The growing gap between what the public had been told R&D costs are, and the actual costs

One of the reasons why drug companies have fought so hard to block transparency of clinical trial costs is the growing gap between what the public had been told those costs are, and the actual costs, particularly for treatments for cancer and rare diseases.

Some of the gap is related to the much lower number of patients enrolled in trials, the growing use of expedited reviews, particularly in the United States, and the shorter duration for some trials.

In April 2019, IQVIA published a report on 2018 FDA approvals, with these observations:

    Over two-thirds of new drugs in 2018 came through the regulatory process under one of several tracks intended to accelerate development and review, but over the past three years only those with accelerated approval and breakthrough status have seen significantly shorter times from patent filing to launch.

    Of the NASs launched in 2018, 46% were also approved based on data from trials with fewer than 500 subjects in total, as drugs types are increasingly specialty, niche and orphan drugs, which typically enroll fewer subjects. In addition, the amount of time a drug has been tested in a patient population at time of approval — patient years at approval — is declining.

    APRIL 2019. The Changing Landscape of Research and Development Innovation, Drivers of Change, and Evolution of Clinical Trial Productivity. IQVIA Institute for Human Data Studies.

Trial size

In his 2003 paper, Joseph DiMasi estimated the average cost of drug development, including trials, pre-clinical outlays, costs of failures and capital costs, to be $802 million. This estimate was based entirely on project level data from clinical trials, which he used as a proxy for pre-clinical outlays also. In 2003, DiMasi estimated the the total out of pocket outlays on trials, in his still secret sample, at $125 million. DiMasi also estimated that the average number of patients in trials for new drug approvals were 5,303.


In a 2016 paper, DiMas more than tripled his estimate of development costs to $2.6 billion. In the new paper, DiMasi claimed that on average, the costs of phase 1-3 trials, for new drug approvals, was $339 million. Like the 2003 paper, the $2.6 billion figure was based entirely on data on trial costs, with a series of adjustments and extensions. DiMasi was asked by the Union for Affordable Cancer Treatment (UACT) and others to disclose the per patient costs and enrollment numbers for his new study, which also had a confidential sample of data he obtained from the industry. But despite fairly intense pressure to be more transparent, DiMasi would not provide any numbers on enrollment or per patient costs, and few journalists pressed him to do so.

Both the 2003 and the 2016 paper were actually announced at press conferences, in 2016 with talking points only and without the text of the paper, and were widely cited, even before the papers were actually published.

KEI’s own work has shown, that the number of patients cited in FDA medical reviews for cancer drugs was systematically much lower than for non-cancer drugs, in some years, less than a quarter of the enrollment for non-cancer drugs. And also, the variances were huge. In 2010, more than half of all patients enrolled in trials cited in medical reviews were patients in trials for just 2 drugs.

In 2015, the FDA began publishing clinical trial snapshots. These included the number of patients enrolled. We looked at this data and found that by 2015, the average number of patients cited in the snapshot was just 1,838, or just 35 percent of DiMasi’s 2003 average estimate of patient enrollment. The median was much smaller, just 768, or 42 percent of the average. The largest enrollment was 11,145 and the smallest was just 4 patients. For “priority” drugs, the median enrollment was 518, and for orphan products, the median was 411.

In 2018, a record year for FDA approvals for novel products, the trial patient enrollment numbers were dramatically lower. The average enrollment for all products was 750, and the median was 441. The largest enrollment was just 2,484, down 78 percent from 2015. In fact, the drug with the highest enrollment had less than half the number of patients in the snapshot that DiMasi’s 2003 estimate of the average enrollment.

For priority drugs, the median enrollment in 2018 was only 318 patients. For orphan drugs, only 229.

There are differences in the DiMasi and FDA snapshot methodologies counting trials, but the enormous differences in averages, and the declines in snapshot number from 2015 to 2018 (an apples to apples comparison) are instructive.

For new cell and gene therapies, the numbers are even smaller.

The first two CAR T treatments are an illustration both of small trials and the industry exaggeration of costs. Kymriah was developed at the University of Pennsylvania (Penn) on NIH grants and patented by Dr Carl June. Penn licensed the patents to Novartis. The 2017 FDA approval for Kymriah involved just 63 patients. On May 17, I asked Dr. June how much CAR T trials cost. He told me, $150,000 per patient, a number that was close to twice as much as the 2015 PhRMA estimate of oncology trial costs per patient. But with the FDA saying the trials had 63 patients, this put the cost at under $10 million. Meanwhile, Novartis claimed outlays of more than $1 billion on the treatment.

Yescarta, the CAR T licensed from the NIH and marketed by Gilead, was approved based upon evidence from 100 patients. Again, Gilead is making all sorts of claims about its R&D outlays.

At a recent conference on cell therapies in Washington, DC, experts predicted that some new treatments would have trials as small as 10 patients.

Finally, on Friday, the US FDA approved Zolgensma, a new $2.1 million per patient gene therapy for spinal muscular atrophy (SMA). Zolgensma was licensed from a children’s hospital where it had been developed on NIH grants, and was approved based upon evidence of just 21 patients treated with Zolgensma. This is from the FDA press release:

    The safety and effectiveness of Zolgensma is based on an ongoing clinical trial and a completed clinical trial involving a total of 36 pediatric patients with infantile-onset SMA between the ages of approximately 2 weeks and 8 months at study entry. The primary evidence of effectiveness is based on results from the 21 patients treated with Zolgensma in the ongoing clinical trial.

There are a number of other issues regarding R&D costs that bear more transparency and analysis, such as the practice of booking asset acquisition costs and stock options as R&D outlays, or sorting out the costs associated with post approval R&D from bring new treatments to market, as well as identifying public subsides, such as tax credits, grants, etc.

It is no wonder that drug companies are keen to keep the actual costs of trials secret, and to use DiMasi and other consultants to continue to spin narratives on R&D costs that justify prices that increase every year. It is time for more transparency, and governments should lead.

FDA Snapshot Enrollment Data

All NDA BLA Orphan Not Orphan Priority Standard
2015:avg 1,838 2,141 1,158 663 2,955 1,721 2,665
2015:median 768 1,022 436 411 2,235 518 1,367
2015:n 39 27 12 19 20 15 12
2015:min 4 4 75 4 152 4 230
2015:max 11,145 11,145 4,226 4,226 11,145 8,442 11,145
2018:avg 750 798 621 307 1,415 529 1,340
2018:median 441 487 199 229 1,153 318 1,060
2018:n 55 40 15 33 22 40 15
2018:min 34 64 34 34 163 34 150


2,484 2,484 2,184 1,472 2,484 2,414 2,484