Last week, FDA Director of CDER, Janet Woodcock, lectured to a full house at the Mission Bay campus of the University of California – San Francisco. She spoke on ways academic research centers could play a larger role in the development of new medical technologies. I very much enjoyed the talk, the slides for which are available here.
One aspect of her talk struck me as surprising. At the start of her presentation, as she set the stage with a description of rising healthcare costs and the significant challenges to developing new medical technologies, she cited the oft-repeated statistic that new drug development requires 10-12 years and costs over $1 billion dollars. It was a casual remark, not controversial in the least, and since she was stating the obvious, the audience did not so much as blink as she proceeded to the main points of her talk.
I found this a little strange because Dr. Woodcock’s use of the statistic, as I interpreted it, was a tacit acceptance that the numbers are true. Keep in mind that the FDA is often criticized as being largely responsible for the significant time and expense of new drug development. Many industry analysts argue that it is due to FDA’s stringent regulatory hurdles that new drug development requires so much time and money. To what extent this is true is subject to much debate, but I think it’s safe to say that, if I were at the FDA, I might not accept so willingly the magnitude estimates of a problem that I supposedly helped to create.
The numbers are indeed staggering: Ten to twelve years and over $1 billion dollars to develop a new little pill. Comparisons to “Hollywood economics” and the film industry are often made, but I think oil field exploration is probably more accurate.
These figures were calculated by researchers at the Tufts Center for the Study of Drug Development (TCSDD), an academic think tank focused on the pharmaceutical industry. The numbers are endorsed by PhRMA, the pharmaceutical industry organization, and they are often used as a justification for the high price tags on new therapies. How else can a company recoup the significant time and investment in a new product but to charge upwards of $100,000 per year for its use?
These estimates have never sat well with me. In my opinion, the 10-12 year, $1 billion dollar notion is misleading at best, and flat wrong at worst. Here are two reasons why.
First, these figures are based on historical data. Specifically, they are based on the R&D expenditures on select new drugs originating in the major pharmaceutical companies, which most recent estimates use data to around 20051. This is misleading because it measures how much it did cost, not how much it should or could cost. So, while the bloated infrastructure and wasteful spending of Pharma companies during the boom years of the 1990s and 2000s is widely criticized, those expense measures are accepted as an accurate metric of “true” drug development costs? Seems strange to me. If a small town of 50,000 residents has a budget of $10 million, does that mean that it costs $10M to run a small town of 50,000 residents? Not necessarily. Does it really cost $250 million to execute an inmate on death row in California? Yes, it can, but it doesn’t necessarily cost that much2.
Second, the estimates will go down in the future. One obvious reason is that Pharma is slashing its R&D budget, so the denominator in the equation is shrinking. Acquisitions and in-licensing have costs too, but not as high as fully-integrated internal R&D and strategic transactions can be structured to mitigate risk (stage of development, option to license, CVRs, ability to terminate, etc.). Furthermore, drug development strategies, especially clinical trial practices, are trending toward smaller trials in select patient populations with greater use of biomarkers. These changes appear to reduce both sunk costs (money actually spent on discovery and development) and the cost of failure (which constitutes a substantial of the time and expense estimates), not to mention a positive impact on regulatory and payer reimbursement strategies.
Frankly, NO entrepreneurs or biotech executives today pitch to investors a 10-12 years, $1 billion drug development proposition. They pitch focused development programs with expedited development timelines and reduced clinical costs. Moreover, business development teams at the major pharmas similarly balk at licensing opportunities that require extra large and long Phase 3 development programs (think drugs for metabolic disease, with Phase 3 trials enrolling, say, 15,000 patients over 4-5 years and costing $300 million dollars). I find it hard to see how the 10-12 years, $1 billion metric applies in drug development today.
Drug development is very expensive and lengthy, no doubt, and the TCSDD authors acknowledge the variability across companies and across therapeutic areas. But my point is that the use of the 10-12 years, $1 billion notion is quickly becoming outdated, if it isn’t already, and is therefore misleading (not to mention self-serving). So should we expect lower cost innovative drugs in the future? Maybe, but I don’t expect the costs savings will be quickly or willingly passed on to consumers. I absolutely believe that the FDA should not lower its safety and efficacy standards, though there is room for updating and validating new clinical endpoints for some diseases. Fortunately, structural changes in how pharma does business in addition to advances in translational medicine means they won’t have to.
Feel free to place your comments below.
References:
1) Kaitlin, K.I. Deconstructing the Drug Development Process: The New Face of Innovation. Nature: Clinical pharmacology & Therapeutics, 87, 356-361.
2) Tempest, Rone, “Death Row Often Means a Long Life”, Los Angeles Times, March 6, 2005
Report this comment
Matt Herper wrote an excellent article several months back in which he simply divided the number of large pharma NDA’s approved by their respective research budgets. It’s a pretty straightforward way of getting at the problem, and the number he comes up with is well in excess of the Tufts number (and without including costs of capital, which are controversial among those without a finance background).
https://www.forbes.com/sites/matthewherper/2012/02/10/the-truly-staggering-cost-of-inventing-new-drugs/
I’m sure that Mr. Herper’s estimate of $4B reflects a period in which ever increasing amounts of money were thrown at declining new technical opportunities, and we will see an improvement over the next several years. But will it decrease to less than $1B (by more than 75%)? I’d be very surprised.
I think it is to some extent fair to characterize high estimates of the cost of developing new drugs as self serving for the pharmaceutical industry. But it would be equally fair to characterize low ball estimates as serving the interests of those attempting to raise capital.
Report this comment
John,
you make a good point that companies hoping to raise capital will paint a picture of expeditious development and lower costs. Investors typically discount those claims to account for expected delays and budget excesses. But I think Herper’s piece suffers from the same issues as the Tufts analysis. The question they were really asking is, “what did pharma get for its R&D money?” Unfortunately, not much. But that’s a different question from, “How much does it cost in time and money to develop a new drug?” The connection between the two is the assumption that pharma did everything that could be done to develop new medicines as fast and as cheaply as possible. I think it’s clear that they didn’t, and I mean this purely from the current perspective on how to optimize clinical designs (e.g., use of biomarkers for patient selection and pharmacodynamics data) and evolution in the choice of therapeutic indications (huge indications like diabetes and CV disease to “niche” indications like ALK+ NSCLC).
My point is that the huge estimates on drug development hurdles are retrospective analyses that are misleading today. Pharma and biotech companies are doing things differently today because the productivity has been so poor; that is, not getting much for their money. So, when I see the <label for=“wpn-comment-body” class=“hidden”>Comment <em>(Required)</em></label><textarea class=“box textarea” id=“wpn-comment-body” name=“comment” placeholder=“Enter your comment here…” required=“1” labelclass=“hidden” autocomplete=“off”>B or B numbers, I’m uncomfortable. -Adam
Report this comment
Fair enough Adam. I certainly saw my fair share of inefficiency in big pharma. But I saw a lot in biotech too. So I’d be interested to see whatever data you have that says that this truly can be done more efficiently.
Many of the small companies that I see seem to be in a situation equally as difficult as big pharma. Arena pharmaceuticals,is about to have the second Adcom for their first drug, and they have a billion dollars in negative retained earnings (e.g., cumulative net losses) on their balance sheet. Optimer, which went with the cheaper alternative of developing antibiotics, had over $200m in negative retained earnings at the end of last year. Curis has over $700M in negative retained earnings. I’m not actually aware of any company that has put a drug on the market without accumulating cumulative losses of over $200M, but maybe you know of some examples? The ones that I am aware of are temptations to despair.
I think there is a distribution of outcomes, but its probably too optimistic to point to the upper 5% and assume that the mean can be moved to that point.
Report this comment
I agree with you in principle, but here are a few thoughts. Accumulated deficits are misleading too, as they include SG&A costs and potentially other non-cash, charges, so will be heavily influenced by the company’s business model. That said, I see ARNA as a great example of my point – drug development done the 1990’s way. OPTR took advantage of shorter trials in acute indications, no doubt. CRIS had funding from Genentech for their approved Hedgehog inhibitor.
As you suggest, it would be interesting to do a Tufts- or Herper-style analysis on biotech companies that achieved an NCE approval, partnered or not. My instinct is that retrospective from today, the average cost would be lower, though the timelines would be just as long, e.g., 10-12 years. But you’re right, biotech companies may have been leaner entities, but they still operated in the same clinical and regulatory environment.
But I still think that Tufts analyses of the future will show a different picture as the efficiencies I’m describing make an impact on productivity. Of the 30 new molecular entities approved by CDER in 2011, 57% were Fast Track/Accelerated Approval, 40% were 1st-in-class, and 37% were for orphan indications. These metrics reflect changing drug development strategies.
Report this comment
Thanks Adam. I certainly hope you are right. I’ve been a pessimist in recent years, and that isn’t much fun at all.