Next Score View the next score


    Tufts pegs cost to develop a cancer drug at $648m, and debate rages on

    Dan Kitwood/Getty Images

    How much does it cost to develop a cancer drug?

    A new analysis finds the magic number is $648 million, which is substantially less than an earlier albeit controversial estimate of $2.6 billion for the cost to develop any and all new medicines, in general.

    As with that earlier estimate by the Tufts Center for the Study of Drug Development, though, this latest analysis is already engendering criticism, a reflection of an ongoing debate over true development costs and how these should be calculated. This is important because the pharmaceutical industry has often used R&D costs to justify its pricing.


    “These results suggest that pharmaceutical drug development is extremely lucrative and the current drug prices are not necessarily justified by the R&D spending on these drugs,” the researchers who conducted the new analysis — Dr. Vinay Prasad of Oregon Health and Science University and Dr. Sham Mailankody of Memorial Sloan Kettering Cancer Center — wrote in JAMA Internal Medicine.

    Get Talking Points in your inbox:
    An afternoon recap of the day’s most important business news, delivered weekdays.
    Thank you for signing up! Sign up for more newsletters here

    To arrive at their $648 million estimate, the researchers chose 10 publicly traded drug makers with only one cancer medicine that, at the time of regulatory approval, had no other treatments on the market. They reviewed eligible candidates during a 10-year period beginning in January 2006, some of which were developed internally and others that were acquired.

    They calculated total R&D spending from the first R&D work to the year of approval and also accounted for failures — the cost of drugs that never made it to market. In addition, the researchers assigned 7 percent opportunity costs, which is the return that investors could be expected to forgo if the money had been invested elsewhere while a drug is being developed. This raised median costs to $757 million.

    “The total revenues from the sales of these 10 drugs after approval were $67 billion, more than 7-fold higher than the total R&D spending,” Prasad wrote in an e-mail. “The median time on the market for these drugs was four years. Since the median duration of market exclusivity for oncologic drugs is about 14 years, these drugs will earn billions more. Nine out of 10 companies had higher revenues than R&D spending and four companies had more than 10-fold higher revenues than spending.”

    He also maintained that the cost to develop cancer medicines is unrelated to the novelty of the mechanism of action or the efficacy of the drugs.


    But the results were criticized by Joseph DiMasi, director of economic analysis at the Tufts Center; he was the principal investigator for the 2014 study that determined the cost to develop a medicine was about $2.6 billion at the time. The Tufts center, by the way, receives industry funding.

    “There are a number of serious flaws with this study,” he wrote to us.

    “Most importantly, it includes only a small set of companies that were relatively successful in development during the period they analyzed. As a result, it inadequately adjusts for risks across the industry and so under counts the costs of investigational cancer drug failures. The methodology also ignores non-R&D costs, and likely overstates revenues for half of the firms by counting the purchase prices paid when these companies were acquired in full by larger firms as sales revenues for the cancer drugs examined.”

    DiMasi also maintained that some development costs were missed, since these occurred prior to the start date the researchers used for certain drugs, and the numbers for opportunity costs were too low. He also objected to the fact that the analysis added acquisition prices to revenue for five drugs that were acquired by other companies.

    “This is inappropriate” because acquisition prices account for the value of the company as a whole, not just the value of the approved cancer compound,” he wrote.


    Prasad and Mailankody criticized the Tufts report for, among other things, a lack of transparency. The Tufts analysis did not disclose the names of the specific drugs. In their own analysis, they included a chart naming all 10 drugs examined, along with the R&D start date, R&D spending, opportunity costs, and subsequent revenue.

    Similarly, Bernard Munos, a former corporate strategy adviser at Eli Lilly who is now a senior fellow at FasterCures, a medical research think tank, wrote that “the study starts with a good intent, but suffers from severe flaws that invalidate its results. “It’s unfortunate, but one must keep in mind that an informed debate on R&D costs and drug prices must rest on rigorous analyses. This one fails the test.”

    How so? He argues that the sample size was unhelpful, because the success rate in clinical development for cancer drugs is between 5 and 7 percent. So it takes 14 to 20 drugs, on average, to yield one approval. Either way this would be a big portfolio, and “the cost of developing so many drugs in parallel would be well beyond the resources of the small companies in the sample,” he wrote us.

    Munos also maintains the analysis omits certain costs. He points to a pair of drugs that were cited in the analysis, but included only the cost of developing updated formulations. He also said the analysis overlooks that development of several drugs was shared with partners whose costs were not included.

    Munos also suggested the method to calculate sales “adds apples and oranges,” since the chosen drugs began earning revenues at different times. “It would have been more meaningful to take the revenues for the first four years,” he said, because typically 25 percent of drugs earn very large revenues, while others struggle to recover costs. “Blockbusters pay for the majority of drugs that are never profitable,” he noted.

    Conversely, a consumer advocate, Jamie Love of Knowledge Ecology International, who has studied patient access and development costs, wrote us that “any study of drug development costs that cites real world data and names the specific drugs or projects is helpful.”

    He pointed out that the Tufts study used a different methodology, where most of the costs come from assumptions regarding risks, and the out of pocket costs were generally lower, but risk and capital costs were adjusted “quite a bit higher.”

    Love added the latest analysis does deal with risks by including costs associated with multiple candidates, not only the ones that are approved, but complained that it also does not say much about how the Orphan Drug tax credit reduces companies net costs.

    “I think this JAMA study is useful,” he concluded, “but it won’t be the last word.”

    Ed Silverman can be reached at Follow him on Twitter @Pharmalot.