Monopoly and public health in the medication market: The case of sofosbuvir

Introduction

Over the last years, the price of prescribed medications has been experimenting double digit growth. New, innovative compounds may arrive to cost tens or even hundreds of thousands of dollars.^1,2 Among the causes of this trend, patent certainly plays an important role.

The most striking example was the marketing of sofosbuvir ³ by Gilead Science Inc. So striking it was to deserve an inquiry of the US Senate Committee on Finance, whose Report ⁴ offers a unique opportunity not only of understanding the pricing process in a pharmaceutical company, but also of having direct cognition of industrial data which rarely are disclosed by pharmaceutical industries.

Gilead commercialized sofosbuvir (Sovaldi) in 2013, after the acquisition of Pharmasset Inc., a start-up company which was developing the compound. In the first one and half years of commercialization, Gilead had a monopoly of the care of Hepatitis C, being the only Direct-Acting Antiviral (DAA) available on the market for interferon free treatment of Genotype (GT) 2 and 3, and later, in combination with ledispavir (Harvoni), of GT 1 infections.

In this paper, using data from the Report and from other open sources, we estimate the premium Gilead gained from being a monopolist, that is the differential between the cost of developing and marketing sofosbuvir and the revenue it has generated, beyond what can be considered a reasonable profit. Being an estimate, and given the magnitude of the difference between cost and price which will be shown later, this exercise does not need very precise figures.

The $11.1 billion Gilead paid for buying Pharmasset ⁵ can hardly be considered R&D cost. It was a financial investment, made betting on a return conditioned by a monopoly situation. Therefore we shall not consider it.

The market

Sofosbuvir market is given by Hepatitis C Virus (HCV)-infected persons with chronic liver disease. Let us limit our exercise to the USA and the European Union, where more reliable data are available, and where Gilead has realized the most of the revenue from the drug.

According to CDC, in the USA there are 3.5 million people with HCV chronic infection; of these, 2.8 million are estimated to have developed a chronic liver disease. ⁶ In Europe, data are not that good. ECDC reports 30,000 official cases of infection in 2012, but it is not known how many of these are incident or prevalent, there are no data from a number of countries and there is a substantial underreporting. ⁷ We may refer to a WHO study which estimated 5,700,000 chronically infected people in the EU, ⁸ and adopting the CDC disease model, this would mean 4.5 million people with chronic liver disease, thus a potential market of 7.3 million people.

Of course not all of them can, nor need to be treated on the first year. A 10-year program can be devised as a good public health strategy, because of the long interval between infection and the onset of the liver disease. Patients severely ill can be treated first, then the others, beginning with those whose infection is known for a longer time. Such a strategy would also allow possible unexpected effects to show up while treating people with higher benefit/risk ratio.

R&D cost

R&D cost has three components: the R&D cost of sofosbuvir itself, the share pertaining to sofosbuvir of Gilead R&D failure cost, and the opportunity cost of the capital.

The Report provides the actual R&D cost, as derived from Pharmasset and Gilead internal documents. Pharmasset spent $62.4 million for pre-clinical studies, included a federal grant of $244,479.25. ⁹ Gilead declared $880.3 million as R&D cost from 2012 to 2014, when FDA license was obtained. However, this amount was not spent by Gilead on sofosbuvir solely, but also on three other different compounds (one of them being sofosbuvir plus lesdipavir). Gilead failed to provide the Senate with the cost of sofosbuvir alone, despite repeated requests to do so. ¹⁰ Thus, one-fourth of this figure can be assumed as a good estimate of the real cost, i.e. $220 million, which is also consistent with the $240 million Pharmasset had forecasted as the cost of phase 2 and 3 studies, before being acquired by Gilead. ¹¹ Post marketing study cost was not disclosed by the company. ¹² It can only be estimated. Taking as a reference phase 3 studies, whose cost was about $80 million, according to data provided by Pharmasset, and considering that post marketing studies last longer, we may double that figure, magnanimously not considering that often phase 4 studies have also a marketing purpose. The R&D cost is then $442 million.

Industries strongly claim that only a fraction of the research investment leads to the marketing of a profitable drug, and that this is a cost they need to recover when commercializing successful products. Thus, also these failure costs need to be included in our simulation.

In the Gilead website, annual reports are published. As of today, reports from 2011 through 2015 are available. ¹³ In all reports (but 2014), a table acquaints us with compounds under investigation as well as the phase in which they are (Pipeline). Putting together this information across the years, we derived a table showing how many compounds were abandoned and how many arrived to the market. In the 2014 report, the Pipeline table is missing, but there is information in the discursive presentation. Thus, data from 2014 are logically inferred (compounds present in the 2013 and in the 2015 reports are assumed to be quoted also in the 2014 one). It must be noted that some information is not accurate. For instance, GS-9820 compound is mentioned as being in phase II in 2013, and not quoted any more afterwards, even though others sources refer to it as being under study in phase I until 2016. ¹⁴ The complete picture is in the Excel file, spreadsheet Pipeline, in the online Supplementary Material.

Over the period 2011–2015, Gilead submitted for approval four compounds and started marketing 9 new products (Excel file, spreadsheet Success, in the online Supplementary Material). On the contrary, compounds which appear to have been abandoned ¹⁵ are 15. This is a 37.5 % success rate. The total cost estimate is $2.071 million (details are shown in the Excel file, spreadsheet Failures, in the online Supplementary Material). Dividing $2.071 million by nine marketed compounds, ¹⁶ we obtain $230 million, which is the failure cost share for sofosbuvir.

The opportunity capital cost, the economic theory says, is the difference between the actual return of a given use of the capital and the best one it could produce, had another and more profitable use been made of it. ¹⁷ However, there is disagreement about including it in the R&D cost of medications ¹⁸ : in fact, one may wonder which could ever be, this more profitable investment, rather than developing drugs. If this opinion may be challenged as a general statement, it certainly holds true in the case of sofosbuvir. In the report, neither in Pharmasset or Gilead documents, nor in claims from firm officials, there is any mention of it. Thus, in this exercise, it will not be considered.

Production and distribution cost

The production cost of sofosbuvir is not revealed by Gilead. The Report documents as Pharmasset forecasted for the testing phase a manufacturing cost of $1 for a 400 mg pill, thus $84 for a 12-week course of therapy. Also, Pharmasset estimated that this cost would be cut to one-third, once mass production had begun, ¹⁹ that is $28. This cost relates only to the pharmaceutical principle and does not include converting it into the Finished Pharmaceutical Product (FPP).

A couple of studies have estimated the manufacturing cost of several DAA.^20,21 One calculated a production cost of sofosbuvir of $101 for a course of therapy for a market target of 5 million patients. The figure includes a markup of 25% as a profit margin of the Active Pharmaceutical Ingredients, and an add-on of 40% for conversion to FPP. The other, based on a production volume of 1 to 5 million courses of therapy, and including the 40% markup for conversion to FPP, gives a cost in the range between $64 and $136. So, let take $101, as evaluated by the first study.

Then distribution cost needs to be added. This cost varies considerably according to the drug, the health system organization, the country, the supply chain, the quantity, and so on. A rough estimate made by Kaiser Health News says that, given a Wholesale Acquisition Cost (WAC) of $250, the drug maker receives $175, i.e. thus with a 43% add-on. ²² Even though WAC included profit and R&D cost, while we are following a different model, we can take that percentage as the best possible estimate. Applying it, the fixed cost for manufacturing and distributing sofosbuvir to the final user would be $144 per course of therapy.

A business plan

We can now put everything together and decide a hypothetic selling price. R&D cost, including the amortization share of failures ($442 plus $230 million equal $672 million), is to be repaid pro quota by any course of therapy sold. Being the market size 7.3 million patients, the unit cost would be $92, to be added to distribution cost. Thus, the overall cost of a single course of therapy would be $236. Table 1 gives a summary of our business plan assuming a 100% profit surcharge, so a selling price of $472 per course of therapy. In the Excel file, spreadsheet Simulation, in the online Supplementary Material, anyone can play with her/his own business plan changing the parameters we have here considered.

OPEN IN VIEWER

Table 1.

Hypotetic 10 year business plan for commercializing sofosbuvir ($)

	No. of patients	No. of treated per year	Yearly manufacturer cost	Yearly profit b	Yearly payor cost	Period profit
Medicare	450,000	45,000	9,205,404	9,205,404	18,410,808	92,054,042
Medicaid a	698,000	69,800	14,278,605	14,278,605	28,557,210	142,786,048
Other USA	1,650,000	165,000	33,753,149	33,753,149	67,506,298	337,531,488
EC	4,500,000	450,000	92,054,042	92,054,042	184,108,084	920,540,422
Total	7,298,000	729,800	149,291,200	149,291,200	298,582,400	1,492,912,000

Gilead business plan was quite different. In addition to the above-mentioned costs, it had to recoup $11,1 billion. Sofosbuvir was sold at $88,000 (WAC) per course of therapy (before discount and rebates, reported however as limited in the first year/year and half), and generated $8.690 billion revenue in the first 12 months of marketing.^25,26 From the third quarter of 2014 on, when Harvoni was commercialized, Gilead wisely is not providing analytic sale figures anymore, packing together revenues from all its antiviral products. However, comparing financial reports across years, including 2015, ²⁷ the income from Harvoni and Sovaldi in the third quarter of 2014 and in 2015 can be easily evaluated as $20 billion approximately. Certainly, a great commercial success.

Discussion

The sofosbuvir case is quite peculiar. Gilead enjoyed a monopoly situation in a large and sensitive market and Sovaldi and later Harvoni have been adding real value to the treatment of HCV-infected patients. This may not be the case with other specialty medications. However, the very fact of being such an extreme case emphasizes some key issues about the price of medications, and made it possible to know firsthand details about its business facts.

One of these is the efficiency of the R&D process, with a success rate much higher than those reported in literature. A Tuft University study ²⁸ found, over 21 years, only 19% success rate. Another study found much lower figures over nine years. ²⁹ Other studies offer more different figures, but always worse than Gilead performance. ¹⁸ It is true we have been observing Gilead over five years only, but probably, rather than the observation period, methodology matters. The R&D failure rate is a political key issue for pharmaceutical industries, which justify that way high prices. More important than the success rate, therefore, is R&D cost, which appears also to be low in Gilead. The above-mentioned Tuft University study calculated an average R&D cost of $2.6 billion. However, this study was heavily questioned. ³⁰ Other studies give a different and much lower figure. One of these for instance (partially financed by a drug company) quotes R&D cost at an average level of $1.6 billion (2011 dollars). ³¹ All studies include the cost of research failures and the opportunity cost of capital, the latter contributing for about half of the overall cost. The opportunity cost of the capital should be regarded as very specific of each investment, and it is difficult to generalize how to include it in a cost model. Presenting average values may be misleading. Had we included in our model the capital cost, we would have had a figure consistent with that of the second study quoted above. Anyway, included or not, the order of magnitude of the difference between the cost estimated here and WAC determined by Gilead does not change.

In this exercise marketing, administration and other costs have not been considered. In the 2015 Gilead financial report, ²⁶ the entry Selling, General and Administrative Expenses is $3,2 billion. To include it in the model, it should be divided by the number of courses of therapy of all drugs commercialized by Gilead, a calculation we have no data to perform, and probably not relevant for the purpose of this simulation.

Gilead, as other pharmaceutical industries, takes in no account the usual cost factors considered in other business sector to determine a product price. In an interview, Gilead executive Jim Meyers, who played a lead part in making the pricing recommendation, did not know the cost of manufacturing the drug. ³² The peculiarities, distortions and inefficiencies of the medication market are widely treated in the literature. ³³ Among motivations for high prices, two, in our view, are most relevant.

The first is the odd way the customer surplus ³⁴ operates in the asymmetric health care market, so peculiar that we could nominate it patient surplus. As the severity of a disease increases, the patient surplus tends to infinity. How much is a person ready to pay, or what pressure can she/he put on a payor, when her/his health or, worse, life is being threatened?

The second is monopoly. Patent is defended as being the only mean to promote research and technology advancement in the pharmaceutical sector. Our data demonstrate that monopoly may add a far larger toll than what it is needed to repay investment. In fact, when patent did not exist, research was carried out and pharmaceutical industries were on the market, earning good profits any way. ³⁵ On the other hand, a patent-free research could be more efficient because of the information circulation and sharing. ³⁶ The question is much ideological. What is the prevalent interest? To provide better care at a reasonable cost, limiting therefore rights currently granted to patent holders under any legislation, or to defend the free market, where actors may pursue the maximum profit using any legal mean? As a matter of fact, antitrust legislation exists, and has been applied several times in the name of the social interest, limiting business freedom. Why a similar principle should not be applied to the medication market? On how to repay investment, a number of solutions have been suggested, had an antitrust legislation be enforced. For instance, the reimbursement of R&D cost or the public sector assuming directly the responsibility or the cost of phase II and III studies.^33,35,36 Another mean we suggest is to oblige a manufacturer to sell to at least three other companies the patent against a price which distribute the investment cost among all of them.

Medication price trend largely exceeds those of almost all other economic indexes, ³⁷ diverting a huge amount of resources from other goals, that is, a rent and not an investment repayment. A number of studies on sofosbuvir economics demonstrate that, even though the cost per treatment is high, the cost per cure decreases. However, we could not find any of these studies that in the disclosure did not report a conflict of interest of the authors. There are a number of issues which can be raised, about these studies. For instance, they do not consider that the cost of a treatment is concentrated over a short period of time while the cost of a cure may be spread over years, and this makes a difference on the burden the payor sustains. They do not consider that only a fraction of infected people develops a severe disease, with the subsequent costs. And, last but not least, resources made available by technology improvement are adsorbed almost completely by drug manufacturers instead of improving social effectiveness.

We close with a question: since Gilead has recouped its investment in Sofosbuvir and Harvoni ad abundantiam, why it maintains the price at such high level instead of lowering it and win competition against other products which have arrived in the market in the meanwhile?