Related Experiment Video
Updated: May 30, 2025
In Vitro Methods for Comparing Target Binding and CDC Induction Between Therapeutic Antibodies: Applications in Biosimilarity Analysis
Published on: May 4, 2017
An Automated Tool for Glycosimilarity Assessment of mAb Therapeutic Biosimilars: Trastuzumab and Bevacizumab as Case
Anuj Shrivastava1, Sanjeet S Patil1, Rohan Shah1
1Department of Chemical Engineering, Indian Institute of Technology Delhi, New Delhi, India.
Background:
With the expiration of patents for multiple biotherapeutics, biosimilars are gaining traction globally as cost-effective alternatives to the original products. Glycosylation, a critical quality attribute, makes glycosimilarity assessment pivotal for biosimilar development. Given the complexity of glycoanalytical profiles, assessing glycosimilarity is nontrivial.
Objective:
This study proposes a Python-based automated tool for rapid estimation of glycosimilarity index (GI).
Materials And Methods:
A comprehensive analytical glycosimilarity comparison of the trastuzumab originator product, Herclon (Roche), with five marketed biosimilars:Trasturel (Reliance Life Sciences), Canmab (Biocon), Vivitra (Zydus Ingenia), Hertraz (Mylan), and Biceltis (Cipla), has been performed. Similarly, a comparison between the bevacizumab originator product, Avastin (Roche), and its five biosimilars: Abevmy (Mylan), Krabeva (Biocon), Ivzumab (RPG LifeSciences), Bryxta (Zydus), and Advamab (Alkem Labs), is presented. Glycan profile has been assessed using liquid chromatography-fluorescence detection, and the data have been integrated using the XGBoost-machine learning algorithm to quantify glycan composition. The GI has been calculated by combining profile similarity and compositional similarity, estimated on the basis of the criticality and tolerance of each glycan.
Results:
The tool enabled rapid GI estimation (< 1 min/sample) with reduced errors compared with Excel (> 10 min/sample). Biosimilars exhibited high GI with several exceeding 95%, while the lowest GI observed were 87.80% for trastuzumab and 92.39% for bevacizumab.
Conclusions:
The Python-based tool offers a high-throughput and a reliable platform for glycosimilarity assessment, outperforming traditional analysis. Minor variations in glycosylation patterns were observed among the biosimilars, suggesting a modest glycosimilarity variation (GI range between 80 and 100%). However, the limited number of innovator batches analyzed constrained the establishment of definitive tolerance limits. Future studies should focus on analyzing larger datasets to improve accuracy and define precise tolerance limits, enhancing the tool's reliability and its potential to accelerate biosimilar development.
