Analytical Data and Single-Dose PK are Sufficient to Conclude Comparable Immunogenicity for Biosimilars: An Ustekinumab Case Study

Abstract

Comparative immunogenicity is a key regulatory requirement for biosimilars. In a streamlined biosimilar development process, absent a comparative clinical efficacy study, the analytical data and clinical pharmacokinetics (PK) study need to provide sufficient evidence for a conclusion of comparable immunogenicity. In this case study, we have reviewed the role of analytical and clinical data in the immunogenicity assessment of all currently available ustekinumab biosimilars and their reference product. Public information for European Medicines Agency–and US Food and Drug Administration–approved biosimilars reveal that the single-dose clinical PK studies were sensitive in detecting differences in terms of immunogenicity between the biosimilar and the reference product, a finding that was replicated in the comparative clinical efficacy studies. The rates for anti-drug antibodies and neutralizing antibodies were comparable albeit numerically lower for all biosimilars compared to their reference product, which correlates with lower levels of non-human glycans such as α-1,3 galactose known to be potentially relevant for immunogenicity. Our study demonstrates that the single-dose clinical PK studies were sensitive in confirming comparable immunogenicity of ustekinumab biosimilars with their reference product. The comparative clinical efficacy studies revealed no additional information. This finding adds on to the evidence that clinical PK and the comparative analytical assessment, specifically the comparison of quality attributes with potential immunogenic relevance, suffice for the evaluation of immunogenicity of biosimilars in general.

Key Points

Single-dose clinical pharmacokinetics (PK) studies were sensitive in detecting anti-drug antibody (ADA) and neutralizing antibody (Nab) rates between ustekinumab biosimilars and the reference product and showed comparable, albeit numerically lower, ADA and Nab rates for the biosimilars. The comparative efficacy studies confirmed the findings of the single-dose PK studies.

While a direct comparison of absolute ADA and Nab rates across biosimilars cannot be made as different assays were utilized, the lower immunogenicity rates correlate with lower levels of non-human glycans such as α-1,3 galactose in each of the biosimilars, which has been shown to have the potential to increase immunogenicity. This finding corroborates the predictive nature of the analytical assessment for comparable immunogenicity, a principle that has been successfully applied in the regulation of process manufacturing changes of biologics for over 3 decades.

This study suggests that, in most cases, immunogenicity data from a single-dose clinical PK study, together with the comparative analytical assessment with a focus on quality attributes with potential immunogenic relevance, are sufficient for evaluating the comparable immunogenicity of a proposed biosimilar. This further underpins the validity of a more efficient, streamlined biosimilar development process without the need for a comparative clinical efficacy study.

Introduction

The evolution of the biosimilar regulatory approval pathway continues to be debated, fostered by better technical capabilities and accumulating experience with biosimilars already approved and in use in multiple jurisdictions. The long-term safety and efficacy of biosimilars shown through their widespread use in multiple jurisdictions has been as expected from experience with their reference products [1, 2].

The UK’s Medicines and Healthcare products Regulatory Agency (MHRA) and the World Health Organization (WHO) have recently revised their biosimilar guidelines to allow the development of biosimilars without ‘confirmatory’ comparative efficacy studies (CES) when based on sufficient evidence for biosimilarity [3, 4]. This requires the biosimilar candidate to be thoroughly characterized and shown to be structurally and functionally similar to the reference product. The MHRA guideline update was also supported by the experience with biosimilar assessments since the creation of the biosimilar pathway in the EU in 2004 [5]. Meanwhile, multiple stakeholders have opined that successful analytical characterization may preclude the need for a CES, in part due to the high specificity and sensitivity of these assays for detecting clinically meaningful differences [5–12]. The European Medicines Agency (EMA) is in an active collaborative process to re-evaluate the need for comparative efficacy studies [13], as are other regulators from around the world [14].

Several retrospective studies have been conducted to estimate the impact of comparative efficacy studies on the approvability of biosimilars [5–8]. These studies included all products approved 2006–2019 [8] or focused on monoclonal antibodies and fusion proteins [5–7]. In no case did clinical studies detect a potential difference that was not already detected by the comparative analytical assessment (CAA), supporting the CAA’s fundamental role as the most sensitive indicator for potential differences between biosimilars and their reference products [15].

In general, products that were comparable in analytical, including physicochemical and functional characterization, and PK studies were approved irrespective of the outcome of the CES, and the CES did not significantly contribute to the regulatory decision-making process, introducing in certain cases more concerns than they solved, and which were unrelated to biosimilarity. Notably, there were cases of successful biosimilar programs despite formally failing the CES [7, 8]. Furthermore, there were cases where successfully conducted CES could not compensate for issues in the CAA [8].

The translation of retrospective knowledge and experience into prospective predictions has its challenges. However, the significant and extensive accumulated evidence from approved biosimilars has already enabled flexibility within the regulatory framework, while confirming the fundamental basis for a scientifically valid and ethical approach that minimizes unnecessary clinical studies. This foundation gives confidence in the streamlined approach, which is essential for maintaining the feasibility of biosimilar development while ensuring broader patient access to quality-assured, safe, and effective biosimilars [16, 17].

Nonetheless, an important question to answer is how comparative immunogenicity can be assured within a streamlined approach while maintaining the highest scientific standards of evidence and full confidence in the products finally approved. The immunogenicity of biotherapeutics, when present, is an issue that can have consequences for patients ranging from harmless to severe. The immunogenicity of biosimilars remains a widely discussed topic, and experience with products approved in the EU and US suggests that analytically similar biosimilars show comparable immunogenicity [1, 2, 18, 19]. Both the EMA and the US Food and Drug Administration (FDA) require manufacturers to include a comprehensive evaluation of immunogenicity in the marketing application based on validated assays to enable a robust assessment of immunogenicity [20, 21].

The regulatory requirement for comparable immunogenicity between a biosimilar and its reference means that the rates of immunogenicity triggered by the biosimilar and the reference medicine need to be in the same range. However, lesser immunogenicity, defined as numerically fewer patients developing anti-drug antibodies (ADAs) at any time with biosimilar treatment, is typically accepted by regulators [22] as it is viewed as a reflection of fewer impurities rather than due to attributes of the drug substance itself.

Our case study of all currently approved ustekinumab biosimilars provides a practical way to consider two major issues often raised as a limitation for streamlined development of biosimilars. The first one is the sensitivity of a clinical PK study in detecting ADAs and neutralizing antibodies (Nabs) when compared with the results of a CES. The second topic addresses the link between the CAA and immunogenicity, considering that all clinical properties of a medicinal product are consequences of its quality, which includes its physicochemical and functional attributes.

Methods

The study compared the immunogenicity data, i.e., the rates of ADAs and Nabs, between the biosimilars and their reference product measured in the comparative PK studies and CES. We also evaluated the comparability of the immunogenicity data between the PK studies and CES for each of the biosimilars separately. As reported in the European Public Assessment Reports (EPARs), the measured ADAs and Nabs over the time of the study, the most suitable rates for enabling the comparison between PK and CES and between the biosimilars and their reference products, were used for this study, as detailed in the “Results” section.

The comparative analytical data packages were screened for differences in those quality attributes known to be potentially relevant for immunogenicity, such as aggregates, host cell proteins, and non-human glycans as described in the literature [23, 24]. They were further evaluated for correlation with the immunogenicity rates shown in the clinical studies. Aggregates were reported to be comparable in all cases. The level of host cell proteins was not publicly available in the data source, whereas the comparative amounts of non-human glycan were assessed in the study (see Table 1).

Table 1.

Content of non-human glycans, terminal α-1,3 gal, and NGNA in comparison to the RP

Biosimilar	Cell line	α-1,3 gal content in biosimilar	α-1,3 gal content in RP	NGNA content in biosimilar	NGNA content in RP
SB17	CHO	Not detected Lower than RP	1.8–2.6% Higher than biosimilar	Not detected	Detected
AVT04	SP2/0	Much lower than RP	Much higher than biosimilar	Similar	Similar
CT-P43	CHO	NA	NA	NA	NA
ABP 654	CHO	Not detected	1.7–4.9%	Not detected	Detected
DMB-3115	SP2/0	Lower than RP	Higher than biosimilar	Lower than RP	Higher than biosimilar
FYB202	CHO	Lower than RP	Higher than biosimilar	Lower than RP	Higher than biosimilar
Bmab 1200	SP2/0	Lower than RP	Higher than biosimilar	Lower than RP	Higher than biosimilar

All data are from the EPARs, with the exception of SB17, which stems from the US FDA action package data [25–33]

α-1,3 gal α-1,3 galactose, CHO Chinese Hamster Ovary, EPAR European Public Assessment Report, FDA Food and Drug Administration, NGNA N-glycolylneuraminic acid, RP reference product, NA not available

The study includes the seven biosimilar ustekinumabs that have been approved in the EU and/or US through March 2025. Their development code names are SB17, AVT04, CT-P43, ABP 654, DMB-3115, FYB202, and Bmab 1200. The study reviewed their development data as published in the EPARs and the FDA action packages when available [25–33]. In one case (Bmab 1200), an additional reference enabled a more detailed comparison of ADA and NAb data [34].

Results

Anti-Drug Antibody Rates

ADAs were measured throughout the clinical PK trials and CES for each of the biosimilars and their corresponding reference product. Different assay systems were applied by the developers, which led to different sensitivities in measuring ADAs and Nabs. All comparative clinical PK studies of the seven biosimilars were single-dose studies in healthy volunteers, using the 45-mg prefilled syringe presentation. The studies were conducted in a three-arm design comparing the biosimilar candidate with EU- and US-sourced reference product. The study size ranged from 201 to 491 healthy volunteers. Two biosimilar developers conducted two PK studies, as the first one failed to demonstrate equivalent PK due to detrimental combinations of operational issues like difference in administered dose, higher variability in clearance due to study population selection, and difference in immunogenicity [28, 32].

The CES were all conducted in psoriasis patients, with sample sizes ranging from 380 to 600 patients for each study.

For the most representative comparison of ADAs between the biosimilars and the reference product, Figure 1 shows the total ADA rates whenever available. The total ADA rate is the percentage of study participants who tested positive for ADAs at least at one time point of the study. For SB17 and DMB-3115, the data at the time point with the maximum antibody rates were used. The ADA and Nab assay used for Bmab 1200 showed a very high sensitivity, testing close to all study participants positive for ADAs and Nabs at least at one time point in the study [26]. To enable a meaningful comparison of the immunogenicity, the Bmab 1200 data shows the proportion of the healthy volunteers who tested positive for ADAs at all time points of the study. For the CES, the Bmab 1200 data reflect the results at the last visit at week 52 [34].

Fig. 1.

Anti-drug antibody rates of the pharmacokinetics (PK) studies and comparative efficacy studies (CES) of the different ustekinumab biosimilars and the reference product

Neutralizing Antibody Rates

The neutralizing antibody rate represents a subset of the ADA rate, i.e., those antibodies that also pose a neutralizing effect on the ustekinumab.

Figure 2 shows the total Nab rates whenever available. The total Nab rate is the percentage of study participants who tested positive for Nabs at least at one time point of the study. As for the clinical PK studies, SB17 and DMB-3115 data show the rate at the time point with the maximum antibody rates. For Bmab 1200, due to the high assay sensitivity, the data shows the study participants who tested positive for Nabs at all time points in the PK study and at the last visit at week 52 in the CES, to enable a meaningful comparison.

Fig. 2.

Neutralizing antibody rates of the pharmacokinetics (PK) studies and comparative efficacy studies (CES) of the different ustekinumab biosimilars and the reference product

Quality Data

The quality data published in the EPARs and the FDA review documents revealed minor differences in some quality attributes, each of which were justified as clinically irrelevant. We specifically looked at components of the drug substance, and also along with the excipients, which together comprise the drug product. Regarding the impurities that could potentially increase immunogenicity, the aggregate and microaggregate levels were described as being similar; no data were published on host cell proteins. There was a difference in host cells used for production—four biosimilars were manufactured using Chinese Hamster Ovary (CHO) cell lines and the reference product and three biosimilars were manufactured using mouse hybridoma (SP2/0) cell line. When considering the drug substance, the levels of non-human glycans and N-glycolylneuraminic acid (NGNA) sialylation were consistently lower in the biosimilars compared with the reference product (see Table 1).

Discussion

As noted below, the comparison of absolute immunogenicity rates between different studies has its limitations; however, relative comparisons within any one given study are meaningful. All biosimilars showed numerically lower ADA and Nab rates than the reference product. These differences in ADA and Nab rates can be consistently observed in both PK studies and CES (see Figs. 1 and 2). This result highlights the main conclusion of our analysis, i.e., that the single-dose comparative PK studies are sensitive at detecting immunogenicity, a finding that in all instances is confirmed by the multidose CES in patients. Kurki et al. also reported consistent immunogenicity results between PK studies and CES for biosimilars in Europe [19]. This cumulative evidence supports the conclusion that in most cases comparative PK studies plus analytical data are sufficient for evaluating comparative clinical immunogenicity between proposed biosimilars and the corresponding reference product. Although single-dose PK studies in healthy volunteers may suffice, biologicals that cannot be compared in healthy volunteer studies for safety reasons may require comparative PK studies in patients, using the appropriate dosing schemes. We acknowledge situations wherein PK studies are not relevant, e.g., products with intraocular or intrathecal administration.

The immunogenicity data depends upon the performance characteristics of the respective assays, such as sensitivity, drug tolerance, and cut-off level to differentiate true signals from noise. Therefore, the actual numerical values typically vary between studies by different developers, which limits the comparison of absolute antibody rates between different products. Notably, the pivotal clinical studies of the reference product [35] showed lower post-baseline rates of ADAs (i.e., 9.5% and 6.8%, respectively) and Nabs (i.e., 4.7%) than were shown in any of the biosimilars studies (see Figs. 1, 2). It is apparent that the ADA and Nab assays conducted by biosimilar companies have higher sensitivity, leading to a higher detection rate of ADAs and Nabs. In the most pronounced case of high assay sensitivity in the Bmab 1200 studies, nearly all Bmab 1200 study participants showed positive ADA at least at one visit. The difference in ADA and Nab rates for the reference product between the biosimilar and the pivotal clinical studies is not new, as it has also been observed for other biological therapies over time [19]. The higher assay sensitivity in the later studies likely also reflects the evolving state-of-the-art in ADA and Nab assay technology.

The intrinsic immunogenicity of a biotherapeutic is a consequence of its quality attributes. It is governed first and foremost by the amino acid sequence, which determines the T cell epitopes that can be presented to and recognized by the immune system. In addition, certain process- and product-related impurities may further increase immunogenicity and must be controlled to sufficiently low levels.

Quality attributes that may be potential impurities that increase immunogenicity include aggregates [36] or any glycans not naturally occurring in the human body [37]. Certain non-human glycans include α-1,3 galactose (α-1,3 gal) and NGNA, both of which can be synthesized in small amounts in the mammalian cell lines used for the production of biotherapeutics. Evidence shows that α-1,3 gal has immunogenic potential; for example, it was linked with allergic reactions after ingestion of red meat and with anaphylaxis of cetuximab in sensitized patients [37, 38]. While cetuximab contains exposed α-1,3 gal epitopes in the fragment antigen-binding (Fab) part of the antibody, it has also been shown that more sterically shrouded α-1,3 gal epitopes in the fragment crystallizable (Fc) part, such as in ustekinumab, can be recognized by anti–α-1,3 gal antibodies and may therefore exhibit immunogenic potential [39].

The EPARs and FDA review information of the seven approved biosimilar ustekinumab products discussed herein reported similar levels of aggregates between the biosimilars and the reference product [25–33]. No quantitative information was published on process-related impurities such as host cell proteins. However, differences in non-human glycans were reported (see Table 1). Notably, the biosimilars show consistently lower levels of α-1,3 gal, and all but one biosimilar also showed lower levels of NGNA than the reference product, regardless of whether the biosimilar was made with CHO or SP2/0 cell lines. While these differences may partly explain the numerically lower immunogenicity rates of the biosimilars, they were deemed acceptable and with no clinically meaningful differences. As discussed, assay sensitivity and study design differences also contribute to the immunogenicity results in the clinical studies and should be considered in the interpretation of data. The ustekinumab case study is a particularly useful example, as we have a significant number of biosimilar candidates and their reviews are available to examine the sensitivity of a single-dose clinical PK study, when compared to a CES, for immunogenicity detection.

The EPARs and FDA review information include antibody rates over time data showing that both ADAs and Nabs develop quickly, often within days or weeks after the onset of the PK study, enabling a robust comparison of immunogenicity well within the study’s duration. Regarding the reported consequences, ADAs and Nabs slightly affect ustekinumab serum concentrations, but without ultimately impacting the PK assessment of the biosimilars. There were no adverse effects in safety or efficacy for patients treated with either product in the CES who developed ADAs and Nabs [25–33].

Overall, our study supports the view that in most cases a CES does not add new information with regards to the totality of evidence for establishment of biosimilarity over what is provided by the results of CAA and comparative PK studies. In addition, the ustekinumab case study supports the conclusion of the predictive nature of the analytical assessment for comparable immunogenicity, a principle that has been applied in the regulation of process manufacturing changes of biologics for over 3 decades [40].

Limitations of the study

This study is limited to the qualitative information that has been published by the FDA and EMA after their formal review and approval of each of the biosimilars, plus additional data in one case from one of the developers. A complete quantitative correlation between the ADA and Nab rates and the quality attributes would require head-to-head analysis of all products on all immunogenicity-relevant impurities.

This study is a retrospective analysis of one monoclonal antibody product class. Furthermore, the absolute ADA and Nab rates can only be compared in one and the same study, as the assay sensitivity varies between different assay set-ups, as described above. Immunogenicity results from different biosimilar studies may not be directly comparable in an absolute manner because the ADA tests used can differ in design and sensitivity. However, this limitation does not compromise the validity to compare relative ADA and Nab rates between studies.

Conclusion

The single-dose PK studies for all seven ustekinumab biosimilars show comparable, albeit numerically lower, rates of ADAs and Nabs to the reference product. These results were consistent with the findings of the CES in all cases, demonstrating that single-dose PK studies were sufficiently sensitive for evaluating comparative immunogenicity. ADAs and Nabs develop quickly, allowing for a robust comparison of immunogenicity within a few weeks after dosing.

While direct comparison of absolute ADA and Nab rates across biosimilars cannot be made, as different assays were utilized, the relative lower immunogenicity rates of the biosimilars correlate with lower levels of immunogenicity-relevant non-human glycans such as α-1,3 gal. This finding corroborates the predictive nature of the analytical assessment for comparable immunogenicity, a principle that has been successfully applied in the regulation of process manufacturing changes for over 3 decades.

As such, we conclude that, in most cases, immunogenicity data from a single-dose clinical PK study together with the comparison of quality attributes with potential immunogenic relevance provides a robust data package to evaluate the comparable immunogenicity of a proposed biosimilar. This allows for more efficient streamlined biosimilar development without the need for a comparative efficacy study.

Declarations

Funding

No funding was received for this review.

Conflict of interest

The authors, with the exception of Elena Guillen, are employed by companies developing and/or marketing biosimilar medicines. Gillian Woollett is an Editorial Board member of BioDrugs. Gillian Woollett was not involved in the selection of peer reviewers for the manuscript nor any of the subsequent editorial decisions. Elena Wolff-Holz is an Editorial Board member of BioDrugs. Elena Wolff-Holz was not involved in the selection of peer reviewers for the manuscript nor any of the subsequent editorial decisions.

Availability of data and material

Not applicable.

Ethics approval

Not applicable.

Consent to participate

Not applicable.

Consent for publication

Not applicable.

Code availability

Not applicable.

Author contributions

MS had the idea for the article. All authors contributed to the drafting, data analysis, review, and final approval of the manuscript.