Cadonilimab

Cadonilimab (AK104) is Akeso's bispecific antibody that hits both PD-1 and CTLA-4 checkpoints on a single molecule, already approved in China as Kaitanni for cervical cancer (2022) and first-line HER2-negative gastric cancer (2024) [1]. Global Phase 3 programs are running across NSCLC, hepatocellular carcinoma, and gastric/GEJ. The commercial bet is straightforward: match the efficacy of ipilimumab plus nivolumab combos while cutting the class-defining immune toxicity that has capped that regimen's use.

Cadonilimab is approved in China and investigational everywhere else. First NMPA approval landed June 2022 for recurrent/metastatic cervical cancer after prior platinum chemotherapy, followed by a first-line HER2-negative gastric/GEJ approval in 2024 [2]. There are no FDA approvals, and no publicly disclosed breakthrough, fast track, or orphan designations in the US. The current global Phase 3 program includes NCT06617416 (unresectable stage III NSCLC consolidation, comparator sugemalimab, estimated primary completion October 2026) [10] and NCT07700979, a three-arm first-line gastric/GEJ study of chemo alone vs chemo plus cadonilimab vs chemo plus nivolumab, n=900, primary endpoint overall survival, sponsored by Akeso, estimated primary completion August 2029 [3]. Hepatocellular disease-focused work is anchored by combinations with lenvatinib and hepatic arterial infusion chemo, with active Phase 2 conversion-therapy studies [4]. Memorial Sloan Kettering is running an investigator-initiated Phase 2 in gastric/GEJ [5], giving the drug some US clinical footprint outside Akeso's own trial network. Ex-China regulatory timeline depends on partnership economics that have not yet been publicly disclosed. Investors should not confuse cadonilimab with ivonescimab, Akeso's PD-1/VEGF bispecific licensed to Summit Therapeutics; these are separate assets on separate commercial paths.

PD-1 is a brake pedal on T cells, the immune cells that hunt tumor cells. When tumors light up PD-L1 on their surface, they press that brake and T cells lose interest. CTLA-4 is a second brake, but it acts earlier in the process: it sits on T cells during the priming stage in lymph nodes, before those cells ever reach the tumor. PD-1 blockade works well in some cancers but fails in many because T cells never got activated in the first place. Adding CTLA-4 blockade (the ipilimumab plus nivolumab regimen) gets more T cells into the fight, but blocking CTLA-4 systemically causes severe colitis, hepatitis, and endocrine dysfunction, which caps how many patients can tolerate the combo. Cadonilimab tries to thread that needle. It is a single IgG1 molecule with two arms: one binds PD-1, the other binds CTLA-4. Because CTLA-4 is expressed at low density on resting T cells but high density on activated tumor-infiltrating T cells, the bispecific preferentially engages CTLA-4 where PD-1 is also present, biasing dual blockade toward the tumor microenvironment rather than the whole body. A 2026 meta-analysis covering more than 1,500 patients across indications reported a favorable immune-related adverse event profile relative to conventional dual checkpoint therapy [6]. A separate cervical cancer Phase 2 report from Clark et al. adds efficacy support in a recurrent small cell neuroendocrine cervical subgroup [7]. Whether that safety-efficacy shift holds up in Western populations at registrational scale is the open question. A critical structural detail: cadonilimab uses a modified IgG1 Fc with ablated ADCC, ADCP, and CDC effector function [6]. This distinguishes it from ipilimumab, whose intact IgG1 Fc is hypothesized to deplete intratumoral regulatory T cells (Tregs) via ADCC as part of its efficacy mechanism. Cadonilimab's Fc-silent design trades away that Treg-depletion activity in exchange for lower systemic immune activation, which is the structural basis for its claimed safety advantage. The tradeoff is not free: in indications where ipi/nivo efficacy is thought to depend on intratumoral Treg depletion (for example MSI-high colorectal cancer), the Fc-silent format could underperform. This is a scientific open question, not a resolved point, and it directly frames the ex-China readout risk.

The two most consequential Phase 3s carry different design profiles. NCT06617416 tests cadonilimab as consolidation after concurrent chemoradiation in unresectable stage III NSCLC, comparator sugemalimab, a PD-L1 antibody already approved for this setting in China. Estimated primary completion is October 2026 [10]. Sugemalimab is a fair active control but a relatively easy benchmark; a positive readout will need clean magnitude to translate to Western regulators who anchor on the PACIFIC/durvalumab standard. NCT07700979 is the more commercially important trial: first-line HER2-negative gastric/GEJ adenocarcinoma, n=900, three arms (chemo, chemo plus cadonilimab, chemo plus nivolumab), primary endpoint overall survival, estimated primary completion August 2029 [3]. That long OS follow-up window means the real global readout window is 2029, not the near term. Three-arm designs are unusually ambitious for a single sponsor because they demand statistical power for two comparisons and expose the sponsor to head-to-head risk against Bristol Myers Squibb's nivolumab. A win against the nivo arm would be commercially transformative; a numerical loss would be hard to explain. Hepatocellular work sits earlier: NCT06187961 is a Phase 2 evaluating cadonilimab with lenvatinib and hepatic arterial infusion chemo as conversion therapy for initially unresectable disease [4]. The MSKCC-sponsored NCT07631000 adds a small US Phase 2 datapoint in gastric [5]. Cervical Phase 3 work continues in parallel to support ex-China filings on the flagship China indication.

Our model estimates a 11% chance this drug is eventually approved. It starts from the historical base rate for Phase 3 drugs in this area (about 48%), then adjusts using ten facts about the trial and sponsor. What moves the number most: it is held back by heavier-than-usual blinding, the sponsor's thin or weak approval record, weak or limited earlier-phase results, and a randomized design. The other facts land near average for this stage, so they leave the estimate roughly where the base rate put it.

Efficacy risk is bounded but real. The dual-checkpoint mechanism is validated, so the question is whether cadonilimab's structural design delivers a meaningfully better therapeutic index than free-drug combos or just matches. If it matches, pricing power collapses in ex-China markets where ipilimumab plus nivolumab is approaching biosimilar exposure. A specific unresolved risk: the Fc-silent format may underperform in indications where intratumoral Treg depletion is a load-bearing part of the ipi/nivo mechanism, notably MSI-high colorectal cancer. Safety risk is on-target: colitis, hepatitis, pneumonitis, and endocrine dysfunction are class effects of CTLA-4 blockade. Chinese trial data suggest lower rates than ipi/nivo [6], but Chinese patient populations differ in HLA distribution, microbiome composition, and comorbidity mix from US and EU populations, and immune-related AE rates could climb in registration studies conducted in Western sites. Competitive risk from same-mechanism programs is real: at least two other PD-1/CTLA-4 bispecifics are in development, including AstraZeneca's volrustomig (MEDI5752) and MacroGenics' lorigerlimab. A Western regulatory win by any of them would set the comparator bar and could preempt cadonilimab's ex-China narrative. Execution risk is the largest factor. Akeso has never independently taken a drug through FDA registration. Its out-licensing strategy for cadonilimab in the West has not been disclosed, in contrast to the ivonescimab deal with Summit Therapeutics [9]. Without a US commercial partner, ex-China filing timelines slip. Commercial risk is severe: Merck's pembrolizumab dominates first-line gastric and NSCLC. Even a positive Phase 3 requires magnitude of benefit large enough to displace entrenched standard of care, and US payers will scrutinize a Chinese-manufactured biologic on both cost and supply chain grounds, particularly given current geopolitical tension over Chinese biotech.

Worth watching, not yet a top-priority signal. The two China approvals establish that the mechanism works in the clinic; the open question is whether the ex-China clinical and commercial thesis will hold. Concrete milestones for the calendar: NCT06617416 NSCLC consolidation primary completion is estimated October 2026 [10], meaning the first Western-relevant Phase 3 readout is a Q4 2026 / early 2027 event. The commercially decisive gastric Phase 3 (NCT07700979) does not primary-complete until August 2029 [3], so the big head-to-head vs nivolumab is a late-decade story, not a near-term one. The specific data point to track: the first head-to-head readout against a PD-1 monotherapy in a Western-enrolling Phase 3, or a licensing deal pairing cadonilimab with a major pharma outside China. Monitor Akeso's HKEX disclosures for partnership news and for balance-sheet updates that would signal how long they can self-fund the global program without a partner. Do not conflate cadonilimab with ivonescimab, which is Akeso's separate PD-1/VEGF bispecific licensed to Summit Therapeutics and covered under SMMT's SEC filings [9]. Cadonilimab's economics are almost entirely China-driven today; commercial upside for global investors depends on a partnership that has not yet materialized. If Akeso signs cadonilimab to a Merck-scale partner, this becomes an immediate priority asset; absent that, it is a China-market story with optional Western upside on a multi-year clock.