Antibodies are proteins produced by the immune system capable of binding to foreign substances in the body, such as bacteria or viruses, and destroying them. The same is true for cancer cells. No wonder the pharmaceutical industry soon captured their potential for drug development.
The first therapeutic monoclonal antibody was approved by the US Food and Drug Administration (FDA) in 1986. Since then, over 100 monoclonal antibodies have been designated as drugs, acting as effective therapeutic agents.
Because of their specificity and affinity for a variety of molecular targets, antibodies are useful for treating many different diseases. In the past few decades, these molecules have been the focus of technological advancements aimed at improving their affinity, production, and isolation abilities while reducing their immunogenicity.
Antibodies are key tools in research, but they also represent the fastest-growing class of biotherapeutics on the market because of their naturally favorable attributes (i.e., potency, robustness, and ease of production). Antibody engineering is revolutionizing the diagnosis and immunotherapy for developing the treatment of various diseases, especially cancer therapy.
In recent years, therapeutic antibodies have become the primary class of drugs in development - and seven of the ten bestselling drugs in 2018 were mAbs. The global therapeutic antibody market was valued at $115.2 billion in 2018 and is expected to generate revenue of $300 billion by 2025. Recently, the 100th monoclonal antibody was approved by the FDA, with hundreds more in various stages of clinical trials
However, the average cost of developing drugs based on monoclonal antibodies is $650-$750 million, and the process takes from 8 to 9 years.
What can companies do to speed up and streamline this process? Computational methods based on AI, machine learning, and deep learning are the industry's best answers. Keep on reading to learn why AI technologies hold so much promise for antibody discovery and what results they have delivered so far.
Antibody engineering focuses on discovering and modifying antibody sequences or structures that bind a therapeutic target, modulating its pathological function. The idea is to produce highly specific antibodies characterized by optimal processing, stability, and tolerance (with some of the functions not necessarily being associated with 'natural antibodies' and having to be engineered in).
To achieve the highest objective response rate in patients and the lowest toxicity, antibodies must be produced to preserve their therapeutic potential while optimizing their purification.
Researchers in academic and industrial laboratories have studied therapeutic antibodies to fight cancer, HIV, and other serious diseases. Using the latest techniques based on AI, scientists can produce fully human monoclonal antibodies that are less likely to provoke an unwanted immune response than early mAbs that were developed using mice.
Take a look here for an overview of techniques used in antibody engineering today: What is antibody engineering and how is it transforming drug discovery today?
Traditionally, researchers begin with an antibody candidate that may have a reasonable affinity with its target. The gene encoding the antibody is then mutated to produce a large number of antibodies related to the original candidate.
With the help of AI, researchers can quickly determine which antibodies best bind the target.
A team of Swiss researchers showed how optimized antibody variants could be identified by predicting antigen specificity via deep learning from a massively diverse space of antibody sequences. A training dataset was generated through random mutations using a CRISPR mutation method, generating about 40,000 related antibodies. The method was screened for binding. They used the ML algorithm to identify improved variants using a well-established antibody cancer drug. Several of them improved binding compared with the original drug. They were also easier to produce and more stable.
The team showed that their method could successfully increase the number of antibody candidates derived from a screen on the human immune system, resulting in the discovery of new variants with improved properties that are potentially better tolerated by the human body.
This example shows that the successful application of AI techniques will result in more effective drugs and less expensive to produce, leading to improved tolerability and better clinical outcomes for patients who receive them.
The development times of drugs based on monoclonal antibodies are increasing. A report that analyzed FDA-approved monoclonal antibodies showed that clinical development times – specifically, the duration of Phase II and III trials – are becoming, while FDA review times remained constant.
The average time from an investigational new drug (IND) filing to market amounted to 6.7 years for 11 MAbs that the FDA approved between 1994 and 2003. Take a look at the period between 2004 and March 2011, and you'll see the period of 8.3 years for 12 mAbs approved in this time, according to Deloitte Recap LLC analysis, Therapeutic Monoclonal Antibodies – Insights, Strategies, and Data.
Another important angle is IP protection for new drugs. Since the clock starts ticking at the discovery phase, the shorter the discovery and development, the longer the protected sales for pharmaceutical companies.
The accelerated speed of drug discovery and development is especially important in the context of global events such as pandemics. During a pandemic, there is no time to waste in the evaluation of therapeutic modalities, including vaccines, nucleic acids, small molecules, and mAbs. Until recently, the evaluation of mAb therapies for this scenario has been slow, and the production capacity was limited.
What has changed to enable rapid evaluation of mAb therapies in this case? The product development timeline from lead mAb identification to phase 1 investigational new drug application (IND) is 10–12 months at many companies today — a dramatic reduction from the 18 months that was standard in the industry 5 or more years ago. A combination of recent technical advances and the acceptance of business (but not product quality or patient safety) risks offers a further acceleration for clinical trials.
By using AI technologies, teams can accelerate these activities and enable production capacity for clinical studies for therapeutic mAbs.
AI helps even more to identify the best candidates for the next stages - improved focus and selection early on in the process are essential for companies to mitigate risks while shortening the development cycle.
Ultimately, a wide range of AI applications is already being explored to tackle the fundamental challenge of developing new drugs, from target identification through clinical trials, requiring years of time and billions of dollars.
Recognizing this potential and hoping that the new technology can also help them develop more-effective and better-targeted drugs, pharmaceutical companies are collaborating with specialized teams to implement these new technologies.
AI can examine vast quantities of existing data and learn patterns that might be too subtle or complex for humans to recognize. This technology could then predict new small molecules with desirable properties, taking the computational screening process to a new level.