Astellas’s R&D head talks strategy ahead of looming patent cliff

Cambridge, Massachusetts—Japanese pharmaceutical company Astellas Pharma is one in a spate of big drugmakers staring down an impending problem: the patent cliff, when blockbuster drugs lose their time-limited chokehold on the market and generic versions can undercut their sales.

Xtandi, a prostate cancer treatment, brought Astellas nearly $4.6 billion in the first 9 months of the Japanese fiscal year 2025. Xtandi will lose a key patent in the US next year. The loss is expected to deal a significant blow to Astellas’s overall revenue. But the company, headquartered in Tokyo and with research and development (R&D) centers around the US, has already been preparing.

Chief R&D officer Tadaaki Taniguchi tells C&EN that Astellas has concentrated its drug development engine in four focus areas: immuno-oncology, targeted protein degradation, genetic regulation, and blindness and regeneration.

“This is our strength, from basic research to clinical research or even commercialization phase,” Taniguchi says. “So this is a perfect fit for us to bring up more product to differentiate any other company that’s working on this space.”

At the Astellas Life Sciences Center in Cambridge, Massachusetts, Taniguchi spoke to C&EN about a few drug candidates that represent the new era of Astellas’s pipeline. ASP2138 and ASP546C both target the transmembrane protein Claudin 18.2 with the goal of treating gastric and pancreatic cancers. ASP2138 is a bispecific antibody, while ASP546C is an antibody-drug conjugate (ADC). ASP3082, or setidegrasib, is a small-molecule protein degrader targeting KRAS G12D for pancreatic cancer—potentially competing with Revolution Medicines’ daraxonrasib, another KRAS-targeting drug that just showed promising results in a Phase 3 pancreatic cancer trial.

On the cell- and gene-therapy side, Taniguchi points to AT845, an adeno-associated virus (AAV) gene therapy for Pompe disease, and ASP7317, a cell therapy for geographic atrophy.

Taniguchi discussed how they fit into Astellas’s strategy and how Astellas’s scientists are using artificial intelligence. This interview has been edited for length and clarity.

It’s a really good time for us to connect, because it really seems that Astellas is at an inflection point. I would love to hear, at a high level, what your R&D strategy is right now.

The Xtandi patent is coming off next year, starting with US. Of course, there’s some different timelines for Europe and Japan, but I think we need to really look at how to accelerate our pipeline development. We started a focus-area approach.

We’re progressing 3082 now—it’s called setidegrasib—very quickly in both pancreatic cancer and also lung cancer. We just started a first-line pancreatic cancer study in combination with chemo.

That’s the Phase 3 trial?

That’s the Phase 3 trial, right. Revolution Medicines recently announced their positive data from their trial targeting second-line or later-line pancreatic cancer. This is also encouraging that a pan-KRAS product is actually showing benefit in treating pancreatic cancer. So this is also encouraging for us to continue to actually push the speed of our trial, more focusing on first-line pancreatic cancer.

Setidegrasib also has huge opportunity, not just in pancreatic cancer, but also in the lung cancer area. We continue to work on bringing our pan-KRAS degrader, which is ASP5834, it’s now in Phase 1 trials in US and Japan.

The expectation for KRAS-targeted products is actually increasing. We differentiate our product based on our science: our products degrade; other companies are working on inhibitors. So this is a great differentiation opportunity.

There’s a lot of excitement coming up. I haven’t talked about biotechnology, but I think not just progressing our internal portfolio from research to clinic but also thinking about partnering, biotechnology, we can actually generate more meaningful product for the patient. We need to have more treatment options.

I wanted to ask about your licensing strategy, focusing a little bit on two recent deals, one with Dyno Therapeutics in the gene-therapy space, and the other with Vir Biotechnology in the cancer space. How do those two deals fit into your broader licensing strategy, and can we expect more deals like that going forward?

Looking back, what we have done in the initial phase of our BD [business development] strategy is actually creating internal capability by acquiring some companies. We acquired Ocata [Therapeutics] in Westborough here [in Massachusetts] to actually try to get more capability to develop our cell-therapy product. Same for gene therapy; we acquired Audentes [Therapeutics] around, I think, 2019 or 2020. We actually have a foundation of gene therapy such that we can actually generate more product. And then same for Universal Cells, which is actually located in Seattle. With that, we can actually generate more cell-therapy products much easier.

Vir Biotechnology has amazing technology. We can eliminate unnecessary safety concerns. This product is targeting prostate cancer. And this is a perfect match for us. We know how bispecific T-cell engagers work, we know who we need to work with for prostate cancer, and we know how to develop the product.

Dyno is slightly different. There are more technology platforms for AAV. . . . In gene therapy, you need to really specifically think about which organ you target and how to access it. Dyno technology will really allow us to go directly to the target with this AAV.

It sounds like you’ve already done a lot of the deals that are forming the foundation of your R&D pipeline. Are you looking to do additional licensing and mergers and acquisitions (M&A)?

I don’t think we are always looking at M&A. If it makes sense, we could do it, but I think our current BD strategy is more filling our portfolio pipeline, because I think we are progressing pretty well in the early-stage programs coming to the late stage.

When you are looking for external partnerships, are you still looking in those four target spaces, or are you potentially expanding beyond that?

Looking at these primary four classes is really our priority for probably the next, at least, 3 to 5 years. But at the same time, the science is progressing, right? Science is progressing, and particularly in these area, you have a lot of opportunities.

We are always evaluating a lot. We have this SakuLab, which is basically open innovation space that we can invite biotechs to come into, or our partners, like Mass General Brigham. So this is a perfect place to take a look at how to see the progress of science, that we can tap into that to expand in the future.

Pivoting a little bit, I would be remiss if I didn’t ask how you’re using AI in your drug discovery and development at the moment.

This is a hot topic now. We have been working with Nvidia’s supercomputing system. One product is already in the clinic. There’s some pros and cons and limitations of the AI to generate a new product, but once you hit, it’s very fast. To bring this product to the clinic, it took less than 2 years.

Recently we also announced a new technology called the Protein Station. [There’s an] opportunity to expand our capability in large molecules like antibodies and bispecifics and so on. The Protein Stations have automations generating the ideal product design but also doing proteomics and finding out which is the best molecule to actually move forward.

How are your researchers adapting to using AI tools?

They love it.

[Laughing] Do they really?

We don’t want to scale up all of a sudden. So we want to limit access to this and have, let’s say, 30 or 50 people working on it. But other researchers are willing to access it. They want to try, because they are researchers; they like innovation. They want to try something new. So that’s what we see. The researchers are actually learning how to use it and creating their own applications.

So each researcher is creating their own individual interface, basically?

That’s what we see. The researchers, sometimes, are also good at the technology side. Now AI can easily generate programming to create a small application. So they do, but we need to sort it out. You know, I don’t want to create 100 apps, right? That would be chaotic. So the digital team is actually sorting out which is working best.

Are you finding that there are specific programs or types of molecules that seem to work better with AI than others?

I think simpler molecule is better. Small molecule is best. Now we try simple large molecules, like antibodies, using this Protein Station.

For AI, ADCs are too complicated. AI is not there yet. We probably need AI to have more information, data to learn how to create the most-optimized product. Now, I think human researchers are doing a much, much better job.