Clinical trial design is a critical factor in the approval of biosimilars and pharmacists will need to be able to correctly interpret clinical data associated with these drugs in order to ensure that safe and effective therapy is being offered to patients
George Dranitsaris BPharm PhD
Department of Haematology,
University of Ioannina, Greece
The emergence of recombinant DNA (rDNA) technology in the 1970s and 1980s created a novel pathway for new drug discovery. rDNA allowed drug developers to direct genes, which were incorporated into living cells, to produce drug therapies that were larger and structurally more complex that traditional small molecules.(1) Even with today’s technology, such biological drugs cannot be developed through chemical synthesis. Approximately 90% of all biological products are produced from three sources; Escherichia coli (E. coli), yeast or Chinese Hamster ovary cells.(2,3)
The important differences between biological drugs and chemically synthesised compounds are molecular size and manufacturing processes. Biologicals are much larger than the latter. As an illustration, the anticancer drug paclitaxel has a size of 854 Daltons (Da), while the commercially available granulocyte colony-stimulating factor (G-CSF) is 18,000 Da.(4) Monoclonal antibodies (mAbs) are even larger,
with sizes in the range of 145,000 to 160,000 Da.(4)
The process of manufacturing a biological agent is also more complex than for a small molecule. The production process of the biological begins with cloning of the relevant gene into a cDNA vector and transferring this into a host cell (such as E. coli or yeast). Once the protein is expressed, the appropriate cell line is selected and expanded in a fermentation medium where it produces the protein defined by the vector.(3,4) A complex process of purification and validation is then required prior to obtaining the purified bulk drug. Quality control in the form of confirmation of the DNA sequence of the cloned gene by either Southern blot analysis of total cellular DNA or sequence analysis of the mRNA is usually conducted both before and after full-scale fermentation.
This drug development process has resulted in the approval and commercialisation of several biological drugs, such as insulin, human growth hormone, erythropoietin, G-CSF, interferon alfa and monoclonal antibodies, such as rituximab and trastuzumab among others.(4,5) The market size of biological drugs is also substantial. In 2009, global sales of such agents were approximately $US130 billion.(6)
Development and approval
Most new drugs that receive regulatory approval are under patent protection, which typically lasts 20 years from the time the patent is awarded. Therefore, the patents of the first-generation biological agents, such as interferon alfa, G-CSF and erythropoietin, have expired. Furthermore, the patents on the larger and more complex mAbs, such as rituximab and trastuzumab, will also be expiring over the next few years. Commercially, this means that, by 2017, approximately US$60 billion of biological therapies will be going off-patent and will be open to competition.(7)
This has stimulated several generic and brand pharmaceutical companies to invest in the development of biosimilars. Biosimilars are intended to offer comparable safety and efficacy to the reference, off-patent biological. They are not generic alternatives per se, and are generally not interchangeable. Given their structural complexity, multi-faceted manufacturing process and risk of immunogenicity, unique regulatory pathways are required for biosimilars.
Biosimilars have been available for clinical use in Europe for a number of years, with the first agent (Omnitrope®, a biosimilar to Genotropin®) being approved in 2006 by the European Medicines Agency (EMA).(8) The EMA has also taken the lead in the drug review process, with well-defined guidelines for the regulatory approval of biosimilars.(3,5)
The EMA has created guidelines specific to particular classes of biosimilar agent related to manufacturing quality, non-clinical pharmacology and toxicology, pharmacokinetics and clinical considerations.(5) Such guidance documents are particularly relevant because the larger and more structurally complex mAbs are soon to lose patent protection and will be open to biosimilar competition.(8)
From a pharmacist’s perspective, it is necessary to understand and critically review the clinical data that will be associated with the newly approved biosimilars. This article reviews the different types of clinical trial design that are relevant to biosimilar drug approval.
For a biosimilar to receive regulatory approval in the major global markets, clinical data generated from a randomised controlled trial is required. The most common design for drugs is the superiority trial. Superiority trials are designed to demonstrate that one treatment is more effective than another. However, such a design is not relevant for biosimilar drug development because the intent is to demonstrate comparability, and not superiority, in clinical endpoints. Equivalence and non-inferiority trials are more relevant.
The objective of the clinical development programme for a biosimilar is to demonstrate no clinically meaningful difference in efficacy relative to the innovator product. To do this, equivalence trials of adequate sample size must be conducted and should ideally be double blinded. One of the first things required in designing an equivalence trial is to establish a ‘minimally clinically important difference (MCID)’ or δ in the primary trial endpoint. The MCID is defined as the minimum difference in a meaningful clinical endpoint between two treatments, beyond which regulatory bodies would consider the two drugs to be non-equivalent. From a trial design point of view, the smaller the MCID selected during the design of an equivalence trial, the larger the final sample size.
The design of equivalence trials is opposite to that of superiority trials. The null hypothesis in equivalence trials is that the two treatments are not equivalent (either better or worse) and the alternative hypothesis is that they are equivalent. Therefore, the intent of an equivalence trial is to reject the null hypothesis of non-equivalence and accept the alternative hypothesis that the treatments are indeed equivalent.(9,10) In an equivalence trial, it is only possible to assess for equivalence. Superiority or inferiority cannot be tested.
A critical component in the design of an equivalence trial is to establish the δ. Several approaches have been used and these have included the results from previous randomised trials, from expert opinion elicited via a Delphi approach or from patients with the condition.(11) However, a more systematic approach preferred by regulators such as the EMA and the US FDA is the 95-95 method.(11,12)
This approach starts with a meta analysis of randomised trials that compared the control group in the planned equivalence study to placebo. The lower 95% CI of the difference between the control group and placebo is known as M1. The next step involves taking a fraction of M1, usually between 50% and 75% (negotiated with the regulator) and this figure (known as M2) becomes the δ in the equivalence trial. With the established δ, power and alpha, the sample size can be determined.
Upon completion of the trial, the key determinate in concluding equivalence is rejection of the null hypothesis of non-equivalence. This can be done by assessing the p value and the lower end of 95% CI of the odds ratio (OR) in the primary endpoint between the two groups. If the 95% CI is greater than – δ, equivalence between the two groups can be concluded.
Equivalence trials and the pre-set value of δ are extremely important with respect to biosimilar clinical development. For the first generation biosimilar to interferon alpha, the δ for clinical response was 15% relative to the reference product.(13) For the biosimilar to filgrastim, the δ was one day for severe neutropenia following chemotherapy.(13) In both cases, equivalence was demonstrated and the products received approval from European regulators.
Superiority trials attempt to determine if the experimental group is better than the control. The intent of an equivalence design is to see if the new treatment is as good as the comparator or if it falls within a predefined clinical margin of effectiveness. In contrast, non-inferiority trials are a hybrid of the two former designs because they can simultaneously test two hypotheses pertaining to the new drug and the control group.(11) The first hypothesis is tested; namely, if the new treatment is at least non-inferior to the control. If it is, the superiority hypothesis can then be evaluated. If the new therapy is neither superior nor equivalent, then it is highly likely to be inferior relative to the a priori trial criteria. Therefore, non-inferiority trials are efficient because a definitive conclusion can be made about a new drug from a single randomised trial.
Non-inferiority trials require that a non-inferiority margin be established, which is analogous to the δ required in equivalence trials.(11,12) The methods to establish δ are identical to those used in an equivalence design. However, the interpretation of the outcomes of a non-inferiority trial can be more complicated than those from an equivalence or superiority trial. The possible outcomes from a non-inferiority trial are guided by the final 95% CI of the OR for treatment response between groups. Note, outcomes can also be evaluated as differences in effect between groups, but OR were chosen in this example because they are easier to interpret.
There are several potential outcomes from a non-inferiority trial. If the lower 95% CI of the OR for treatment success lies above + δ, then the new treatment is statistically superior to the control and the benefit is clinically meaningful. If the lower 95% CI of the OR is above 1.0 or below + δ, the new treatment is superior to the control group, but the benefit may not be clinically significant. The new treatment would be considered non-inferior to the control if the lower 95% CI lies above – δ. If the upper 95% CI falls below – δ, the conclusion would be that the new product is inferior to the control group. The most important element in the correct interpretation of a non-inferiority trial is the placement of the 95% CI for the OR of efficacy relative to ± δ.
The US FDA released its guidance document on biosimilars in February 2012.(14) Among the key issues covered were clinical trial design and reference product selection that comprises the control arm of such studies. The guidance document also stated that, when multiple biosimilars are available in international markets, the manufacturer must demonstrate that their biosimilar is clinically comparable to a referenced product that has been previously approved by the FDA. Through their clinical trials programme, the guidance document stated that the manufacturer must demonstrate that there are “no clinically meaningful differences between the biological product and the reference product in terms of safety, purity and potency”.(14) The guidance indicates that the scope and number of clinical trials will be assessed on a case-by-case basis and will depend on the amount of residual uncertainty about the biosimilar and the reference product after the preclinical data have been evaluated.(14)
The FDA guidance specifically addresses clinical study design issues. The FDA recommends that the manufacturer conduct a non-inferiority study with a two-sided test with the null hypothesis being that the biosimilar product is inferior to the reference drug based on a pre-specified equivalence margin. The margins should be supported by available scientific data to allow clinically meaningful differences in safety and efficacy to be detected. Under certain circumstances, the FDA would allow a one-sided, non-inferiority design, which would generally require a smaller sample size than a two-sided trial. Such a design should test the hypothesis that the biosimilar poses no more risk in terms of safety and immunogenicity, while demonstrating no clinically meaningful differences in efficacy between the investigational biological and the reference product.(14)
The FDA document also provides guidance on endpoints for such trials. The endpoints must be clinically relevant and sensitive in their ability to detect clinically meaningful differences in safety and efficacy between products. A manufacturer may use endpoints that were different from those used in the clinical development of the reference product. However, the selection must be scientifically justified. If surrogate endpoints are used, they must be correlated with clinical outcomes.
The EMA has also made biosimilar guidance documents available to manufacturers.(3) Most recently, guidelines for the development of the more structurally complex and larger mAbs have been developed.(15) The EMA has a preference for adequately powered randomised, parallel group equivalence trials involving endpoints that measure drug activity. Using oncology as an example, the EMA indicates that a suitable endpoint for comparable activity between a biosimilar and the reference product would be overall tumour response. According to the EMA, such an endpoint would be affected to a lesser extent from patient- and disease-related factors than endpoints such as progression-free and overall survival.(15) Similar to the stance taken by the FDA, the EMA stops short of providing specific equivalence margins for the various types of clinical endpoints.
Biosimilars will continue to enter clinical practice, and over the next five years will include some of the more structurally complex mAbs. Biosimilars offer a potential to save healthcare costs, but they present some uncertainty related to safety and efficacy, particularly when the reference biologic has multiple indications. Coupled with the abbreviated approval pathway, which requires fewer patients than with the original reference drug, clinical trial design will be a critical factor in the approval of these products and in building confidence with healthcare professionals and patients.
Pharmacists need to be aware that these products are not generic versions of the original biological and there will always be uncertainly, even after equivalence or non-inferiority has been demonstrated. This will be especially relevant with the new mAb biosimilars because the endpoints for the trials of these products may not be the same as with the original branded biologicals. Therefore, pharmacists will need to be able to correctly interpret the clinical data associated with a biosimilar in order to ensure that safe and effect therapy is being offered to patients.
- Biological drugs are distinct from traditional drug agents because of their unique manufacturing process, large molecular weight and potential for immunogenicty. As a result, generic copies of biological drugs cannot be made.
- Instead of generic copies, biosimilars of biological drugs can be manufactured.
- Biosimilars are a new class of drugs intended to offer comparable safety and efficacy to the reference, off-patent biological.
- For biosimilars to be approved for clinical use, the manufacturer must conduct either equivalence or non-inferiority trials against the off-patent biological.
- A biosimilar will receive regulatory approval if the difference in efficacy relative to the branded biological is no greater than a predefined non-inferiority margin ‘δ’ on a clinically acceptable trial endpoint.
- However, there will always be uncertainly with biosimilars, even after equivalence or non-inferiority has been demonstrated.
- Johnson IS. The trials and tribulations of producing the first genetically engineered drug. Nat Rev Drug Discov 2003;2:747-51.
- Woodcock J et al. The FDA’s assessment of follow on protein products: a historical perspective. Nat Rev Drug Discov 2007;6:437-42.
- Mellstedt H, Niederwieser D, Ludwig H. The challenge of biosimilars. Ann Oncol 2008;19:411-9.
- Beck A. European medicines workshop on biosimilars monoclonal antibodies: Perspective from the EU. MAbs 2009;1:406-10.
- Dranitsaris G, Amir E, Dorward K. Biosimilars of protein based drug therapies: regulatory, clinical and commercial considerations. Drugs 2011;71:1527-36.
- Kueppers E. Follow on biologics: how to develop a competitive advantage. Business Devel Licensing J 2010;12:17-18.
- Zacks Investment Research. Biosimilars: The next big thing. www.dailymarkets.com/stock/2012/01/13/biosimilars-the-next-big-thing/ (accessed 30 May 2013).
- Ahmed I, Kaspar B, Sharma U. Biosimilars: impact of biologic product life cycle and European experience on the regulatory trajectory in the United States. Clin Ther 2012;34:400-19.
- Blackwelder WC. Proving the null hypothesis in clinical trials. Control Clin Trials 1982;3:345-53.
- Ware JH, Antman EM. Equivalence trials. N Engl J Med 1997;16;337:1159-61.
- Schumi J, Wittes JT. Through the looking glass: understanding non-inferiority. Trials 2011;12:106.
- Rothmann M et al. Design and analysis of non-inferiority mortality trials in oncology. Stat Med 2003;22:239-64.
- Fox A. Biosimilar medicines-new challenges for a new class of medicine. J Biopharm Stat 2010;20:3-9.
- US Food and Drug Administration. www.fda.gov/newsevents/newsroom/pressannouncements/ucm291232.htm (accessed 30 May 2013).
- European Medicines Agency. Guideline on similar biological medicinal products containing monoclonal antibodies. www.ema.europa.eu/docs/en_GB/document_library/Scientific_guideline/2010/11/WC500099361.pdf (accessed 30 May 2013).