USING QUANTITATIVE SYSTEMS TOXICOLOGY (QST) TO ASSESS DRUG SAFETY: THE EXPERIENCE OF THE DILISIM INITIATIVE

Paul B. Watkins

Eshelman School of Pharmacy, University of North Carolina, Chapel Hill, USA

Abstract

Translating safety data obtained in vitro systems into quantitative risk assessments in specific patient populations remains a challenge. The DILIsim Initiative is a six year old public-private partnership that has been supported by scientists from 16 or the top 20 pharmaceutical companies, the FDA, and academia. The Initiative has taken a “middle out” modeling approach and uses differential equations to recapitulate key processes whereby drugs can cause liver injury. Simulated patient populations have been created by varying parameters in the model to reflect genetic and non-genetic variation. The Initiative has produced software (Dilisym®) that is being increasingly employed in decision making within Pharma. To use the software, in vitro assays are performed on new drug candidates (and sometimes also on major metabolites) and these data are entered, along with estimates of liver exposure anticipated during a specified dosing regimen. The software then predicts the incidence and severity of liver injury anticipated in the target patient population. The model has been built by successive modifications to recapitulate the known liver safety outcomes of “exemplar compounds”. The model is now being tested in a validation cohort of drugs and model predictions have so far been shown to be accurate in over 90% of the validation drugs tested to date. Importantly, the results of Dilisym modeling have increasingly been included in regulatory submissions, including a recent NDA submission and presentation at a recent FDA Advisory Committee Meeting. The experience with the DILIsim Initiative has strongly supported an important future role for QST modeling in assessing and improving liver safety.