Eleni Kotsampasakou

Title of the Doctoral Thesis: Predicting liver toxicity on basis of transporter interaction profiles

Publishing year: 2016

Tags: drug-induced liver injury (DILI) / cholestasis / hyperbilirubinemia / liver transporters / organic anion transporting polypeptides (OATPs) / OATP1B1 / OATP1B3 / bile salt export pump (BSEP) / breast cancer resistance protein (BCRP) / P-glycoprotein (P-gp) / classification models / machine learning / in silico modeling / imbalanced datasets / toxicity predictions


Drug-induced liver injury (DILI) is currently a major challenge for drug development in pharmaceutical industry: it is one of the main causes for attrition during clinical and pre-clinical studies and the primary reason for drug withdrawal from the market. Subsequently, there is great need for recognizing or foreseeing potential hepatotoxicity issues as early as possible. Unfortunately, predicting hepatotoxicity is not an easy task, due to the complexity of the endpoint and potential idiosyncratic phenomena. In recent years, liver transporters attracted lots of attention regarding their role in development of drug induced hepatotoxicity. There are many reports in literature for several transporters, including among others bile salt export pump (BSEP), breast cancer resistance protein (BCRP), P-glycoprotein (P-gp) and organic anion transporting polypeptide 1B1 and 1B3 (OATP1B1 and OATP1B3). Main topic of the current thesis is modeling liver toxicity endpoints, as well as general drug-induced hepatotoxicity, by combining information of the liver transporters’ inhibition aforementioned and molecular descriptors. Due to lack of in vitro data, predictions of transporters’ inhibition were used instead. For this cause, classification models for OATP1B1 and OATP1B3 inhibition were initially developed, while for the rest of BSEP, BCRP and P-gp in silico models already available in-house were used. The studied endpoints were drug-induced liver injury, hyperbilirubinemia and cholestasis. Apart from modeling, also the role of hepatic transporters’ inhibition was investigated for the cases of the toxicity endpoint. Mainly human, - and in some cases also animal - data were used. They come primarily from public sources – thus, extended careful curation was provided - while some of the animal in vivo data were provided from the eTOX consortium. Several models were developed, both for transporters and toxicity endpoints, with some of them yielding very satisfactory performance. In general, the modeling of the transporters was a comparably easier task and gave better results with simpler classification schema. For toxicity endpoints with a more straightforward mechanistic basis, like cholestasis, association between transporter inhibition and toxicity was also shown. For more general forms of toxicity, like DILI, there was no clear trend. Of course, there are more hepatic transporters, as well as enzymes, playing an important role and their inclusion in a further study would be interesting.