Cervical cancer (CC) is the fourth greatest global burden in terms of both incidence and mortality in women, with the leading cause of cancer death in 37 countries (1). In 2022, Global Cancer Observatory (GLOBOCAN) estimated 661,021 new CC cases and 348,189 CC-related deaths. Notably, China accounts for 22.8% of the worldwide incidence and 16.0% of CC-related mortality, resulting in a tremendous medical burden (2). Although early-stage CC is often amenable to radical surgery or chemoradiotherapy, recurrent or metastatic cervical cancer (R/M CC) patients are incurable and have a dismal prognosis, with a five-year survival rate of only 17% (3). Few effective therapeutic options are left for R/M CC (4). Recent research also emphasizes the urgent need for novel therapeutic strategies in metastatic cancers, as conventional treatments often fail to improve survival outcomes (5).

Recent advances in cancer immunotherapy, which manipulates the immune system to recognize and attack cancer cells (6, 7), have revolutionized the paradigms of CC management (3, 7). Multiple types of immune checkpoint inhibitors (ICIs) targeting programmed death protein-1 (PD-1) and its ligand PD-L1, such as pembrolizumab, camrelizumab, nivolumab, have entered clinical trials successively and exhibited improved efficacy in monotherapy or combination with chemotherapy, radiotherapy, and targeted therapy (8–12). Compared with platinum-based chemotherapy of objective response rate (ORR) ranging from 20% to 30% (13), anti-PD-1 monotherapy could achieve an ORR of 12.2%-33.3%, and reach up to 65.9% when combined with other therapies in late-line treatment of R/M CC (14). Despite this progress, some R/M CC patients still fail to respond to immunotherapy, especially in PD-L1-negative tumors. Hence, considerable strides are urgently needed to improve treatment outcomes.

Dual-targeted immunotherapy is a clinically validated strategy for enhancing antitumor activity compared to anti-PD-1 monotherapy. However, the associated side effects can sometimes be intolerable (15). Cadonilimab (AK-104) is a novel bispecific antibody targeting PD-1 and CTLA-4 with favorable safety profile. It features a unique symmetric tetravalent structure that enhances binding activity within the tumor microenvironment, thereby inhibiting immunosuppressive pathways and augmenting T-cell-mediated responses (16). The emerging class of bispecific antibodies has shown promising results in multiple cancers by enhancing T-cell responses and overcoming tumor immune evasion (17). Cadonilimab became the first bispecific antibody approved for patients with R/M CC who were resistant to platinum-based chemotherapy in China in June 2022, marking a significant advancement in cervical cancer therapy. This approval was based on the COMPASSION-03 trial conducted by Gao et al., in which 111 patients who had failed platinum-based chemotherapy were treated with cadonilimab monotherapy. Cadonilimab achieved an overall response rate (ORR) of 32.3% (32/99), with a median progression-free survival (PFS) of 3.71 months. The ORR in the PD-L1+ cohort was 43.8% (28/64), compared with 16.7% (3/18) in the PD-L1- cohort (18). Additionally, grade ≥3 immune-related adverse events (irAEs) were reported in only 4.5% of patients. Furthermore, modulation of the tumor microenvironment through nanotechnology and molecular engineering may synergize with such bispecific antibodies (19).

Despite such encouraging results, no real-world studies have yet focused on cadonilimab in R/M CC. Against this backdrop, our study presents the first real-world evidence from 13 academic medical centers in China, evaluating the effectiveness and safety of cadonilimab in treating R/M CC. Additionally, we investigate biomarkers such as PD-L1, tumor mutational burden (TMB), and integrated genomic profiling to predict response sensitivity to immunotherapy in R/M CC.

Cadonilimab yielded favorable response rates without compromising safety profiles in R/M CC management. To our knowledge, this multicenter study represents the first real-world assessment of cadonilimab in R/M CC patients. Cadonilimab exhibited an ORR of 38.8% and a DCR of 72.1%. Among those with PD-L1 CPS ≥1, the ORR improved to 56.5%, and the median PFS reached 14.0 months. Notably, patients with PD-L1 CPS <1 also demonstrated significant benefits, with an ORR of 35.7% and a median PFS of 12.8 months, indicating cadonilimab could also provide encouraging anti-tumor activity even in PD-L1 CPS <1 subgroup. Moreover, only 56 (40.3%) patients developed grade 3–4 adverse events in our study, comparable to or lower than the previous study (14), indicating its well-tolerated nature. These results are consistent with prior findings that novel immune agents can offer a favorable safety profile compared to traditional checkpoint inhibitors (26). However, the long-term evaluation of cadonilimab needs an expanded sample and further follow-up.

Cadonilimab treatment led to significantly longer PFS (12.4 months) than cisplatin-based combination therapy (5.8 months), followed by non-platinum drugs (2–3 months) among R/M CC patients (27). Compared with the investigator’s choice of cisplatin-based chemotherapy, cadonilimab treatment resulted in a 9.7% higher ORR (27). Similarly to our study, the COMPASSION-03 study of the CC cohort also demonstrated cadonilimab with a high ORR of 32.3% in R/M cases (18). Compared with another generic drug (QL1706), cadonilimab also showed higher ORR in CC treatment (28). Hence, the anti-PD-1/CTLA-4 bispecific antibodies, like cadonilimab, have become an important component of the treatment regimen for R/M CC. Moreover, compared to dual PD-1 and CTLA-4 checkpoint blockade combinations in CC treatment, such as the CheckMate 358 study (29) and a Phase II trial of balstilimab and zalifrelimab combination (NCT03495882) (30), cadonilimab showed a more favorable anti-tumor power with higher ORR and lower toxicity, underscoring its enormous potential in CC management.

The significant antitumor activity of cadonilimab is likely due to its tetravalent design. Preclinical studies have suggested its tetravalent structure could enhance the binding avidity of high densities of PD-1 and CTLA-4, leading to increased immune responses to antitumors (16). Mechanistically, blocking the binding of PD-1 and PD-L1 could maintain the active status of tumor-reactive T cells, which would be becoming inactivated after persistent chronic stimulation. On the other hand, CTLA-4 blockade could activate effector T cells and inactivate regulatory T cells, thereby enhancing antitumor immunity. The unique complementary mechanisms of blockade of the PD-1 and CTLA-4 pathways underlie the improved antitumor activities of cadonilimab (31).

An increasing body of evidence also supported the synergistic effect of radiotherapy and immunotherapy in treating malignant tumors, such as non-small cell lung cancer (32) and esophageal or gastroesophageal junction cancer (33), as it can eliminate the primary tumors and induce host immunity to control distant metastases (34). Consistent with previous studies, we found that patients treated with cadonilimab in combination with radiotherapy demonstrated a higher ORR of 52.6% and a DCR of 86.8%, further identifying the combination of radiotherapy as an independent prognostic factor for both OS and PFS in R/M CC. Previous studies have also confirmed that combination strategies, including immunotherapy and radiation, can enhance anti-tumor efficacy via synergistic mechanisms (35). Preclinical studies show that radiotherapy can induce tumor cell death and antigen release, enhancing the immune response by altering the tumor microenvironment and making tumor cells more recognizable to immune cells (36). In addition, ICIs could promote tumor vascular normalization in a T-cell-dependent manner through interferon (IFN)-mediated signaling between T cells and endothelial cells, improving tissue perfusion and reducing intratumoral hypoxia and acidosis, thereby sensitizing tumors to ionizing radiation (37). Hence, the combination of cadonilimab and radiotherapy has shown significant potential in treating R/M CC and might emerge as an important therapeutic strategy in the future.

The safety profile of cadonilimab was consistent with that previously reported for the drug and combined therapy in patients with different tumor types (38–40). In our study, cadonilimab, combined with other therapies, generally reflects a manageable safety profile with mostly grade 1–2 TRAEs, even under the circumstance of prolonged exposure to cadonilimab. It is worth noting that the incidence of grade 3–4 irAEs among participants treated with cadonilimab monotherapy was only 31.6% (6/19), comparable to the 28% reported in the COMPASSION-03 study and lower than MEDI5752 (38%), another bispecific anti-PD-1/CTLA-4 antibody (18, 41). When combined with other treatments, only 46.6% of patients experienced grade ≥3 TRAEs, which is also lower than the reported rate of 73.3% (18), underscoring cadonilimab’s favorable safety profiles. Mechanistically, cadonilimab’s favorable safety profile is likely attributed to its unique Fc-null design. The Fc-null design inhibits binding to Fc receptors, significantly reducing antibody-dependent cellular cytotoxicity (ADCC), antibody-dependent cellular phagocytosis (ADCP), and the release of interleukins-6 and -8 (42). Immune-related toxicity profiles might also be influenced by monocyte/macrophage repolarization, a mechanism gaining attention in cancer immunotherapy (43). These properties collectively contribute to the low toxicity observed in clinical settings.

In-depth analysis of the PIK3CA mutation landscape in our cohort revealed that all alterations were restricted to canonical helical domain hotspots—E542K and E545K—which are among the most frequently reported oncogenic mutations in solid tumors, including cervical cancer (44). These mutations are known to promote constitutive activation of the PI3K-AKT pathway by relieving inhibitory interactions with the p85 regulatory subunit, thereby enhancing downstream proliferative and survival signaling (45). Both E542K and E545K have been implicated in increased immune evasion and tumor aggressiveness (46). Furthermore, KEGG pathway enrichment analysis in our cohort demonstrated that PIK3CA mutations often co-occurred with COL6A6 alterations and were significantly enriched in the PI3K–AKT signaling pathway. This pathway not only drives cell cycle progression and metabolic adaptation but also shapes the tumor immune microenvironment by regulating apoptosis resistance, cytokine secretion, and immune cell recruitment (47, 48). Notably, our findings are consistent with a prior report by Xu et al. (49), which demonstrated that patients harboring PIK3CA mutations exhibited a significantly higher objective response rate (ORR) to anti-PD-1 therapy (91.7%) compared to those with wild-type PIK3CA (46.2%; P = 0.012). In our study, all four patients with PIK3CA mutations responded to cadonilimab (CR or PR), whereas no mutations were detected in the non-responder group. Although this trend did not reach statistical significance (P = 0.084), likely due to the limited sample size, the observed pattern suggests a potential predictive role of PIK3CA mutations for immunotherapy responsiveness in selected patients. These findings are exploratory in nature and warrant validation in larger, independent cohorts. Emerging evidence also suggests that traditional compounds, including herbal derivatives, may modulate tumor immunogenicity and complement checkpoint-based approaches (50). Furthermore, future mechanistic studies are needed to explore whether aberrant PI3K–AKT pathway activation influences cadonilimab sensitivity by modulating immune evasion or the tumor microenvironment.

Of note, to our knowledge, this study represents the largest real-world data set to date evaluating cadonilimab as a therapeutic modality in patients with R/M CC. On the one hand, compared to the limited use of ICIs targeting PD-1 in PD-L1-negative populations with a low ORR ranging from 0-16.7% (9, 51–54). Cadonilimab demonstrated significant antitumor activity even in a PD-L1-negative population, with an ORR of 35.7% and an mPFS of 12.8 months, indicating the broad application of cadonilimab in the PD-L1-negative population. On the other hand, 8.5% of the population achieved CR, giving us hope that recovery from disseminated tumors and maintaining a normal life are no longer a luxuries for late-stage CC patients in the era of immunotherapy. Further in-depth research and fundamental experiments are necessary to explore the potential roles of cadonilimab in the low PD-L1 expression subgroup and its antitumor mechanism.

Despite the encouraging results, several limitations should be acknowledged. First, this was a retrospective study conducted in China, which led to selection bias and limited generalizability. Second, the long-term benefits and late toxicities were unavailable due to the relatively short follow-up time. Third, many challenges remain in the clinical use of cadonilimab due to financial constraints. However, with growing real-world evidence, we believe cadonilimab will soon be covered by Chinese medical insurance and benefit more R/M CC patients. Moreover, treatment regimens in our study were heterogeneous, and baseline characteristics varied across subgroups. Due to the limited sample size, we did not perform propensity score matching or multivariable adjustment. Therefore, subgroup analyses—such as those based on PD-L1 CPS—should be considered exploratory and interpreted with caution. Lastly, the limited number of patients who underwent PD-L1 testing and whole-exome sequencing reduced the statistical power to assess the predictive value of biomarkers such as PD-L1 expression, TMB, and PIK3CA mutations. These exploratory findings should be interpreted with caution and validated in larger prospective studies.

In sum, R/M CC is a life-threatening disease with limited treatment options available. In our study, cadonilimab demonstrated promising efficacy with acceptable safety profile, even in the low PD-L1 population. Cadonilimab has the potential to become a new standard of care in treating CC. Further investigations involving larger-scale randomized clinical trials and real-world studies are warranted.