There’s an article in the latest Nature Reviews Drug Discovery on recent drug attrition rates that caught my eye. The authors are looking over 2006-2007 trials and approvals, comparing the biotech industry with traditional pharma. ("Biotech" is defined as a company that's included in either the American Stock Exchange's biotech index and/or the NASDAQ's). In that period, the biotechs scored 47 FDA approvals (45% of the total approvals), but had 68 Phase III failures, which is 74% of that total. Pharma companies had only 5 Phase III failures during that stretch – the other 18 were biotech/pharma joint ventures, and those had a corresponding 16 approvals.
That’s food for thought, all right. The authors make much of the comparatively higher success rate for the biotech/pharma alliance compounds versus the biotechs that went it alone. I have to say, though, that the first explanation that came to my mind was one that they mention, but refer to as “cynical”: that the products which got partnered were disproportionately drawn from the list of those more likely to succeed in the first place.
But is “higher success rate for alliances” really the way to look at the data? Coming at the figures from another direction, I’d argue that “lower success rate for anything labeled biotech” would be a better fit. After all, the FDA approval/Phase III failure numbers are 47/68 for biotech, and 16/18 for biotech/pharma codevelopment, and I’d argue that those ratios are a lot closer to each other than either one is to the ratio for pure pharmaceutical companies, which was 36/5. Look at it this way: if the biotech-alone success rate was as good as the alliance one, you’d expect maybe 53 failures for those 47 successes instead of the 68 that really took place. But if biotech had the same success rate as pharma alone, those 47 winners would have been accompanied by only about 7 failures.
Cynics with a different orientation might wonder if the higher failure rate comes from a higher number of attempts on innovative drugs in biotech, as opposed to follow-ups and me-toos. But looking at another table in the same paper, where the authors split such compounds out, the me-too data in the pharma industry shows 15 FDA approvals versus 1 Phase III failure. The corresponding biotech figures show 20 approvals and 17 failures, so even the follow-on drugs have a harder time of it. (In case you're wondering, the figures from the opposite end of the spectrum, the new compound/new indication class, are 17 approvals versus 4 failures for pharma, as opposed to a toe-curling 9 approvals and 42 failures for biotech). Breaking down the numbers in another way, biotech companies had 37 out of 115 compounds in the me-too class (32%), while pharma had 16 out of 41 (39%), which isn't that big a difference.
This sort of thing is particularly interesting for someone of my age or older, because it brings back memories of the 1980s and the first big biotech boom, back when Genentech and Biogen went public and Cetus was still a going concern. The pitch back then was that biotech products were actually going to have a higher success rate, because they were, after all, mostly proteins that were already in use by the body, right? The definition of "biotech" has changed a lot since then, though - if you look at those companies in the two indices linked above, you'll notice that many of them don't work on biological products at all, but would be better classified as "small pharma". But I'm not sure if the general public appreciates that distinction. . .