Severe drug-induced liver injury (DILI) remains a major safety issue due to its frequency of occurrence, idiosyncratic nature, poor prognosis, and diverse underlying mechanisms. Numerous experimental approaches have been published to improve human DILI prediction with modest success. A retrospective analysis of 125 drugs (70 = most-DILI, 55 = no-DILI) from the Food and Drug Administration Liver Toxicity Knowledge Base was used to investigate DILI prediction based on consideration of human exposure alone or in combination with mechanistic assays of hepatotoxic liabilities (cytotoxicity, bile salt export pump inhibition, or mitochondrial inhibition/uncoupling). Using this dataset, human plasma Cmax,total ≥ 1.1 µM alone distinguished most-DILI from no-DILI compounds with high sensitivity/specificity (80/73%). Accounting for human exposure improved the sensitivity/specificity for each assay and helped to derive predictive safety margins. Compounds with plasma Cmax,total ≥ 1.1 µM and triple liabilities had significantly higher odds ratio for DILI than those with single/dual liabilities. Using this approach, a subset of recent pharmaceuticals with evidence of liver injury during clinical development was recognized as potential hepatotoxicants. In summary, plasma Cmax,total ≥ 1.1 µM along with multiple mechanistic liabilities is a major driver for predictions of human DILI potential. In applying this approach during drug development the challenge will be generating accurate estimates of plasma Cmax,total at efficacious doses in advance of generating true exposure data from clinical studies. In the meantime, drug candidates with multiple hepatotoxic liabilities should be deprioritized, since they have the highest likelihood of causing DILI in case their efficacious plasma Cmax,total in humans is higher than anticipated.

The hepatic risk matrix (HRM) was developed and used to differentiate lead clinical and back-up drug candidates against competitor/ marketed drugs within the same pharmaceutical class for their potential to cause human drug-induced liver injury (DILI). The hybrid HRM scoring system blends physicochemical properties (Rule of Two Model: dose and lipophilicity or Partition Model: dose, ionization state, lipophilicity, and fractional carbon bond saturation) with common toxicity mechanisms (cytotoxicity, mitochondrial dysfunction, and bile salt export pump (BSEP) inhibition) that promote DILI. HRM scores are based on bracketed safety margins (<1, 1−10, 10−100, and >100× clinical C max,total ). On the basis of well-established clinical safety experience of marketed/withdrawn drug candidates, the background analysis consists of 200 drugs from the Liver Toxicity Knowledge Base annotated as Most-DILI-(79), Less-DILI-(56), No-DILI-(47), and Ambiguous-DILI-concern (18) drugs. Scores were generated for over 21 internal and 7 external drug candidates discontinued for unacceptable incidence/magnitude of liver transaminase elevations during clinical trials or withdrawn for liver injury severity. Both hybrid scoring systems identified 70−80% Most-DILI-concern drugs, but more importantly, stratified successful/unsuccessful drug candidates for liver safety (incidence/severity of transaminase elevations and approved drug labels). Incorporating other mechanisms (reactive metabolite and cytotoxic metabolite generation and hepatic efflux transport inhibition, other than BSEP) to the HRM had minimal beneficial impact in DILI prediction/stratification. As is, the hybrid scoring system was positioned for portfolio assessments to contrast DILI risk potential of small molecule drug candidates in early clinical development. This stratified approach for DILI prediction aided decisions regarding drug candidate progression, follow-up mechanistic work, back-up selection, clinical dose selection, and due diligence assessments in favor of compounds with less implied clinical hepatotoxicity risk.

Given a particular descriptor/method combination, some quantitative structure–activity relationship (QSAR) datasets are very predictive by random-split cross-validation while others are not. Recent literature in modelability suggests that the limiting issue for predictivity is in the data, not the QSAR methodology, and the limits are due to activity cliffs. Here, we investigate, on in-house data, the relative usefulness of experimental error, distribution of the activities, and activity cliff metrics in determining how predictive a dataset is likely to be. We include unmodified in-house datasets, datasets that should be perfectly predictive based only on the chemical structure, datasets where the distribution of activities is manipulated, and datasets that include a known amount of added noise. We find that activity cliff metrics determine predictivity better than the other metrics we investigated, whatever the type of dataset, consistent with the modelability literature. However, such metrics cannot distinguish real activity cliffs due to large uncertainties in the activities. We also show that a number of modern QSAR methods, and some alternative descriptors, are equally bad at predicting the activities of compounds on activity cliffs, consistent with the assumptions behind “modelability.” Finally, we relate time-split predictivity with random-split predictivity and show that different coverages of chemical space are at least as important as uncertainty in activity and/or activity cliffs in limiting predictivity.

Malaria, in particular that caused by Plasmodium falciparum , is prevalent across the tropics, and its medicinal control is limited by widespread drug resistance. Cysteine proteases of P. falciparum , falcipain-2 (FP-2) and falcipain-3 (FP-3), are major hemoglobinases, validated as potential antimalarial drug targets. Structure-based virtual screening of a focused cysteine protease inhibitor library built with soft rather than hard electrophiles was performed against an X-ray crystal structure of FP-2 using the Glide docking program. An enrichment study was performed to select a suitable scoring function and to retrieve potential candidates against FP-2 from a large chemical database. Biological evaluation of 50 selected compounds identified 21 diverse nonpeptidic inhibitors of FP-2 with a hit rate of 42%. Atomic Fukui indices were used to predict the most electrophilic center and its electrophilicity in the identified hits. Comparison of predicted electrophilicity of electrophiles in identified hits with those in known irreversible inhibitors suggested the soft-nature of electrophiles in the selected target compounds. The present study highlights the importance of focused libraries and enrichment studies in structure-based virtual screening. In addition, few compounds were screened against homologous human cysteine proteases for selectivity analysis. Further evaluation of structure-activity relationships around these nonpeptidic scaffolds could help in the development of selective leads for antimalarial chemotherapy.

The identification of inhibitors effective against mutated PfATP6 suggests ways in which artemisinin resistance may be overcome.