IPT 37 2011 9/6/11 09:42 Page 34 Discovery Technology Improving the Developability of Biopharmaceuticals By Jesús Zurdo, The combination of computational and protein engineering tools with Rebecca Michael and novel orthogonal methods of measuring aggregation can be used to Yvette Stallwood at Lonza Biologics, and assess the developability of biopharm products in the early stages Karin Hedman and of development, thereby reducing the level of risk in later stages. Teodor Aastrup at Attana AB Biopharmaceutical development is marred by As a result, there is an urgent need for methodologies multiple risks. High levels of attrition during that will facilitate the assessment of potential preclinical and clinical development are posing a therapeutic efficacy and identification of development significant challenge, and pushing drug development risks in a more flexible, rapid and inexpensive way. costs to levels that are no longer sustainable. This has motivated the adoption of a number of Diverse strategies for failing early and cheaply are strategies as an alternative to traditional lengthy and currently being explored, with an emphasis on expensive clinical development pathways – such as the translational medicine, predictive technologies, application of translational medicine – to obtain rapid scale-down models and further up-front product proof of concept in patients, and the identification characterisation. One of these approaches involves of suitable biomarkers as surrogate endpoints of the incorporation of a developability risk assessment clinical efficacy. In addition, the development of – focusing on manufacturability and safety – in the pharmacogenomics and companion diagnostics in the early phases of development, in order to help with definition of patient groups, or the utilisation of the selection and design of products with the right microdosing and Phase 0 studies, to achieve rapid quality attributes. One such attribute is aggregation, information on efficacy and safety, are also actively which constitutes a significant hurdle for pursued (2,3). biopharmaceutical development and a potential risk of increased immunogenicity. In this article, we IMPLEMENTING AN EARLY RISK ASSESSMENT discuss how protein-engineering approaches based on predictive computational methods, together with One other tool growing in importance in drug early analytics, can be utilised to select candidates development is the assessment of ‘developability’; this with enhanced developability. We also provide looks into aspects related to the manufacturing, examples of new orthogonal strategies to assess stability and safety of therapeutic candidates that could aggregation and stability at an early stage. jeopardise their success during preclinical and clinical development. Developability assessment can be used to CHALLENGES IN THERAPEUTIC DEVELOPMENT identify development risks as well as to incorporate critical quality attributes (CQAs) early on in the design For every hundred therapeutic candidates that enter the of therapeutics, and it is bringing a rejuvenated interest clinic, fewer than 10 will ever be registered. This high in establishing better links between discovery and attrition rate – particularly acute in late phases of development stages. The discovery phase tends now clinical development – is pushing the costs of drug to move beyond maximising biological activity development to levels higher than ever before, currently and also looks specifically into aspects of approaching US$2 billion on average. Causes for manufacturability and developability (productivity, attrition are often very diverse and range from aggregation, stability, solubility, viscosity) and safety insufficient therapeutic efficacy to poor bioavailability, (primarily immunogenicity) (4-6). The ultimate goal of safety concerns (toxicology, immunogenicity), such approaches is to assess critical properties in the formulation problems or high development costs (1). product at a very early stage – ideally allowing the The current situation is posing a significant challenge to ranking of different candidates based not only on their the development of novel therapies and brings into biological activity, but also their developability question the blockbuster model as a viable option for readiness. A number of initiatives have been put the pharmaceutical industry. forward that emphasise the need for implementation of 34 Innovations in Pharmaceutical Technology IPT 37 2011 9/6/11 09:42 Page 35 predictive technologies, scale-down models and further significant negative financial impacts in later up-front characterisation methods in therapeutic development stages. development, with special attention to in silico methodologies (7,8). AGGREGATION ASSESSMENT METHODS ENGINEERING DEVELOPABILITY Monitoring stability and aggregation in INTO A PRODUCT biopharmaceuticals is not an obvious task. The main challenge comes from the fact that aggregation can occur In silico protein characterisation can be used to identify at very different stages in the biomanufacturing and potential developability and safety risks in development processes, and can also manifest itself in biopharmaceuticals, and to ‘design in’ desirable very different ways, including low host viability, low properties. Aggregation and low stability are two productivity, the presence of inclusion bodies and the specific problems imposing severe restraints in the development of opalescent solutions or precipitates. manufacture and development of biopharmaceuticals, Aggregation is not only influenced by the polypeptide and potentially contribute to an increased risk of sequence itself, but also by the nature of the host and undesired immune responses in patients (9-11). process used for its production, the final formulation and Predictive computational tools can be used to identify a history of the batch preparation, storage and fill- structural or sequence elements that could potentially finish conditions. compromise the stability and developability of a given biopharmaceutical. Such potentially deleterious Size-exclusion chromatography (SE-HPLC) is perhaps elements can then be eliminated by protein the most widely utilised standard analytical engineering, resulting in molecules with improved technology for assessing the degree of aggregation in a manufacturing and development properties (12,13). given preparation. SE-HPLC is straightforward to implement, extraordinarily robust, and easy to Examples of such an approach include the application transfer across different laboratories. However, there of computational tools to re-engineer antibody are two main limitations linked to the use of SE- molecules with reduced aggregation, improved HPLC in early analytics. The first relates to the fact stability and increased productivity, as shown in that SE-HPLC can only accurately assess the level Figure 1. Other complementary platforms also aimed of aggregates within a limited window-size at improving the developability and safety of corresponding to the small oligomers that can be biopharmaceuticals include T-cell epitope mapping included in the column. This means that SE-HPLC and preclinical immunogenicity risk assessment, and misses other types of aggregates, including sub-visible make use of either computational or in vitro particles, and may in some cases not be predictive methodologies (14). Such technologies are examples enough to describe the overall stability of a of new and powerful strategies that can be applied to polypeptide in solution. The second major limitation mitigate development and safety risks at the beginning of SE-HPLC is its still relatively low throughput, of drug development, reducing the probability of which limits its applicability as a broad screening tool encountering problems that could have very during lead selection or in the early phases of Figure 1: Engineered antibodies with enhanced A B C developability using Lonza’s AggreSolveTM platform. The graph shows 50 Protein loss (aggregation)104 Binding activity (pM) Productivity (relative to WT) the relative performance 7 45 of engineered variants 40 6 compared with parental 103 35 5 wild type antibody. A) 30 4 Accelerated stability at 25 102 (pM) 60°C; B) Binding activity 50 3 20 of engineered versus wild 15 EC 101 Relative yield 2 type antibody molecules; 10 Percentage protein loss 1 and C) Relative 5 productivity of variants 0 100 0 WT #4 #5 #6 #7 WT #4 #5 #6 #7 WT #4 #5 #6 #7 compared with wild type sequence assessed in Engineered variants Engineered variants Engineered variants stable pooled transfections. www.iptonline.com 35 IPT 37 2011 9/6/11 09:43 Page 36 Figure 2: A) ODA A B measures the relative abundance of secondary- binding sites in a given Percentage monomer (SE-HPLC) polypeptide (such as an ODA relative Detection antibody). Monomeric Labelled-mAb1 100 100 polypeptides only display 80 80 a single binding site that is covered by the capture 60 60 antibody. However, Monomer oligomers exhibit on their 40 40 surface multiple additional 20 20 ODA (relative units) binding sites that are Percentage monomer Aggregates susceptible to being 0 0 recognised by the labelled mAb1 Ab1 Ab2 Ab3 Ab4 antibody. Detection can + Aggregation - be done directly or via a conjugated molecule (such as streptavidin- HRP). B) Comparison of the aggregation development. To compensate for the former, a broad therapeutic antibody). This monoclonal antibody is assessment of a collection range of orthogonal technologies are often used, such used both for capturing the biopharmaceutical