True or False: Only Two Out of Ten Drugs Earns Back Its Development Costs?

Originally published 3/28/2016

Drug pricing has been much in the news lately, with patients, pundits, and politicians all decrying the astronomical costs of many medicines. In response to this contretemps, trade organizations representing drug makers have provided explanations of why their prices are actually reasonable and provide good value to patients. According to a recently published white paper from BIO, the biotechnology industry’s trade association, “only two of every 10 drugs on the market ever earn back enough money to match the costs of R&D and the FDA approval process before their patent expires.”

Is this really true?Drug development is a remarkably difficult undertaking. It’s bad enough that only about one drug in ten entering clinical trials actually winds up being approved for sale by the FDA. Of these apparent “winners”, eight out of ten of them will actually turn out to be money losers, according to the BIO quote above. The net result: only about two percent of all drugs entering clinical trials will actually be profitable for their developers before their patents expire. Put another way, ninety-eight percent of molecules that look promising enough in animal studies to enter the clinic will eventually either fail due to safety and/or efficacy concerns, or they will win FDA approval and be sold at a loss. This enormous failure rate is a key driver of alternative drug assessment methods, such as organ-on-a-chip technologies.

Is the industry’s overall combined clinical and financial success rate really this low? Estimating drug sales is an inexact process, but are industry folks so bad at it that they continually choose to develop what end up being money-losing drugs? The winners would have to earn billions in excess profits for this trend to be sustainable. Actually, this must generally be true; after all, when was the last time you heard of a Big Pharma or Big Biotech company losing money?

Let’s do some back-of-the-envelope calculations. How much money does it cost to develop and obtain FDA approval for the “average” drug? The most recent estimate is $2.6 billion. The largest part of this expense, $1.1 billion, is the “time cost of money” or the opportunity cost. This is the theoretical amount that could have been earned in some other financial investment had the money not been used to fund drug development. And how long is the average drug sold before it’s patent expires? The industry estimate is about 10 years. This means that a drug “needs” to earn, on average, about $260 million per year just to break even, assuming zero sales after patent expiration. Is this difficult to do? Let’s look at the yearly worldwide revenues of the world’s 100 top selling drugs in 2014-2015. Sixty-six of these drugs earned over $1 billion, and even the lowest earning drug (#100 on the list) brought in $707 million. Despite the fact that about 88 percent of all prescriptions written these days are for generic drugs, the industry apparently has no difficulty in finding and developing big winners.

Having said that, not all marketed drugs are successful; some really struggle to earn money. Worst-case example that I could find: CTI BioPharma’s Pixuvri (pixantrone). An FDA advisory panel unanimously rejected this drug in 2010, so the company decided to sell it in Europe. After winning conditional marketing authorization from the European Commission in May 2012, the company only sold $12.7 million of the drug during 2013-2015 (about $4.2 million per year). This was actually much less than the $17.8 million paid to CEO Jim Bianco as compensation over that interval. Pixuvri sales are actually going down; in 2015 were only half of what they were in 2014. It’s a pretty safe bet this drug won’t ever earn back its development costs. Another sound wager: this drug didn’t cost anywhere near $2.6 billion to develop.

Biopharma, as an industry, outspends virtually every other industry in the percentage of revenue devoted to R&D. For many large pharma companies, this percentage is in the 15 to 20 percent range. How much a drug needs to bring in as revenue depends on the size of the company. Organizations with tens of thousands of employees must earn substantially higher revenues than those obtained by a virtual biotech company with six employees. Drugs for rare diseases can be developed for much lower costs than medicines for more common disorders, such as diabetes or heart disease. There are two primary reasons for this. The companies that develop these orphan drugs are generally smaller in size than Big Pharma members, and clinical trials can be done with many fewer patients and yet still yield statistically valid results.

Clinical trials for drugs to prevent heart attacks or strokes might need to include hundreds of times as many patients as ones set up for a rare disease, and would therefore be a lot more expensive. A drug to treat scorpion-inflicted wounds was approved after a clinical trial of only 15 patients. Drugs intended to treat rare diseases often have less than 200 patients in their pivotal clinical trials. In contrast, the ALLHAT trial, which studied the ability of anti-hypertensive and lipid lowering agents to prevent heart attacks, had 33,357 patients that completed the approximately five-year trial.

Not all drugs cost a fortune to develop. Genzyme (prior to its acquisition by Sanofi), BioMarin, and Alexion each developed rare disease treatments for less than $1B, sometimes way less (BioMarin spent $134 million per drug). Forbes’ Matthew Herper compiled a list in 2013 of how much different companies spend on R&D per drug. He identified a dozen companies that spent less than $100 million. According to my back-of-the-envelope calculation above, their drugs would only need to earn about $10 million per year to be profitable. These drugs turned out to be quite attractive to other companies. According to my research, half of those companies on Herper’s list had been acquired by 2015.

Are Pharma Sales Executives Really That Stupid?

Given how expensive it is to produce new medicines, it’s hard to see why companies would knowingly develop drugs that will not earn back their costs. These decisions, of course, are not made by “companies”, but by executives within those organizations whose primary goal is, generally speaking, to maximize shareholder returns. It’s easy to find examples of “exuberant” drug sales predictions that weren’t matched by subsequent sales. Take Exubera, an inhaled form of insulin developed by Sanofi-Aventis/Pfizer. It was brought to market in 2006 because some marketing wizards thought that diabetics would prefer an inhalable insulin product instead of the standard injections. Projected earnings were estimated to be in the range of $1.5 billion per year. However, after two years the drug had garnered less than 1 percent of the insulin market, and it was discontinued in 2008. Even worse, Sanofi later partnered with MannKind to bring another inhaled insulin, Afreeza, to market in 2015. Less than a year later, lousy sales ($7.5 million in 2015) led Sanofi to bail on this deal as well. Insanity, it’s been said, is doing something over and over and expecting a different result. Maybe Sanofi never heard this quote.

On the other side of the coin, drug revenue estimators often guess way too low. Novartis was initially quite reluctant to bring the anti-cancer drug Gleevec (imatinib) to market; peak sales were predicted to top out at only $100 million per year. Two years after it launched in 2001, the drug earned blockbuster status (sales over $1 billion/year). In 2015, the drug generated $4.7 billion in worldwide sales. The Novartis development team is no doubt quite happy that it decided to move forward with taking this drug to market, even if they had to be dragged into making that decision.

Some Companies Are Better Than Others in Choosing Drugs to Develop

In a recent article reviewing Pfizer’s checkered history of mega-mergers, Bernard Munos compared Pfizer’s lack of R&D productivity with Johnson & Johnson. Writing about Pfizer, Munos said, “From 2005 through 2014, it spent $80 billion in R&D, and got 10 drugs approved by FDA. In comparison, Johnson & Johnson spent $51 billion and got 15 approvals.” Not all drug approvals, of course, are created equal, and different drugs will generate dissimilar revenue streams. This comparison might actually skew heavily towards Pfizer’s favor if all ten of their approved drugs were blockbusters, and none of J&J’s were. These analyses can be tricky, because one also has to calculate R&D spending for drugs that were developed internally as well as those that came from acquired companies.

Drugs Can Be Highly Profitable Even After Their Patents Expire

Note that the “only two out of ten….” sentence ends with the qualifier “…before their patents expire.” This qualifying phrase suggests that some drugs do eventually turn a profit after their patents expire. Many drugs can earn money for decades after their patents have ended. We all learned this when Turing Pharmaceuticals acquired and then jacked up the price of pyrimethamine (Daraprim), a drug used for treating toxoplasmosis (and later HIV) that’s been on the market for 62 years (it received FDA approval back in 1953). Drugs don’t need to recoup their R&D and clinical costs before their patents expire; it’s just a nice idea. Pyrimethamine likely turned a profit way back in the 1950s, although finding that data now would be a real challenge. Turing CEO Martin Shkreli claimed that his company could not make a profit selling the drug for $13.50 a pill, which was the price that CorePharma (and later Impax Laboratories), the previous seller(s) of the drug in the U.S., were charging. The veracity of his statement is likely tied to the fact that he paid $55 million to acquire the drug. Shkreli didn’t mind grossly overpaying for the drug because his plan was to raise the price to $750 per pill. As many critics noted when this scandalous price hike became public, this same drug is available in the UK for about 66 cents per pill, and costs only about 2 cents per pill in Brazil.

Revenues always take a huge hit when a generic form of a blockbuster medication becomes available, but the original drug can continue to earn very healthy revenues even with a generic successor(s) on the market. Lipitor, which lost patent protection in 2011, was still ranked as the 36th best selling drug in 2014, with worldwide sales of $2.7 billion (most of that revenue comes from outside the U.S.). That’s down more than $10 billion from the drug’s peak sales year, but it hardly qualifies as chump change. Pfizer’s plans to sell Lipitor over the counter (i.e. without a prescription) were scrapped in 2015 when a clinical trial revealed that many patients could not take the drug correctly without help from their doctors.

Drugs Strategically Removed from the Market Before Patent Expiration

Some highly profitable drugs are deliberately taken off the market prior to their patent expirations. They are replaced by “superior” (and usually more costly) versions of the drug that may offer minor advantages, such as less frequent dosing, and are covered by newer patents. The idea (referred to by critics as evergreening) is to force patients to switch over from the old drug to the new version before generic copies of the older drug hit the market. The goal: keep patients on the new, improved drug from switching to the now “inferior” generic version of the old medicine. Extra revenues generated by the replacement drug more than make up for the loss of revenue when the original medicine is pulled from the market.

Drugs Taken Off the Market Before Patent Expirations for Other Reasons

Some highly profitable drugs are yanked off the market due to toxicity issues, disappearing before their patents expire. Think Vioxx and Phen-Fen. A few medicines are initially approved for the market based on early clinical data, but eventually the drug gets pulled when they’re shown to have no real benefit. For example, the FDA removed the breast cancer indication from Genentech’s bevacizumab (Avastin) for a lack of safety and efficacy. Gemtuzumab ozogamicin (Mylotarg; used to treat acute myeloid leukemia) was voluntarily taken off the market by Pfizer due to ineffectiveness eight years after receiving accelerated approval from the FDA.

U.S. Patent Expirations Have Historically Not Been As Important for Biologics

That’s because, until very recently, there was no path available in the United States for developing biosimilars, which are essentially generic versions of these proteins. This process finally changed in March 2015, when the FDA approved filgramstim-sndz (Zarxio), a copy of Amgen’s filgramstim (Neupogen). Neupogen originally won FDA approval way back in 1991, some 24 years earlier and at a time when U.S. patents were only good for 17 years (it’s since become 20). Biologics, of course, have been among the most profitable drugs worldwide.

What About Altruism and Good Will?

There’s no requirement that drug companies develop only those medicines that are guaranteed to make money. Good publicity can help to burnish a company’s ethical reputation (something that’s been badly damaged as of late in the pharma industry). There are many examples of drugs being donated overseas by pharma companies to treat various tropical diseases, such as leprosy, guinea worm disease, and sleeping sickness. The classic example is Merck’s gifting in 1987 of its anti-parasitic drug ivermectin (Mectizan), which is used to treat river blindness in Africa, South and Central America, and the Middle East. The donation program is still ongoing after 29 years. The drug was originally introduced in 1981 for veterinary use. I don’t know if it recouped its development costs in the veterinary market, but it’s been incredibly valuable as a drug in the third world.

How Exactly Are R&D Costs Per Drug Calculated?

It’s nearly impossible to assess the validity of this “two out of ten drugs” claim without access to tightly held industry R&D data. One big question arises as one thinks about this value proposition: how are costs for R&D and the FDA approval process actually quantified? These numbers are apparently difficult even for industry leaders to calculate. Dr. John Lechleiter, the CEO of Eli Lilly, argues strongly that requiring increased disclosures by drug companies is a misguided effort, and he assigns the blame for high drug prices to greedy insurance companies. In this same op-ed piece, he refers to the “task of assembling drug-by-drug R&D-spending figures” as “nearly impossible.” This is the excuse as to why most companies don’t ever report their actual expenditures for each drug project. Instead, they share “average numbers” (e.g. the $2.6 billion cited above) that include the cost of failed drug development efforts. Are these R&D numbers based, in part, on drugs that get acquired by other companies for big bucks? Let’s say that a small biotech spends $20 million dollars to develop a drug to the point where it is about to enter phase II trials. If a Big Pharma company buys the small biotech for $8 billion to acquire that drug, does that price (plus any additional funds expended to finish bringing the drug to market) represent the new “costs of R&D and the FDA approval process?” If true, this would point to M&A activity as a potential key contributor to the high cost of drug development. If Big Pharma A purchases Big Pharma B for $100 billion, and acquires a developmental pipeline of 40 drugs, how does one calculate the actual development costs for every individual drug? The short answer is that it's impossible to say, because those of us outside of the company will never see those numbers.

Where Did The “only two out of ten drugs…” Phrase Come From?

Interestingly, I found nearly this same quote in a 2008 uncredited editorial in the Wall Street Journal. And a 2008 statement from PhRMA said, “only two of every 10 drugs that reach the market ever earn back enough money to match or exceed the average R&D cost of getting them to the marketplace.” Notice that the patent expiration issue was not part of this quote back then, and the qualifier instead focused on average, not individual, R&D costs. Going back even further, I found a slightly different version that stated, “only three of every 10 drugs that reach the market ever earn back enough money to match or exceed the average R&D cost of getting them to the marketplace.” The source of the datum used to construct this quote looks to be Figure 5 in a 1994 study by Duke University economists. The Figure plots the after-tax revenues for marketed drugs against their drug development costs in after-tax 1990 dollars for drugs introduced from 1980-1984, expressed in deciles. This outdated analysis suffers from some significant problems:

Incomplete and out of date: According to this chart, there were 103 new chemical entities (NCEs) approved in the time period 1980-1984, yet the analysis only includes 67 drugs. Note: only one biotech drug (insulin in 1982) was approved during those years as the industry was still in its early days.

Cherry picking: The sample size was small (only 67 NCEs) and included “all NCEs originating from and developed by the pharmaceutical industry, except those in a few therapeutic classes where the R&D process is likely to be non-representative.” It’s not clear how these “few therapeutic classes” were determined to be “non-representative”, but it appears that the total number of NCEs in these classes was 36. Why weren’t they included?

Inappropriate comparator: Average R&D costs were given as $202 million. How was this number calculated? “While we do not have R&D specific data for each of the NCEs in our sample, a major study of R&D costs for pharmaceuticals is now available. The study, by DiMasi et al. (1991) is based on a representative survey of U.S. drug firms, and utilizes cost data on a random sample of 93 drugs first tested in humans between 1970 and 1982. These drugs had an average introduction date of 1984.” “Representative surveys” and “random samples” make it hard to discern the validity of the average R&D cost figure.

The authors go on to tweak the analysis by redoing their calculations after the removal of Zantac from the list, which at the time was the world’s top selling drug. This calculation dropped the average returns of each of the drugs by 33 million, and the remaining 66 drugs had an average net present value of minus $10.7 million. Another way to say this would be that without Zantac, each of the remaining 66 new drugs introduced between 1980-1984 would have been predicted to lose money. This defies credulity.

You don’t need to have a PhD in economics to know that this analysis, to put it charitably, was problematic at best. According to Fortune magazine, the drug industry has historically been more profitable than all other industries. Industry profits as a percent of revenue increased from about 9 percent in 1980 to over 18 percent in 2000 (despite a continuous rise in R&D costs over that period). All other industries never exceeded 6 percent during this same time period. While coming up with new drugs is clearly getting harder over time, monopoly pricing remains the huge engine that drives this industry. Even in a sea of clinical drug failures, it’s clear that biopharma’s ability to generate profitable drugs is much greater than the “only two out of ten drugs….” phrase would have you believe.

Author’s note: I’d like to thank Greg Heberlein for compiling the financial data on CTI BioPharma