Prediction of human drug-induced liver injury (DILI) in relation to oral doses and blood concentrations

Simple item page
dc.contributor.advisorHengstler, Jan G.
dc.contributor.authorAlbrecht, Wiebke
dc.contributor.refereeWatzl, Carsten
dc.date.accepted2021-06-07
dc.date.accessioned2021-07-05T09:47:31Z
dc.date.available2021-07-05T09:47:31Z
dc.date.issued2021
dc.description.abstractDrug-induced liver injury (DILI) is the leading cause for acute liver failure in the USA and in Germany and one of the most common reasons for withdrawal of drugs from the market or failure of a drug candidate during development. Since DILI cannot be accurately predicted by animal models, a reliable in vitro test system for the prediction of human hepatotoxicity would be a valuable asset for drug development as well as for regulatory purposes. In this thesis an in vitro/in silico approach for the prediction of human hepatotoxicity in relation to blood concentrations and oral doses was established. This approach combines in vitro effective concentrations derived from a cytotoxicity assay, in vivo concentrations obtained by physiologically based pharmacokinetic (PBPK) modelling and a support vector machine (SVM) classifier based on these concentration pairs to predict the risk for hepatotoxicity for specific exposure scenarios. For systematic test system evaluation and optimization two novel performance metrics, the Toxicity Separation Index (TSI) and Toxicity Estimation Index (TEI), were utilized. These indices eliminate the need for a priori defined in vitro and in vivo concentrations and foster the systematic evaluation of the benefit of additional readouts. As a first step the feasibility of the in vitro/in silico approach was tested for primary human hepatocytes (PHH) and a training set of 28 compounds with in total 30 different in vitro/in vivo concentration vectors, yielding a sensitivity of 100%, a specificity of 88% and an accuracy of 93% in the leave-one-out classification with the SVM based classifier. A SVM based classifier utilizing all vectors was then applied to derive in combination with reverse PBPK modelling an acceptable daily intake (ADI) for the hepatotoxicant pulegone. The derived ADI was comparable to published ADIs based on two rodent studies. Next, the compound set was extended to a total of 80 compounds with 82 distinct in vitro/in vivo concentration pairs. The SVM leave-one-out classification resulted in a sensitivity of 77.8%, a specificity of 59.4% and an accuracy of 70.1%. Furthermore, the feasibility of the approach substituting HepG2 cells for the PHH and a combination of both cell culture systems for the extended compound set was evaluated. The obtained sensitivity was 88.9% and 86.7% and the specificity 62.5% and 65.6%, respectively. The accuracy was in both cases 77.9%.en
dc.identifier.urihttp://hdl.handle.net/2003/40290
dc.identifier.urihttp://dx.doi.org/10.17877/DE290R-22163
dc.language.isoenen
dc.subject.ddc610
dc.subject.rswkLeberde
dc.subject.rswkLebertoxizitätde
dc.titlePrediction of human drug-induced liver injury (DILI) in relation to oral doses and blood concentrationsde
dc.typeTextde
dc.type.publicationtypedoctoralThesisde
dcterms.accessRightsopen access
eldorado.secondarypublicationfalsede

Files

Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
Prediction of human drug-induced liver injury (DILI) in relation to oral doses and blood concentrations.pdf
Size:
6.7 MB
Format:
Adobe Portable Document Format
Description:
DNB
Download
License bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
license.txt
Size:
4.85 KB
Format:
Item-specific license agreed upon to submission
Description:
Download

Collections

Lehrstuhl für Umweltchemie und Analytische Chemie