168
Z. Liu et al.
9.4.2 Drug-Induced Liver Injury (DILI) Annotations
To further investigate whether the extrapolation between assay systems is endpoint
dependent, we used well-established drug-induced liver injury (DILI) annotation
data sets. DILI is a major concern for drug developers and contributes to many drug
attritions [43]. In this chapter, three different ways to classify DILI were utilized:
NCTR DILI classification scheme [34], Xu’s label [44], and Sakatis’ label [45].
The NCTR DILI classification scheme was based on FDA approved drug labeling
information, which classified the drugs into three categories including Most-DILI
concern, Less-DILI concern and No-DILI concern. In this chapter, we used drugs
belonging to Most-DILI concern. Xu’s data set was refined by using images of the
cells to assess toxicity endpoints in human liver cultures. Sakatis’ data adopted in vitro
biological activation to detect DILI potential in over two hundred compounds. Only
the drugs with positive DILI from Xu and Sakatis’ data were mapped to the Open
TG-GATEs list.
Aside from different DILI annotation data sets, we also employed different hepatotoxic manifestation data for further investigation of our PRank method. Specifically, the SIDER database (http://sideeffects.embl.de/), which stands for Side Effect
Resource, is an online database listing reported side effects for both over-the-counter
and prescription drugs. The side effect terms in the SIDER database were standardized with preferred terms (PTs) by using the Medical Dictionary for Regulatory
Activities (MedDRA) ontology. The hepatotoxic-related side effects were extracted
by mapping the PT terms onto the Society of organ (SOC) level. The PTs with primary
SOC hepatobiliary were considered as hepatotoxic-related side effects. The extracted
hepatotoxic-related side effects were further classified by the domain experts and
generated the following five categories: liver hepatobiliary abnormality, transaminase elevations, histologic findings, liver injury patterns, and the severity of liver
injury [37].
9.4.3 Therapeutic Categories
We also examined concordance among TGx assay systems for different therapeutic categories. The WHO Anatomical Therapeutic Chemical (ATC) Classification
System was used to categorize the compounds in different therapeutic classes. ATC
consists of a five-level ontological structure. The second level representing the therapeutic subgroup was used in this study.
Z. Liu et al.
9.4.2 Drug-Induced Liver Injury (DILI) Annotations
To further investigate whether the extrapolation between assay systems is endpoint
dependent, we used well-established drug-induced liver injury (DILI) annotation
data sets. DILI is a major concern for drug developers and contributes to many drug
attritions [43]. In this chapter, three different ways to classify DILI were utilized:
NCTR DILI classification scheme [34], Xu’s label [44], and Sakatis’ label [45].
The NCTR DILI classification scheme was based on FDA approved drug labeling
information, which classified the drugs into three categories including Most-DILI
concern, Less-DILI concern and No-DILI concern. In this chapter, we used drugs
belonging to Most-DILI concern. Xu’s data set was refined by using images of the
cells to assess toxicity endpoints in human liver cultures. Sakatis’ data adopted in vitro
biological activation to detect DILI potential in over two hundred compounds. Only
the drugs with positive DILI from Xu and Sakatis’ data were mapped to the Open
TG-GATEs list.
Aside from different DILI annotation data sets, we also employed different hepatotoxic manifestation data for further investigation of our PRank method. Specifically, the SIDER database (http://sideeffects.embl.de/), which stands for Side Effect
Resource, is an online database listing reported side effects for both over-the-counter
and prescription drugs. The side effect terms in the SIDER database were standardized with preferred terms (PTs) by using the Medical Dictionary for Regulatory
Activities (MedDRA) ontology. The hepatotoxic-related side effects were extracted
by mapping the PT terms onto the Society of organ (SOC) level. The PTs with primary
SOC hepatobiliary were considered as hepatotoxic-related side effects. The extracted
hepatotoxic-related side effects were further classified by the domain experts and
generated the following five categories: liver hepatobiliary abnormality, transaminase elevations, histologic findings, liver injury patterns, and the severity of liver
injury [37].
9.4.3 Therapeutic Categories
We also examined concordance among TGx assay systems for different therapeutic categories. The WHO Anatomical Therapeutic Chemical (ATC) Classification
System was used to categorize the compounds in different therapeutic classes. ATC
consists of a five-level ontological structure. The second level representing the therapeutic subgroup was used in this study.