Analytical and functional characterization serves to demonstrate that the active component of a biosimilar medicine is structurally and functionally similar to the active component of the reference medicine, and the biosimilar would thus be expected to behave in the same way as the reference medicine [7, 22, 38]. As noted above, analytical and functional characterization constitutes the foundation and the largest body of evidence of the ToE [22], accordingly represented as the ‘base’ of the biosimilarity development pyramid (Fig. 1).

Methods and assessment criteria of analytical similarity are based on characterization of the reference medicine and its critical quality attributes (CQAs) [39]. QAs are measurable physicochemical characteristics of a biologic medicine that determine its principal properties, and can vary owing to either inherent or manufacturing variability [40]. In biosimilar development, it is therefore fundamental to identify which QAs are CQAs (e.g., oxidation, glycosylation, and binding to Fcγ receptors); that is, have a direct impact on the medicine’s pharmacokinetics, safety, efficacy, and immunogenicity and must be demonstrated to closely match those of the reference medicine, laying within prespecified thresholds [21, 22, 38, 39, 41, 42]. QAs are categorized into very high, high, moderate, low, or very low criticality. All QAs ranked with very high, high, or moderate criticality are considered as CQAs [43].

The biosim-NTZ comparative analytical and functional characterization comprised multiple investigations [35, 36, 44], where the structure and function of biosim-NTZ was compared with that of US-ref-NTZ and EU-ref-NTZ, using quality ranges calculated based on standard deviations from the range of all tested US-ref-NTZ batches (Table 1). The criticality of QAs was assessed based on a risk ranking approach using a preliminary hazards analysis, as per the International Conference on Harmonization (ICH) guidelines (ICH Q9) [36, 45]. Standard deviation multipliers were applied according to risk: 2.5 for ‘high’ risk attributes (those directly related to the mechanism of action), 3 for ‘moderate’ risk attributes (product variants not directly related to the mechanism of action or those attributes directly related to the mechanism of action in which orthogonal methods were included in the comparative analytical assessment), and 4 for ‘low’ risk attributes [35]. For QAs with ‘low’ and ‘very low’ criticality, the data were compared descriptively (a ‘low-risk’ category was also applied for QAs that are not amenable to a statistical evaluation, e.g., if a certain attribute can only be assessed qualitatively) [35].

The first aspect of the analytical assessment was a comparison of the physicochemical, biophysical, and in vitro functional properties of natalizumab. This included structural characterization, assessment of product-related variants, and studies investigating the Fab-arm exchange under physiologically relevant conditions.

Extensive characterization demonstrated matching outcomes between biosim-NTZ and US-ref-NTZ and EU-ref-NTZ in terms of structural characteristics (Table 1). Peptide mapping followed by liquid chromatography with tandem mass spectrometry confirmed the primary amino acid sequence of biosim-NTZ was identical to US-ref-NTZ and EU-ref-NTZ [35, 36, 44].

The results from Fourier-transform infrared spectroscopy, near and far ultraviolet circular dichroism, fluorescence emission spectroscopy, Förster resonance energy transfer (FRET), and X-ray crystallography demonstrated that the higher order structures of biosim-NTZ, US-ref-NTZ, and EU-ref-NTZ also matched (Table 1) [35, 36, 44]. Comparison of batches of biosim-NTZ and US-ref-NTZ showed that the levels of natalizumab dimers in biosim-NTZ were low and similar to that of US-ref-NTZ, as determined by size-exclusion chromatography [35, 36]. High-molecular-weight impurities were not found in biosim-NTZ or US-ref-NTZ. The purity of biosim-NTZ in terms of the amount of antibody fragments (low-molecular-weight impurities), as determined by capillary electrophoresis under non-reducing and reducing conditions, were within range of US-ref-NTZ [35, 36].

Natalizumab is an immunoglobulin G4 (IgG4) monoclonal antibody that does not mediate Fc-associated activities owing to the low affinity of IgG4 towards Fcγ receptors [46, 47]. Assessments confirmed the lack of Fc-associated effector function for both biosim-NTZ and ref-NTZ (data not shown) [35, 36].

Natalizumab, as an IgG4 monoclonal antibody, is known to undergo a heavy-light chain recombination and therefore a Fab-arm exchange [46]. The Fab-arm exchange takes place when the Fab-arm fragment of a therapeutic antibody exchanges with the Fab-arm fragment of an endogenous antibody, resulting in the formation of a bispecific IgG4 antibody [48]. The kinetics of the Fab-arm exchange between biosim-NTZ and US-ref-NTZ and EU-ref-NTZ was evaluated in real time using Förster resonance energy transfer. The results demonstrated that similar Fab-arm exchange rates were observed for biosim-NTZ, US-ref-NTZ, and EU-ref-NTZ (Supplementary Fig. 1 in the ESM) [35, 36, 44].

The molecules’ performances were then compared under different conditions, including a comparative forced degradation and stability study to ensure biosim-NTZ and ref-NTZ were comparable under long-term (5 ± 3 °C, inverted), accelerated (25 ± 2 °C/60 ± 5% relative humidity, inverted), and stress (40 ± 2 °C/75 ± 5% relative humidity, inverted) conditions through 6 months [35, 36].

Biosim-NTZ, US-ref-NTZ, and EU-ref-NTZ responded in a similar way to applied stress conditions, namely thermal stress (50 °C for 28 days), oxidative stress (5% 2-amidinopropane dihydrochloride at 20 °C for 48 h), light stress (526 W/m2 ultraviolet-A light-hour + 1200 klux-hour visible light), freeze-thaw stress (five cycles of −80 °C for 22 h followed by 20 °C for 2 h), pH stress at acidic pH (pH 4.0 at 30 °C for 21 days), and mechanical stress such as agitation (750 rpm at 20 °C for 24 h) [35, 36, 44].

Comparative stability studies revealed minor differences between biosim-NTZ and ref-NTZ for stress conditions with no expected clinically meaningful impact, while no significant differences were observed under long-term and accelerated storage conditions up to the 6-month timepoint, confirming similarity between biosim-NTZ and both US-ref-NTZ and EU-ref-NTZ [35, 36, 44].

The potency of biosim-NTZ was assessed through the detection of any differences in the molecule’s interactions with its known targets and used as a sensitive tool for confirmation of similarity regarding the mechanism of action of biosim-NTZ and ref-NTZ on a functional level [35, 36]. Binding affinity to α4β1 and α4β7 integrins for biosim-NTZ versus US-ref-NTZ and EU-ref-NTZ was assessed via an indirect enzyme-linked immunosorbent assay (ELISA) [Table 1]. The results of the indirect ELISA demonstrated that the binding affinity of biosim-NTZ was within the range of US-ref-NTZ and EU-ref-NTZ [35, 36].

The ability of biosim-NTZ to block the interaction of α4β1 integrin and α4β7 integrin with its cognate receptors VCAM-1 and MAdCAM-1 was also tested by means of a competitive ELISA. Biosim-NTZ, US-ref-NTZ, and EU-ref-NTZ were comparable in terms of blocking the interactions of α4β1 integrin with VCAM-1 and α4β7 integrin with MAdCAM-1 (Fig. 2a, b). The α4β1 integrin and α4β7 integrin indirect ELISA together with the VCAM-1 and MAdCAM-1 competitive ELISAs demonstrated equivalent potency of biosim-NTZ to US-ref-NTZ and EU-ref-NTZ within the predefined quality range [35, 36, 44].

Potency of natalizumab samples by competitive enzyme-linked immunosorbent assay (ELISA). A Inhibition of interaction between vascular cell adhesion molecule 1 (VCAM-1) and α4β1; B Inhibition of interaction between mucosal vascular addressin cell adhesion molecule 1 (MAdCAM-1) and α4β7 integrin. Biosim-NTZ biosimilar natalizumab, ref-NTZ reference natalizumab. The boxes indicate quartiles and the horizontal line in each box represents the median value; the whiskers show the data distribution; the circles represent the individual batches of US-ref-NTZ, EU-ref-NTZ, and biosim-NTZ