Prediction of clinical risks by analysis of preclinical and clinical adverse events

Abstract

This study examines the ability of nonclinical adverse event observations to predict human clinical adverse events observed in drug development programs. In addition it examines the relationship between nonclinical and clinical adverse event observations to drug withdrawal and proposes a model to predict drug withdrawal based on these observations. These analyses provide risk assessments useful for both planning patient safety programs, as well as a statistical framework for assessing the future success of drug programs based on nonclinical and clinical observations. Bayesian analyses were undertaken to investigate the connection between nonclinical adverse event observations and observations of that same event in clinical trial for a large set of approved drugs. We employed the same statistical methods used to evaluate the efficacy of diagnostic tests to evaluate the ability of nonclinical studies to predict adverse events in clinical studies, and adverse events in both to predict drug withdrawal. We find that some nonclinical observations suggest higher risk for observing the same adverse event in clinical studies, particularly arrhythmias, QT prolongation, and abnormal hepatic function. However the lack of these events in nonclinical studies is found to not be a good predictor of safety in humans. Some nonclinical and clinical observations appear to be associated with high risk of drug withdrawal from market, especially arrhythmia and hepatic necrosis. We use the method to estimate the overall risk of drug withdrawal from market using the product of the risks from each nonclinical and clinical observation to create a risk profile.

Considerations in the extrapolation of drug toxicity between humans and dogs

“…Clark [5] also comments on the importance of considering false negatives. Using a likelihood ratio, he observed that while the precipitation of cardiac arrhythmias, liver damage and renal failure in animals were highly indicative of the likelihood of a similar event in humans, the lack of canine toxicity did not share the same prognostic value.…”

Recent efforts to elucidate the scientific validity of animal-based drug tests by the pharmaceutical industry, pro-testing lobby groups, and animal welfare organisations

“…One of these reports [2], as we have already discussed in our work, did not estimate specificity, without which the evidential weight toward likelihood of human toxicity/non-toxicity provided by the animal models—which is precisely what we need to know—cannot be calculated. As the authors of the cited study themselves acknowledged, “A more complete evaluation of this predictivity aspect will be an important part of a future prospective survey.” Another such cited report [15] showed human predictability for some therapeutic areas to be over 90%—yet it also showed many other areas where results from animal studies failed to significantly correlate with human observations, which were overlooked. Importantly, this analysis also utilised Likelihood Ratios (LRs), and the author argued why this is superior and necessary— much as we did in our own papers.…”

Improving the Safety Assessment of Chemicals and Drug Candidates by the Integration of Bioinformatics and Chemoinformatics Data

“…As soon as proprietary and public clinical data can be leveraged on a broader scale and subjected to automated translational analyses (‘which finding in a specific preclinical species is translated to what extent into human beings?’), the prediction from animal to humans will reach another step. First approaches of such big‐data analyses for 3815 drugs have recently been published based on PharmaPendium's data‐mined dossier documents from FDA and EMA containing both preclinical and clinical data . The results of such analyses will eventually cross‐fertilize hazard and risk assessment also for industrial or consumer chemicals, where the human data are only available for a few chemicals of specific concern from biomonitoring studies.…”