Application of Bayesian approaches in drug development: starting a virtuous cycle.

The pharmaceutical industry and its global regulators have routinely used frequentist statistical methods, such as null hypothesis significance testing and p values, for evaluation and approval of new treatments. The clinical drug development process, however, with its accumulation of data over time, can be well suited for the use of Bayesian statistical approaches that explicitly incorporate existing data into clinical trial design, analysis and decision-making. Such approaches, if used appropriately, have the potential to substantially reduce the time and cost of bringing innovative medicines to patients, as well as to reduce the exposure of patients in clinical trials to ineffective or unsafe treatment regimens. Nevertheless, despite advances in Bayesian methodology, the availability of the necessary computational power and growing amounts of relevant existing data that could be used, Bayesian methods remain underused in the clinical development and regulatory review of new therapies. Here, we highlight the value of Bayesian methods in drug development, discuss barriers to their application and recommend approaches to address them. Our aim is to engage stakeholders in the process of considering when the use of existing data is appropriate and how Bayesian methods can be implemented more routinely as an effective tool for doing so.

Voluntary exploratory data submissions to the US FDA and the EMA: experience and impact.

Heterogeneity in the underlying mechanisms of disease processes and inter-patient variability in drug responses are major challenges in drug development. To address these challenges, biomarker strategies based on a range of platforms, such as microarray gene-expression technologies, are increasingly being applied to elucidate these sources of variability and thereby potentially increase drug development success rates. With the aim of enhancing understanding of the regulatory significance of such biomarker data by regulators and sponsors, the US Food and Drug Administration initiated a programme in 2004 to allow sponsors to submit exploratory genomic data voluntarily, without immediate regulatory impact. In this article, a selection of case studies from the first 5 years of this programme - which is now known as the voluntary exploratory data submission programme, and also involves collaboration with the European Medicines Agency - are discussed, and general lessons are highlighted.

Informed toxicity assessment in drug discovery: systems-based toxicology.

Technological advances in the biological, chemical and in silico sciences have transformed many scientific disciplines, including toxicology. A vast new palate of toxicity testing tools is now available to investigators, enabling the generation of enormous amounts of data using only small amounts of test sample and at relatively low cost. In addition to these tools, the pharmaceutical industry has an urgent need for toxicity testing earlier in the process, based on the recognition that safety issues are the single largest cause of drug candidate attrition from development portfolios and the marketplace. However, along with the opportunity provided by new testing tools comes the dilemma of deciding which tools to use and, equally as important, when and why to use them. It may well be unwise to apply a new toxicity test or screening system simply because one can, as both false positive and false negative outcomes can quickly negate the value of a toxicity test system and may even have a net negative impact on drug discovery productivity. This can be true even of test systems that are considered to be 'validated' in the traditional sense. How then is an investigator or drug discovery organization to decide which of the new tools to use, and when to use them? Proposed herein is a strategy for identifying high-value toxicity testing systems and strategies based on program knowledge and informed decision-making. The decision to apply a certain toxicity testing system in this strategy is informed by knowledge of the pharmacological target, the chemical features of molecules active at the pharmacological target, and existing public domain or institutional learning. This 'fit-for-purpose' approach limits non-targeted or 'uninformed' toxicity screening to only those few test systems with high specificity, strong outcome concordance and molecular relevance to frequently encountered toxicity risks (eg, genotoxicity). Additional toxicity testing and screening is then conducted to address specific known or potential toxicity risks, based on existing knowledge of the target pharmacology and secondary pharmacology or chemical attributes with known or suspect risk, and by active 'interrogation' of both the target and active chemical moieties during the drug discovery process. This model for toxicity testing decision-making is illustrated by two case studies from recent experience.

Dual-action hypoglycemic and hypocholesterolemic agents that inhibit glycogen phosphorylase and lanosterol demethylase.

Diabetic dyslipidemia requires simultaneous treatment with hypoglycemic agents and lipid-modulating drugs. We recently described glycogen phosphorylase inhibitors that reduce glycogenolysis in cells and lower plasma glucose in ob/ob mice (J. Med. Chem., 41: 2934, 1998). In evaluating the series prototype, CP-320626, in dogs, up to 90% reduction in plasma cholesterol was noted after 2 week treatment. Cholesterol reductions were also noted in ob/ob mice and in rats. In HepG2 cells, CP-320626 acutely and dose-dependently inhibited cholesterolgenesis without affecting fatty acid synthesis. Inhibition occurred together with a dose-dependent increase in the cholesterol precursor, lanosterol, suggesting that cholesterolgenesis inhibition was due to lanosterol 14alpha-demethylase (CYP51) inhibition. In ob/ob mice, acute treatment with CP-320626 resulted in a decrease in hepatic cholesterolgenesis with concomitant lanosterol accumulation, further implicating CYP51 inhibition as the mechanism of cholesterol lowering in these animals. CP-320626 and analogs directly inhibited rhCYP51, and this inhibition was highly correlated with HepG2 cell cholesterolgenesis inhibition (R2 = 0.77). These observations indicate that CP-320626 inhibits cholesterolgenesis via direct inhibition of CYP51, and that this is the mechanism whereby CP-320626 lowers plasma cholesterol in experimental animals. Dual-action glycogenolysis and cholesterolgenesis inhibitors therefore have the potential to favorably affect both the hyperglycemia and the dyslipidemia of type 2 diabetes.