The shockwave of animal experiments is forcing AI drug development to accelerate its evolution

The rhesus monkey, mouse, and other experimental animals have long been the "unsung heroes" of new drug development, using their bodies to ensure the safety of human medications.

With the emergence of relevant policies from regulatory authorities, this is quietly changing.

On April 24, the Ministry of Industry and Information Technology and six other departments issued the "Implementation Plan for the Digital and Intelligent Transformation of the Pharmaceutical Industry (2025-2030)," which distilled 41 typical scenarios for technological innovation development, providing a reference for the digital and intelligent transformation of pharmaceutical companies.

It mentioned "data mining of animal model data and virtual animal experiments," which addresses issues such as high demand for animal alternatives and discrepancies in consistency with human results, suggesting the use of data mining and simulation technology to establish computer simulation models for animal modeling.

A more significant change comes from across the ocean, 14 days prior.

On April 10, the FDA released the "Roadmap for Reducing Animal Testing in Preclinical Safety Studies" (hereinafter referred to as the "Roadmap"), stating that it will gradually eliminate animal testing in preclinical safety studies for monoclonal antibodies (mAbs) and other drugs, opting instead for more new alternative technologies such as organoids.

This transformation is reshaping the rules of the game in China's pharmaceutical industry.

To this end, Xinfeng visited and researched multiple CXO, model animal companies, AI pharmaceutical companies, and organoid-related enterprises, discovering that the impact of this wave of animal testing may bring multiple changes:

First, the short-term impact is limited, while the long-term direction is the development of the industry.

In the short term, the accuracy of new technologies compared to animal testing still requires more data validation;

The development of new technologies such as organoids is indeed a trend, but whether they can completely replace animal testing to have a disruptive impact remains unclear.

Second, alternative solutions are being vigorously developed.

Many CXO, model animal, and AI pharmaceutical companies are laying out plans for new technologies such as organoids, with cooperation methods including self-research and strategic investment.

Third, the rapid development of new technologies such as "organoids" still requires more experimental data validation.

Domestic companies that have independently developed organoids using "organoids + AI" can achieve over 80% accuracy, but at the same time, some AI pharmaceutical companies have reported to Xinfeng that there are significant differences in drug feedback between organoids and animal testing.

From Monkeys to Mice

As a necessary step before new drug clinical trials, "non-clinical safety evaluation services" (hereinafter referred to as "safety evaluation") comprehensively assess the safety of candidate drugs through experimental monkeys and other animal models both in vivo and in vitro.

However, experimental monkeys are expensive in the U.S.

For example, in monoclonal antibody development projects, it typically requires the use of 144 experimental monkeys, with the average price of an experimental monkey in the U.S. reaching $50,000, which is a heavy burden for drug developers.

Due to ethical, cost, and other considerations, the FDA has aimed its "first shot" at experimental monkeys in this roadmap.

The FDA plans to shorten the toxicology testing cycle for monoclonal antibodies using experimental monkeys over the next three years: for mAbs that do not show concerning signals in one month of research and testing, the conventional six-month toxicology test using experimental monkeys will be shortened to three months.

This has raised market concerns about the prospects of businesses related to experimental monkeys, with Zhaoyan New Drug (603127.SH) being the first to feel the impact of "monkey hoarding." In 2022, JOINN acquired control of Yunnan Yingmao Biotechnology Co., Ltd. and Guangxi Weimei Biotechnology Co., Ltd. for 1.805 billion yuan, obtaining 19,600 experimental monkeys.

As of the end of 2024, the value of biological assets, including experimental monkeys, amounted to 1.069 billion yuan, accounting for over 10% of JOINN's total assets.

In recent years, the decline in domestic experimental monkey prices has made JOINN experience the pain of "stockpiling monkeys."

In 2024, the decline in the fair value of biological assets has already brought JOINN a net loss of 114 million yuan.

Non-clinical evaluation services for drugs, including safety assessments, are JOINN's core business, generating 1.917 billion yuan in revenue in 2024, accounting for over 90%.

The potential reduction in the use of experimental monkeys in the future may pose a challenge of declining fees for JOINN.

The core business of Yinuosi (688710.SH), which also engages in "stockpiling monkeys," is safety assessment, achieving 889 million yuan in revenue in 2023, accounting for over 80% of total revenue.

From 2021 to 2023, the total amount of biological assets, including experimental monkeys purchased by Yinuosi, reached 706 million yuan.

WuXi AppTec (603259.SH) has also acquired a company dealing with experimental monkeys.

In November 2019, WuXi AppTec's subsidiary Suzhou WuXi AppTec New Drug Development Co., Ltd. acquired 100% of Suzhou Kanglu Biotechnology Co., Ltd. for 804 million yuan, obtaining approximately 20,000 experimental monkeys.

However, due to WuXi AppTec's larger scale, the proportion of biological assets related to experimental monkeys is limited.

As of the end of 2024, WuXi AppTec's biological assets amounted to 2.018 billion yuan, accounting for only 2.51% of total assets.

Reducing the use of experimental mice in preclinical research is also one of the FDA's goals.

As experimental animals that are highly homologous to humans and possess characteristics such as strong reproductive ability, short generation cycles, and low breeding costs, the genome of experimental mice serves as an important transition connecting large animal experiments.

In practical applications, CXO companies replace specific gene sequences in experimental mice with human genes through gene editing technology to evaluate drug efficacy and toxicity.

However, this business is currently facing challenges.

The FDA's roadmap indicates that within the next three years, it will pilot projects using non-animal models for drug development submissions.

For example, one of the animal models used in monoclonal antibody experiments is transgenic mice. In the future, the FDA may allow applicants to submit drug development using a series of human in vitro tests or multi-parameter assessments combined with physiologically-based pharmacokinetic models (one form of computer simulation) as non-animal models to replace transgenic mouse experiments.

Currently, the "three giants of domestic model animals," YK Biotechnology (688046.SH), SMOC (688265.SH), and BIOCYTOGEN-B (2315.HK), all provide mouse experimental services, and potential challenges are emerging.

Vision and Reality

Limiting animal experiments has always been the policy direction of the FDA.

As early as 2022, the U.S. Congress passed the FDA Modernization Act 2.0, explicitly allowing the use of non-animal alternative methods to support new drug trials, and since then, various parties have been exploring new technologies as alternative solutions This roadmap is more like an upgraded version of the previous one, further providing pharmaceutical companies with clearer directions for the use of new technologies.

The FDA clearly stated that it encourages pharmaceutical companies to use various methods such as organoids, computer simulations, and in vitro human tissues to assess immunogenicity, toxicity, and pharmacodynamics that are highly relevant to human biology.

For example, the FDA allows pharmaceutical companies to use results from organoid or computer simulation studies as supportive data in drug clinical trial applications/biologics license applications.

However, in the short term, the impact of the new policy on the industry remains limited.

Because under the complexity of biological organisms, the idea of quickly and completely replacing animal experiments with new technologies is clearly unrealistic.

"In the short term, the impact on the company's business is not significant, as many new methods currently only serve as supplements and cannot fully simulate biological organisms," a person close to Yaokang Biotechnology told Xinfeng.

"This policy is not new, nor is it as significant as most people think; there are still challenges to actual implementation, but it is a step in the right direction, clarifying the attitude towards embracing AI predictive models, organoid tissues, and other alternatives to animal testing, and has taken an important step towards reducing regulatory burdens," explained a person close to Yingsi Intelligent.

"The FDA's new policy does not completely negate the 'ban' on animal testing, but represents a more realistic and gradual regulatory reform. Although the FDA advocates that breakthroughs in drug safety evaluation may be achieved through alternative technologies in the future, there is still a long way to go. The mechanisms of diseases are highly complex, and preclinical pharmacodynamic evaluations will still rely heavily on animal models for a considerable time," a person close to BIOCYTOGEN stated to Xinfeng.

This individual further pointed out that with the simplification of the FDA's requirements for preclinical safety evaluation tests, the pressure on new drug development companies regarding time and costs will be significantly alleviated, allowing more research resources to be directed towards innovative pipelines.

Such a trend will further release the market's actual demand for high-fidelity animal models, especially humanized mice in pharmacodynamic validation.

In the long run, reducing animal testing is indeed a trend in the industry.

Many CXOs and model animal companies are exploring the application of new technologies such as organoids and computer simulations.

Xinfeng learned from a person close to Yaokang Biotechnology that the company is currently focusing on and laying out organoid technology.

"In the medium to long term, it is impossible to determine whether new technologies will have a disruptive impact. What we see now can only serve as a supplement. However, strategically, we definitely need to lay out plans, such as organoid technology, which can also provide customers with more options," this individual stated.

Coincidentally, the day before the FDA released the roadmap, Yaokang Biotechnology announced changes to its previous IPO fundraising projects.

It added a project for an "AI-driven organoid and animal disease model multimodal preclinical drug research platform," with an intended investment of 200 million yuan; at the same time, it reduced the investment in the "integrated base construction project for model animal mouse research and breeding" from 600 million yuan to 430 million yuan.

JOINN is expanding its tumor organoid-drug sensitivity platform based on its experience in developing organoid platforms and has established in vitro drug sensitivity testing for osteosarcoma organoids, but it has not yet been commercialized "Currently still in the research and development stage, not yet commercialized," a person close to JOINN confirmed to Xinfeng.

The aforementioned person close to BIOCYTOGEN also stated to Xinfeng: "The company has always adhered to a forward-looking vision. With the localized deployment of AI platforms like DeepSeek and the exploration and accumulation of more new technologies, combined with the company's rich experience and product resources accumulated in the industry, we will gradually release more information on the progress of intelligent drug discovery and development."

Although animal testing remains a necessity in the short term, the industry is betting on the future.

Welcoming New Entrants

According to the FDA's planning, new technologies mainly cover organoids and microphysiological systems, computer simulations, etc., which can play an important role in reducing animal testing.

Among them, organoids are similar to 3D cells, derived from body cells, and cultured in vitro to form mini versions of organs that can simulate the structure and function of natural tissues, including liver and intestinal organoids.

This is expected to drive the industry to welcome more entrants.

Currently, two well-known AI pharmaceutical companies in China, XTALPI (2228.HK) and Insilico Medicine, are making strides in this area.

XTALPI has incubated organoid-related companies, including Xige Biotechnology (Shenzhen) Co., Ltd. (hereinafter referred to as "Xige Biotechnology") and Shenzhen Yaosu Technology Co., Ltd. (hereinafter referred to as "Yaosu Technology").

Xige Biotechnology is the world's first "organoid + AI" drug research and development company.

According to Xige Biotechnology's introduction to Xinfeng, taking cardiac toxicity testing as an example, the traditional gold standard for cardiac toxicity testing, "hERG," can only assess potassium ion channels, with a clinical toxicity prediction accuracy of about 40%; through its independently developed "cardiac organoid + AI" technology, it tested over 120 FDA-approved or phase II clinical drugs and found that the accuracy can reach 85%.

"After extensive real-time comparative validation, organoids can become reliable preclinical models in the future, at least partially replacing animals in the efficacy and safety evaluation tests for new drugs and biosimilars," said Zhang Haisheng, founder and CEO of Xige Biotechnology, to Xinfeng.

XTALPI has also reached a strategic cooperation with Yaosu Technology to integrate AI drug development technology with organoid models and organ-on-a-chip technology, including using organ-on-a-chip's preclinical liver toxicity model for toxicological evaluation, thereby achieving key efficacy validation, etc.

This year, Yaosu Technology also established a cooperative project with the FDA's Center for Drug Evaluation and Research to jointly promote the application of microphysiological system technology in drug safety testing, focusing on drug-induced liver injury assessment.

Insilico Medicine, on the other hand, is entering the field through self-research and plans to combine AI for toxicological property prediction in the future.

According to a person close to Insilico Medicine who spoke to Xinfeng, during the AI-generated molecular structure process, it has already incorporated drug-like properties and metabolism-related options into its self-developed chemistry engine, chemistry42, and will also establish predictions for small molecule toxicological properties in the future.

Three Major Challenges to Address

Despite the advantages of new technologies like organoids, it is still difficult to effectively replace animal testing in the short term.

First, the accuracy of new technologies like organoids still has room for improvement. The aforementioned person close to Insilico Medicine revealed to Xinfeng that, taking organoid molecules as an example, the performance of small molecule drugs in the human body differs from that in organoid models.

Currently, when Insilico Medicine evaluates small molecule drugs designed by AI, it not only considers the ranking performance of various screening items in the AI model and the performance of molecules in organoid models but also combines animal experiments for research.

This means that even with the most powerful AI, real animal experiments are still needed for support.

Second, there is a lack of sufficient real-world data feedback.

Currently, building AI models for predicting experimental results requires a large amount of high-quality toxicology and safety data, which is difficult to obtain publicly.

"We hope to engage in extensive cooperation with pharmaceutical companies, research institutions, and regulatory authorities to promote data sharing and transparency, accelerating the implementation of technology-driven new policies," Insilico Medicine pointed out.

Third, specific technical guidance standards have yet to be established, requiring continuous exploration from all parties.

"From a short-term perspective, standard formulation and regulatory recognition remain key challenges. The company is working with top research institutions such as Peking University and Tsinghua University, as well as regulatory agencies, to promote the standardization of organoid technology, aiming to provide more precise, efficient, and ethical solutions for global drug development," Xige Biotechnology stated to Xinfeng.

Currently, there is also exploration of organoid technology development in China, but there are still no more detailed guidelines.

The direction in which this productivity revolution in innovative drug development will ultimately head is something the market is eagerly watching