Biosimilar Characterization

Comparative studies between authentic batches of Herceptin®, manufactured in different geographical locations (the EU and the United States), and a possible biosimilar candidate of trastuzumab have been carried out by Firth et al. (paper submitted) using LC-MS and a series of enzymatic digestions, as described in Section 3.1, to quantify the glycosylation profiles. The relative MS responses, presented as a function of percent, are shown in Table 10.1. The trastuzumab biosimilar sample clearly contained significantly less afucosylated glycan. With the addition of surface plasmon resonance binding analysis and

Table 10.1 Averaged relative MS responses for the quantification of nonglycosylated, afucosylated, and fucosylated Fc/2 subunits of EU and US Herceptin® and a trastuzumab sample following digestion with IdeS, EndoS, and EndoS2.


Mean relative MS response (%)


Fc/2 + GlcNAc

Fc/2 + GlcNAc + Fuc

US Herceptin®




EU Herceptin®








ADCC potency studies, the trastuzumab sample was demonstrated to have significantly lower binding affinities toward the FcyIIIa receptor and reduced potency of approximately 50% compared with the authentic Herceptin® batches. It was observed that there were also differences between the two batches of Herceptin®, with the EU lot containing a higher level of afucosylation compared with the US lot. With afucosylation playing a role in the secondary mechanism of action, it was interesting that different lots of the approved, authentic Herceptin® contained different proportions of this important attribute, but with little effect on potency. The findings illustrated the importance of evaluating multiple lots of innovator product in order to understand the structure and the effect of variability in attributes such as glycosylation toward defining acceptance criteria for biosimilarity.

Work published by Damen et al. also reports differences in the glycosylation profiles between different batches of the originator trastuzumab [20]. Their analysis was carried out at the intact level using LC-ESI-MS, and deconvoluted masses were closely matched to the calculated masses for the main glycoforms. Although qualitatively the different batches were the same, quantitatively they were not. Comparison of a candidate biosimilar with the equivalent innovator mAb characterized by LC-MS highlighted how MS can be used to identify mass differences and therefore differences in amino acid sequence that ultimately decide whether a proposed biosimilar is likely to meet acceptance criteria even at the intact level [68]. The authors report observing a difference of 64 Da between the two samples. Following reduction with DTT, the location of the mass difference could be attributed to the HC subunits (32 Da on each). Additional nanoLC-MS/MS identified variation of two amino acids within the HC, which accounted for the change in mass. From this information, it could be concluded that the biosimilar mAb was derived from a different allotype altogether. The ability to identify key quality attributes, such as differences in amino acid sequence, in a quick and simple experiment, is essential for mitigating risk from a biosimilar campaign within the early stages. At the intact level, however, the mass measurement accuracy is not capable of differentiating small mass differences or minor modifications such as deamidation or oxidation. Therefore bottom-up MS approaches have been studied by Chen et al. [159]. A three-step strategy was employed including peptide sequencing via collision-induced dissociation/electron transfer dissociation (CID/ETD), alongside nontargeted comparisons of the tryptic map and targeted comparisons of minor modifications for three samples: originator trastuzumab and two biosimilar products, one without known variants and one with two amino acid variants. Varying levels of modifications were detected across all three samples.

< Prev   CONTENTS   Source   Next >