LC5o (Lethal Dosage 50/Lethal Concentration 50)
Shayne C Gad
~ 2005 Elsevier Inc. All rights reserved.
Introduction
The 50% lethal dose (LD50 or LD50) is the statis-
tically calculated dose (or concentration) of a mate-
rial (generally expressed as the amount of material
per unit of body weight) that would be expected to
cause the death of half the members of the target
species receiving it. The 50% lethal concentration
(LC.so) is the equivalent statistical projection for in-
halation. Until the mid-1980s, these figures were
considered perhaps the basic component of a toxicity
profile for any chemical or drug. However, after
many years of controversy and debate on a number
of fronts, including objections from animal rights
advocates, three alternative animal tests have been
developed to replace the LD50. They are the fixed-
dose procedure, the acute toxic class method and the
up and down (or 'up/down') method, and their use
has led to significant improvements in animal wel-
fare. These new tests have undergone revision and
refinement to improve their scientific performance
and to increase their regulatory acceptance. Further,
research into replacements for test animals, such as
cellular cultures, organs harvested from slaughter-
houses, in silico (computer) modeling, and physical/
chemical systems, has been extensive. While these
approaches will not be able to completely replace the
use of animals in the foreseeable future, they have a
bright future.
The in vitro cytotoxicity tests that have been
developed can already help reduce the number of an-
imals used in acute oral toxicity testing. For example,
cytotoxicity data are being used to determine the
starting dose for in vivo testing by applying a stand-
ard regression between cytoxicity and acute oral
LD50 values. A database of correlations between
cytotoxic responses and the acute oral LD50 of rats or
mice has been determined for hundreds of chemicals.
Using this approach, it has been proposed that a
tiered in vitro/in vivo testing process will reduce an-
imal use in the up/down method. For example, the in
vitro cytotoxicity of a new chemical is determined as
the first step, and the LDs0 value (mgkg -1) is pre-
dicted from the cytotoxicity data. The predicted LD50
dose is then used as the starting dose in the up/down
protocol.
In silico models (or 'expert systems') have also
been developed. These are computer software-based
structure-activity relationship and quantitative struc-
ture-activity relationship analyses of data libraries of
acute toxicity data developed for use in evaluating
and predicting the acute oral and inhalation toxicity
potential of a chemical or drug.
While most of the focus has been on the potential
for oral toxicity, in vitro testing and computer mod-
eling for the evaluation and prediction of respiratory
toxicity are also being developed. For example, one
strategy consists of checking the existing data avail-
able for the test material itself, or on related sub-
stances, followed by acquiring knowledge on the
physicochemical properties of the test material. These
steps are followed by the use of computer modeling
techniques to try to predict the likely toxic effects and
target sites. In vitro tests could then follow to identify
likely target cells and evaluate the specific effects on
the cells (e.g., morphology could be determined and
assessments of the cellular energy status). A further
phase of in vitro tests could then be conducted on the
basis of results obtained in the first phase of in vitro
testing, choosing from tests using various types of
respiratory tract cells.
When did modern Western society become con-
cerned with lethality testing? For what reasons were
protocols developed for describing lethality in ani-
mals in quantitative terms for the purposes of
making scientific, regulatory, or marketing decisions?
Interestingly, in this age of genetic engineering, few
people realize that biologically derived materials
were the subject of regulations well before the
passage of the Pure Food and Drug Act in the Unit-
ed States in 1906. In 1901, a diphtheria epidemic
broke out in St. Louis, MO, because of improperly
manufactured antidiphtheria toxin. In response to
the resulting public outcry, the US Congress passed
the Virus Act of 1902. It regulated all viruses, se-
rums, toxins, antitoxins, and other such products
sold for the prevention or cure of disease in man.
Among other things, the bill eventually established
consistent potency criteria. In fact, by World War II
the US FDA was requiring batch-to-batch certifica-
tion and release for biologicals, a policy that remains
in effect for certain drugs. Hence, the earliest lethal-
ity testing was for the purpose of establishing con-
sistent potencies of biologicals, such as diphtheria
toxin, and not for evaluating synthetic chemicals.
One of the earliest publications discussing lethality
testing was an investigation into the lethality of
diphtheria toxin in guinea pigs. The publication de-
scribed lethality empirically in terms of percentage of
dead animals at each dosage because methods for
calculating lethality curves and the median lethal
dosage had not yet been developed. The authors re-
ported that lethal response to a given dosage of toxin
varied with the time of the year. Hence, years before
the term LDs0 came into parlance, supposedly as an
exact indicator of toxicity, data had been published
attesting to the volatility and imprecision of this cal-
culated parameter.
Because the first use of lethality testing was in
describing the potency of biologicals, it only makes
sense that the same methods were soon applied to
extracted botanicals. (Note: There is no doubt that
both the Germans and the English tested in animals
the various poison gases employed during World War
I. Little of this work, however, appears to have been
published in the open scientific literature, although
portions of it have recently been made public.} In
1926, de Lind van Wijngaarden published on the le-
thality of digitalis extracts. Interestingly, he did not
plot his data as mortality versus dose. He delivered
his extracts intravenously and titrated the dosage
until he achieved complete heart stoppage. He was
thus able to determine the precise lethal dosage for
each animal and noted that these followed a bell-
shaped or Gaussian distribution. His experiments
took 5 years and used more than 500 cats, an effort
that would have been excessive and expensive by to-
day's standards. However, he did conclude that no
more than nine cats would normally be required to
'calibrate' an extract of digitalis. Trevan, in a pivotal
paper (1927), described the lethality of strophanthin,
cocaine, and insulin. Modern reviews have focused a
great deal of attention on the large number of frogs
used by Trevan. Most of the data he discussed,
however, were derived from experiments in mice
using cocaine or insulin. Perhaps so little attention
was given to this aspect of Trevan's paper, even
though it comprised the major portions of his work
(which ran to 31 pages and contained 11 figures and
six tables of data), because it has never been repli-
cated. For some of the lethality curves reported by
Trevan, well over 900 mice were dosed. Again, such
efforts would be excessive and expensive by today's
standards but were necessitated, in part, by the less
rigorous method of deriving lethality curves and ca-
lculating the median lethal dosage (LDs0). Modern
methods of data transformation and statistical anal-
ysis were, at that time, still in their infancy. He also
recognized that it was not necessary to describe an
entire dosage-response curve to calculate an LD50.
He, in fact, recommended that lethality determina-
tions start with small groups of two or three animals
each and that larger groups be used for confirmatory
purposes.
Behren confirmed the observations of both de Lind
van Wijngaarden and Trevan. It is clear from his ar-
ticle that the use of animals for standardizing digita-
lis extracts was accepted to the point of being
incorporated into the German and Dutch pharma-
copoeias. The objective of his paper was to compare
the cat and frog methods and develop a basis for
using fewer animals. He concluded that the frog
method was superior and that no more than 44 frogs
needed to be used, which was considerably less than
the 100-200 frogs prescribed in the German phar-
macopoeia of that period. Interestingly, these early
papers are often criticized with regard to the num-
bers of animals used, but the objectives and conclu-
sions are often ignored.
Both Trevan and Behrens noted that when the
percentage of animals that died at specific dosages
was plotted against the logarithm of the dosage, the
resulting curve (the lethal dosage curve) had a sigmo-
idal shape slope and range that was 'characteristic'
for the species and the test substance. Shackell (1925)
first pointed out that such curves are integrated or
cumulative frequency curves (or ogives) and coined
the term 'dose-response ogive' (curve). Trevan noted
that these curves owe their shape to the fact that
different individual animals require different quanti-
ties of poison for death to occur. It was also Trevan
who identified the midpoint on this curve as being
the dosage that would kill 50% of the animals ex-
posed. He designed that point as the median lethal
dose, or LD50, and, thus, is widely credited with
having developed the classical LD50. Trevan and
Behrans essentially read the LD50 directly from their
mortality dose-response curves.
Lethality testing of biologicals and botanicals was
essentially a response to governmental regulation. It
was only natural that similar methods would be
applied to synthetic chemicals. Major chemical com-
panies started establishing toxicity or industrial
health laboratories during the 1930s; the lethality
testing of synthetic chemicals was established by the
1930s. However, there were no regulatory require-
ments for such tests. In fact, there was no premarke-
ting toxicity testing of synthetic chemicals required at
all. In 1937, an elixir or sulfanilamide dissolved in
ethylene glycol was introduced into the market. Over
100 people died as a result of ethylene glycol toxicity.
The public response to this tragedy helped prompt
the US Congress to pass the Federal Food, Drug, and
Cosmetic Act of 1938. It was this law that mandated
the premarket testing of drugs for safety in experi-
mental animals. By the mid-1940s, most chemical
and pharmaceutical companies were routinely testing
new chemicals for lethality. In fact, until the 1960s,
preclinical or premarketing toxicity data packages
normally consisted of little more than acute lethality
data. Recently, new laws, increased scientific sophis-
tication, and greater societal concern over sublethal
chronic toxicity has led to more extensive and ex-
pensive preclinical or premarketing toxicity testing
packages, where acute lethality is a small, but still
real, concern.
The protocols used to assess lethality have changed
considerably since the 1920s. While the principles
originally described by Trevan have never been ques-
tioned, the methods for calculating the LDs0 have
become more sophisticated and the need for a high
degree of precision has been questioned. The prac-
tical result is that by using modern protocols, re-
latively few animals are species (generally rats and
mice) are employed and only two routes of admin-
istration are used. At least one route must be the
intended or the most probable human exposure
route. Hence, such protocols generally result in the
generation of eight lethal dosage curves (one/route/
sex/species). In the drug industry (where this ap-
proach is common), the two routes are generally oral
and intraperitoneal for an oral drug and oral and
intravenous for an intravenous drug.
Protocol Design Considerations
Whatever type of experimental protocol one chooses
to use in a lethality test, there are certain principles
and criteria that should be universally applied. The
principles are especially relevant in studies in which
small numbers of animals are used.
First, a wide variety of intrinsic and extrinsic fac-
tors can influence the outcome of a lethality test.
These include species, strain or substrain, age, weight,
and sex of the animals; husbandry practices (e.g., type
of bedding and cage population}; environmental con-
ditions; feed and water quality; nutritional state; and
volume and vehicles of test substance delivery. The
point to be made here is that the criteria for all these
factors should be specified in detail in the protocol
and strict adherence to the protocol observed. Oth-
erwise, intrastudy comparisons are invalid. Small
differences in protocols can cause large differences
in the LDs0 and are probably the major cause of
the considerable laboratory-to-laboratory variation
in the LDs0s.
Second, because the animals will generally receive
a single exposure, great care must be given to the
preparation and delivery of the test articles. In a
chronic study, occasional miscalculations or mis-
delivery of the dosage would not generally greatly
affect the study outcome but would clearly have a
greater effect on the conclusions of a lethality screen.
One should always include appropriate safeguards.
Third, one must make sure that all animals are
successfully dosed and that accidental deaths are
identified as such. In acute rodent studies, we rou-
tinely assign spare animals to a dosing group. Per-
manent numbers are not assigned until we are certain
that the dose has been successfully delivered (e.g.,
Was the supposedly intraperitoneal dose accidentally
delivered intravenously? Was there reflux from the
site?). Spare animals not dosed are returned to the
pool of animals available for the study. Animals
found dead should be examined for evidence of ac-
cidental trauma. For example, it is not uncommon
for a rat to suddenly move while being gavaged. This
may result in a torn esophagus that may take 1 or
2 days to become evident. Depending on the admin-
istration route, one must pay close attention to do-
sing techniques and the volume limitations imposed
by these techniques. For example, 20 mi kg- 1 is the
maximum volume that should be given orally to a
rodent. Deaths that are clearly accidental should not
be considered in the final conclusions.
Fourth, lethality protocols, by the nature of the
question they address, do not specify all dosages.
This can sometimes result in a study in which ab-
surdly high dosages are administered. Hence, all
protocols should clearly state what the ceiling or
limit dosage will be and the reasons for selection.
