At the dawn of biosimilar insulins

If a classical, small molecule drug loses patent and data protection, generics may be marketed based on relatively simple regulatory requirements. The manufacturer has to show that the active substance of the generic is identical to the active substance of the original drug and has comparable pharmacokinetics in a bioequivalence trial. Clinical efficacy and safety studies are not needed for a marketing authorisation of a generic.

This generic pathway has been the basis of the introduction of many affordable safe and effective alternatives to innovative medicines. However, the generic regulatory pathway is not suitable for biological, including therapeutic proteins as interferons, monoclonal antibodies and insulin. The main reasons are the complexity of protein pharmaceuticals and the lack of analytical technology to completely characterise these products. The complexity is mainly caused by their heterogeneity resulting from modifications by the natural biochemical processes in the host used in production. Also production, purification, formulation and storage may introduce modified molecules. In addition, DNA and protein host cells may be present in this mixture of molecules, as well as resins and monoclonal antibodies used for purification.

It is impossible to show that two protein products are identical. Therefore, the European Union introduced the term ‘biosimilars’ in their 2004 legislation providing a regulatory pathway for marketing authorisation of ‘copies’ of biological therapeutics. On the basis of this legislation, the European Medicines Agency (EMA) was the first regulatory agency in the world to issue a comprehensive set of guidelines designed by the CHMP, the scientific committee of the EMA.¹

The most important requirement for a marketing authorisation submission of a biosimilar, which sets it apart from the generic regulatory pathway, is the need for comparative clinical data.

The EMA/CHMP not only issued general guidelines concerning quality, clinical and non-clinical testing but also immunogenicity. In addition, for many products, including insulin, the EMA/CHMP issued product-specific guidance. At the core of these guidelines is the need to show similarity in quality, safety (including immunogenicity), and efficacy between the biosimilar and the original reference product.

Since 2006, the EMA/CHMP/EU has approved biosimilar versions of recombinant somatropin, recombinant human erythropoietin (rHuEPO), filgrastim, follitropin and infliximab. A number of products (including insulins) were rejected or withdrawn by the submitting companies after an unfavourable first review by the EMA/CHMP.²

Many industrial countries, including Australia, Canada and Switzerland, have implemented legislation and/or guidelines based on the example set by the EMA. Also, the Food and Drug Administration (FDA) in the US is implementing a marketing authorisation regulatory pathway pioneered by the EU.

The World Health Organization (WHO) has also adopted the principles of the EU approach for its biosimilar guideline.³ The WHO guideline provides guidance for low- and middle-income countries, of which many are in the process of introducing their regulatory approach for biologics and biosimilars. Because of the lack of patent protection in most of these countries, many intended copies of biologics are available.

Sometimes these products were allowed under the existing regulatory requirements for generics. The quality of some of these products has been reported to be poor and the documentation showing their safety and efficacy is not available.

Because use of the term biosimilar should be restricted to products evaluated by a regulatory procedure requesting comparative clinical data and questions remain regarding their clinical effects, the term ‘bioquestionables’ has been suggested for these products.⁴

On 10 September 2014, the European Commission, on the basis of a positive opinion, authorised the marketing of Abasria®, a biosimilar insulin glargine from Eli Lilly/Boehringer Ingelheim, for the treatment of type 1 and type 2 diabetes.⁵ No other biosimilar insulin has yet been authorised by the regulatory agencies in the industrialised world.

At the time of writing, the CHMP had not yet published the European Public Assessment Report (EPAR) of this new biosimilar insulin glargine. The EPAR is a scientific discussion about its quality, safety and efficacy and the similarity to the reference product, Lantus. But considering the high quality of the other biosimilars approved by the EMA/CHMP until now and the lack of any product-specific safety issue notwithstanding an intense pharmacovigilance, Abasria^® can also be expected to be of a high standard. Lilly/Boehringer Ingelheim have presented some data about their product during recent scientific meetings, confirming this expectation.

In the past, EMA/CHMP has evaluated submissions of biosimilar insulin candidates developed by Marvel Life Sciences. The submissions were withdrawn in 2007 on the basis of a negative first evaluation by the CHMP; the same happened after resubmission in 2012. These withdrawals were, in principle, voluntary, but on the basis of major concerns expressed by the EMA. The three biosimilar candidate insulin products were: Solumarv^® (soluble rapid-acting insulin); Isomarv Medium® (intermediate-acting isophane insulin); and Combimarv^® (a mixture of the other two products, 70% long-acting isophane insulin + 30% soluble rapid-acting insulin).

The major concerns are discussed in the respective withdrawal assessment reports issued by the EMA and were mainly related to immunogenicity issues.⁶ The doubts about the quality issues of the biosimilar candidates and differences in pharmacodynamics were identified in a clamp study in volunteers. The Marvel biosimilar insulin candidates also showed a higher incidence of hypersensitivity reactions and a lower efficacy during a 24-week period based on Hb1c levels compared with the original products.

The EMA/CHMP suspected association between the high level of impurities and a higher titre of insulin antibodies in the patient treated with the Marvel insulins, leading to a lesser clinical efficacy. The EMA/CHMP also had problems with the validation of the assays measuring the antibodies to the impurities. No data were submitted on the antibody status of the patients after the 24-month treatment. The EMA considered the risk management plan insufficient.

These rejections also show that immunogenicity of insulins can still be a clinical problem and is also a major regulatory concern. And immunogenicity was also the main safety issue in the discussions on how to regulate the marketing authorisation of biosimilars. This is reflected in the 13 different guidelines of the CHMP/EMA on how to evaluate immunogenicity of biologics and biosimilars.² This emphasis on immunogenicity results from a number of incidents with therapeutic proteins caused by an increased induction of antibodies after a simple change in formulation.

According to the regulations, the amino acid sequence of the biosimilar and the original product should be exactly the same; but they have been shown to differ in formulation, level of impurities, including aggregates, type and level of glycosylation, manufacturing process and other factors with the original products. All these factors have been identified as drivers for the immunogenicity of therapeutic proteins rather than the amino acid sequence.

Also for insulins, a strong association exists between immunogenicity and the formulation and level of impurities (especially aggregates).⁷ The first commercial porcine- and bovine-derived insulin products introduced in 1922 contained less than 1% insulin and were comprised mainly of impurities. Serious adverse effects caused by immunogenicity were common, such as anaphylaxis, hypersensitivity and insulin resistance.

The introduction of new purification technologies dramatically increased purity and reduced the adverse effects, although insulin-related and unrelated impurities were still present. In the 1970s, highly purified, mono-component, animal-derived, insulin products were introduced. And although the incidence of antibody formation was shown to be high, these products rarely showed immune-related adverse effects in rare cases.

The introduction of completely human insulin products, either semi-synthesized or produced by recombinant DNA technology, brought the expectation that immunogenicity problem would be solved. However, antibodies are still identified in about 40% of the patients, and the introduction of modified insulin preparation such as glargine more than two decades ago was not associated with an increase in antibody formation, although protein modification is considered to increase the risk for immunogenicity.

So over the years, insulins have become less immunogenic, and this reduction seems to be the result of an increase in the purity rather than the switch from animal to human insulins. For most other therapeutic proteins, product quality attributes have been identified as the main driver for immunogenicity.

So antibodies to insulin can be assumed to be induced by the same immunological mechanism, leading to reduced tolerance as reported for other therapeutic proteins.⁸ This immunological response has been shown to be mostly T-cell independent and the result of direct activation of B cells leading to the production of antibodies. This direct activation of B cells can be achieved by presenting proteins complexes in a repeated, array form.

Aggregates that may be formed during manufacturing and storage apparently act as these supra-molecular structures and are capable of directly activating B cells. An additional issue with insulins is their tendency to form crystal structures depending on the excipients that also may act as supra-molecular structures capable of inducing an antibody response by breaking tolerance.⁹

The clinical effects associated with insulin antibodies have been reviewed extensively by Fineberg and colleagues.⁷ Insulin resistance is the most common adverse effect and was first described as early as 1922, soon after the introduction of the first generation of impure insulin products. With the current highly purified products, the incidence of immunogenicity is still high; however, the clinical effects of antibodies to insulin is at present less clear, for many reasons: for example, the assays for antibodies have become more sensitive, leading to more positive patients; and traditionally the assay used for insulin antibodies only detected binding antibodies. For other therapeutic products, loss of efficacy is only seen in patients with persisting high titres of neutralising antibodies. Studies investigating the clinical consequences of antibodies to insulins, including biosimilars, should be based on assays for neutralising antibodies and should stratify the patients on the basis of antibody titre and persistence.

As with the other biosimilars, another contentious issue for insulins will be substitution and interchangeability. The generic status for a small molecule implies interchangeability and interchangeability is a condition for safe substitution. In the case of generics, these different terms are often used as synonyms. However, in the case of biosimilars, substitution and interchangeability not only have different meanings, but different rules also apply. In most European countries substitution of biologicals and biosimilars at the pharmacy level without consent of the treating physician is discouraged or forbidden by law. The reason behind this restriction is traceability in case of side effects of the product and is not based on lack of interchangeability of products.¹⁰

Interchangeability is a property of a product and although there is yet no evidence showing clinical or biological effects of switching to a biosimilar, the consensus in most European countries is to keep a patient on a specific product if he/she responds well and there is no reason to change. If switching is necessary, patients should be monitored because biological products differ and therefore the response of the patients.¹¹