AZD0901

AZD0901 (international nonproprietary name: sonesitatug vedotin) is AstraZeneca's antibody-drug conjugate (ADC) aimed at Claudin-18.2 (CLDN18.2), a protein that normally sits sealed between stomach lining cells but gets exposed on the surface of most gastric cancers. AstraZeneca paid $63 million upfront plus up to roughly $1.1 billion in milestones to in-license the compound (then called CMG901) from China-based Keymed Biosciences in February 2024 [1]. The deal followed positive Phase 1 data from Keymed's KYM901 study, which is the most direct precedent for the program (see Status) [15]. The lead Phase 3 study, NCT06346392, randomized 592 patients with second-line or later metastatic gastric or gastroesophageal junction (GEJ) adenocarcinoma expressing CLDN18.2, comparing AZD0901 monotherapy to investigator's choice chemotherapy [2]. The trial is active but no longer recruiting, with progression-free survival (PFS) as the primary endpoint. This matters commercially because Astellas already proved the target works. Zolbetuximab (Vyloy), a naked anti-CLDN18.2 antibody added to chemotherapy, won first-line FDA approval in October 2024 after the SPOTLIGHT and GLOW Phase 3 studies hit overall survival [3]. AstraZeneca is betting that bolting a cytotoxic payload onto the antibody produces a stronger drug, one that either competes in first-line or salvages patients who progress on zolbetuximab.

Novel compound, first regulatory exposure pending. The Phase 3 second-line monotherapy study (NCT06346392, n=592) has closed enrollment, and a Phase 2 monotherapy study supporting that program (NCT07143604, n=33) is also active and not recruiting [2][4]. AstraZeneca has pushed AZD0901 into perioperative resectable gastric cancer (NCT07069712, Phase 2, recruiting) and a large Phase 3 combination program with capecitabine plus or minus rilvegostomig, AstraZeneca's PD-1/TIGIT bispecific antibody, sized at 2,130 patients (NCT07431281) [5][6]. CMG901 received FDA fast track designation in April 2022 for monotherapy in relapsed/refractory metastatic gastric/GEJ cancer, a designation that carries forward under the AZD0901 name [16]. There is also a CLARITY-PanTumor01 Phase 2 basket study (NCT06346392-adjacent) extending CMG901 into other CLDN18.2-expressing solid tumors [17]. Critically, CMG901 was not a clinical blank slate at the time of the AstraZeneca in-license. The Keymed KYM901 Phase 1 study generated dose-escalation and dose-expansion data in Chinese patients with advanced gastric/GEJ cancer: in the confirmed CLDN18.2-positive dose-expansion cohort, the objective response rate (ORR) was 33 percent (95% CI 23.0 to 43.3 percent) with a disease control rate of 70 percent; the most common grade 3 or higher adverse events were neutrophil count decreased (18.6 percent) and anemia (13.3 percent), and the most common any-grade events were anemia (62.8 percent), vomiting (57.5 percent), and hypoalbuminemia (57.5 percent) [15]. Those numbers, not the Western Phase 2 study, set the efficacy bar AstraZeneca was paying for. Based on enrollment completion timing for NCT06346392, a Phase 3 PFS readout is plausible in 2026 to 2027, with overall survival data following later. The combination Phase 3 readout will come later still given enrollment scale. AstraZeneca's 2024 annual report and subsequent pipeline disclosures position the CLDN18.2 franchise as a meaningful oncology growth contributor, though specific peak-sales guidance for AZD0901 is not broken out separately from total oncology revenue [7].

Claudin-18.2 is a protein that normally acts like glue between cells in the stomach lining, sealing them together so digestive juices do not leak through. In gastric and GEJ cancers, the same protein gets exposed on the cell surface where antibodies can reach it. Roughly 30 to 40 percent of gastric tumors express CLDN18.2 at levels considered targetable [8]. AZD0901 is a humanized anti-CLDN18.2 IgG1 antibody conjugated via a protease-cleavable linker to monomethyl auristatin E (MMAE), with a drug-to-antibody ratio (DAR) of approximately 4 [15][18]. The antibody binds CLDN18.2 on tumor cells, the complex is internalized, and the cleavable linker releases MMAE inside the lysosome. MMAE is a microtubule-disrupting cytotoxin (the same payload used in Adcetris/brentuximab vedotin, Polivy/polatuzumab vedotin, and Padcev/enfortumab vedotin), so its toxicity profile is well characterized: peripheral neuropathy (nerve damage in hands and feet) and low blood counts are the standard problems [9]. A cleavable linker plus a membrane-permeable payload like MMAE produces a bystander killing effect: once MMAE is released inside a CLDN18.2-positive cell, it can diffuse across the plasma membrane and kill adjacent tumor cells that do not themselves express the target. This is mechanistically central to why an ADC might outperform a naked antibody in gastric cancer, where CLDN18.2 expression is patchy rather than uniform across the tumor. Zolbetuximab can only engage cells it directly binds; AZD0901, in principle, can reach the CLDN18.2-low pockets that flank positive regions. The case for hitting the target is genetics-light but clinically strong. Zolbetuximab plus chemotherapy improved overall survival in two Phase 3 studies (SPOTLIGHT and GLOW), confirming that CLDN18.2 expression on tumor cells translates to a survival benefit when an antibody engages it [10][11]. AZD0901's bet is that adding a cytotoxic payload with bystander reach increases tumor kill rates compared with the naked antibody, which could either deepen response rates in first-line or rescue patients who progress on zolbetuximab.

The lead Phase 3 study is NCT06346392, a randomized open-label comparison of AZD0901 monotherapy against investigator's choice chemotherapy (typically paclitaxel, ramucirumab plus paclitaxel, or irinotecan) in 592 patients with CLDN18.2-positive metastatic gastric or GEJ adenocarcinoma who failed at least one prior line of systemic therapy [2]. Primary endpoint is PFS in all randomized participants, with overall survival as a key secondary. CLDN18.2 positivity is defined by immunohistochemistry (IHC, a staining test that detects protein on tumor tissue). The exact threshold (cell positivity percentage and staining intensity) is not publicly disclosed in the registry record, which is a meaningful gap because the threshold directly determines both enrolled population size and expected response rate. SPOTLIGHT and GLOW required at least 75 percent of tumor cells showing strong (2+/3+) membranous staining using the 43-14A antibody clone (the Vyloy companion-diagnostic-class assay), which restricted the eligible population to roughly 38 percent of CLDN18.2-tested gastric cancers [10][13]. If AZD0901 uses a similar 75 percent / 2+/3+ bar, it will compete in the same population zolbetuximab is approved in; if it uses a looser threshold (lower percent positivity or 1+ staining permitted), the eligible population expands but per-patient response rates may drop because more enrolled tumors have lower target density. The KYM901 Phase 1 expansion cohort that produced the 33 percent confirmed ORR was selected for CLDN18.2 positivity but used the Keymed assay parameters; whether the AZD0901 Phase 3 uses the same, the SPOTLIGHT-style threshold, or a more permissive cutoff is one of the most important undisclosed details in the program [15]. The trial design is conventional and the comparator arm is appropriate for second-line gastric cancer. The 592-patient sample is powered to detect a meaningful PFS improvement, and active-not-recruiting status confirms enrollment is complete, which removes execution risk from the recruitment side.

Our model estimates a 11% chance this drug is eventually approved. It starts from the historical base rate for Phase 2 drugs in this area (about 13%), then adjusts using ten facts about the trial and sponsor. What moves the number most: it is helped by a non-randomized design, its light or open-label blinding, and the sponsor's strong record of getting drugs approved; it is held back by weak or limited earlier-phase results. The other facts land near average for this stage, so they leave the estimate roughly where the base rate put it.

Efficacy risk centers on the biomarker. CLDN18.2 IHC has shown variable inter-laboratory concordance, and the threshold that defines positive directly determines both eligible population size and expected response rate [13]. Set the bar high and the addressable market shrinks. Set it low and the trial may miss its endpoint due to diluted target expression. Safety risk is the MMAE payload. Across approved MMAE-based ADCs (brentuximab vedotin, polatuzumab vedotin, enfortumab vedotin), peripheral neuropathy and neutropenia are dose-limiting [9]. Stomach lining cells also express low levels of CLDN18.2, so on-target gastric toxicity is biologically plausible, and the bystander killing that strengthens the efficacy case also raises the toxicity question for adjacent normal tissue. Zolbetuximab's safety record suggests on-target gastric toxicity is manageable with antibody alone, but the ADC version concentrates a cytotoxin at the target tissue. KYM901 Phase 1 showed high rates of vomiting (57.5 percent) and hypoalbuminemia (57.5 percent) consistent with GI insult [15]. Whether these tolerability signals scale acceptably in a Western Phase 3 population is an open question. Competitive risk is real. Zolbetuximab plus chemotherapy is now standard first-line care for CLDN18.2-positive gastric cancer, which shrinks AZD0901's second-line opportunity if patients have already seen anti-CLDN18.2 therapy and are biologically pre-treated against the target. Multiple competing CLDN18.2 ADCs are in development: IBI343 (Innovent) reported encouraging Phase 1 data, and LM-302 (LaNova) is advancing [14]. Note that CARsgen's CT041 is a CLDN18.2 CAR-T cell therapy and not a direct ADC competitor in modality, even though it targets the same antigen. Program risk if NCT06346392 fails: the combination Phase 3 NCT07431281 (AZD0901 plus capecitabine, with or without rilvegostomig) provides a partial salvage path, but a monotherapy second-line miss would force AZ to defend the program almost entirely on chemo-combination data, which would compress competitive differentiation versus zolbetuximab plus chemotherapy and likely cap peak sales at the lower end of sell-side ranges [6]. Commercial risk: even with a successful Phase 3, payers may demand head-to-head data against zolbetuximab-containing regimens rather than investigator's choice chemotherapy before granting preferred status.

Worth watching closely. AstraZeneca paid up to $1.1 billion in potential milestones for this asset after seeing KYM901 Phase 1 data, the Phase 3 has finished enrolling, and FDA fast track is in place [1][2][16]. The most informative near-term signal is Phase 2 monotherapy ORR from NCT07143604, expected at a major oncology meeting (ASCO or ESMO) within the next twelve months. The relevant benchmark is the KYM901 33 percent confirmed ORR in the Chinese expansion cohort [15]. If single-agent ORR in the Western Phase 2 clears or matches that bar in CLDN18.2-high patients, the Phase 3 PFS readout becomes high-probability. If monotherapy ORR drops materially in the Western population (a known risk for China-origin oncology assets), the second-line opportunity narrows and AZD0901's value shifts almost entirely to the combination Phase 3 program (NCT07431281), which reads out later. AstraZeneca's FY2024 total revenue was $54,073 million per its Form 20-F filing [7], so this asset is a meaningful but not company-defining bet, with sell-side peak sales models typically in the $1 to $3 billion range depending on first-line versus second-line positioning. Patent and data exclusivity timing for AZD0901 is not publicly broken out in the AZ pipeline disclosures reviewed here; the composition-of-matter patent estate originated with Keymed and the relevant filings are not enumerated in the in-license press release [1]. This is a gap worth filling before any peak-sales discounting work. Check back at ASCO 2026 or ESMO 2026 for the Phase 2 monotherapy data drop, and watch for any 8-K disclosure of Phase 3 interim analysis results. The competitive dynamics with zolbetuximab will shape labeled indications even if the Phase 3 hits its primary endpoint, since the second-line population is being reshaped by first-line CLDN18.2 exposure.