AstraZeneca reported that sonesitatug vedotin, also known as Sone-Ve and previously designated AZD0901 or CMG901, produced a statistically significant and what the company described as highly clinically meaningful improvement in overall survival in the global Phase 3 CLARITY-Gastric01 trial. The investigational CLDN18.2-directed antibody-drug conjugate was tested against investigator-selected therapy in patients with previously treated, locally advanced or metastatic gastric cancer, gastroesophageal junction cancer or esophageal adenocarcinoma whose tumors expressed CLDN18.2 in at least 25% of tumor cells.
The result is important because overall survival is the clearest patient-outcome endpoint in advanced cancer. It also provides AstraZeneca with the first pivotal survival evidence from sonesitatug vedotin after earlier-stage studies had shown antitumor activity but could not establish comparative benefit.
However, the topline announcement is not an uncomplicated clean sweep. The trial met its overall survival primary endpoint in the third-line and later population and a key secondary overall survival endpoint across the second-line and later population. The other primary endpoint, blinded independent central review-assessed progression-free survival in the broader population, showed a favorable trend but did not reach statistical significance.
That endpoint split gives the program a credible regulatory foundation, but it also makes the full dataset essential. AstraZeneca has not yet disclosed the hazard ratios, median survival values, confidence intervals, p-values, duration of follow-up, response rates, treatment exposure or detailed safety tables. Until those data are presented at a medical meeting, the magnitude, consistency and clinical context of the reported survival advantage cannot be independently assessed.
Why does the overall survival result matter even though progression-free survival missed?
Overall survival carries particular weight in later-line gastric and gastroesophageal junction cancer because patients have aggressive disease, limited treatment options and relatively short expected survival after progression on established regimens. Demonstrating that patients lived longer in a randomized Phase 3 trial can therefore be more clinically persuasive than showing only a radiographic delay in tumor progression.
AstraZeneca said the survival benefit was statistically significant in the third-line and later primary analysis and in the broader second-line and later population. This suggests that the result was not confined to the most heavily pretreated subgroup, although the exact size and consistency of benefit across treatment lines remain undisclosed.
The PFS miss nevertheless matters. A therapy that improves overall survival without meeting a progression-free survival endpoint may still be clinically valuable, but the apparent disconnect requires explanation. The eventual presentation will need to show whether the survival curves separated early or late, whether the proportional-hazards assumption held, how censoring was handled, whether subsequent therapies differed between the study arms and whether results were consistent across regions, comparator choices and biomarker-expression levels.
None of those explanations should be assumed from a press release. The PFS result could reflect the statistical design, radiographic assessment dynamics, treatment sequencing or genuine differences between control of measurable disease and longer-term survival. It could also become less puzzling once the actual effect sizes and survival curves are available.
For regulators and clinicians, the decisive issue will not be that one endpoint missed in isolation. It will be whether the complete efficacy and safety package supports a favorable benefit-risk profile in a clearly defined patient population.
What does the Phase 3 trial design reveal about the strength and limits of the evidence?
CLARITY-Gastric01 is a randomized, open-label, sponsor-blinded, global Phase 3 study. The trial enrolled 592 participants and was conducted across 175 centers in 19 countries.
During the dose-selection stage, patients were randomized equally to sonesitatug vedotin at 2.2 mg/kg every three weeks, 1.8 mg/kg every three weeks or investigator’s choice of therapy. The confirmatory stage continued with the selected 2.2 mg/kg dose.
Randomization and blinded central assessment of progression-free survival strengthen the comparative evidence, although the open-label treatment assignment can still influence treatment management, discontinuation decisions and the reporting of subjective adverse events. Overall survival is less vulnerable to assessment bias than PFS, but it can be affected by post-progression treatment and regional differences in access to subsequent therapies.
The investigator-choice control arm also has practical and analytical consequences. Permitted options included regimens such as ramucirumab plus paclitaxel, taxane monotherapy, irinotecan, trifluridine-tipiracil and apatinib, depending on geography and prior treatment. This makes the study more representative of the fragmented later-line treatment landscape, but it may also create heterogeneity in control-arm outcomes.
The full publication will need to clarify the distribution of comparator treatments and whether the survival advantage remained consistent against the main control categories. A favorable result driven primarily by one comparator or one region would have different implications from a broadly consistent effect.
The evidence is currently topline rather than peer-reviewed. AstraZeneca has said that the results will be presented at a forthcoming medical meeting and shared with global regulatory authorities. The announcement does not establish approval, determine the final label or resolve how regulators will interpret the missed progression-free survival co-primary endpoint.
Could the broader CLDN18.2 threshold materially expand the eligible population?
The most commercially and clinically distinctive part of the program may be its biomarker threshold. CLARITY-Gastric01 enrolled patients whose tumors expressed CLDN18.2 in at least 25% of tumor cells at any staining intensity.
AstraZeneca estimates that this definition may cover roughly 60% of gastric and gastroesophageal junction cancers. If confirmed through the full efficacy dataset and an approved diagnostic strategy, the threshold could make sonesitatug vedotin relevant to a substantially broader population than therapies requiring high-intensity, high-percentage CLDN18.2 expression.
That expansion is biologically plausible because an antibody-drug conjugate delivers a cytotoxic payload after target binding and internalization. Sonesitatug vedotin combines a humanized anti-CLDN18.2 antibody with a cleavable linker and the microtubule-disrupting payload monomethyl auristatin E.
Earlier translational work described direct cytotoxicity, antibody-dependent cellular cytotoxicity, complement-dependent cytotoxicity and bystander killing as potential contributors to its antitumor activity. These mechanisms provide a rationale for studying the ADC across different levels of target expression, but they do not prove that low-expression tumors receive the same clinical benefit as high-expression tumors.
The Phase 3 presentation will therefore need to show results by CLDN18.2 expression level, staining intensity, anatomical site and treatment line. If benefit is concentrated in higher-expression subgroups, the practical population could be narrower than the trial’s inclusion rule suggests. If survival benefit remains consistent down to the 25% cutoff, the result would support a substantially broader precision-oncology proposition.
The companion diagnostic is central to that argument. The study is evaluating the clinical performance of the Ventana SP455 assay for identifying patients who may benefit from sonesitatug vedotin. Reproducibility across laboratories, tissue quality, scoring consistency, turnaround time and concordance between archival and newly collected samples could all influence real-world patient identification.
How does sonesitatug vedotin fit into a market already shaped by zolbetuximab?
CLDN18.2 is no longer an experimental target without regulatory validation. The United States Food and Drug Administration approved Astellas Pharma’s zolbetuximab with fluoropyrimidine and platinum chemotherapy in October 2024.
The approved indication covers first-line treatment of adults with locally advanced unresectable or metastatic HER2-negative gastric or gastroesophageal junction adenocarcinoma whose tumors are CLDN18.2 positive under an approved companion diagnostic. The associated Ventana assay uses a cutoff of at least 75% viable tumor cells demonstrating moderate-to-strong membrane staining.
Sonesitatug vedotin is not a direct replica of that approach. It is an ADC rather than a cytolytic monoclonal antibody, it is being developed initially from a later-line randomized trial, and CLARITY-Gastric01 used a materially broader expression definition.
Those differences could position the drug for patients who are not eligible for high-expression first-line zolbetuximab therapy, patients whose disease progresses after earlier treatment, or patients for whom another targeted later-line option is needed. The current Phase 3 result does not establish any of those potential uses as an approved indication.
AstraZeneca is also moving the asset into first-line development. The Phase 3 CLARITY-Gastric02 study is evaluating sonesitatug vedotin with capecitabine, with or without the PD-1 and TIGIT bispecific antibody rilvegostomig, in CLDN18.2-positive, HER2-negative advanced gastric, gastroesophageal junction and esophageal adenocarcinoma.
That trial began dosing in February 2026 and has an estimated enrollment of 2,130 participants. It represents a much larger and longer-term test of whether the ADC can move earlier in treatment and compete within an increasingly biomarker-fragmented first-line market.
Competition is also developing among other CLDN18.2-directed ADCs, including IBI343, SHR-A1904 and tecotabart vedotin. Peer-reviewed early-phase studies have reported antitumor activity for these programs, but their designs, biomarker thresholds, payloads, doses and patient populations differ substantially.
Cross-trial response rates cannot establish superiority. AstraZeneca’s advantage is that sonesitatug vedotin now has a reported Phase 3 overall survival result, while the durability, safety and regulatory significance of that result still require full disclosure.
What does the safety language reveal, and what remains undisclosed?
AstraZeneca said sonesitatug vedotin was well tolerated, that its safety profile was consistent with prior experience and that no new safety signals were identified. That wording is reassuring at a high level, but it is not a substitute for event rates, severity, treatment discontinuations, dose modifications, deaths and exposure-adjusted analyses.
The published Phase 1 KYM901 study provides useful historical context. Among 107 patients in the gastric or gastroesophageal junction dose-expansion cohort, 68% experienced grade 3 or worse treatment-emergent adverse events and 50% experienced serious adverse events.
The most common grade 3 or 4 events included decreased neutrophil count, anemia and vomiting. One treatment-related death was reported. The confirmed objective response rate across the relevant 2.2 to 3.0 mg/kg gastric and gastroesophageal junction population was approximately 28%, with a median follow-up of nine months.
Those Phase 1 findings cannot predict the exact Phase 3 safety profile because the patient mix, dosing, follow-up and comparator framework differ. They do indicate which questions deserve attention when the pivotal data are released.
Clinicians will want to see rates of neutropenia, anemia, gastrointestinal toxicity, peripheral neuropathy, infection, treatment discontinuation and treatment-related mortality. Because the payload is MMAE, cumulative tolerability and the ability to maintain dose intensity could be particularly relevant for patients already weakened by several rounds of treatment.
“No new safety signals” means that the disclosed follow-up did not reveal a previously unrecognized pattern. It does not mean that the treatment is free from clinically important toxicity. The practical question will be whether the survival improvement is sufficiently large and consistent to justify the treatment burden compared with available later-line regimens.
Why will companion diagnostics and biomarker consistency influence regulatory review?
The regulatory package will involve more than the medicine. Because treatment eligibility depends on CLDN18.2 expression, AstraZeneca and Roche Diagnostics will need to support the analytical and clinical performance of the Ventana SP455 assay used in the development program. The trial’s efficacy analyses are intended to contribute to that validation.
A lower expression threshold may increase the number of identifiable patients, but it can also make assay precision near the cutoff more important. Small differences in tissue handling, staining intensity or pathologist interpretation could move a sample above or below eligibility.
Regulators may examine whether the 25% threshold was prespecified, how missing or unevaluable samples were handled, whether outcomes were robust across central and local testing and whether the assay performs consistently across gastric, gastroesophageal junction and esophageal adenocarcinoma specimens.
The existing zolbetuximab companion diagnostic creates useful clinical awareness because many oncology centers already recognize CLDN18.2 testing. However, the assays, scoring systems and therapeutic cutoffs are not interchangeable.
A patient who is negative under the 75% moderate-to-strong criterion could potentially qualify under the sonesitatug vedotin trial definition. Only a future approved assay and product label can determine whether that broader eligibility applies in clinical practice.
What does the result mean for AstraZeneca’s ADC strategy and investor sentiment?
AstraZeneca controls global development, manufacturing and commercialization rights to sonesitatug vedotin through an exclusive license from KYM Biosciences, a joint venture owned by Keymed Biosciences and Lepu Biopharma.
The original agreement included a $63 million upfront payment, up to $1.1 billion in development and sales milestones and tiered royalties reaching the low double digits. A further $45 million milestone was triggered in early 2026 after first-patient dosing in CLARITY-Gastric02.
Keymed Biosciences said AstraZeneca now estimates potential peak sales of $3 billion to $5 billion for the product. That range should be treated as a company forecast rather than an established market outcome.
Reaching it would require regulatory approvals, a sufficiently broad biomarker-defined label, reliable diagnostic adoption, competitive first-line results, manufacturing scale and successful positioning against established regimens and emerging CLDN18.2 therapies.
The market reaction to AstraZeneca on July 27 was positive but cannot be attributed solely to CLARITY-Gastric01. London-listed shares rose 1.7% to close at £128.88 on a day when the company also reported second-quarter financial results that exceeded profit expectations.
The closing price remained approximately 18% below AstraZeneca’s £157.32 52-week high, while Reuters reported that the shares were down about 7% for 2026.
AstraZeneca reported second-quarter revenue of $15.38 billion, core earnings per share of $2.63 and 15% growth in oncology sales. The company maintained its 2026 outlook and its ambition of reaching $80 billion in annual revenue by 2030.
For investors, sonesitatug vedotin is best viewed as a meaningful late-stage oncology de-risking event rather than a completed commercial victory. The survival result supports AstraZeneca’s wider ADC expansion and offers a potential new growth asset. The missed PFS endpoint, undisclosed effect sizes and untested first-line commercial strategy prevent the topline announcement from resolving the asset’s full value.
Which data will determine whether sonesitatug vedotin becomes a new treatment standard?
The forthcoming medical presentation must answer several questions at once. The hazard ratios and absolute median overall survival improvement will show whether AstraZeneca’s description of a highly clinically meaningful benefit is supported by a substantial patient-level gain.
The PFS curves and landmark analyses may help explain the failed endpoint. Subgroup results will indicate whether the reported survival advantage extends across second-line and later settings, lower and higher CLDN18.2 expression, geographic regions, comparator regimens and the three included anatomical cancer types.
Detailed safety tables will determine whether the treatment could reasonably displace existing later-line chemotherapy rather than merely add another toxic option. Diagnostic analyses will show whether the 25% threshold can be implemented consistently across pathology laboratories.
Regulatory authorities will also need to assess how the successful overall survival analyses and unsuccessful PFS co-primary endpoint fit within the trial’s prespecified statistical hierarchy and multiplicity framework.
The topline result moves sonesitatug vedotin materially closer to regulatory consideration and validates CLDN18.2-directed ADC development at the Phase 3 level. It does not yet establish an approved therapy, an exact eligible population or a new standard of care.
AstraZeneca’s strongest case will emerge only if the full dataset shows a sizable survival advantage, an acceptable treatment burden and consistent efficacy across the broader biomarker population that makes the program clinically and commercially distinctive.
