KOTARO FUKUOKA, TAKURO KUSASHIO and YAOYU ZHANG, Nikkei staff writers
TOKYO — Artificial intelligence is transforming the landscape of drug discovery and development. The technology is helping to slash the time and money needed to develop new drugs for COVID-19 and other serious diseases by quickly identifying promising drug candidates.
In the case of COVID-19, the application of AI helped one company come up with a treatment that was approved in the U.S. in a lightning-fast nine months.
British AI startup BenevolentAI identified baricitinib, a drug developed by Elli Lily for the treatment of rheumatoid arthritis, as a potentially effective COVID-19 drug in just a few days. The medication has been approved as a COVID-19 treatment in the U.S. and Japan. The European Medicines Agency has also begun evaluating baricitinib for use against the coronavirus.
A BenevolentAI specialist team was tasked with using the company’s state-of-the-art AI to find an already approved drug that could be repurposed as a COVID-19 treatment. This approach made it possible to win emergency use authorization by the U.S. Food and Drug Administration to treat hospitalized COVID-19 patients in just nine months, instead of the several years typically required. Drug discovery usually involves a long search for candidates and animal testing to evaluate their safety.
BenevolentAI’s technology identifies potential drug candidates using data from clinical trials, academic papers, and its own database on diseases, genes and pharmaceuticals. When a target protein is identified, AI finds candidate drugs that act on it.
Applying AI to drug discovery and development is expected to sharply reduce in the time required to create new drugs, a process that usually takes nine to 17 years. That time could be cut in half for approved drugs that are repurposed for other uses.
In February 2020, soon after the World Health Organization declared the COVID-19 outbreak to be a public health emergency of international concern, BenevolentAI’s first paper on baricitinib as a candidate COVID-19 treatment, published in the British medical journal The Lancet, found that the drug may inhibit the ability of the virus to infect lung cells and cause inflammation in patients.
Eli Lilly, which owns the rights to baricitinib, and the National Institute of Allergy and Infectious Diseases launched a study in the U.S. to examine the efficacy and safety of the drug as a potential treatment for hospitalized COVID-19 patients. Because the study found that baricitinib can shorten recovery times and improve clinical outcomes for patients, the FDA granted emergency use authorization for the drug in November last year. The drug has been shown to reduce mortality in hospitalized patients by 38% when used in combination with remdesivir, an antiviral medication, according to data released by Eli Lilly.
BenevolentAI is also developing drugs on its own, focusing on treatments for more than 10 diseases, including atopic dermatitis and amyotrophic lateral sclerosis (ALS), also known as motor neurone disease or Lou Gehrig’s disease.
The company started a clinical trial on a potential treatment for atopic dermatitis in February. It is also working with AstraZeneca to develop a treatment for chronic kidney disease.
Use of AI in drug discovery and development is spreading around the world. Sumitomo Dainippon Pharma, in partnership with Exscientia, an Oxford, England-based AI drug discovery startup, has found a candidate treatment for obsessive-compulsive disorder. Last year, the Japanese drugmaker began clinical trials of the drug candidate in Japan to evaluate its safety.
“We found the candidate in less than a year using AI for the process, which typically takes four and a half years,” said an executive at Sumitomo Dainippon Pharma. In May, the company started Phase 1 clinical trials in the U.S. on an Alzheimer’s disease psychosis drug candidate designed using Exscientia’s AI technology.
AI does not do all the work. It is used to find candidate drugs and narrow down the design of new drugs by crunching huge amounts of data from scientific papers and experiments. People must work out which direction to take the research and development itself.
Exscientia is attracting the attention of pharmaceutical and biotechnology companies around the world. Evotec, a German biomedical company, has jointly developed a new cancer treatment with Exscientia. Human clinical trials of the A2a receptor antagonist began in April, according to Evotec. The candidate drug was discovered eight months after the two companies launched the project.
Taisho Pharmaceutical of Japan and Insilico Medicine, a Hong Kong-based AI startup, started a joint research project last fall to identify therapeutic compounds that may slow the cellular effects of aging. Insilico is using its AI networks to identify therapeutic targets and find druglike molecules that target senescent cells. The accumulation of these cells as people age is thought to be behind a variety of diseases.
Insilico’s job is to identify the role senescence plays in specific cells, tissues and diseases, with different proteins implicated for each, and to design molecules to tackle those targets. Taisho will validate the computer-generated compounds through in vitro and in vivo testing.
Some Japanese AI startups are playing catch-up with such overseas leaders in the field. Hacarus, a Kyoto-based AI startup, and the University of Tokyo in June announced the start of a joint research project to develop cures for Alzheimer’s disease and Parkinson’s disease. Both are caused by the accumulation of certain proteins in the brain. Using AI to develop drugs for these types of disease is still rare.
The project is aimed at creating a system in one year to efficiently search for compounds that could be candidate drugs. The AI-driven approach will “dramatically improve the speed and accuracy of the research process, which has traditionally depended on human hands and eyes,” said Taisuke Tomita, a professor at the University of Tokyo’s Graduate School of Pharmaceutical Sciences.
More than 120 Japanese companies and universities have joined the Life Intelligence Consortium, an industry-academia collaboration aimed at applying AI to the life sciences. The consortium has already created around 20 prototype AI programs for drug discovery and development.
“AI will soon become an essential technology for drug discovery and development,” said Yasushi Okuno, a Kyoto University professor. The technology offers the opportunity to “reconsider the conventional wisdom that developing a new drug takes 10 years.”
Coming up with new has become hugely expensive. The average cost to develop a prescription drug that makes it to market soared to some $2.9 billion in the 2000s, up from $180 million in the 1970s, according to an estimate by Tufts University in the U.S. The rise in drug development costs has been far steeper than inflation overall.
Many new drugs have been developed over years, especially treatments for cancers and “lifestyle diseases.” Many substances that act on important molecules implicated in the development of diseases have already been identified and developed into drugs. As a result, it is becoming increasingly difficult to develop effective new drugs. And a greater emphasis on safety has stretched out the time needed for clinical trials.
Only one in about 30,000 candidate substances actually becomes a new drug, according to the Japan Pharmaceutical Manufacturers Association. The process can take nine to 17 years. R&D spending by pharmaceutical companies is equal to roughly 10% of annual sales, compared with around 4% for the manufacturing sector as a whole.
In their efforts to come up with new treatments, drug companies are devoting more of their resources to the development of biopharmaceuticals — complex medicines made from living cells or organisms, often created using cutting-edge technologies, including antibody drug conjugates. Developing and manufacturing biopharmaceuticals is thus complicated and costly.
One such drug, Nivolumab, is sold under the brand name Opdivo. It was first used in Japan in 2014 to treat a variety of cancers and initially cost some 35 million yen ($318,000) a year to administer. This led to complaints that its use would further strain the government’s already huge and growing health care spending.
AI is among the technologies that could, by drastically shortening drug development time and cost, help to keep prices in check, thereby improving access to new treatments for previously intractable diseases and enhancing the quality of life for everyone.