Which Pharma Spent The Most R&D On Each Drug?
8 CommentsBy Ed Silverman // February 10th, 2012 // 9:11 am
As the feared patent cliff loomed, drugmakers began talking up their R&D priorities and efficiencies with an eye toward easing investor anxiety and positioning themselves for presumably better times. At the same time, pharma continually trumpted the high cost – still higher failure rates – involved. But which drugmaker made the best use of their R&D dollars? And which did the worst job?
Well, the winner was Amgen. Between 1997 and 2011, the biotech spent $33.2 billion on R&D while nine drugs were approved, which worked out to $3.7 billion per drug, adjusted for inflation, according to Forbes, which crunched data provided by InnoThink Center For Research In Biomedical Innovation and Thomson Reuters Fundamentals via FactSet Research Systems.
At the other end of the spectrum was AstraZeneca. During the same period, the drugmaker spent nearly $59 billion on R&D, while five drugs were approved, which worked out to almost $11.8 billion for each approval. Combine this result with patent expirations for the Crestor cholesterol pill, the Nexium acid reflux med and the Seroquel antipsychotic, and some 30,000 jobs get cut.
However, AstraZeneca is now spending $4.5 billion on stock buybacks, which is about what Eli Lilly and Abbott Laboratories spent on R&D for each drug approved, according to Forbes. Theoretically, if AstraZeneca had greater success – which is, of course, unpredictable and difficult to achieve – perhaps there would have been enough money for both another new drug and appeasing investors.
The point Forbes tries to raise, though, is that failures should be included in the accounting but, at the same time, this high cost is nothing to crow about when pharma tries to explain the big price tags slapped on some medicines. Of course, drugmakers may point to this data to justify recent decisions to eliminate or outsource large swaths of R&D in the name of efficiency.
To what extent the ranking will change appreciably in, say, five years will be interesting to see, given the vagaries of R&D. But the ability to compare over time can also allow, perhaps, for some additional analysis into managerial decisions that shape all those priorities and bets.
benjamins pic thx to amagill on flickr
Gil Roth
Forbes won’t let me register to leave a comment, so I’ll post this here instead. The “R&D spend / new drugs” formula ignores the amount of R&D expenditure that goes toward conducting new trials for existing drugs, for expanding labels to new indications or conducting massive post-marketing safety trials.
And then there’s the fact that R&D spend now is for drugs that may be approved 5 years from now, which is somewhat mitigated by choosing the long time scale, but still, it’s a little like choosing a cheap arbitrary starting point because there happened to be a bunch of approvals in the early going.
Sure wish Forbes’ registration page worked.
Observer
@Gill – sound familiar … or at least close to accurate?
“WE are each entitled to our own opinion,
but no one is entitled to his own facts.”
– Danile Patrick Moynihan
Matthew Herper
Gil — could you send me an email so that I can have our tech people check out what went wrong for you? It’s mherper@forbes.com.
The original Munos analysis I worked off of simply used one year’s R&D spend and the annualized rate of drug approvals. So what I did was to use annualized, inflation-adjusted R&D over the same time period as the NMEs. Yes, NMEs today result from yesterday’s research dollars. But the research budgets, adjusted for inflation, didn’t change that much, unless there was a big merger. By taking the total research money spent over a long period and the total number of drugs, I think you do get a pretty good idea of what is being spent per drug. Keep in mind that a lot of the D costs will be back-loaded.
Sure, this doesn’t include line extensions, but most line extensions are spent on recent NMEs. I actually think it’s pretty fair to include the entire Lipitor program in the cost for Pfizer’s Lipitor NME. Again, taking a long time frame helps here.
Thanks for the comment, and sorry you had trouble on our site!
original industry insider
There’s one metric that is never published but is actually an impotant metric for companies to adjust headcount. That is sales/employee. We used to use the metric $200,000/FTE, meaning that a company of 100,000 should bring in roughly $20 billion in top line sales to have a decent ROI. That metric can also be used for downsizing, thus as sales go down you downwardly adjust the headcount to maintain that ratio of $200K/FTE.
Outside the Box
Frankly, a more valuable analysis would be spend per approved indication. Drugs with only one indication but lots of failed post-marketing trials would push the number up; drugs with lots of successful line extensions would push the number down – but the net effect would be a metric that would actually have some meaning in terms of identifying the success ratios of these companies. I would also include an element for international approval and remaining patent life at launch – then we would be looking at some truly useful analysis. Nice try, but in this form the analysis is pretty meaningless.
Stewart Lyman
I also do not find much value in these published numbers as I am not sure how to interpret them. Let’s focus on Amgen as an example. Amgen at present sells a total of 10 drugs, according to their website. Let’s examine these drugs in detail. Two drugs were sold before the 15 year time frame mentioned, Neupogen and Epogen, and both were blockbusters. Neulasta is simply a longer lasting version of Neupogen, just as Aranesp is a longer acting version of Epogen. Aranesp was specifically developed by Amgen to get around a contractual agreement that they had with J&J, and it enabled them to move into a market from which they had previously been precluded from entering. Two of the drugs, Prolia and Xgeva, are actually the same molecule, and should really be counted as one drug. Vectibix and Enbrel (and a significant amount of the Prolia and Xgeva intellectual property) were all acquired by Amgen when they purchased Immunex in 2002. Nplate, a true NME, hit the market in 2008, and Sensipar was acquired from NPS Pharmaceuticals. Not present in their current line up are three drugs that they sold/outlicensed in 2008, which were Kepivance, Stemgen, and Kineret. Combined sales of these in 2007 were a mere $70 million, which is why they were dumped. So what we have here is an amalgam of line extensions, acquisitions, and divestitures, mixed in with some in-house research produced drugs. How were these counted in the analysis that was done? The number of employees over the past 15 years probably doubled or tripled, especially since Amgen has acquired a number of companies in that time frame. It is not clear to me exactly how one can readily compare these numbers to those of other companies, or even to calculate values on a per employee basis. And Matthew Herper: I, too, was unable to leave a comment on the Forbes website. You have some IT issues to sort out.
Matthew Herper
Thanks to both of you for saying so. I’ve forwarded this thread to our IT people. There is supposed to be a new reg system on its way.
original industry insider
To extend Stewart’s point, I can tell you in fact that Abbott didn’t spend a plug nickel on the original R&D for it’s biggest seller, Humira. It was develped entirely by my former company, Knoll Pharmaceutical Co., a subsidiary of BASF Pharma. Not realizing the potential of Humira, BASF practically gave the drug away when Abbott acquired the rights to the product back in 2000. Since that time Humira has been the ultimate cash cow for Abbott, and their only investment has been to do the sNDA’s for the subsequent indications.