9 A Pair Ranking (PRank) Method for Assessing Assay …
161
and promoted by REACH legislation [3] and the 7th Amendment of the Cosmetics
Directive [4]. In the USA, the Tox21 program is led by the EPA [5–7] and ToxCast
was initialized by a cross-agency effort [8] to advancing regulatory science [9]. This
program was proposed by the US Food and Drug Administration (FDA) to promote in vitro and in silico approaches for enhancing risk assessment and eliminating
animal use.
Toxicogenomics (TGx) is a sub-discipline of pharmacology that offers a groundbreaking addition to conventional toxicology approaches [10]. TGx approaches have
been widely applied in addressing different toxicological questions at the molecular
level [11]. For example, Fielden et al. [12] employed a five-day rat in vivo TGx
model to predict the nongenotoxic carcinogenicity, which generated better prediction performance with a mechanistic assessment of underlying mechanism. In the
following studies, researchers expanded the studies for both size and treatment duration to achieve a better prediction performance with accuracies in the range 75–80%
[13–16]. Furthermore, some studies utilized cell-based in vitro (HepG2 cells) TGx
assay to uncover different mechanisms between genotoxic and nongenotoxic hepatocarcinogens [17] and further developed cell-based in vitro TGx prediction models [18]. Herwig et al. [18] developed human in vitro TGx models in hepatomaderived cells and hTERT-immortalized renal proximal tubule epithelial cells to predict nongenotoxic carcinogens. Huang et al. [19] employed a serum-based TGx assay
to predict drug-induced liver injury, which yielded a 92.1% accuracy with several
important pathways including Toll-like receptor signaling, apoptosis and mitochondrial damage-related DILI mechanism enriched. Advances in emerging technologies
such as next-generation sequencing and bioengineering including iPSC cell culture
have also been integrated into TGx field and provide a promising approach for risk
assessment [20, 21].
Unlike decades ago, several large public available TGx data sets such as open
TG-GATEs [22, 23], DrugMatrix [24] and PredTox [25] have been generated and
currently provide tremendous opportunities for formulating hypotheses to advance
toxicology researches. Considering the different assay types, species, and genomics
technologies in the preclinical setting, a comprehensive assessment among different
TGx assay systems in the preclinical setting is urgently needed for selecting the
fit-for-purpose approach.
To fill the gap, we developed a Pair Ranking (PRank) method to assess the transferability among the different TGx assays and utilized our method to address several
key questions in the TGx field. In this chapter, we will first lay out the key questions
for promoting in vitro TGx systems in risk assessment. Then, we will elaborate on
the PRank method with a few case studies. Final, we will summarize the roadmap for
further positioning the proposed PRank method toward potential regulatory applications.
161
and promoted by REACH legislation [3] and the 7th Amendment of the Cosmetics
Directive [4]. In the USA, the Tox21 program is led by the EPA [5–7] and ToxCast
was initialized by a cross-agency effort [8] to advancing regulatory science [9]. This
program was proposed by the US Food and Drug Administration (FDA) to promote in vitro and in silico approaches for enhancing risk assessment and eliminating
animal use.
Toxicogenomics (TGx) is a sub-discipline of pharmacology that offers a groundbreaking addition to conventional toxicology approaches [10]. TGx approaches have
been widely applied in addressing different toxicological questions at the molecular
level [11]. For example, Fielden et al. [12] employed a five-day rat in vivo TGx
model to predict the nongenotoxic carcinogenicity, which generated better prediction performance with a mechanistic assessment of underlying mechanism. In the
following studies, researchers expanded the studies for both size and treatment duration to achieve a better prediction performance with accuracies in the range 75–80%
[13–16]. Furthermore, some studies utilized cell-based in vitro (HepG2 cells) TGx
assay to uncover different mechanisms between genotoxic and nongenotoxic hepatocarcinogens [17] and further developed cell-based in vitro TGx prediction models [18]. Herwig et al. [18] developed human in vitro TGx models in hepatomaderived cells and hTERT-immortalized renal proximal tubule epithelial cells to predict nongenotoxic carcinogens. Huang et al. [19] employed a serum-based TGx assay
to predict drug-induced liver injury, which yielded a 92.1% accuracy with several
important pathways including Toll-like receptor signaling, apoptosis and mitochondrial damage-related DILI mechanism enriched. Advances in emerging technologies
such as next-generation sequencing and bioengineering including iPSC cell culture
have also been integrated into TGx field and provide a promising approach for risk
assessment [20, 21].
Unlike decades ago, several large public available TGx data sets such as open
TG-GATEs [22, 23], DrugMatrix [24] and PredTox [25] have been generated and
currently provide tremendous opportunities for formulating hypotheses to advance
toxicology researches. Considering the different assay types, species, and genomics
technologies in the preclinical setting, a comprehensive assessment among different
TGx assay systems in the preclinical setting is urgently needed for selecting the
fit-for-purpose approach.
To fill the gap, we developed a Pair Ranking (PRank) method to assess the transferability among the different TGx assays and utilized our method to address several
key questions in the TGx field. In this chapter, we will first lay out the key questions
for promoting in vitro TGx systems in risk assessment. Then, we will elaborate on
the PRank method with a few case studies. Final, we will summarize the roadmap for
further positioning the proposed PRank method toward potential regulatory applications.