Drug Approvals Rorschach: What Do You See?

Derek Lowe over at In The Pipeline posted some observations today about a recent Nature Reviews Drug Discovery article, authored by Bear Sterns analysts Elizabeth Czerepak and Stefan Ryser, regarding the origin of drugs that are approved by FDA and drugs that fall at the last clinical hurdle.

In this analysis of a total of 103 FDA approvals and 91 Phase III failures between January 2006 and December 2007 drug candidates straight outta pharma seemingly performed best: 36 approvals vs only five failures--adding an additional four approvals for pharma based on acquired programs takes the figure to 40 (whereas biotech-pharma alliances yielded 16 approvals vs 18 failures, and biotech-only figures were 47 and 68). But there is a caveat: these are not all novel chemicals and biologics--line extensions and me-too drugs are included in the tally and fairly evenly spread between pharma and biotech.

Still, when looking purely at the approvals (63 generated by biotech, compared to 40 by pharma), the upshot of the original analysis is no suprise: that biotechs ought to continue to enjoy good leverage over their would-be pharmaceutical partners when it comes to filling pharma pipelines via acquisitions and licensing deals.

Though as the analysts themselves point out (and Derek & co concur), if biotech-pharma collaborations boast solid results it is potentially because pharma has already cherry-picked the most promising biotech drug candidates. You'll find no argument from us there--that's one reason Phase II deal values are on such a dramatic up-swing. It also begs the question: for pharmas, mightn't it be a prudent strategy to do more alliances?

Maybe, maybe not: if biotechs sport the most success, they also sport the most failure, and in the process drag down the stats of their pharmaceutical partners. Czerepak and Ryser hypothesize that the abundance of clinical failure by biotechs is a result of under-funding, leading back to their "More Alliances, please" conclusion.

Could it be, as others have suggested, that the failure rate among biotechs can best be explained by their lack of regulatory and late-stage clinical development experience? That probably plays a role. We also considered the possibility that biotechs' high failure rate during the '06-'07 time frame could be an artifact of the relative abundance of financing available, even for mediocre projects, back during the biotech bubble, when many of these molecules would have been entering the clinic.

And on top of that, even though there are many more biotech projects in this sample than pharma--149 for biotech and biotech/pharma collaborations vs. 45 pharma-only ventures--many smaller firms are likely to only have one or maybe two clinical projects. Killing off a lead compound can mark the death of a small firm: that means even iffy Phase II compounds can find their way to pivotal trials. Despite their recent pipeline woes, Big Pharma have the luxury of attrition.

Finally, as Derek and his commenters point out, it should be noted there is some semantic confusion around what is and what is not a biotech product. For the purposes of the Bear Sterns analysis a 'biotech' drug is one developed by a company listed on the AMEX or Nasdaq biotech indices--not the more traditional (and accurate) definition related to biologics vs small molecules. But that argument is for another post; besides, isn't everyone just specialty pharma these days anyway?

What do the numbers tell you? Write back and tell us.