18
Z. Wang and J. Chen
toward a predictive science that would increasingly rely on in vitro tests based on
human tissues and cells was advocated. in silico/computational models were suggested to characterize the toxicity pathways and to assess the exposure, hazard, and
risk of chemicals, in order to reduce the time and expense spent, and the number of animals sacrificed and to extend knowledge on the mechanisms of toxic
effects.
Toxicity pathways generally refer to biochemical/cellular signaling pathways that
if improperly perturbed would eventually lead to adverse health effects. Traditionally, descriptive toxicity end points or apical end points such as individual death
or abnormal behavior could be thus reduced or attributed to multiple nodes or key
events anchored along the toxicity pathways or networks. The nodes could then be
examined with specifically designed in vitro tests. Furthermore, efficiency of in vitro
biomacromolecule or cell-based assays has been significantly enhanced by automatic
technology, which resulted in so-called high-throughput screening (HTS) technology capable of performing over 1500-well-plate level operations and readouts with
just a single run [9, 10]. HTS technology paved the way for projects such as ToxCast
[11] and Tox21 [12] that aim at screening relatively large libraries of chemicals and
generate the so-called toxicological big data [13].
Novel/alternative experimental methods such as HTS indeed promote the development of toxicology. However, they have met their own problems [14, 15]. Historically, in vivo end points have formed the basis for chemicals risk assessment and are
deemed so-called “golden standards” by some toxicologists and regulators. In fact,
most cell-based in vitro end points cannot be simply mapped onto the traditional
in vivo end points [16]. False positive hits resulting from in vitro tests that are inconsistent with available in vivo evidences hence become a nuisance. It is thus necessary
to clearly delineate the relationship between in vitro and in vivo end points, which
emphasizes in vitro–in vivo extrapolation (IVIVE) that aims at applying the in vitro
results for evaluating in vivo effects [17, 18]. Optimistically, adverse outcome pathways (AOPs) coined by Ankley et al. [19] appear to be a straightforward conceptual
framework to explain the specifically designed in vitro end points and apical in vivo
end points as molecular initiating events (MIEs)/key events (KEs) and adverse outcomes (AOs), respectively. However, a quantitative AOP that is applicable to IVIVE
is still far from real practice. Besides, the types of in vitro end points that can be
implemented on an HTS platform are still limited [12]. The implementation of HTS
that relies on specific apparatus cannot be adequately cost-effective. Moreover, the
number of chemicals screened by Tox21 during 2008–2013 is ca. 10,000 [14], which
is de facto much less than the number (>15,000) of newly registered chemicals in a
single day on the chemical abstract service (CAS, www.cas.org) system in 2014. In
conclusion, the current experimental system for toxicology can meet neither the need
for chemicals risk assessment nor the requirement for development of toxicology.
Thus, an auxiliary yet critical field for complementing the experimental sector of
chemicals risk assessment has emerged: computational toxicology [20].
Z. Wang and J. Chen
toward a predictive science that would increasingly rely on in vitro tests based on
human tissues and cells was advocated. in silico/computational models were suggested to characterize the toxicity pathways and to assess the exposure, hazard, and
risk of chemicals, in order to reduce the time and expense spent, and the number of animals sacrificed and to extend knowledge on the mechanisms of toxic
effects.
Toxicity pathways generally refer to biochemical/cellular signaling pathways that
if improperly perturbed would eventually lead to adverse health effects. Traditionally, descriptive toxicity end points or apical end points such as individual death
or abnormal behavior could be thus reduced or attributed to multiple nodes or key
events anchored along the toxicity pathways or networks. The nodes could then be
examined with specifically designed in vitro tests. Furthermore, efficiency of in vitro
biomacromolecule or cell-based assays has been significantly enhanced by automatic
technology, which resulted in so-called high-throughput screening (HTS) technology capable of performing over 1500-well-plate level operations and readouts with
just a single run [9, 10]. HTS technology paved the way for projects such as ToxCast
[11] and Tox21 [12] that aim at screening relatively large libraries of chemicals and
generate the so-called toxicological big data [13].
Novel/alternative experimental methods such as HTS indeed promote the development of toxicology. However, they have met their own problems [14, 15]. Historically, in vivo end points have formed the basis for chemicals risk assessment and are
deemed so-called “golden standards” by some toxicologists and regulators. In fact,
most cell-based in vitro end points cannot be simply mapped onto the traditional
in vivo end points [16]. False positive hits resulting from in vitro tests that are inconsistent with available in vivo evidences hence become a nuisance. It is thus necessary
to clearly delineate the relationship between in vitro and in vivo end points, which
emphasizes in vitro–in vivo extrapolation (IVIVE) that aims at applying the in vitro
results for evaluating in vivo effects [17, 18]. Optimistically, adverse outcome pathways (AOPs) coined by Ankley et al. [19] appear to be a straightforward conceptual
framework to explain the specifically designed in vitro end points and apical in vivo
end points as molecular initiating events (MIEs)/key events (KEs) and adverse outcomes (AOs), respectively. However, a quantitative AOP that is applicable to IVIVE
is still far from real practice. Besides, the types of in vitro end points that can be
implemented on an HTS platform are still limited [12]. The implementation of HTS
that relies on specific apparatus cannot be adequately cost-effective. Moreover, the
number of chemicals screened by Tox21 during 2008–2013 is ca. 10,000 [14], which
is de facto much less than the number (>15,000) of newly registered chemicals in a
single day on the chemical abstract service (CAS, www.cas.org) system in 2014. In
conclusion, the current experimental system for toxicology can meet neither the need
for chemicals risk assessment nor the requirement for development of toxicology.
Thus, an auxiliary yet critical field for complementing the experimental sector of
chemicals risk assessment has emerged: computational toxicology [20].