OR WAIT 15 SECS
Proving a biosimilar's pharmacokinetic 'equivalence' requires adherence to several unique factors.
Unlike classic drugs, which have relatively simple molecular sizes and structures (typically up to 300 Da), biologics are large molecules (up to 270,000 Da) with highly complex three-dimensional tertiary structures.1 To help put the magnitude of this difference into context, the chemical structures of aspirin, a small classic drug with a molecular weight of 180 Da, and Humira®, a novel biologic with a molecular weight of approximately 148,000 Da, are presented in Figure 1.
Although classic drugs such as aspirin are usually easily characterized using sensitive analytical methods, biologics are created using genetically modified cell lines, and often undergo significant post-translational modifications (such as glycosylation), which can give rise to considerable heterogeneity. Consequently, producing an exact replica of an existing biologic is an almost impossible task.
The first patented biologics were recombinant versions of endogenous human proteins, such as insulin, in the 1980s, followed by more complex products, such as monoclonal antibodies, in the late 1990s.2 Since biologics were first introduced, sales have grown considerably year-on-year, and by 2011 global sales had reached approximately 142 billion USD (equivalent to 19% of the global biopharmaceutical market), with more than a third of this (37.6%) attributed to the top 10 biologics.3 However, the period of exclusivity for these top 10 biologics is fast approaching, with a "patent cliff" anticipated between now and 2019 (see Figure 2).
This patent cliff creates an enormous opportunity to develop generic versions of biologics. Advances in technology have led to higher production yields, reduced production times, and lower costs, while the global recession has significantly increased the pressure on national governments to drive down the cost of health provision;3 and, perhaps most significantly, the European Medicines Agency (EMA) and the U.S. Food and Drug Administration (FDA) have both introduced regulatory pathways for the review and approval of "generic" biologics,4-9 thus paving the way to market for what have become widely known as "biosimilars."
Guidelines for the approval of biosimilars were first developed by the EMA during 2005 to 2006, with the first biosimilar approved in the EU in 2006 (Omnitrope®, a somatropin biosimilar). Since the approval of Omnitrope®, a further 19 biosimilars have been approved by the EMA for use in the EU, as of January 2014.10; however, two of these approvals have subsequently been withdrawn—one for filgrastim in April 2011 (Filgrastim Ratiopharm®) and one for somatropin in May 2012 (Valtropin®)—leaving a total of 18 biosimilars currently approved for use in the EU.
Although the route to market for classic generics is well defined, and has been successfully applied to many drugs over the years—typically a small number of studies in healthy volunteers are sufficient to prove physiochemical and pharmacokinetic (PK) equivalence—the corresponding route to market for biosimilars is relatively new and considerably more complex. Comparative PK and pharmacodynamic (PD) studies are essential; efficacy trials are usually also required, the safety profile must be well defined, and immunogenicity must always be investigated. Ahead we focus on some of the key issues affecting PK studies used to support biosimilar drug applications:
The most common study designs associated with standard bioequivalence studies are crossover designs, with a relatively small number of subjects receiving both test and reference drugs in separate treatment periods (in a randomized fashion), with a short wash-out period between each period. However, using crossover designs in the development of biosimilars can often prove to be difficult for a number of reasons: for example, biologics tend to have much longer half-lives than classic drugs, therefore, a crossover approach is generally not practical (as the required wash-out period is often prohibitively long), and the potential for biologics to illicit an immune response can also limit the use of crossover studies (if a patient was to develop an immune response in the first period of a crossover study, the patient's ability to participate in the second treatment period would be compromised).
To negate these issues, it is common to use parallel designs when conducting biosimilar studies; only one treatment period is required for each subject, removing the need for a wash-out period, and the potential knock-on effect caused by an immune response is also limited. However, it is important to note that parallel designs are not without issues of their own: large sample sizes are often required to ensure that there is sufficient statistical power to prove biosimilarity, and, as treatment differences are estimated between subjects (rather than within subjects), it is important to stratify treatment groups by certain covariates (i.e., age, weight, and sex) to avoid potential imbalances in baseline characteristics between treatment groups.
Recent examples of successful Phase I biosimilar studies include Remsima®11 and Inflectra®12—both Remicade® (infliximab) biosimilars—which were approved by the EMA in 2013. To investigate the PK of these biosimilars, randomized, double-blind, multi-center, parallel-group studies were carried out using a large sample size of 250 patients (125 per treatment group).11,12 The biosimilar and innovator biologics were subsequently shown to be comparable by demonstrating that the 90% confidence intervals for the key PK endpoints (AUC and Cmax) were within the bioequivalence limits (0.80, 1.25).13
While large parallel-design studies such as these are often required in early phase PK comparability studies, it is perhaps worth noting that this is not always the case; all studies should be based on the PK and PD characteristics of the comparator, or reference, biologic. Indeed, Grastofil® (approved in 2013)14 and Nivestim® (approved in 2010)15 —both Neupogen® (filgrastim) biosimilars—have recently been approved for use by the EMA following completion of a series of Phase I cross-over trials: one for Grastofil® (using 35 patients), and two for Nivestim® (using 44 and 48 patients, respectively). In these instances, crossover studies were possible as the terminal half-life of Neupogen®, the comparator biologic, is in the region of three hours, which is considerably shorter than the half-life associated with most biologics (thus compatible with a crossover approach). However, this is not typically the case. Of the top 10 biologics listed in Figure 2, only two have half-lives which are not measured in days, Novolog® and Lantus®, which are insulin-based biologics and have short half-lives in the region of one to three hours; the other eight have half-lives ranging from approximately 3 days (Enbrel®, Lucentis®, Neulasta®) up to 50 or 62 days (Avastin® and Rituxan®, respectively). Therefore, the use of crossover studies in the development of biosimilars for most of the top 10 biologics is likely to be impractical.
In classic bioequivalence studies, PK equivalence is demonstrated using bioequivalence limits of (0.80, 1.25); the test and reference products considered to be equivalent if the 90% confidence interval of the ratio of geometric least squares means lies entirely within (0.80, 1.25).13 There are currently no such limits defined for biosimilars; indeed, the EMA guidance on similar biological medicinal products4 currently states that "The acceptance range to conclude clinical comparability with respect to any pharmacokinetic parameter should be based on clinical judgement, taking into consideration all available efficacy and safety information on the reference and test products. Hence, the criteria used in standard clinical comparability studies, initially developed for chemically derived, orally administered products may not be appropriate and the clinical comparability limits should be defined and justified prior to conducting the study." Therefore, might it be possible to justify wider acceptance limits?
A review of the clinical development of the 20 biosimilars which have (at some stage) been approved for use in the EU prior to January 201410 would suggest not. Of the 20 approved biosimilars, at least 19 applied bioequivalence limits (0.80, 1.25) as the key criteria for establishing PK comparability at some stage of their clinical development (the approach used for Somatropin Biopartners® is unclear, based on the European Public Assessment Report16). In some instances (e.g., Retacrit® and Silapo®) the limits were applied post-hoc (perhaps following regulatory advice), and in one instance (Valtropin®) a wider limit (0.70, 1.43) was used for Cmax. This is a common approach used in standard bioequivalence studies when comparing highly variable drugs.13 Therefore, in the absence of acceptance criteria specific to biosimilars, adhering to the guideline on investigation of bioequivalence13 is recommended.
Anti-drug antibodies (ADA)
Due to their nature (recombinant proteins, some with non-human origins) and a complex manufacturing process that often results in impurities, biologics have the potential to illicit an immune response. Indeed, nearly all biologics induce anti-drug antibodies (ADA); however, the incidence of ADAs differs widely among products and between individuals.17 ADA may have no clinical significance; however, high levels can interfere with the PK and PD properties of the drug (e.g., increasing clearance, and thus reducing the extent of systemic exposure and desired effect of the drug). It is, therefore, important to have an assay in place to test for the presence of ADA at suitable intervals during the study, so that subjects who illicit an immune response can be identified and the impact assessed.
As defined in a recent guideline issued by the EMA, "differences that could have an advantage as regards safety (for instance, lower levels of impurities or lower immunogenicity) should be explained, but may not preclude biosimilarity."4 While differences in immune response for a true biosimilar seem unlikely, equivalence in immune response cannot be assumed and is an integral part of the PK assessment; indeed, care needs to be taken when dealing with subjects who develop an immune response. One approach, which has been ratified by the EMA and FDA in one of our recent studies, is to perform a secondary analysis which incorporates ADA titer results as a covariate. Although the statistician may find it tempting to include terms for ADA and ADA with treatment interaction in the primary analysis—so that estimates of difference can be constructed separately for each sub-population, and averaged over both—since ADA is not a truly independent covariate, the overall estimate may be misinterpreted. In particular, an overall treatment difference would balance for similar levels of ADA response in each treatment, when it would be more appropriate to reflect the relative problems with immune response, and not correct for an imbalance. Thus, assessing the effect of ADA as a secondary analysis ensures that the ADA issue is addressed, without compromising the validity of the primary analysis.
One additional consideration that should be taken into account when evaluating suitable ADA blood sampling times concerns the presence drug. High levels of drug can interfere with ADA assays, therefore, it is good practice to test for ADA levels at time points where drug levels are expected to be at their lowest (e.g., at the end of the PK sampling period following a single dose).
In the future, the exact nature of statistical assessments to be performed in terms of ADA activity may be dictated by regulators; however, until this topic has been investigated further and authorities clarify how best to deal with the PK analysis of ADA positive subjects as standard, it is imperative that this issue is not overlooked.
Whereas bioequivalence studies are typically limited to the area under the concentration-time curve (AUC) and maximum concentration (Cmax) as measures of the overall extent of systemic exposure and rate of absorption, respectively, elimination characteristics must also be considered when comparing biologics due to their long half-lives and potential to develop immune responses. Indeed, current EMA guidelines state that "the design of comparative PK studies should not necessarily mimic that of the standard "clinical comparability" design (CHMP/EWP/QWP/1401/98), since similarity in terms of absorption/bioavailability is not the only parameter of interest. In fact, differences in elimination characteristics between products, e.g. clearance and elimination half-life, should be explored."4
In this context, it is important to note that, to fully characterize elimination kinetics in biosimilarity studies, the sampling schedule guidelines associated with equivalence (or clinical comparability) studies should still be applied: i.e., "The sampling schedule should also cover the plasma concentration time curve long enough to provide a reliable estimate of the extent of exposure which is achieved if AUC(0-t) covers at least 80% of AUC(0-8). At least three to four samples are needed during the terminal log-linear phase in order to reliably estimate the terminal rate constant, which is needed for a reliable estimate of AUC(0-8)."13 Thus, the last sampling time point may be many weeks or even months after dosing.
Relative purity or protein content
Although there have been significant advances in the manufacture and production of biologics in recent years, the process is still considerably more complex than the chemical synthesis of classic drugs, and often results in impurities. Whereas "the assayed content of the batch used as test product should not differ [by] more than 5% from that of the batch used as reference product"13 in bioequivalence studies (unless otherwise justified), it may not be possible to meet this criteria for biologics; therefore, the total protein content of the test and reference drugs need to be considered when assessing biosimilarity.
The importance of this issue is perhaps best illustrated by referencing a recent study performed to compare the PK of Retacrit®, a biosimilar, to that of Eprex®, an innovator biologic.18 This study highlighted differences in the amount and type of glycoforms between the biosimilar and the reference drug, which is to be anticipated given the different production processes. More notably, however, Eprex comprised more total protein (μg/mL) than Retacrit, which appeared to contribute to the potency (IU/mL) of Eprex being 10% higher than labelled. The EMA accepted that the PK of Eprex and Retacrit were comparable, based on introducing a correction factor to allow for the difference in protein content. Once this was taken into account, the comparison of PK parameters was well within the defined equivalence margins.
Region-specific reference drugs
We have recently encountered an issue in a Phase I study whereby the FDA would not accept the use of an European Economic Area (EEA)-licensed drug as a reference, and similarly, the EMA would not accept the sole use of a U.S.-licensed reference. As the ultimate aim was to license the biosimilar in both the EU and the U.S., both regulatory agencies advised that it would be necessary to include two reference drugs in the study, and conduct a three-way comparison between the test drug, the US-licensed drug, and the EEA-licensed product. Should similarity be proven for all three treatments in the Phase I studies, subsequent Phase II and III trials may then be limited to just one reference.
This is in line with the recent EMA guideline on Similar Biological Medicinal Products, which states that "...with the aim of facilitating the global development of biosimilars and to avoid unnecessary repetition of clinical trials, it may be possible for an applicant to compare the biosimilar in certain clinical studies and in vivo non-clinical studies with a non-EEA authorized comparator (i.e., a non-EEA authorized version of the reference medicinal product), which will need to be authorized by a regulatory authority with similar scientific and regulatory standards as EMA (i.e., ICH countries). In addition, it will be the applicant's responsibility to establish that the comparator authorized outside the EEA is representative of the reference product authorised in the EEA."4
It is difficult to believe that this will always be a requirement for biosimilar studies; perhaps the stance of the EMA and FDA will soften as more biosimilars come to market and the difference between EU and U.S. versions of the same drug are better characterized. However, until then, the inclusion of multiple reference drugs, at least in pivotal Phase I studies, may be essential.
Although this article focuses primarily on the PK characteristics of biosimilars, the importance of investigating to support these early-phase PK studies should not be overlooked. PD markers should also be assessed and compared to the innovator biologic; indeed, comparative PK and PD studies are essential for biosimilar approval. In certain cases, where the PK and PD properties of the reference drug are well defined, and at least one PD marker is accepted as a surrogate marker for efficacy, comparative PK/PD studies alone might be sufficient to demonstrate clinical comparability; however, efficacy trials are usually also required. In addition, the nature, seriousness, and incidence of all adverse events must be well defined, to ensure that there are no safety concerns associated with the biosimilar, and immunogenicity must always be investigated, as anti-drug antibodies can alter both the PK and PD effect.
There are a number of characteristics of biologics that set them apart from traditional generic drugs, as summarized in Table 1. These characteristics make the route to market for biosimilars more complex than it may at first seem, with differences in study design between bioequivalence and biosimilarity studies.
Due to their long half-lives and the potential to illicit an immune response, biologics may not be compatible with simple crossover designs; therefore, parallel study designs are often utilized. Acceptance limits for the 90% CI of the ratio of PK parameters must be predefined and approved by regulators; although traditional equivalence limits of (0.80, 1.25) tend to remain favorable. Anti-drug antibodies have the potential to affect pharmacokinetic characteristics and should, therefore, be measured and their impact considered. Unlike standard equivalence studies, the assessment of comparability must extend beyond absorption (Cmax) and overall exposure (AUC), with half-lives (t1/2) and clearance (CL) also investigated. Furthermore, the relative protein content of the test and reference drugs needs to be considered, with a suitable correction factor applied, if necessary. In addition, region-specific versions of reference drugs may exist in some instances (e.g., Humira®); therefore, if the intention is to ultimately license the new biosimilar across multiple regions, it may be necessary to incorporate more than one reference drug in at least one bridging study.
When faced with the prospect of proving PK "equivalence" between a biosimilar and an innovator biologic, it may be natural to assume that standard bioequivalence study designs and analysis methods would be sufficient to meet the objectives of the study; after all, such study designs have been successfully employed for many years to bring generic drugs to market. However, due to various complexities of biologics, special consideration needs to be given to the study design (parallel versus cross-over), potential immunogenicity (anti-drug antibodies), elimination characteristics (half-life and clearance), and relative protein contents of the test and reference drugs before commencing the study. Furthermore, guidance should be sought from regulators to determine the most suitable acceptance limits to conclude clinical comparability, and to assess the potential benefits of using multiple region-specific reference drugs in pivotal early-phase clinical studies. Only then, with these factors fully and adequately addressed, can an investigator be confident that a biosimilar trial will be a success.
Yvonne Moores is Executive Vice President, Operations, Quanticate, email: [email protected].
1. Shellenkens, H. "When Biotech Proteins Go Off-Patent." Trends Biotechnol. 2004, 22, 406–410.
2. Calo-Fernandez, B. Martinez-Hurtado J. L. "Biosimilars: Company Strategies to Capture Value from the Biologics Market." Pharmaceuticals. 2012, 5, 1393-1408.
3. EvaluatePharma. World Preview 2018 Embracing the Patent Cliff. Technical report, Evaluate Pharma, London, UK, 2012.
4. EMA. Guideline on Similar Biological Medicinal Products. 2013 (draft).
5. EMA. Guideline on Similar Biological Medicinal Products Containing Biotechnology-derived Proteins as Active Substance: Non-clinical and Clinical Issues. 2013.
6. EMA. Guideline on Similar Biological Medicinal Products Containing Biotechnology-derived Proteins as Active Substance: Quality Issues (Revision 1). 2012.
7. FDA. Quality Considerations in Demonstrating Biosimilarity to a Reference Protein Product. 2012.
8. FDA. Scientific Considerations in Demonstrating Biosimilarity to a Reference Product. 2012.
9. FDA. Biosimilars: Questions and Answers Regarding Implementation of the Biologics Price Competition and Innovation Act of 2009. 2012.
10. GaBI Online - Generics and Biosimilars Initiative. Biosimilars use in Europe [www.gabionline.net/]. Mol, Belgium: Pro Pharma Communications International. 2014.
11. EMA. Remsima® European Public Assessment Report (EMA/CHMP/589317/2013). 2013.
12. EMA. Inflectra® European Public Assessment Report (EMA/CHMP/589422/2013). 2013.
13. EMA. Guideline on the Investigation of Bioequivalence. 2010.
14. EMA. Grastofil® European Public Assessment Report (EMEA/CHMP/303037/2013). 2013.
15. EMA. Nivestim® European Public Assessment Report (EMA/262762/2010). 2010.
16. EMA. Somatropin Biopartners® European Public Assessment Report (EMA/CHMP/229458/2013). 2013.
17. Schellekens, H. (2002) "Bioequivalence and the Immunogenicity of Biopharmaceuticals." Nat. Rev. Drug Discov. 1, 457–462.
18. Schellekens, H. (2009) "Assessing the Bioequivalence of Biosimilars." Drug Discov. Today. 14, 9–10.
Related Content:Trial Design