Network meta-analysis of immunotherapy for persistent, recurrent, or metastatic cervical cancer

The first-line treatment for persistent, recurrent, or metastatic cervical cancer continues to pose significant clinical challenges. While the application of immune checkpoint inhibitors (ICIs) has transformed the existing treatment paradigm, the lack of direct comparisons between the different immunotherapy regimens limits the selection of optimal strategies for diverse patient populations. We have searched through the PubMed, Embase, and Web of Science databases for phase III randomized controlled trials (RCTs) for persistent, recurrent, or metastatic cervical cancer. The primary endpoints were overall survival (OS) and progression-free survival (PFS). The objective response rate (ORR) and the incidence of grade ≥3 treatment-related adverse events (TRAEs) were the secondary endpoints. Data were analyzed using a frequentist network meta-analysis and Bucher’s indirect comparison method. Four phase III RCTs (1,924 patients) were enrolled. Combinations with bevacizumab, pembrolizumab, and atezolizumab demonstrated comparable efficacy to each other [OS relative hazard ratio (HR) = 1.11, 95%CI = 0.77–1.62; PFS relative HR = 1.09, 95%CI = 0.78–1.52), both significantly superior to chemotherapy. In the treatment setting without bevacizumab, cadonilimab outperformed pembrolizumab plus chemotherapy for OS (relative HR = 0.75, 95%CI = 0.45–1.25) and PFS (relative HR = 0.64, 95%CI = 0.40–1.03). Patients with programmed death-ligand 1 (PD-L1)-positive tumors [combined positive score (CPS) ≥ 1] derived clear benefit from immunotherapy, whereas the PD-L1-negative (CPS < 1) populations experienced limited improvements overall. However, cadonilimab showed a trend toward PFS improvement in this subgroup (HR = 0.65, 95%CI = 0.42–1.03, p = 0.06) and reached significance for PFS in the metastatic subgroup (HR = 0.70, 95%CI = 0.54–0.92, p < 0.05). The ORR evaluation confirmed that all the combinations with ICIs greatly increased the tumor response rate. In addition, the risk of grade ≥3 TRAEs was comparable between the pembrolizumab and cadonilimab regimens (OR = 1.07, 95%CI = 0.57–2.02). In the bevacizumab combination setting, pembrolizumab and atezolizumab demonstrate equivalent efficacy. Cadonilimab excels in settings without bevacizumab, showing efficacy in PD-L1-negative patients and metastatic subgroups. Thus, clinical decision-making should integrate multiple pieces of information such as the PD-L1 expression status and bevacizumab eligibility, and individual safety profiles should be considered to create personalized treatment routes. In the future, head-to-head comparative studies are urgently needed to precisely identify optimal treatment regimens for certain types of patient subgroups. https://www.crd.york.ac.uk/PROSPERO/view/CRD420251158152, identifier CRD420251158152.