Drug Developers at PCOC: As AI Speeds Discovery, Testing and Access Gaps Remain

The final session on Day 1 of Patient-Centered Oncology Care® (PCOC) 2026 in Nashville, Tennessee, asked a panel of drug developers: What gives you hope?

For several, hope was distilled in a moment shared in Chicago a few months prior, when they stood and cheered for daraxonrasib.

At the 2026 American Society of Clinical Oncology (ASCO) Annual Meeting in May, the plenary audience rose for 42 seconds when the Kaplan-Meier curves from the phase 3 RASolute 302 trial (NCT06625320) appeared on screen.¹⁻⁴ Among 500 patients with previously treated metastatic pancreatic ductal adenocarcinoma, the oral RAS(ON) multiselective inhibitor nearly doubled median overall survival vs chemotherapy (13.2 vs 6.7 months).¹^,² It was a result many had anticipated, but the visual of the evidence represented more than just a successful trial.

For David Spigel, MD, president and chief medical officer of Sarah Cannon Research Institute, the daraxonrasib tale served as an example of what is possible in drug discovery, and he referenced it as he moderated the panel, “The Changing Face of Drug Development: How, Where, and Why.”

For Peg Squier, MD, PhD, being in the lecture hall for the ovation marked a milestone, as she recalled her days working in a cancer center. “The very first patient I ever put on a clinical trial was a pancreatic cancer patient,” said Squier, now group vice president of US medical affairs for hematology-oncology at Incyte Corporation. “It was with a drug that, like all the other drugs for so many years, didn't work,” she said. “And I think it's really important to acknowledge the significance of a breakthrough that fantastic for a disease that miserable.”

On August 26, 2026, FDA approved daraxonrasib, to be sold as Rasonque by Revolution Medicines, as the first-in-class targeted therapy for adults with metastatic pancreatic adenocarcinoma, after at least 1 prior systemic therapy or who cannot receive multiagent therapy. As the panel would discuss, the news from ASCO had patients lining up for the drug, and FDA approved it 6.5 months ahead of its goal date.⁵

As Akshara Richhariya, PhD, head of oncology portfolio marketing at Pfizer, explained, this is the challenge today: modern tools and regulatory flexibility allow for more speed than ever. And yet, some patients will always be savvier and gain access before others. Once the ASCO data hit the front page, patients wanted the drug, but payers would not cover it. “What do you do in that distance between the drug approval [and] when the FDA approval comes?” she asked.

The drug did not appear in guidelines from the National Comprehensive Cancer Network (NCCN) until September 11, 2026, following FDA approval.⁶ 

From 20 molecules to “in silico”

Squier recalled that a decade ago, a fast discovery cycle meant chemists handing 20 new molecules to biologists every couple of weeks. Today, companies are building AI capabilities and partnering with AI firms to mine internal and external data for new targets.

“We used to talk about in vitro and then in vivo. Now we talk about in silico,” Squier said. “So, you go through additional orders of magnitude of potential molecules before the chemists get busy, and so the idea is by the time you’re actually synthesizing something, it's an optimized molecule, and so you can reduce the number of cycles, compress the number of cycles, and make better decisions earlier.”

Spigel said his team had recently raised the idea of “ghost trials,” which simulate a study before launch. “Could we simulate how that trial would perform in an artificial world?” he said. “Not how they would perform in terms of efficacy or toxicity, because you don't know those things. But could you even find the patients? What do you think would be the likelihood that they would stay on the study for how many cycles?” Such a simulation could refine a study quickly, he said, but “it's a make-believe scenario, and so how do you know any of that even matters? But I think we have the tools now to do these things pretty easily.”

Who owns the data?

Ramon Tiu, MD, head and vice president of oncology clinical services at Takeda Oncology, called data “one of the most important currencies” in drug development and said ownership shapes how AI can be used to find trial patients. Hospital electronic medical records are a rich source, he said, but “oftentimes the hospital would love to keep those data for patients within their own system so that the patients stay in.”

The same tension applies to molecular testing results. “If it is the company [that] has that particular data set, the main focus would be to try to identify particular patients that would be amenable to therapies that are available for the company,” Tiu said. “While on the other hand, if it is the patient who actually owns the data, then they can have the ability to be able to use that data; they can offer this up to a specific platform, allowing … therapies that are coming from different companies to be applicable for them.”

“I've always thought the patient owns the data, and the data come from the patient,” Spigel said.

Richhariya agreed, but said testing, health record, trial, and payer data sit at different points in the patient journey. “Data is king, but it's only good or useful when you can connect these disparate pieces. And I think that's one of the biggest challenges,” she said.

Vicky Brown, head of US oncology at Boehringer Ingelheim, warned that AI models are only as good as their inputs. “While we’re still not at the place where all patients are tested at the time of their diagnosis, … we're basing our hypothesis on those that are already tested, and are we then biasing our viewpoints?” she asked. “Are we designing trials for those where we know more about it, rather than all these patients that are not tested?”

Getting new therapies into the community

As trials target narrower biomarker-defined subsets, Brown said results must reach the record before treatment or trial decisions are made. “Testing is great, but if no one is looking for those results and making those treatment decisions based on it, those patients are not identified, and even if they are, they're not benefiting from those therapies that come through,” she said.

Richhariya said trials still happen largely in academic centers, where testing is readily available, even though some estimates say up to 80% of patients are treated in the community. “The biggest challenge we have right now is how do we make these trials and these drugs, even if they’re approved, available to patients who are in the community,” she said. “Pharma has a responsibility here as well to see what we can do to really step up and start that process much sooner—not wait for the drug to be approved before the community starts getting to use it or get to know about it.”

Tiu said decentralized trials that rely on telemedicine and digital devices are not yet ready for oncology at full scale. “In cancer patients, it's a much sicker population, very fragile patients. The types of therapies that are being used are much more toxic in some ways,” he said, favoring a hybrid approach with local laboratory work and monitoring close to home.

Squier said she has seen progress in simpler observational and investigator-initiated studies. “I think we're getting closer to fulfilling the promise, which is it doesn't matter where you are,” she said. “But yeah, it's baby steps.”

The business case for rare subsets

Asked whether investors will keep funding therapies for small populations, Squier pointed to the targeted therapy imatinib (Gleevec; Novartis). “I have some mentors who were involved with the development of Gleevec, and they were scolded by some investors who told them that it was a much narrower patient population than colon, lung, or breast cancer, and that they would never make any money,” she said, to rumbling in the room.

“Those kinds of successes really demonstrated that profound benefit to a narrower patient population is every bit as viable a commercial story as a marginal benefit to a broader patient population.”

Brown said advocacy groups and registry data help ensure that “every eligible patient counts” in rare-population trials. “Being able to bring advocacy and patients to the table before a go/no-go decision—I think in my mind, then that business case or that case for it solves itself.”she said. “It’s really understanding that little rare tumor type or biomarker that you go into first; then when you show that, it opens up all these other areas.”

Access between data and approval

Spigel used daraxonrasib to frame the access question, noting that he had prescribed it that day for a patient with lung cancer after it received an NCCN Compendium listing about 10 days earlier. “It's been a crazy summer, right? The data came out in May or so, and then we have a drug that's now available to patients,” he said. “It's a great story. It's just everybody sees that news and wants access, and how do you do that?”

FDA issued a “safe to proceed” letter for a daraxonrasib expanded access treatment protocol this spring, 2 days after receiving the request.^5,7 Richhariya said such programs matter before payers are willing to offer reimbursement. “I feel very strongly about EAPs. I think that's an incredibly important program to make sure that drugs that really have impact on patients, patients are not waiting for them for that long,” she said. “But there as well, I think there is this disparity, because people who use those EAP programs more are the ones who are aware of the product. They are aware of the trial, and the trials are happening in the academic centers.”

Even after approval, she said, testing remains a hurdle. “If the patient is not even identified as eligible for a drug, that's a different kind of barrier. It's not a reimbursement or cost barrier. It's a barrier in diagnosis.”

Reasons for hope

Tiu cited the convergence of new technology, new approaches, and regulatory change. “As much as there's a lot of uncertainty because people may not be very familiar with it, I think it's helping accelerate drug development,” said Tiu, who worked on Takeda’s approval for rusfertide (Mimrylo), which treats erythrocytosis in adults with polycythemia vera.⁸

“Ultimately, I think this would lead to better drugs that will be approved, more appropriate ones, and I think this will be a win for patients.”

“What gives me hope is the fact that there isn't a them and an us,” Brown said. “Our biggest challenge now that we can't do alone, we need to do together, is how do we ensure that amazing science gets in the hands of those that need it—not just some of them, but all of them.”

Richhariya said the same tools driving discovery can fix delivery. “How do we make sure that all patients have equal access to products that can help them? They have access to trials, no matter where they live, whether they live in a big city or they live in rural America,” she said. “The tools that we have today will really empower us to solve the problems that we've been grappling with on the delivery side.”

References

1. O'Reilly EM, Wainberg ZA, Hendifar AE, et al. Daraxonrasib or chemotherapy in previously treated metastatic pancreatic cancer. N Engl J Med. 2026;395(4):325-337. doi: 10.1056/NEJMoa2605555
2. Wolpin BM, et al. Daraxonrasib, a RAS(ON) multi-selective inhibitor vs chemotherapy in previously treated metastatic pancreatic adenocarcinoma (mPDAC): primary and final analysis from the phase 3 RASolute 302 study. Presented at: 2026 American Society of Clinical Oncology Annual Meeting; May 31, 2026; Chicago, IL. Abstract LBA5.
3. Caffrey M. RASolute 302 brings a "transformative" moment in pancreatic cancer: a 60% improvement in overall survival. AJMC. May 31, 2026. Accessed November 2, 2026. https://www.ajmc.com/view/rasolute-302-brings-a-transformative-moment-on-pancreatic-cancer-a-60-improvement-in-overall-survival
4. Daraxonrasib nearly doubles survival in previously treated metastatic pancreatic cancer. The ASCO Post. June 2026. Accessed November 2, 2026. https://ascopost.com/news/june-2026/daraxonrasib-nearly-doubles-survival-in-previously-treated-metastatic-pancreatic-cancer/
5. FDA approves first-in-class targeted therapy for metastatic pancreatic cancer. News release. FDA. August 26, 2026. Accessed November 2, 2026. https://www.fda.gov/news-events/press-announcements/fda-approves-first-class-targeted-therapy-metastatic-pancreatic-cancer
6. National Comprehensive Cancer Network. NCCN Clinical Practice Guidelines. Pancreatic adenocarcinoma. Version 4.2026. September 11, 2026. Accessed November 2, 2026. https://www.nccn.org/professionals/physician_gls/pdf/pancreatic.pdf
7. FDA permits expanded access for investigational pancreatic cancer drug. News release. FDA. April 30, 2026. Accessed November 2, 2026. https://www.fda.gov/news-events/press-announcements/fda-permits-expanded-access-investigational-pancreatic-cancer-drug
8. Takeda receives US FDA approval of Mimrylo (rusfertide), marking a potential shift in the treatment paradigm for polycythemia vera. News release. Takeda. August 28, 2026. Accessed November 2, 2026. https://www.takeda.com/newsroom/newsreleases/2026/fda-approval-mimrylo/