LeMone Bio's Tong Dudu Discusses Clinical Data and Costs of Metabolism-Enhanced CAR-T at 1‰ Dose

CAR-T cells that have been continuously stimulated by tumor antigens and are utterly exhausted no longer respond to additional immune stimulation signals. LeMone found a "counterintuitive" approach.

English text is being prepared.

English text is being prepared.

[Editor's Note] The Prix Galien USA, known as the "Nobel Prize of the pharmaceutical world," announced its 2026 award winners on October 29.

Notably, innovative companies incubated by XtalPi have been shortlisted for this award for two consecutive years. Last year, Signet Therapeutics was nominated for "Best Biotechnology Product" for SIGX1094, the world's first targeted drug for diffuse gastric cancer developed in collaboration with XtalPi; this year, LeMone Bio and MetaNovas were both nominated for "Best Startup." The former, with XtalPi's R&D support, achieved complete remission and discharge (100% CR) in dozens of patients with relapsed/refractory hematologic malignancies and lupus erythematosus using a metabolically enhanced CAR-T therapy at one-thousandth of the conventional dose; the latter collaborated with XtalPi to develop MP-5342, the world's first oral lactate dehydrogenase A (LDHA) inhibitor, whose preclinical data show a safety window exceeding 600-fold, targeting the global autoimmune market worth over one hundred billion dollars.

One-thousandth.

This is the lowest dose currently used in early clinical studies of LeMone Bio's core product META 10-19, only 1‰ of the dose administered for conventional CAR-T products, equivalent to 2,000 cells per kilogram of patient body weight.

The principle of CAR-T therapy is not complicated: doctors extract T cells from patients, engineer them into "special forces" that recognize specific target antigens, and then reinfuse them into the body to precisely attack cancer cells.

According to industry common knowledge, the efficacy of CAR-T depends on the reinfusion of a sufficient number of cells; cutting the dose to one-thousandth should leave little efficacy. Yet LeMone's metabolically enhanced CAR-T product demonstrated breakthrough efficacy even at extremely low doses: in the IIT (investigator-initiated clinical study) for this product, dozens of patients with relapsed/refractory leukemia and lymphoma achieved a complete response rate of 100% (100% CR) after treatment; among patients with diffuse large B-cell lymphoma (DLBCL), the 12-month complete response rate remained at 77.8%.

"I think what the judges value most is the leap in efficacy brought by breakthrough technology," said Tong Dudu, Vice President of Business Development at LeMone Bio, who was involved throughout the Galien Award application process. "Underlying breakthrough technology, excellent clinical data, plus better safety, lower production costs, and shorter production cycles brought by low doses—combining these advantages, I think is the main reason we have attracted industry attention, especially this Galien Award shortlisting."

Interleukin-10 (IL-10) is an immunosuppressive factor in immunology textbooks, but LeMone's founding team discovered that it can specifically act on terminally exhausted T cells, reversing T cell exhaustion through oxidative phosphorylation metabolism and restoring their proliferative capacity, killing activity, and immune memory. LeMone Bio was founded by Professor Tang Li's team at the Swiss Federal Institute of Technology Lausanne (EPFL) together with XtalPi, developing metabolically enhanced immunotherapy based on the core mechanism of immunometabolic reprogramming; related basic research findings and clinical results have been published in top international journals such as Nature, Nature Biotechnology, Nature Immunology, Lancet Haematology, and JAMA Oncology.

"A T cell in a terminally exhausted state can be understood as a worker who has been working continuously and is already very tired. At this point, if you paint him a rosy picture and give him positive stimulation, he no longer responds; you have to do metabolic regulation for him, let him rest a bit and recover his expansion capacity and sustainable killing ability," Tong Dudu said by way of analogy.

He told AI Tech Review that when this mechanism was first explained publicly, there was constant skepticism: Isn't IL-10 an immunosuppressive factor? How could it promote CAR-T's anti-tumor effect? But LeMone ultimately responded forcefully to the skepticism with real-world clinical data.

At the beginning of the clinical study, LeMone's research team planned to start at 5% of the conventional CAR-T dose and gradually escalate upward, but found that in the first dose group, after reinfusion, the peak expansion proportion of CAR-T cells in the body exceeded 80%, with excellent efficacy. The team then adjusted the clinical study protocol and explored lower doses downward: 2.5%, 1%, 0.25%, all the way down to 1‰.

Dose reduction directly rewrote the economics of CAR-T. The production cost of a single early CAR-T treatment was as high as five to six hundred thousand yuan, and the end price to patients reached about 1.2 million yuan per "shot." With technological advances and domestic substitution, CAR-T production costs across the industry have generally dropped to around two to three hundred thousand yuan. Through its low-dose strategy and production process optimization, LeMone has already kept costs under 100,000 yuan with manual production; LeMone has also independently developed a set of small, fully enclosed automated production equipment, which is expected to further reduce production costs to below 50,000 yuan in the future while greatly shortening the production cycle, providing room for CAR-T to become affordable and accessible.

At the registration and pipeline level, LeMone's pace accelerated markedly after it launched IND filings at the end of 2024: in the first half of 2026 it successively obtained two FDA IND clearances (for hematologic malignancies and autoimmune indications), and in September it received three CDE IND implicit approvals, completing dual China-U.S. filings. Beyond the first pipeline, its BCMA CAR-T targeting multiple myeloma and its solid tumor pipelines targeting liver cancer, gastric cancer, and glioma have all entered the IIT clinical research stage.

Its collaboration with XtalPi spans LeMone's entire chain from early design to clinical retrospective analysis. For affinity optimization of the metabolic enhancer factor IL-10, XtalPi's AI platform delivered the first batch of design results in just two weeks; in the solid tumor CAR-T project targeting colorectal cancer, faced with the subtle difference that "a tumor mutant antigen differs from the normal antigen by only one amino acid," XtalPi's AI model likewise, after iterative optimization, produced antibody sequences capable of high-precision specific binding and optimized selection.

More noteworthy is the application of AI in clinical sample analysis. LeMone is using AI to conduct high-throughput omics retrospective analysis of clinical samples from relapsed patients, attempting to answer an important question: why is it that with the same manufacturing process, some patients remain relapse-free for three years while others relapse after one year?

English text is being prepared.

The above data and judgments come from Tong Dudu, Vice President of Business at LeMone Bio. A computational chemistry background, he first was responsible at LeMone for AI pipeline R&D and liaison with XtalPi, and later shifted to business BD and financing. This experience spanning the laboratory and the negotiating table allows his answers to cover both technical and commercial logic.

01

A "Galen Prize" nomination and a "counterintuitive" discovery

Tong Dudu: The specific application was executed by a colleague on my team responsible for BD, and I participated in the entire application process, including material preparation. I think what the judges valued most was still the leapfrog improvement in efficacy brought by breakthrough technology. In fact, all the applying companies have excellent team backgrounds; the key is still whether the technology has sufficient original breakthroughs, and whether, after applying this technology to the pipeline, good results have already been seen in the clinic.

▎AI Tech Review: There are many companies doing CAR-T, and the "metabolic enhancement" label is also used by others. Where do you differ?

Tong Dudu: Most marketed CAR-T products are relatively early second-generation designs without various enhancing factors; later companies all use various methods to make the effect stronger, but different companies refer to different pathways. We add the IL-10 element, letting it secrete IL-10 autologously to reverse T-cell exhaustion. Immunometabolism is a very large field, and some companies are even the opposite of us—we enhance the immune system's capability, while some want to inhibit immune system activity; both are regulation of T-cell metabolism.

▎AI Tech Review: IL-10 is traditionally considered an immunosuppressive factor. How do you use it against tumors? How were external doubts dispelled?

Tong Dudu: In traditional understanding that is indeed the case, so when we first went out to present it, everyone would question: Isn't IL-10 immunosuppressive? How can you use it against tumors? It is very abnormal, counterintuitive.

But our research found that within a population of terminally exhausted T cells, IL-10 can reverse T-cell exhaustion through immunometabolic reprogramming. You can understand it as: a worker who keeps working is already extremely, extremely tired; at this point, drawing him a pie in the sky or giving him positive stimulation no longer works—you have to regulate his metabolism, let the CAR-T cells rest a bit and recover their proliferative capacity and sustainable killing ability.

This mechanism is actually somewhat disruptive, and currently very few teams at home and abroad are following up. In particular, for cell therapies that secrete IL-10, currently only we and our partner company are carrying out related R&D.

Early on, everyone indeed held an uncertain attitude and wanted to see more clinical data. As the IIT advanced, we disclosed 12 cases of data at the end of 2023, more data in 2024, and in 2025 expanded the indications to SLE and others; besides the earliest work in Hefei, later multi-center clinical trials in Hangzhou and elsewhere were also carried out, and all the result data were very good, so everyone gradually accepted this mechanism.

▎AI Tech Review: How was the 1‰ dose determined? Why can such a low dose still be effective?

Tong Dudu: Initially, following the conventional clinical protocol, we started dose escalation from 5% of the conventional dose, first using a low dose to ensure safety and gradually increasing after confirming patient tolerance. But after using 5%, we found that the peak expansion proportion of CAR-T cells in patients exceeded 80%, and the efficacy was very good.

Since 5% could achieve such a good effect, could lowering the dose also achieve the same result? So we explored all the way down—2.5%, 1%, 0.25%—and tried as low as 1‰. Going lower is of little significance, because some fixed costs in CAR-T production are unavoidable.

Because CAR-T cells are very special—they are a "living" drug, unlike small-molecule drugs that are metabolized and gone once ingested. After encountering tumor cells in the body, CAR-T cells activate and proliferate. Although the dose we administer is very low, their in vivo expansion capacity is very strong; at peak, both the absolute number and the proportion are no worse than with conventional doses—in fact, even higher. Their Cmax may be higher than the Cmax of the expansion curve of conventional CAR-T infused at 100% dose.

Tong Dudu: Very large. Our current manual process can already achieve a cost of under 100,000 yuan. We have also developed a small, fully enclosed automated production device that occupies very little space—10 units can fit in a single room—greatly reducing the amortized plant cost. Combined with process optimization, based on current estimates, automated production is expected to bring the cost down to below 50,000 yuan.

In the early days, the production cost of a single CAR-T treatment was as high as 500,000 to 600,000 yuan, and the end price to patients reached about 1.2 million yuan per "shot." Now, with technological advances and domestic substitution, the industry has generally reduced the production cost to 200,000 to 300,000 yuan per treatment. We can achieve under 100,000 yuan, or even under 50,000 yuan, while the production cycle is also much shorter. Dose reduction, combined with automated process optimization, has played a very key role.

Tong Dudu: After seeing very good clinical results for our first pipeline, we officially initiated IND filing at the end of 2024, completed CMC and toxicology/safety evaluation by the end of 2025, and obtained two FDA approvals in early 2026—one for an oncology indication and one for an autoimmune indication. In September this year, we also obtained three IND approvals from the CDE.

Tong Dudu: We have treated nearly 40 patients in total; only one was PR (partial response), and all the others were CR (complete response), for a complete response rate approaching 100%. In DLBCL, the 12-month complete response rate still remains at around 77.8%. Long-term durable efficacy is also very good.

Tong Dudu: Autoimmune diseases such as systemic lupus erythematosus are caused by the immune system attacking the body's own tissues, so deep immune reset can be achieved by clearing B cells to produce a therapeutic effect. Traditional CAR-T therapy requires lymphodepletion pretreatment (chemotherapy). Many lupus patients are young women, and losing all their hair from chemotherapy is very hard for them to accept. Moreover, autoimmune diseases are relatively chronic and do not progress as rapidly as cancer, so the safety requirements are actually higher. Our lymphodepletion-free regimen, combined with an extremely low dose, has very good safety.

The treatment logic for autoimmune diseases is also different from that for hematologic malignancies. In treating autoimmune diseases with CAR-T, all pathogenic B cells are cleared to achieve a deep immune reset. The newly generated B cells are relatively normal, so patients can remain in remission even after stopping treatment. But previous CAR-T therapies required pretreatment such as lymphodepletion; our lymphodepletion-free approach plus low dose is more tolerable for patients.

Tong Dudu: Our solid tumor CAR-T products targeting liver cancer, gastric cancer, and glioma have all entered the IIT clinical research stage; we have obtained IIT ethics approval and initiated clinical trials.

Tong Dudu: The headquarters is in Shenzhen, mainly because of the upstairs-downstairs ecosystem with XtalPi. Zhongshan is the production base, located at the first stop off the Shenzhen-Zhongshan Link. Hangzhou and Hefei are clinical centers—the hospitals for IIT collaboration at the time were selected in these two cities. The Hangzhou team is now relatively large and, in addition to clinical work, also undertakes part of the early R&D. We have set up a subsidiary in Switzerland to connect with overseas clinical and industrial resources, and are currently collaborating with Lausanne University Hospital on clinical research.

We have also received very concrete policy support in Shenzhen. We applied for the city's science and technology research program; the biomedicine and medical device task force helped us connect with clinical resources; and institutions such as the Greater Bay Area Drug Review Sub-center and the Guangdong-Hong Kong-Macao Greater Bay Area Clinical Trial Center have facilitated our communication with regulators. This is very critical for a startup company.

03

Tong Dudu: XtalPi's AI platform has played a very large role in our early drug design. For example, in the optimized design of IL-10, the most core metabolic enhancement factor, we obtained many mutants with greatly improved affinity through XtalPi's platform, applied for patents, and saw very good results in CAR-T cell validation.

Throughout the collaborative R&D process, we put forward requirements—for example, if we wanted the affinity of the IL-10 factor to become higher, we would submit that to XtalPi's computational team. They would use internal models, combined with the experimental data we provided, to fine-tune and output design results to us, and we would then conduct further validation. This accelerated our early drug design; at the time, we obtained the first batch of results in two weeks and directly selected some sequences for validation—as long as they were expressed, the affinity was very good. Without this process, it might have taken two months.

Tong Dudu: Yes, for example, in a solid tumor CAR-T project targeting colorectal cancer, solid tumors present an on-target off-tumor problem: the target recognized by CAR-T is also expressed at low levels in normal tissues, causing toxic side effects.

We prioritized a tumor mutation target—among colorectal cancer patients treated with first- and second-line antibody therapies, a large proportion of resistant patients have mutations in tumor antigens. Developing therapies against mutant antigens is safer. In this project, we collaborated with XtalPi to engineer and design based on existing antibody sequences, enabling it to recognize tumor mutant antigens with very high resolution while not recognizing the original antigen.

This is a very high-precision challenge: the difference between tumor mutant antigens and normal antigens is only a single amino acid. Traditional manual experimental screening may never be able to find them; AI is needed to explore a broader sequence space, find entirely new sequences, and then continuously feed wet-lab data back to the AI model for multiple rounds of optimization and iteration.

Tong Dudu: We are also now applying AI to retrospective analysis of clinical samples. For example, some patients experience relapse, and we need to answer: why is it that with the same manufacturing process, some patients remain in complete remission for more than three years, while others relapse after one year? Is it due to differences in the patient-derived cells, or differences in the in vivo immune microenvironment?

On the other hand, if a certain factor is found to cause relapse, targeted improvements can be made in the clinical protocol or manufacturing optimization. In our early research, we were more focused on discrete points—for example, knocking out a single factor in a pathway to indirectly validate a mechanism—but that was not systematic enough. Using higher-dimensional data for global analysis will give us a deeper understanding of the entire core technology. We have also recruited a new team for this direction.

04

Commercialization and Next Steps

Tong Dudu: I think the current stage is still relatively overheated. Early on, there were specialized models like AlphaFold; now large models and agents have emerged, which can connect the R&D chain together and improve efficiency. Companies that originally built AI infrastructure are also entering biopharma.

Tong Dudu: Our positioning is as an AI-empowered metabolism-enhanced immunotherapy company. CAR-T is the product category where we have the most implementation and the richest clinical data, but the technology itself is not limited to CAR-T—T-cell-related therapies such as TCR-T, TILs, and TCEs have expansion potential from a mechanistic standpoint.

Commercialization has two paths. One is the IND registration pathway; we have just obtained approval, and according to our plan, it will take about 3 to 4 more years to advance to commercialization. In between, for example after reaching Phase II clinical trials, it is very likely that we will license it out, or adopt a co-development approach and find a large MNC or a major domestic pharmaceutical company to help us commercialize it.

Another path is that after the implementation of the "818 Regulations," a pathway for the translation of new biomedical technologies has been provided—after our newer pipelines are validated in IIT clinical trials, there is an opportunity to directly conduct biotherapy at the clinical end.

Note: The "818 Regulations" implemented in 2025 (the Regulations on the Administration of Clinical Research and Translational Application of New Biomedical Technologies) provide a translational application channel other than IND for new technologies such as cell therapy, allowing medical institutions to lead clinical research and translational application under specific conditions.

▎AI Technology Review: How is Leimang's BD progress? How do MNCs view your technology?

Tong Dudu: MNCs already recognize the technology, but after all they have decision-making costs, so they want to see that we can be used across different tumor indications, or that our hematologic tumor product has already advanced to a later stage so that after acquisition it can quickly advance to market. Currently some MNCs have given TS (Term Sheet, letter of intent to invest); if the solid tumor data are good, I believe they will have a stronger willingness to collaborate.

Tong Dudu: The clinical results in autoimmune diseases will be disclosed fairly completely, and if things go quickly, the results can be published in clinical journals; formal Phase I clinical trials for hematologic tumors and autoimmune diseases will soon be launched and steadily advanced; there will be some preliminary data for solid tumors in the coming year. These milestones will also support financing; our financing has always been small steps with rapid progress, and we maintain continuous contact.