At AstraZeneca, Occasion for Introspection

May 19, 2014

By Aaron Krol 

May 19, 2014 | On Friday, a frank assessment of AstraZeneca's R&D practices appeared in Nature Reviews Drug Discovery. The review, written by members of the pharmaceutical company's Innovative Medicines and Early Development division, examined projects in preclinical, Phase I and Phase II trials from 2005-2010, and reached some stark conclusions about the corporate culture's influence on which compounds advanced through successive stages of testing. At the same time, the review offers new reasons for optimism, and may be seen to bolster AstraZeneca in its ongoing battle to resist acquisition by Pfizer.

While the top-line figures are glaring – in particular, the observation that AstraZeneca was letting an exceptional number of compounds through Phase I trials, compared to the industry as a whole, only to see disproportionate failures in Phase II – the authors' substantive look at the causes of failure is enlightening. A lack of confidence in the underlying biology of therapies seems to have been a frequent reason for failure, even in later stages of a project. In 40% of cases where lack of efficacy led to a project's shutdown, teams reported that they "lacked data demonstrating a clear linkage of the target to the disease or access to a well-validated animal model of the disease," and as late as Phase IIb trials, over 40% of teams reported "low confidence" in their molecular targets. In these cases, weak evidence of efficacy in preclinical models may have been enough to move projects forward, despite no strong hypothesis linking a lead drug to the disease pathology.

The authors also noted that a more well-rounded view of the biology was a powerful predictor of success. At the time of the review, only 27% of Phase II trials had failed where teams reported that their drug target had a known genetic link to the disease in humans, and only 18% of Phase IIb trials had failed if teams reported that they began testing with a biomarker for efficacy in mind. "Overall," the authors write, "these data highlight that, throughout every phase of early R&D, it is crucial for scientists and clinicians to gain an understanding of, and confidence in, the disease biology, the relationship of the target to the disease indication, and the proposed mechanism of action of a potential drug."

While the problems identified in this review will look familiar to many working in the pharmaceutical industry, they seem to have grown particularly entrenched at AstraZeneca. The authors note that projects were frequently advanced because teams were evaluated by volume of compounds in active testing, creating a perverse incentive to promote any candidate drug that could meet minimal standards of evidence. One indicator of this culture is the lack of diversity in back-up molecules, meant as potential replacements for a lead candidate in certain key projects. The authors found that back-up molecules were often only minor tweaks of the lead candidate, advanced with no reason to believe they would differ in activity or outperform the lead drug on any metric. "In one extreme case," they write, "we identified a project with seven back-up molecules in the family... [which] all failed owing to the same preclinical toxicology finding."

Although the review was completed in 2011, it has been published at a fortuitous time for AstraZeneca, which has fought off repeated takeover attempts by Pfizer, including an offer of $117 billion over the weekend. There has been frequent speculation that AstraZeneca's shareholders may rebel against the wishes of the board and accept a high bid by Pfizer, which is offering a high premium over the market value of AstraZeneca shares. While AstraZeneca's board members insist that Pfizer is still undervaluing their drug portfolio, skeptics note that AstraZeneca has struggled to bring new drugs to market in recent years.

This report offers reasons to believe that the company's R&D practices may have changed since 2011, with a new emphasis on biological understanding over volume. The authors note that the company's early development portfolio is significantly smaller than it was three years ago, and that teams are now expected to articulate their hypotheses relating drugs to disease targets, and more encouraged to seek out relevant biomarkers, companion diagnostics, and patient subpopulations. While acknowledging that "it is too early to see tangible benefits to the AstraZeneca pipeline," they also observe that both preclinical and Phase II success rates have risen since the new R&D philosophy was implemented, while inflated Phase I success rates have fallen.

While the timing may be coincidence, for AstraZeneca boosters looking for evidence to point to when arguing that Pfizer's takeover bids undervalue the flagging company, this review will be a welcome addition to the public record. For everyone else, it is an instructive glimpse into misguided incentives at the root level of the drug industry.