Discussion
It may be possible to profitably manufacture biosimilars of recombinant human insulin and insulin NPH for a price of US$72 per patient per year or less. Biosimilars of insulin analogues—lispro, aspart, glargine, glulisine and degludec—may be profitably manufactured for prices of US$133 per patient per year or less. With high levels of competition, our estimates suggested that prices could fall to US$48 per year for biosimilars of human insulin and US$78–98 for insulin analogues (table 1). Comparison of estimated prices with recent government procurement prices suggests that robust competition in the human insulin and insulin analogue market would lead to sizeable savings in most countries and that current manufacturers could set significantly lower prices while still making a profit (figure 2).
Price estimates were based on observed prices of API exported from India for RHI, NPH and insulin glargine. For other analogues, assumptions for API costs were based on quotes from biosimilar manufacturers and similarities in structure and process. Given the high level of similarity in molecular structure and manufacturing process, the relatively high price quoted by biosimilar manufacturers for insulin aspart compared with RHI is likely due to differences in demand volumes. Nevertheless, we conservatively assumed that other analogues for which export data were not available were priced similarly. Our estimated prices included a profit margin, a margin to account for transportation and import costs, and a conservative estimate for biosimilar development cost, which we assumed would be recouped equally in resource-limited and high-income markets (rather than at differentiated margins). As a point of comparison, we also estimated the cost of production per item using expense reports from Novo Nordisk (online supplementary appendix), which yielded estimates that were similar to the estimates made based on API prices.
The assumed API costs were in the middle of the range of production costs estimated for monoclonal antibodies (US$20 000–300 000/kg), which are much larger molecules and whose manufacture is more complex.18 31 The prices estimated for RHI were in some cases higher than the prices reported in the ACCISS study (figure 2A).4 This finding likely reflects a combination of conservative assumptions and ‘front-loading’ of the costs of bringing a biosimilar to the market (we assumed that these sunk costs were to be recouped in the first 5 years of sales), which the three big current manufacturers will likely have already recouped many years ago. This finding also suggests that in a small number of countries, insulin biosimilars may initially be marginally more expensive than the lowest prices offered by originators in some contexts. In some countries, where originators offer low prices, in order to encourage a competitive and sustainable market for insulins, new market entrants may need to be supported by procuring at prices that are slightly above the lowest prices available from an originator (such strategies are well established in HIV, tuberculosis and malaria32).
Our estimated prices demonstrate that the price of biosimilar of insulin analogues (except detemir) would not be expected to be more than two times that of RHI, even given conservative assumptions regarding analogue API prices based on low demand volumes (table 1). Estimated prices for insulin analogues were 1.9–2.0 times higher than the estimated price for RHI (except detemir: 5.1–5.9 times higher). This price differential is far smaller than the differentials currently seen in LMIC markets (figure 2). Due to the high level of similarity in molecular structure and manufacture processes, the difference in API price between RHI and analogues is likely explained by the relatively young biosimilars market, differences in demand volumes, different levels of process optimisation, and other overheads such as regulatory costs. If this is true, then the price of API for analogues could rapidly fall to near the price of RHI API. Due to increasing rates of type 1 and 2 diabetes, global demand for insulin API could rise to more than 16 000 kg per year by 2025.1 3 In order to meet this increasing demand, new manufacture facilities will likely need to be built. The cost of these facilities is included in our estimated prices, as it is included in the assumed costs of bringing a biosimilar to market. Governments may offer additional incentives to develop local biosimilar manufacturing capacity, for example, in the form of tax incentives.
Our estimates suggest that yearly treatment with biosimilars of human insulin or insulin analogues could have drug costs similar to, or lower than, current lowest prices for first-line treatments for HIV, for which major international treatment programmes have led to 19.5 million people receiving treatment in 2016.33 34 At the estimated prices, comprehensive treatment of an estimated 19 million people living with type 1 diabetes would cost US$0.9–1.4 billion per year using biosimilar human insulin and insulin NPH, and US$1.5–2.5 billion using biosimilar insulin analogues other than detemir. For comparison, LMIC antiretroviral sales were estimated to be US$1.8 billion in 2014.35 With the global insulin market estimated to be US$32 billion in 2019,4 significant savings would appear to be possible if insulins became available at the prices estimated in this analysis.
Prices for the API are falling: even at the low volumes currently being exported from India, the linear regression models showed 18% yearly decrease in price for exported human insulin API and 27% yearly decrease for insulin glargine. It would be reasonable to expect that with increasing biosimilar production, API prices will continue to fall. More than half of relevant patents protect devices related to insulin treatment, rather than API, and most patents on insulin analogue APIs have expired or are expiring by 2020 in the USA.36 37
The need for price reductions for insulin analogues is made more pressing by the fact that analogues are replacing RHI in high-income countries and a growing number of middle-income countries, despite a lack of evidence for significant clinical benefit.1 If this trend continues, markets for RHI may collapse. Simultaneously, the volume demand for insulins is expected to rise dramatically in the coming decades.3
To stimulate the development of a competitive biosimilars market for insulins, a range of procompetitive policy approaches are available to governments and the international community. These include tenders and special incentives for the first non-originator market entrant. International pooled procurement mechanisms like the Pan American Health Organization’s Revolving Fund and Strategic Fund could also play a role in supporting a competitive biosimilar insulin market for supply of LMICs. Although current analyses suggest that intellectual property (IP) does not pose a significant barrier to biosimilar insulins, if IP barriers did become apparent in some cases, voluntary or compulsory licences could be sought to enable biosimilar production.1 5 Lastly, technology transfer from originator companies to LMIC manufacturers—ie, the transfer of know-how regarding manufacturing processes and/or the transfer of materials such as cell banks—would likely significantly lower the barrier to developing a biosimilar. Other measures to control insulin prices, which are discussed in more detail elsewhere, would remain important, including consideration of import taxes, and controls on distributor and pharmacy mark-ups.1
Earlier commentators have predicted that the price reductions offered by biosimilars will be modest, with many giving estimates for expected price reductions in the range of 10%–70%38; however, the biosimilar market is very young, with a low number of competitive manufacturers for individual products. Recently price reductions as great as 70% have been seen for biosimilar infliximab.39 Price reductions for the first insulin biosimilar available in the USA or Europe—a biosimilar of insulin glargine—have also arguably been modest (a second biosimilar has been approved in the European Union in the past few months, but pricing data are not yet available),40 although large price reductions would not be expected with the first non-originator market entrant. As a point of comparison, the price estimated in this analysis for insulin glargine is about 60%–80% lower than prices of originator insulin glargine in high-income countries.40 Our cost-based estimates thus lie at the higher end of the range of the price reductions predicted earlier by commentators.38
In some jurisdictions, there are insulin-specific regulatory requirements that are less demanding than for biosimilars more broadly.41 However, as for all biosimilars, there are challenges with impurities and within-product variability.38 42 Indian companies have previously sought, and failed, to gain market approval for biosimilar insulins in the European market.7 To address these issues, the WHO is exploring the possibility of using the prequalification programme for insulins.43
Limitations
The key limitation of our analysis is the large number of assumptions that were necessary in order to arrive at estimates of biosimilar prices.
There are expenses in biosimilar manufacture that are not individually considered in this analysis, such as capital expenditures, quality assurance and control, registration costs, costs associated with adherence to manufacturing regulations, and maintaining an aseptic manufacturing process. However, we have included a conservative assumption for the total costs of bringing a biosimilar to market. In addition, many of these costs are likely to be already represented in the estimated costs of API, which are based on completed sales and/or price quotes. By removing the cost of paying a profit margin to a third-party API manufacturer, inhouse production of API or domestic rather than international procurement would likely reduce costs and result in lower prices, or afford manufacturers higher profit margins, than the prices estimated in this analysis. Similarly, the assumed cost of formulation for the ‘conservative’ formula is based on sales prices of finished phial products, which would also include capital, quality control and operating expenditures.
Our estimated prices assume purchase directly from the manufacturer, including a mark-up for transportation costs. Final prices to patients or health systems may include additional mark-ups added during the supply chain.1 These mark-ups will vary between countries, as they may be subject to local negotiations, regulations and other factors. In some cases, these additional mark-ups can be very large. Beyond a margin for import tariffs and transport, we have not included estimates for these mark-ups as our estimates (and comparisons with current prices) are made from the perspective of government procurement. High supply chain mark-ups, where they exist, would be a priority target for cost-containment measures.
We estimated sustainable prices for the 10 mL phial formulation and did not estimate the price of pen or cartridge formulations. This limits the ability to compare estimated prices with current lowest prices as reported by the ACCISS study, as the figures in the ACCISS study are reported for insulins in any formulation4 (figure 2). Our analysis is primarily focused on resource-limited countries, where the majority of insulins are formulated in phials.4 The added costs for producing other formulations, such as self-injections pens, could be the subject of future work.
