▎ WuXi AppTec content team editor
Image source: Amgen official website
Accelerate drug discovery and development with human data
Amgen says human data will have a significant impact on multiple aspects of new drug discovery and development. For example, relying on human genetic data to select innovative targets that are more likely to succeed, and selecting the right patients can facilitate clinical trials to achieve more consistent results faster. 65 percent of the targets in the company's non-cancer pipeline are currently validated with human genetic data.
Among them, the development of AMG 133, a potential "first-in-class" multispecific molecule for the treatment of obesity and other metabolic disorders, reflects the application of human genetic data. Multiple human genetic studies in populations in Japan, Europe, and the United Kingdom have revealed an association between decreased gastric Inhibitory Polypeptide Receptor (GIPR) expression levels and lower body mass index (BMI).
Based on this insight, Amgen developed the multispecific molecule AMG 133, which couples glucagon-like peptide-1 (GLP-1) analogues to monoclonal antibodies that inhibit GIPR. This research-in-the-case therapy can inhibit GIPR activity while activating GLP-1 receptors, thereby modulating multiple signaling pathways associated with obesity and metabolic disorders.
In early clinical trials, the therapy has shown promising clinical efficacy in the treatment of obese patients, with high doses of the drug reducing the weight of patients by more than 8 kg.
Leverage multi-specific molecules to address the core challenges of drug development
Amgen executives pointed out that future drug development faces three core challenges: the redundancy of biological signaling pathways, the limited window of drug treatment due to toxic side effects on healthy tissues, and the difficulty of many targets to be drugged. The development of multi-specific molecules is expected to solve these problems.
For example, targeting multiple disease-related signaling pathways in a molecule can overcome the redundancy of biological systems for better therapeutic outcomes. Rozibafusp alfa (AMG 570) developed by Amgen is a potential "first-in-class" antibody peptide-coupled drug. It is able to simultaneously inhibit the ICOSL signaling pathway that stimulates T cells and the BAFF signaling pathway that stimulates B cells, thus more effectively treating a variety of autoimmune diseases.
In terms of improving the drug treatment index, multi-specific molecules allow drugs to have an effect on specific disease areas, thereby reducing toxic side effects on healthy tissues. For example, the company's AMG 193 is a small molecule inhibitor of PRMT5 that is being developed to treat a variety of solid tumors. An important feature of it is preferential binding to PRMT5 in the presence of methotioadenosine (MTA), exhibiting a stronger killing ability in MTA-positive cells. Between 10 and 20% of solid tumors are MTA positive, which may enhance the anti-cancer effects of AMG 193 while reducing toxic side effects on healthy tissues.
Multi-specific molecules can bring two different molecules closer together, and protac protein degradation therapies developed based on this principle are expected to target targets that were previously difficult to target. Amgen has also recently entered into a research and development collaboration with Arrakis Therapeutics, a company that develops small molecule therapies that target RNA, to jointly develop small molecule drugs that degrade RNA. This strategy promises to greatly expand the genome that can be targeted.
Amgen also introduced two techniques for developing multi-specific molecules, in which the DNA-coding compound library allows researchers to quickly screen billions of compounds to discover innovative multi-specific small molecules.
Single-chain antibodies allow the construction of more complex, multispecific biologics. By linking single-stranded antibodies that target different antigens, "smart" biologic therapies can be generated, such as when combined with two different antigens on a cell. However, it is more difficult to produce such a complex biological product because it is not just modified on the basis of the proteins already in the body. So how can we accelerate the speed and success rate of protein engineering and produce innovative multi-specific biological products?
Accelerate biologics development with artificial intelligence
Amgen executives said that the company's protein engineering platform still relies on continuous iteration and experimental testing of candidate products to promote the research and development of biological products. This process usually takes 2-4 years to obtain a product that enters clinical testing, and even when it does, it is still not possible to determine that the activity of the biologic is as expected, especially multi-specific antibodies and proteins.
To accelerate protein engineering and the development of innovative multispecific biologics, Amgen has embarked on a project called Biologics NExT to accelerate antibody discovery and protein engineering using artificial intelligence and structural prediction based on amino acid sequences.
Dr. Alan Russell, the company's vice president of biologics, said that last year's breakthrough in AlphaFold2's accurate prediction of protein structure based on amino acid sequences changed the development of biological products. Previously, one of the main reasons why protein engineering processes required constant modification and experimental testing to find the best candidate therapies was that scientists could not predict protein structure and related functions based on amino acid sequences. Breakthroughs in artificial intelligence have made structural predictions based on amino acid sequences possible.
A few weeks after the AlphaFold2 results were released, Amgen scientists used artificial intelligence to solve the synthesis of an E3 ligase that could not be solved in the previous two years in two months, providing a new tool protein for the development of innovative therapies.
Using AI-assisted, Amgen's antibody discovery time was reduced by 50%, the success rate of protein engineering was tripled, and the time required for protein engineering was reduced by 70%. Dr. Russell said that the combination of data science and life sciences will reveal the universal principle of protein sequences associated with structure and function, further promoting the development of biological products.
Focus on potential "first-in-class" projects
In 2021, Amgen's two "first-in-class" therapies, KRAS inhibitor Lumakras and TSLP inhibitor Tezspire, were approved by the FDA. In addition to expanding the range of applications for Lumakras and Tezspire, Amgen also introduced other key "first-in-class" R&D projects in the R&D pipeline.
Among them, tarlatamab is a bispecific T cell binder (BiTE) that targets DLL3, which is expressed in more than 85% of small cell lung cancers and is almost not expressed in normal tissues. Tarlatamab kills cancer cells by recruiting T cells around them and activating them.
In early clinical trials, this biTE molecule with a prolonged half-life showed a promising effect, with an initial median response lasting more than 1 year.
The company's efavaleukin alpha (AMG 592) is a potential "first-in-class" IL-2 mutant fusion protein designed to selectively expand regulatory T cells (Treg) to restore the body's immune balance and treat a variety of autoimmune diseases.
In patients with systemic lupus erythematosus, efavaleukin alpha is able to selectively amplify the number of Treg cells circulating in the bloodstream.
In addition, Amgen is developing several potential "first-in-class" bispecific biologics for the treatment of prostate cancer, where the DIP1-targeted bispecific T-cell binder AMG 509 has shown activity to significantly reduce PSA levels in patients.