Ivonescimab Before the FDA’s November 14 Decision: What Summit’s HARMONi Data Show

The FDA’s target date for deciding whether to approve ivonescimab is November 14, 2026. The drug, licensed by Summit Therapeutics (SMMT) from China’s Akeso, is filed with chemotherapy for patients with EGFR-mutant non-squamous non-small cell lung cancer (NSCLC) whose disease has progressed after EGFR-targeted pills; according to Summit’s 10-K, no PD-1-based bispecific antibody has been approved in the US. The difficulty is on record. Summit reports that the FDA said “a statistically significant OS benefit is necessary to support marketing authorization in this setting”, and the overall survival result of Summit’s HARMONi trial missed significance at its main analysis (Summit 10-Q for the second quarter of 2026; press release, May 30, 2025).

The stakes are large for a company with no approved product. At the October 16 close of $14.63, Summit’s 799.7 million shares were worth about $11.7 billion, or about $13.3 billion counting the 109 million shares that AstraZeneca’s new preferred stock converts into (SMMT on BioRadar; our arithmetic). Chairman and co-CEO Robert Duggan controls 71.8% of the stock. On October 16, a November 20 call and put at the $15 strike together cost about 32% of the share price (BioRadar options data), a measure of how large a move the options market expects by then from all news, not only the FDA decision.

Jump to the three scenarios ↓

Key points

• The FDA’s bar is overall survival, and the formal test missed it. In HARMONi’s April 2025 analysis, OS had a hazard ratio of 0.79 with p=0.057, against the 0.0448 the statistical plan required. Two later analyses (HR 0.78 and 0.76) carry only “nominal” p-values: they support the trend but are not formal tests.
• Most of the evidence comes from China, and it is not independent twice. 273 of HARMONi’s 438 patients were enrolled in China in 2022 and were also part of Akeso’s HARMONi-A trial, so HARMONi-A does not confirm HARMONi independently. In the 165 North American and European patients, the latest OS hazard ratio is 0.76 with a 95% confidence interval of 0.52 to 1.10.
• The first market is a few thousand patients a year and already contested. Summit cites more than 14,000 eligible US patients; our illustrative estimate from public inputs is about 6,500, with a range of about 3,400 to 12,300. Amivantamab plus chemotherapy was approved in the same setting in 2024.
• The money is now there, the control stays with Duggan. Summit had $690.7 million on June 30 and flagged substantial doubt about its ability to continue as a going concern; AstraZeneca’s $2.0 billion closed on October 5. Akeso is owed up to $4.6 billion in milestones plus low double-digit royalties.

The drug and the deal

Summit licensed ivonescimab from Akeso in December 2022 for the US, Canada, Europe and Japan for $500 million upfront, adding Latin America, the Middle East and Africa in 2024 for $15 million. Akeso kept China, where the drug has three approvals: in May 2024 for the same EGFR-mutant setting, in April 2025 alone as first-line treatment for PD-L1-positive NSCLC, and in August 2026 with chemotherapy in first-line squamous NSCLC (Summit 8-K, December 6, 2022; Summit 10-K for 2025; Akeso announcements). The shares jumped 194% on the day the license was announced and 272% on May 30, 2024, when Akeso’s HARMONi-2 trial in China beat Keytruda on PFS; they peaked at $36.70 in April 2025 and fell 31% on May 30, 2025, when HARMONi’s survival result came in short of significance (Yahoo Finance closes; Summit 8-Ks).

What HARMONi showed

HARMONi enrolled 438 patients whose EGFR-mutant non-squamous NSCLC had progressed on a third-generation EGFR TKI. They received carboplatin and pemetrexed chemotherapy plus either ivonescimab or placebo, double-blind, with PFS and OS as the two primary endpoints. The 273 Chinese patients enrolled in 2022; the 165 in North America (93) and Europe (72) from May 2023 to October 2024. Summit’s 10-K says the patients enrolled in China were also part of Akeso’s HARMONi-A trial (Lancet Oncology 2026; Summit 10-K for 2025; Summit WCLC 2025 presentation).

Progression-free survival was clearly better. At the primary PFS analysis, median PFS was 6.8 months with ivonescimab against 4.4 months with placebo, HR 0.52 (95% CI 0.41 to 0.66). That analysis had a July 2024 cutoff, before western enrollment finished: 273 of its 345 patients (79%) were Chinese, and only 18 of 275 PFS events came from western patients. With all 438 patients, the HR was 0.57 overall and 0.67 (0.45 to 1.00) in the western group. The response rate was 45% against 34% (Summit WCLC 2025 presentation; Lancet Oncology 2026).

Overall survival trended the same way but did not clear the formal bar.

Swipe the table sideways to see every column.

Sources: Summit press releases of May 30, 2025, September 7, 2025, July 22, 2026 and September 15, 2026. “Nominal” means the p-value is reported for information, outside the pre-planned tests and their error budget.

Three details matter for reading the table. About 215 of the 262 deaths at the April 2025 analysis (82%) were in Chinese patients, whose data have not changed since; the later improvement comes from longer follow-up of the western group, and the medians, 16.8 against 14.0 months, stayed the same in all three cuts (our arithmetic from Summit’s forest plots). The western confidence interval still includes 1.0, so the western data alone do not establish a survival benefit; a subgroup of 165 patients was never designed to, and its estimate has moved toward the overall result. And because HARMONi-A’s significant OS result (HR 0.74, p=0.02) comes largely from the same Chinese patients, it supports consistency within that population rather than adding an independent second trial (JAMA 2026; Summit 10-K for 2025).

Side effects were higher with ivonescimab. Serious treatment-related adverse events occurred in 28% of patients against 15% with placebo; grade 3 or worse events linked to VEGF blockade (such as high blood pressure, protein in the urine and bleeding) in 7.3% against 3.2%; severe treatment-related bleeding in 0.9%. Treatment-related deaths were 1.8% and 2.3% (Summit WCLC 2025 presentation; Lancet Oncology 2026).

What the FDA will weigh

The bar it set. Summit has repeated the FDA’s position in every filing since May 2025. From the first-quarter 2026 10-Q on, it adds that the FDA said HARMONi’s PFS results “may not be sufficient” for approval, a sentence that is not in the 2025 10-K. Summit filed anyway, saying patients can benefit “despite the lack of a statistically significant overall survival benefit”, and points to HARMONi-A’s OS result (Summit 10-Q for the first quarter of 2026; Summit 10-K for 2025).

The precedent cuts both ways. In September 2024 the FDA approved Johnson & Johnson’s (JNJ) amivantamab with the same chemotherapy in the same setting on a PFS benefit; at the prespecified OS analysis “there was no statistically significant difference in OS”, HR 0.73 (FDA approval notice, September 19, 2024). The FDA has not said publicly why it asked more of ivonescimab. Two pieces of context: PD-1 drugs added to chemotherapy have failed in this setting before (Merck’s KEYNOTE-789: PFS HR 0.80 and OS HR 0.84, neither significant), and a Chinese trial of a PD-1 antibody with a VEGF antibody and chemotherapy, ORIENT-31, improved PFS (HR 0.51) but not OS (HR 0.98) (J Clin Oncol 2024; Lancet Respir Med 2023).

Data from one country. The FDA can approve on foreign data if they apply to US patients and can be verified, as Summit’s 10-K describes. In 2022 it did not approve Lilly and Innovent’s PD-1 antibody sintilimab and “recommended an additional multiregional clinical study be performed” (Lilly 10-Q for the first quarter of 2022). HARMONi is multiregional, but most of its patients and events are Chinese, and they were shared with another trial.

Process. The review is on the standard clock, and Fast Track was granted in 2024. As of October 19, no advisory committee meeting for ivonescimab had been announced on the FDA’s calendar or in the Federal Register, where the FDA must give at least 15 days’ notice of one (21 CFR 14.20). Summit sent the June 2026 OS update to the agency during the review; any extension would require the FDA to treat a submission as a major amendment, which can move the goal date by three months, and none has been disclosed (Summit press release, January 29, 2026; press release of July 22, 2026).

The first market: smaller than the headline

Summit cites “over 14,000 U.S. patients eligible for treatment each year in this setting” but does not show its arithmetic (press release, January 29, 2026). Our illustrative estimate multiplies public rates, and three of the six inputs are our assumptions, shown so readers can change them:

Swipe the table sideways to see every column.

Sources: ACS, SEER, Lin et al. 2022, Sabari et al. 2025. The Sabari figure counts patients still on first-line treatment at the data cutoff, so it does not give the second-line share directly.

Summit’s figure is above our high case. Not all of these patients would be candidates for ivonescimab either: AstraZeneca estimates that about 34% of tumors develop high MET levels after osimertinib, and its osimertinib-plus-savolitinib combination improved both PFS and OS in that group in the SAFFRON trial (AstraZeneca, August 17, 2026). These are scenario numbers, not a validated forecast, and the range is not a confidence interval.

The competition in this setting:

Swipe the table sideways to see every column.

Sources: FDA approval notices; Summit filings; NEJM 2026; AstraZeneca 20-F for 2025 and press release of August 17, 2026; Merck 10-K for 2025. Separate trials cannot be ranked against each other.

What it could be worth. At Medicare’s October 2026 payment limit, which is the average sales price plus 6% and not the maker’s net price, amivantamab at the maintenance dose for a patient under 80 kg costs about $358,000 a year for the drug alone; a year of Keytruda costs about $216,000 (CMS payment limit file; our arithmetic). Suppose ivonescimab were priced at a flat monthly equivalent of amivantamab’s, about $29,900, and patients stayed on it for seven months, an assumption rather than a measured duration (HARMONi’s median PFS was 6.8 months). A quarter of our base case would then mean about $340 million of sales a year and half about $680 million. Even the higher figure would be modest next to a market value of $11.7 billion, which suggests investors are mainly paying for first-line lung cancer and other tumors; the filings cannot say how investors weigh them.

The bigger prize: first-line lung cancer and the race

Akeso’s Chinese trials have produced the class’s only head-to-head wins. In HARMONi-2, ivonescimab alone beat Keytruda alone in first-line PD-L1-positive NSCLC on PFS (HR 0.51) and, at an interim analysis, on OS (HR 0.73, p=0.009). In HARMONi-6, ivonescimab plus chemotherapy beat tislelizumab plus chemotherapy in squamous NSCLC on OS (HR 0.66, one-sided p=0.0017). Both were run only in China (Summit press release, September 13, 2026; May 31, 2026).

The global tests are still ahead. HARMONi-3 compares ivonescimab plus chemotherapy with Keytruda plus chemotherapy in about 1,600 first-line patients; enrollment is complete, and Summit has guided the squamous PFS readout, with an early OS look, for the second half of 2026 and the non-squamous readout for the first half of 2027. An earlier squamous interim look in the second quarter of 2026 did not stop the trial. HARMONi-7, ivonescimab against Keytruda in patients with high PD-L1, has a primary completion date of April 2028 (Summit Q2 2026 release and 10-Q; ClinicalTrials.gov).

The prize explains the crowd. Keytruda’s main US patent expires in 2028, Merck expects biosimilars “could begin in December 2028”, and Keytruda was 49% of Merck’s 2025 sales (Merck 10-K for 2025). Bristol Myers Squibb with BioNTech ($1.5 billion upfront plus $2.0 billion in fixed payments), Pfizer ($1.25 billion), AbbVie ($650 million) and Merck ($588 million) have each paid for their own PD-(L)1 x VEGF bispecific; none has a head-to-head result yet, and their main comparative trials are due from 2028 on (company 10-Ks for 2025; ClinicalTrials.gov). Ivonescimab’s lead is real but measured in years, and the data that would establish it outside China are not in yet.

The money behind it

Summit spends more every quarter. Cash used in operations rose from $61 million in the first quarter of 2025 to $141 million in the second quarter of 2026, as HARMONi-3, HARMONi-7 and a colorectal trial enrolled. Cash and short-term investments were $690.7 million on June 30, and the 10-Q said Summit’s situation raised “substantial doubt” about its ability to continue as a going concern, the fourth 10-Q since mid-2025 with that language. In June, insiders bought $52.2 million of shares through the company’s at-the-market program, including Duggan’s 3.81 million shares at $13.12 on June 12, two days after Summit withdrew a $500 million share offering “due to market conditions” (Summit 10-Q for the second quarter of 2026; 8-K, June 11, 2026; Forms 4).

AstraZeneca changed the picture. On September 28 it agreed to buy $2.0 billion of convertible preferred stock at the equivalent of $18.36 a common share, about 19% above that day’s close, together with trials combining ivonescimab with its own cancer drugs; the deal closed on October 5 (Summit 8-K, September 29, 2026; AstraZeneca 6-K, October 5, 2026). The terms worth knowing:

• Conversion into about 109 million common shares, roughly 12% of the company, needs a shareholder vote on November 4 to double the authorized shares and antitrust clearance. Duggan’s stake alone is enough to pass the vote. If the needed change to the charter is not in effect by April 5, 2028, Summit must buy the preferred back in cash at market value.
• Governance: AstraZeneca votes with the board’s recommendation, has a one-year lock-up and, if the board engages with a takeover proposal, gets 10 business days to make its own offer (securities purchase agreement).

Adding the AstraZeneca money and later share sales to the June balance, and assuming third-quarter spending near the second quarter’s, gives roughly $2.6 billion to $2.7 billion in early October. At the second-quarter burn rate that would last more than four years, though spending is still rising and a launch would add to it (our arithmetic; the third-quarter 10-Q is not yet filed).

Three more claims on the value:

• Akeso. Summit still owes up to $4.555 billion in milestones ($1.05 billion regulatory, $3.5 billion commercial) and low double-digit royalties on sales. The license lists a payment for full US approval; its amount is blacked out, and its definition of “full” excludes accelerated approval. Akeso is also the only supplier of the drug substance until Summit sets up other sources, and the 10-K flags the BIOSECURE Act, tariffs and data-transfer rules as China-related risks (Summit 10-Q for the second quarter of 2026; license agreement; 10-K for 2025).
• Control. The proxy statement reports beneficial ownership, which includes options exercisable within 60 days: Duggan 71.8%; his wife, co-CEO Maky Zanganeh, 7.3%, of which about 23.6 million options (her shares alone are about 4.5%); and Akeso 4.0%, counted under its chief executive, Summit director Michelle Xia. Directors and officers together report 81.8%; counting shares only, about 19% of the stock is held outside that group, and Summit is a Nasdaq “controlled company” (proxy statement, October 19, 2026; our arithmetic).
• Dilution. 118 million stock options were outstanding on June 30, many at strikes of $1.06 to $1.68; with them and AstraZeneca’s conversion shares, the share count would pass 1 billion. A new $380 million at-the-market program had $354 million left on October 1 (Summit 10-Q for the second quarter of 2026; proxy statement).

Where this could go: three scenarios

This section is analysis, not a forecast. It describes paths the filed facts make possible and what each would require. We do not assign probabilities.

1. Approval, with or without conditions

The FDA reads HARMONi’s PFS benefit, the improving survival trend and the Chinese data as enough, and approves ivonescimab with chemotherapy. The label may be broad (“after EGFR TKI therapy”) or limited to patients treated with a third-generation TKI, and it may come with post-marketing requirements. A full approval would trigger Akeso’s US-approval milestone; an accelerated approval would not, though nothing in Summit’s filings shows the FDA has offered that route. What to look for: the label’s patient definition, any required confirmatory trial, the price, and the share move against what the options market priced.

2. A longer review

The FDA treats new information, such as answers to late questions or further analyses, as a major amendment and moves the goal date by three months, to around mid-February 2027. The program stays alive; the AstraZeneca money covers the wait, and the HARMONi-3 squamous data could arrive before the new date. What to look for: an 8-K announcing a new goal date, and whether Summit says what the FDA asked for.

3. A complete response letter

The FDA holds to the overall survival bar it set in 2025, possibly also citing the share of Chinese data, and does not approve. The AstraZeneca money gives Summit time, and the company’s US future would then rest on HARMONi-3 and HARMONi-7 in first-line lung cancer. What to look for: the reasons in the letter (survival, regional data or both), whether the FDA asks for a new trial, the HARMONi-3 squamous readout and Summit’s spending plans.

What to watch

• November 4: the shareholder vote on authorized shares, needed for AstraZeneca’s conversion.
• The third-quarter 10-Q: cash, spending and whether the going-concern language goes away after the AstraZeneca money.
• November 14: the FDA’s action, the label wording and any approval conditions.
• HARMONi-3 squamous PFS and early OS, guided for the second half of 2026.
• Competitor data in the same setting: AstraZeneca’s TROPION-Lung15 (guided for the second half of 2026) and a possible US filing for osimertinib plus savolitinib.
• Antitrust clearance for the AstraZeneca conversion and any new share sales.

Bottom line

Ivonescimab reaches the FDA with a clear PFS benefit and a survival trend that has strengthened with follow-up, but without the statistically significant survival result the FDA said it needed, and with most of its evidence from Chinese patients shared with another trial. The first market is a few thousand US patients a year next to an approved competitor; most of Summit’s value rests on first-line lung cancer, where the global data are still to come. AstraZeneca’s $2.0 billion removed the near-term funding question; control stays with Duggan.

FAQ

When is the FDA decision on ivonescimab?

The FDA’s target action date (PDUFA goal date) is November 14, 2026, for ivonescimab plus chemotherapy in EGFR-mutant non-squamous NSCLC after EGFR TKI therapy. The FDA can act earlier; as of October 19, 2026, no advisory committee meeting had been announced.

Did ivonescimab improve survival in HARMONi?

It improved progression-free survival (HR 0.52). Overall survival favored ivonescimab (HR 0.79) but missed statistical significance at the main analysis (p=0.057 against a required 0.0448); later analyses showed HR 0.78 and 0.76 with nominal p-values only.

Why does the FDA want an overall survival benefit?

Summit reported in May 2025 that the FDA said a statistically significant overall survival benefit is necessary for approval in this setting, and later that PFS results may not be enough. The FDA has not explained its reasons publicly; PD-1 drugs have failed to improve survival in this setting before.

Who controls Summit Therapeutics and how much cash does it have?

Chairman and co-CEO Robert Duggan reports beneficial ownership of 71.8% of the shares and his wife, co-CEO Maky Zanganeh, 7.3%. Summit had $690.7 million on June 30, 2026, and received $2.0 billion from AstraZeneca on October 5, 2026.

Sources

• Summit press release, January 29, 2026: FDA accepts the BLA, PDUFA date November 14, 2026; January 12, 2026: BLA submitted
• Summit press release, May 30, 2025: HARMONi topline and the FDA’s OS position
• Summit press release, September 7, 2025 (WCLC 2025) and presentation
• Summit press release, July 22, 2026: updated HARMONi OS; September 15, 2026: WCLC 2026 OS tables
• Le, Passaro et al., Lancet Oncology 2026: HARMONi; JAMA 2026: HARMONi-A overall survival
• Summit 10-Q for the quarter ended June 30, 2026; 10-Q for the first quarter of 2026; 10-K for 2025
• Summit 8-K, September 29, 2026: AstraZeneca investment, press release and securities purchase agreement; proxy statement, October 19, 2026; proxy statement, April 17, 2026
• Summit 8-K, June 11, 2026: offering withdrawn; prospectus supplement, July 23, 2026: at-the-market program
• Summit 8-K, December 6, 2022: Akeso license; license agreement
• Summit press release, September 13, 2026: HARMONi-2 OS; May 31, 2026: HARMONi-6 OS
• FDA approval notice, September 19, 2024: amivantamab with carboplatin and pemetrexed; June 23, 2025: datopotamab deruxtecan
• Lilly 10-Q for the first quarter of 2022: sintilimab; KEYNOTE-789, J Clin Oncol 2024; ORIENT-31, Lancet Respir Med 2023; OptiTROP-Lung04, NEJM 2026
• AstraZeneca press release, August 17, 2026: SAFFRON; AstraZeneca 20-F for 2025
• Merck 10-K for 2025; Bristol Myers Squibb 10-K for 2025; Pfizer 10-K for 2025; AbbVie 10-K for 2025
• American Cancer Society: lung cancer key statistics, 2026; SEER stat facts; Lin et al., JTO Clin Res Rep 2022; Sabari et al., J Thorac Oncol 2025
• CMS Medicare Part B payment limit files, October 2026; 21 CFR 14.20
• ClinicalTrials.gov: NCT06396065 (HARMONi), NCT05899608 (HARMONi-3), NCT06767514 (HARMONi-7)

Sources read October 19 and 20, 2026. Prices are Yahoo Finance and BioRadar daily closes through October 16, 2026. The options figure is BioRadar’s snapshot at 17:09 UTC on October 16 of the November 20 call and put at the $15 strike (mid prices $2.18 and $2.53) against a share price of $14.75. Market estimates are our own illustrative arithmetic from the sources above. This article is for information only and is not investment advice. The author held no SMMT shares when this was written.

BioRadar analyses take no bull or bear side, make no buy or sell calls and are never paid for. Read our editorial policy.