Abstract
Why do a number of pharmaceutical companies work on the development of biosimilar insulins (BIs)? Simply because insulin is a large market; for example, the Danish insulin manufacturer Novo had sales slightly below 5 billion dollars in the year 2011 with an annual growth rate of 12.8%. The global insulin market is around 16.5 billion dollars (
Having this motivation in mind, one can assume that a market that was relatively stable for many years (the main players are still the classic insulin manufacturer Novo, Sanofi, and Eli Lilly) is in the focus of other pharmaceutical companies. To what extent and how dynamically this market will change also depend on the price difference that a “copy” of, for example, insulin glargine will offer in comparison with the originator product. It is clear right away that the reductions in price will be no way in the same range as we have seen with generic drugs; most probably the reduction in price with BIs will be in the range of 20–40%. This relatively small advantage in price reflects the fact that biologics not only require an expensive production process, but also a costly approval process, marketing activities, and building up of distribution channels. This might be the reason why the insulin manufacturers from China or India (or Korea/Ukraine/Poland/Israel…) that are the driving force behind the BIs story are looking for appropriate partners in the United States or European Union. In this context, it would be of interest to know why the world largest pharmaceutical company (=Pfizer) has broken off its cooperation with one of the pharmaceutical giants of India (=Biocon) after some years and investment of some hundreds of millions of dollars. Nevertheless, it might also be that the established insulin manufacturers will come along with BIs of insulin formulations of their competitors once the respective patent has expired. For example, it might very well be that Eli Lilly is one of the first companies that come with a copy of insulin glargine to the market in the United States/European Union. They are currently in Phase III trials with a generic insulin glargine (with a different formulation in comparison with Lantus). This in turn might also mean that they do not apply to the regulatory route used by others for getting the status of a BI for their insulin glargine, but that they try to get the status of an originator insulin; otherwise, it is not clear why they run Phase III studies (at least in the European Union this is not required).
Despite that fact the insulin is a huge market and that BIs will offer a certain price advantage, it is not clear if these agents will become a big market success once the companies get market approval. We have seen with other drugs that within weeks after the expiration date of a patent, the market was taken over more or less completely by generic versions of this drug. It is of note to see that in India—in this country human insulin formulations and insulin glargine formulations not manufactured by the originator are available for years by now, and they offer a price advantage—have a market share of only <10%. In India it has also happened in the past that BI batches have had to be withdrawn from the market as requested by the regulatory agency because of safety concerns. Clearly it is tricky to transfer experience gained in one market to others; however, it is clear that patients (and also physicians) have a tendency to stick to insulin formulations (or more to the names?) to which they are accustomed. In addition, the classic insulin manufacturers do offer additional support to patients and physicians in many respects; they also “guarantee” a good batch-to-batch consistency and a high reliability of supply (see below).
Clearly the major reason why we have no BIs on the market in the United States/European Union is the regulatory hurdles established by the regulatory agencies (Food and Drug Adminstration [FDA]/European Medicines Agency [EMA], respectively). It is clear that BIs can never be a 100% copy of the originator insulin; because of inherent differences in the manufacturing process, each insulin has its fingerprint. The crucial question is, are these differences relevant from a clinical point of view, yes or no?
To demonstrate that these are not relevant, bioequivalence of pharmacokinetic and pharmacodynamic parameters has to be demonstrated along with a good immunological tolerability. To demonstrate bioequivalence (the statistical methods described in the European Union guideline for bioequivalence are applied to BIs) with, for example, insulin glargine requires a sample size in the range of >100 subjects (with two glucose clamps) because of the known considerable variability in the metabolic effect induced, but it is also an interesting question whether the statistical requirements raised for a short-acting insulin are appropriate (=can be fulfilled at all?) with a long-acting insulin. The issue (which clinically is their advantage) with long-acting insulin analogs is that their time–action profiles have no clear peak. It is remarkable that until now no data from any glucose clamp study were published in a scientific journal presenting data from one of the companies aiming to get their insulin formulation approved as a BI in the European Union or United States. The only publication in which data from a glucose clamp study are presented compared the metabolic effects induced by an insulin glargine manufactured by a Chinese company with that of the originator formulation; this study has several limitations (e.g., small sample size) and most probably does not fulfill the expectation the EMA/FDA has for such glucose clamps. 2 However, as indicated in a press release of Biocon (see below), it appears to be possible to demonstrate bioequivalence with respect to pharmacodynamic aspects; however, the pharmacokinetic analysis is hampered by the fact that no specific assay for glargine is commercially available.
The EMA has established a set of highly sophisticated guidelines for biosimilar agents already in 2004; this includes an annex for soluble human insulin. A draft for a revision of this annex was expected to become available in Q1 of 2012, but the EMA has not published this until now. One reason for this revision was that the current version deals specifically with soluble human insulin only, and many questions were addressed to the EMA about adequate handling of insulin analogs, especially about long-acting insulin analogs. In the United States it was not until relatively recently that guidelines for biosimilar agents were published; interestingly enough, these do not cover insulin for some years to come. 3 The European Union guidelines for biosimilar agents and BI are used by several other countries as a template for their own guidelines, but it is also worth to note that there are still several countries without established guidelines.
Differences between the originator insulin and a copy of such an insulin can exist not only due to differences in the manufacturing process used more in general, but they might also vary between the batches of insulin manufactured. Consistency of each batch might sound easy for the uninformed; in practice, there are numerous factors that can have an impact of the product manufactured in a given batch (“The process is the product”). So, in contrast to generics it is not easy to manufacture insulin in large quantities in excellent quality in each and every batch (=Good Manufacturing Practice). Certain batch-to-batch differences are showing up with each insulin (also those approved in the United States/European Union). As long as these differences are small and within the range allowed by the regulatory agencies, the insulin is released by the quality control in the companies and can be marketed. It is not known how often batches are rejected by quality control (or at an earlier stage in the production process) because of too large differences. For companies that start with a complex production process, this might happen more often than in a situation where a given process is running for several years. In view of certain limitations the currently available analytical laboratory techniques have to characterize insulin batches (this is one reason why we need data from glucose clamps studies during the approval process), one wonders how large the differences between batches have to be in order to induce clinically meaningful differences and what good criteria for characterization of batch quality are. In line with these considerations one can foresee situations in which back orders of specific batches happen or out of stock announcements are made, and this would require that patients be shifted from one insulin brand to another. However, the fact that Biocon performed a study in Germany with an insulin glargine formulation (
Despite the hurdles mentioned before, the manufacturers of potential BIs are making good progress, and several of them have already or will apply to the regulatory agencies to get their insulin formulations approved as BIs, but it is not publicly known which companies are currently within this process. The only information we have is from an approval process that has failed a couple of years ago: the Marvel case. This Indian company tried to get three insulin formulations approved in the European Union in 2007. However, not only was the quality of the applications bad, also the data from the glucose clamp data presented indicated relevant differences between the reference products and the test insulin formulations. 4 At the end Marvel withdrew its applications in early 2008. It is of interest to note that many of the Phase I and Phase III studies performed worldwide by insulin manufactures like Biocon are not appropriately designed/performed to support an application for market approval in highly regulated markets like the United States/European Union. 5
In view of novel ultra-fast insulins and ultra-long-acting insulin analogs that are in development (some are in late phases of the clinical development, for example, degludec by Novo under review by the FDA and a new basal insulin by Lilly in Phase 3 studies), the development of insulins has clearly not come to end. In contrast, one might say, even if we will have copies of insulin lispro, insulin aspart, insulin glulisine, insulin glargine, and insulin detemir on the market in some time, this provides a push to the interest in insulin and its further optimization. 6 It is hoped that this will allow us to mimic endogenous insulin secretion even better with subcutaneous insulin administration.
In summary, it is only a question of time when the first BI will be approved and come to the market. So, most probably BIs will become a part of the insulin portfolio soon in the near future also in the United States/European Union. However, it is not clear right now what the word “soon” means in the end: 1 year or 5 years?
One concern is that this change of the insulin market will be associated with a massive increase in the already existing plethora of insulin products, which in turn raises the risk of confusion of patients and physicians. Imagine a situation in which, for example, not just one insulin glargine is on the market but instead 10 with similar names. In the case where a patient is switched from, for example, one insulin glargine formulation from to another and probably sometime thereafter again, and suddenly an increase in insulin antibodies is detected, how can the question be clarified which insulin formulation induced this immune response? Traceability is a highly relevant question in a situation where this shifting from one formulation to the other was done by the pharmacist (=substitution) according to the instruction he or she might have from a given insurance company to provide the patients always the formulation with the lowest price. Such switches might happen without any notice to the treating physician. One also wonders how reliable in such an environment pharmacovigilance system should work appropriately (i.e., be able to detect relative subtle issues that might show up in immune/metabolic responses). Differences in the induced responses might be due to certain differences in the insulin formulations as it is clear that BI will never be identical to the originator product, but will be more or less similar to this as the consequence in the different production processes used (see the review by Owens et al. 1 ). One has to acknowledge, though, that with the highly purified insulins that we are using, in practically all patients no or only moderate immunological responses are induced. Even in cases an increase in insulin-binding antibodies was observed, like with inhaled insulin, no clinically relevant changes in metabolic control/insulin requirements were observed. Therefore, one might argue that relatively small differences are overemphasized by companies that want to protect their market share/business; however, in view of the experience gained in the past with biosimilar agents of other therapeutic proteins (in one case only a change in the manufacturing process induced serious consequences), there are reasons for at least a careful approval process and some long-term data.
The call for action raised by the authors of this BI review
1
that well-defined and globally harmonized preclinical and clinical studies should be performed worldwide in the interest of the safety and efficacy of patients with diabetes for each and every BI should be supported by the diabetes community (see the World Health Organization guideline from 2010:
One of the fundamental issues with BI is the lack of truly scientific publications about this topic; until now more “reviews” about this topic (also editorials…) have been published than data from clinical studies. Therefore, such reviews mainly represent the opinion of the authors (supported by some non-publicly available knowledge) and are not evidence driven but more eminence driven.