The defining question of the next decade is not how many drug candidates can be generated, but who decides which ones move forward, and why

At the 2026 Annual Meeting of the American Society of Clinical Oncology in Chicago, results from the phase III HARMONi-6 trial were presented at the plenary session and published simultaneously in The Lancet.¹ The HARMONi-6 trial tested a novel bispecific antibody that engages PD-1 and VEGF (ivonescimab), given with platinum-doublet chemotherapy to patients with previously untreated advanced squamous non-small cell lung cancer. 

The comparator was an antibody that engaged PD-1, but not VEGF (tislelizumab) plus the same chemotherapy, and ivonescimab appeared to be superior, with a median overall survival of 27.9 months compared to 23.7 months (hazard ratio for death 0.66, suggesting the average improvement in OS is considerably larger than 4.2 months).

Results from the HARMONi-6 trial were widely reported, as this was the first regimen that improved OS compared to an anti-PD-(L)1 antibody plus chemotherapy in a randomized trial for first-line treatment of advanced squamous non-small cell lung cancer. In previous years, the presentation and subsequent publication would typically be associated with the sponsor’s submission of a marketing application to the US Food and Drug Administration and to other international agencies such as the European Medicines Agency. While ivonescimab is approved for use in China,² it is not yet available to lung cancer patients in the United States or Europe.

Why not? 

The HARMONi-6 trial was conducted entirely in China, at 50 hospitals; 93% of the 532 randomized patients were men, and all had squamous cell carcinoma, a histology strongly associated with tobacco exposure.¹ This calls into question the applicability of the results to lung cancer patients in other parts of the world, particularly the U.S. where non-smoking related adenocarcinoma is the predominant lung cancer type. Another challenge to interpreting the HARMONi-6 results is the choice of control arm. Tislelizumab plus chemotherapy is considered standard of care in China, and tislelizumab is approved in Europe and Australia for first-line squamous cell lung cancer. In the United States, however, while tislelizumab is approved for several other tumor types, it is not approved for this indication, either as a single agent or in combination.³

Further complicating the interpretation of this study, results were reported earlier this year from the HARMONi-3 trial, a multiregional phase III trial comparing ivonescimab plus chemotherapy against pembrolizumab plus chemotherapy—a control arm that would be considered acceptable by the U.S. FDA.⁴ In contrast to the HARMONi-6 trial, the separately powered cohort of patients with squamous cell carcinoma in HARMONi-3 did not cross the efficacy threshold at an interim analysis of progression-free survival conducted on a minimal alpha allocation. The independent data monitoring committee recommended that the trial continue as designed and remain blinded.

Collectively, the clinical benefit of ivonescimab remains promising, but it is not completely clear that these benefits will extend to patients in the United States, or other Western economies, compared to their current standards of care. Indeed, the clinical development of ivonescimab is illustrative of both the opportunities and the pitfalls of cancer drug development in a global market.

More medicines but more complexity

More new cancer drugs are reaching more patients than at any point in the history of oncology. At the same time, the path from a promising molecule to a patient who actually receives it has never had more moving parts.

The scale of the cancer drug development enterprise is now well documented. Kang and Ji recently examined early-stage drug development programs worldwide—each defined as the development of a specific drug for a specific indication, at the discovery, preclinical, or phase 0–2 stage, catalogued in a commercial development database and attributed to the headquarters country of the originating company.⁵ Across all diseases, the number of such programs nearly doubled between 2015 and 2024, from roughly 10,400 to about 19,000, and the U.S. share fell from 48.2% to 37.4%.

The shift is even more dramatic for cancer drugs. Early-stage cancer programs rose from 3,720 to 11,115 over the same period; although U.S.-origin programs substantially grew (from 1,811 to 4,051) during this time, they fell as a share of the global total from 48.7% to 36.5%, largely driven by the major increase in China-origin programs from 381 to 4,152, in turn increasing China’s portion from 10.2% to 37.4%. China now originates more early-stage development programs for cancer drugs than the United States. In an accompanying editorial, Richard Pazdur and Steven Usdin observed that in development of antibody-drug conjugates, bispecific antibodies, and cell therapies, China now rivals or surpasses the United States in scale and, in some areas, in innovation.⁶

A cross-national analysis by Friends of Cancer Research reported that from 2020 through 2025, FDA approved 87 novel oncology drugs and China’s National Medical Products Administration approved 94.⁷ From 2023 onward, NMPA’s annual volume exceeded FDA’s. FDA remained faster—review times a median of 182 days shorter, approvals a median of 164 days sooner—and first-in-class approvals remained predominantly at FDA. Of those 181 approvals across six years, only 14 drugs were approved by both agencies. Some of that divergence reflects differences in epidemiology, in unmet need, and in the standards of care against which a new agent must be judged, while some of it reflects where a sponsor chose to run the trial.

The coming challenges in cancer drug development

The use of artificial intelligence is generating candidate drug targets and molecules faster than any prior method, and in greater volumes than the downstream system can absorb. A generation ago, a well-equipped screening operation might evaluate a few hundred compounds against a target in a week. The constraint today is not the number of plausible structures. It is the number of chemists who can synthesize them, the number of pre-clinical models that can reliably triage them, and, perhaps most importantly, the number of patients and clinical trial sites available to test them.

This means that the defining question of the next decade is not how many drug candidates can be generated, but who decides which ones move forward, and on what basis. In practice, that decision is made company by company, and it is driven substantially by commercial opportunity: novelty of the target, size of the population, unmet medical need, competitive density, likelihood of reimbursement. But this calculus reliably rewards risk reduction—the third or fourth agent against a validated target in a more narrow indication—over the novel, unvalidated mechanism of action that might have greater impact.

The current environment sustains this paradigm. When a highly innovative small company spins out of an academic laboratory, the principals quickly learn what they must demonstrate to justify the next round of investment. Unfortunately, the direction chosen is not typically “prove the new biology.” It is more often “find a niche indication where the regulatory path is proven and predictably short.” If one can imagine these conversations occurring among boardrooms across the biotechnology ecosystem, the result is cancer therapy opportunities that look enormous, but can be remarkably narrow in some instances.

Patients will bear the cost if we are not efficient

Patients will ultimately bear the cost when the same scientific question is asked and answered multiple times in multiple jurisdictions because drug sponsors and regulators are not aligned on the trial designs. They bear it when a dose that was never optimized produces toxicity that limits treatment tolerance and duration of therapy. They bear it when a development program for a novel agent is abandoned not because the biology failed but because the commercial case was not compelling or the large-scale manufacturing challenges could not be overcome.

Intrinsic and extrinsic factors affecting the acceptability of foreign clinical data—differences in patient populations, in medical practice, and in health systems—are a real scientific issue, addressed in ICH E5(R1) and revisited regularly since.⁸ Nonetheless, fragmentation has a cost, and that cost is paid in patient-years.

A field with that many degrees of freedom and no forum for candid, cross-sector discussion becomes more expensive, slower, and less equitable and accessible to patients. 

That cost is not hypothetical, and the same molecule we used to introduce this story illustrates both the cost and the remedy. Ivonescimab was first tested in patients with EGFR-variant nonsquamous NSCLC whose disease had progressed on a tyrosine kinase inhibitor in HARMONi-A, a randomized, double-blind phase III trial conducted at 55 centers, all of them in China. An interim progression-free survival analysis supported NMPA approval in May 2024.⁹ The final overall survival results, published in JAMA in June, showed a median of 16.8 months with ivonescimab plus chemotherapy versus 14.1 months with chemotherapy alone (hazard ratio 0.74).¹⁰ 

The investigators were candid about the constraint: because the trial was conducted exclusively in China, generalizability to non-Asian populations remains unproven, and they pointed to the ongoing global HARMONi study as the trial that may resolve it. They also noted that clinical practice in this patient population had changed during the study’s conduct, so the population in which the question is now clinically relevant is not quite the population that was enrolled.

That global study—HARMONi, sponsored by Summit Therapeutics, which licensed ivonescimab outside China—asks essentially the same question in a multiregional population, and Summit has expanded the study from a single-region design. An updated analysis reported consistent overall survival benefits in Western and Asian patients, and the resulting Biologics License Application was accepted for filing in January 2026, with an action date of November 14, 2026.¹¹ 

Whether that sequence represents appropriate regulatory caution or avoidable duplication is a question worth discussing, and reasonable people in this field might answer it differently. Nonetheless, roughly 30 months separate the Chinese approval from the U.S. action date, and patients outside China have been waiting.

One opportunity to convene, debate, and learn: the AAADV Workshop

Guidance documents, advisory committees, and consensus papers each provide valuable information and perspectives. What none of them does is to enable direct dialogue—a biostatistician asking a regulator a question about trial design; a company medical director describing, candidly, a dose-optimization decision that turned out to be wrong; the sponsors of drugs from the same class sharing and highlighting common problems; an investigator hearing how an identical data package was received in Silver Spring, Amsterdam, and Beijing; or a patient advocate saying out loud that a multi-year gap between approval in one country and approval in her own is not acceptable to the people she represents.

The unscripted exchange—the questions that arise on the fly during discussion, and the recognition that a common problem may exist while individuals in the room each grappled with it individually—is what the AAADV (Accelerating Anticancer Agent Development and Validation) Workshop has enabled since 2004. The workshop, held in collaboration over the years with the National Cancer Institute, FDA, AACR, ASCO, and multiple stakeholders, has supported and enabled these conversations. In 2025 the AAADV Workshop reorganized as an independent 501(c)(3) foundation in order to broaden access, expand virtual participation, and support year-round educational programming.

Twenty years ago the key features of a registration program where the trial runs, what the control arm is, which primary endpoint, which agency sees it first—were largely settled questions. Today the selection of drug, target, biomarker assay, patient population, control treatment, clinical endpoint, and region of the world is more complex than ever. A field with that many degrees of freedom and no forum for candid, cross-sector discussion becomes more expensive, slower, and less equitable and accessible to patients.

The 2026 AAADV Workshop will convene November 5–6, 2026, at the North Bethesda Marriott Hotel & Conference Center, in person and virtually. The program takes up exactly the questions this article raises: multiregional clinical trials and global development strategy; benefit-risk assessment in regulatory decision-making, including the sufficiency of a single pivotal trial and the role of post-market evidence; dose optimization for novel agents, combinations, radiopharmaceuticals, and cell therapies; “me too or not me too,” discussion on follow-on agents, biosimilars, and novel formulations; failure to launch, a case-based examination of what FDA complete response letters reveal and teach us; and immune-related adverse events and their implications for treatment tolerance and acceptability. Because of its increasing importance to the field, an evening session is devoted specifically to cancer drug development in China.

Keynote addresses will be delivered by Monica Bertagnolli, president of the National Academy of Medicine; Robert Califf, former FDA Commissioner; Peter Marks, former director of FDA’s Center for Biologics Evaluation and Research; and Thomas G. Roberts, Jr., of Farallon Capital Management. Richard Pazdur, former director of FDA’s Oncology Center of Excellence, will join one of us (RLS) for a fireside chat on the globalization of cancer drug development. Registration and the full program are available at www.AAADV.org.

The question for the cancer care community today is not whether the scientific opportunities remain. It is whether the global development system can convert innovative science into benefit for patients everywhere, within a reasonable time, without requiring each region to relearn what another has already established. 

Many of these questions will get worked out in rooms where the people who design, conduct, review, and fund these trials talk to one another, and to patients and their advocates directly—including about the things that did not work.

References

1. Lu S, Liu B, Luo Y, et al. Ivonescimab plus chemotherapy versus tislelizumab plus chemotherapy in advanced squamous non-small-cell lung cancer (HARMONi-6): interim overall survival analysis of a randomised, double-blind, phase 3 trial in China. Lancet. 2026;407(10548):2620–2629. doi:10.1016/S0140-6736(26)00966-9
2. National Medical Products Administration, People’s Republic of China. Ivonescimab injection approved for marketing by China NMPA. (Approval for EGFR-mutated non-squamous NSCLC after EGFR-TKI therapy.) February 19, 2025. Accessed July 30, 2026. https://english.nmpa.gov.cn/2025-02/19/c_1073516.htm
3. TEVIMBRA (tislelizumab-jsgr) injection, for intravenous use. Prescribing information. BeOne Medicines USA, Inc.; revised December 2025. Accessed July 30, 2026. https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=08ef1e3e-496f-4b0b-94ee-fbba3cc1985a
4. Summit Therapeutics Inc. Summit Therapeutics reports financial results and operational progress for the first quarter ended March 31, 2026. April 30, 2026. Accessed July 30, 2026. https://smmttx.com/news/press-releases/news-details/2026/Summit-Therapeutics-Reports-Financial-Results-and-Operational-Progress-for-the-First-Quarter-Ended-March-31-2026/default.aspx
5. Kang SY, Ji Y. Geographic shifts in early-stage biopharmaceutical innovation. JAMA. 2026;335(15):1355–1356. doi:10.1001/jama.2026.1962
6. Pazdur R, Usdin S. The geography of pharmaceutical innovation: the US and China in a new world order. JAMA. 2026;335(15):1309–1310. doi:10.1001/jama.2026.2390
7. Collins G, Pandita D, Andrews H, et al. A cross-national review of novel oncology approvals in the United States and China (2020-2025). Health Aff Sch. 2026;4(7):qxag183. doi:10.1093/haschl/qxag183
8. International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use. Ethnic factors in the acceptability of foreign clinical data: E5(R1). February 5, 1998. Accessed July 30, 2026. https://database.ich.org/sites/default/files/E5_R1__Guideline.pdf
9. HARMONi-A Study Investigators. Ivonescimab plus chemotherapy in non-small cell lung cancer with EGFR variant: a randomized clinical trial. JAMA. 2024;332(7):561–570. doi:10.1001/jama.2024.10613
10. HARMONi-A Study Investigators. Bispecific antibody ivonescimab added to chemotherapy in EGFR-variant non-small cell lung cancer: the HARMONi-A randomized clinical trial. JAMA. 2026;336(4):306–314. doi:10.1001/jama.2026.7745
11. Summit Therapeutics Inc. Ivonescimab plus chemotherapy shows consistent, favorable overall survival results in Western and Asian patients in updated analysis from global Phase III HARMONi study. July 22, 2026. Accessed July 30, 2026. https://smmttx.com/news/press-releases/news-details/2026/Ivonescimab-Plus-Chemotherapy-Shows-Consistent-Favorable-Overall-Survival-Results-in-Western-and-Asian-Patients-in-Updated-Analysis-from-Global-Phase-III-HARMONi-Study/default.aspx