Cervical cancer remains a major contributor to cancer-related morbidity and mortality worldwide, particularly in low- and middle-income countries. Locally advanced cervical cancer (LACC), defined as FIGO stages IB3 to IVA, accounts for approximately 50%–60% of newly diagnosed cases (Sung et al., 2021; Small et al., 2017). The current standard of care for LACC is concurrent chemoradiotherapy (CCRT), combining external beam radiation therapy with weekly cisplatin chemotherapy, followed by intracavitary or interstitial brachytherapy. Although this regimen has improved survival, outcomes remain suboptimal. Recurrence rates range from 30% to 40%, and 5-year overall survival varies between 50% and 70%, depending on disease stage and regional nodal status (Mileshkin et al., 2023). Therefore, new therapeutic strategies are urgently needed to enhance the curative potential of frontline treatment.

Persistent human papillomavirus (HPV) infection plays a central role in cervical carcinogenesis, contributing to immune evasion via upregulation of immune checkpoint molecules, particularly PD-1 and PD-L1 (Koh et al., 2019; Han et al., 2022; Schelz et al., 2024). Pembrolizumab, a PD-1 inhibitor, received FDA approval for recurrent or metastatic cervical cancer based on the KEYNOTE-158 trial (Chung et al., 2019). This provides a compelling biological rationale for the use of immune checkpoint inhibitors (ICIs) in cervical cancer. Recent trials have begun exploring the integration of ICIs into first-line treatment for LACC, either concurrently with CCRT or as sequential consolidation therapy. For example, the KEYNOTE-A18 trial showed significant improvements in progression-free and overall survival with the addition of pembrolizumab to CCRT and maintenance therapy, while the CALLA trial did not demonstrate a statistically significant benefit with durvalumab. Moreover, early-phase and retrospective studies suggest that the timing, choice of agent, and treatment duration may influence outcomes and toxicity, but consensus is lacking.

Although several individual trials and observational studies have investigated the role of ICIs in LACC, their findings are heterogeneous and sometimes conflicting. Recent meta-analyses on cervical cancer have either focused broadly on systemic therapies or pooled data across early-stage, advanced, and metastatic settings, lacking specificity for LACC and immunotherapy. Given the emerging but fragmented evidence, a focused and quantitative synthesis is essential. This meta-analysis aims to evaluate the efficacy and safety of immune checkpoint inhibitors combined with standard CCRT in patients with LACC, providing clinicians and policymakers with evidence-based insights to guide treatment strategies in this high-risk population.

The risk of bias for randomized controlled trials included in this meta-analysis was assessed using the Cochrane Risk of Bias 2 (RoB 2) tool. Risk of bias for nonrandomized prospective studies was assessed using the ROBINS-I tool across seven domains. Sensitivity analyses excluding studies with ‘some concerns’ or higher risk yielded consistent results, supporting the robustness of the findings. Each study was independently evaluated by three reviewers for domains including randomization process, deviations from intended interventions, missing outcome data, measurement of outcomes, and selection of reported results. Based on these assessments, an overall risk of bias judgment was assigned to each study as low risk, some concerns, or high risk. Discrepancies among reviewers were resolved through discussion or consultation with a fourth investigator.

This meta-analysis, synthesizing five prospective studies including 1,987 patients with LACC, demonstrates that the addition of ICIs to CCRT significantly improves PFS and ORR. Tumor response endpoints were harmonized across trials by focusing on RECIST v1.1–based post-CCRT assessments. Sensitivity analyses restricted to comparable imaging timepoints confirmed the consistency of pooled ORR and CR estimates, indicating that inter-study variation in response timing had minimal influence on the overall results. The pooled hazard ratio for PFS was 0.76 [95% CI: 0.64–0.91], while the pooled ORR showed a significant benefit (OR = 1.28 [95% CI: 1.06–1.56]). The CR rate showed an improving trend but was not statistically significant. Due to the absence of disaggregated data, subgroup analyses by PD-L1 status, stage, nodal involvement, histologic subtype, or brachytherapy parameters could not be performed. These variables may modulate the immune response to PD-1/PD-L1 blockade, underscoring the need for future individual patient data analyses and biomarker-driven stratification. Although the random-effects model (Hartung–Knapp adjustment) yielded wider confidence intervals leading to loss of statistical significance for PFS and ORR, this reflects the conservative nature of variance estimation under small-study conditions (n = 2) rather than genuine inconsistency between studies. The point estimates remained directionally consistent with the fixed-effect results, supporting a robust trend toward benefit.

In the included studies, ICIs combined with CCRT in LACC patients were generally well tolerated. Safety analysis found no significant increase in grade ≥3 AEs, though irAEs were significantly more common (OR = 3.00 [95% CI: 1.68–5.34]). The most frequent irAEs were thyroid dysfunction, skin rash, and elevated liver enzymes, most of which were grade 1–2 and manageable without interrupting treatment. Severe irAEs (grade ≥3) were less frequent, included pneumonitis, hepatitis, and colitis, which were the main reasons for discontinuation in trials such as CALLA and KEYNOTE-A18 (Monk et al., 2023a; Lorusso et al., 2024). Although moderate statistical heterogeneity (I ² = 26.8%) was observed for irAEs by timing, this level likely reflects clinical variation rather than inconsistency in direction of effect. Random-effects sensitivity analysis confirmed stability of the pooled estimate, indicating that the observed heterogeneity has limited impact on the validity of conclusions (Supplementary Figure S2).

Overall, discontinuation rates due to AEs were modest (5%–7%), and most patients were able to complete or resume therapy. Similar toxicity profiles were seen in the Duska and Mayadev trials, with no unexpected high-grade events reported (Duska et al., 2025; Mayadev et al., 2025a). These findings suggest that while ICIs introduce distinct immune-related toxicities, they are typically controllable with monitoring and supportive management, and severe events remain relatively infrequent.

Cervical cancer, especially when driven by HPV infection, exhibits high PD-L1 expression and a suppressive immune microenvironment, making it an ideal target for immunotherapy (Schelz et al., 2024; Monk et al., 2023b; Tan et al., 2024; Montroy et al., 2024). Radiotherapy promotes immunogenic cell death, enhances antigen presentation, and facilitates T-cell infiltration, creating a synergistic setting for ICIs (Procureur et al., 2021). The NRG-GY017 trial further demonstrated that neoadjuvant ICIs significantly increased T-cell receptor (TCR) clone expansion, supporting the biological synergy of ICIs and radiotherapy (Mayadev et al., 2025a). These mechanisms likely underpin the observed improvements in clinical outcomes.

Our findings are align with the KEYNOTE-A18 trial, which showed a promising trend toward improved OS (HR = 0.73; p = 0.058), and confirmed significant PFS and OS benefits in high-risk LACC patients, without compromising health-related quality of life (Lorusso et al., 2024). In contrast, the CALLA trial did not meet its primary PFS endpoint, potentially due to differences in ICI type (PD-L1 vs. PD-1 inhibitors), biomarker selection, and statistical power (Mayadev et al., 2025b). Differences in biomarker assessment and patient selection likely contributed to the discordant efficacy outcomes between KEYNOTE-A18 and CALLA. The higher prevalence of PD-L1 positivity (CPS ≥1) in KEYNOTE-A18 versus the broader, biomarker-unenriched CALLA population may have enriched for immunologically responsive tumors, supporting PD-L1 as a potential predictive biomarker for PD-1 blockade efficacy in LACC. Furthermore, a recent meta-analysis ranked pembrolizumab plus CCRT as one of the most effective regimens for LACC (Petrelli et al., 2025), reinforcing the potential of immunotherapy combined with standard chemoradiotherapy. Additional studies, including those by Duska et al. and Mayadev et al., provide supportive real-world and immunologic evidence for feasibility and safety, despite heterogeneous efficacy results (Duska et al., 2025; Mayadev et al., 2025a). Patient-reported outcomes from KEYNOTE-A18 remained stable or improved over time, reinforcing the favorable benefit–risk profile of integrating ICIs into definitive treatment for LACC (Randall et al., 2025).

Studies that could not be included in meta-analysis still provide valuable insights. Duska et al. compared concurrent vs. sequential pembrolizumab administration and reported no significant OS difference (HR = 0.77 [95% CI: 0.32–1.85]) (Duska et al., 2025), while the NRG-GY017 trial highlighted differences in immunologic activation based on timing (Mayadev et al., 2025a). The KEYNOTE-A18 trial’s interim OS analysis suggests a potential survival benefit with ICI + CCRT, though long-term data remain immature (Lorusso et al., 2024). Together, these findings highlight the importance of treatment timing and immune monitoring in optimizing ICI-based regimens. Given the immaturity of OS data, particularly from KEYNOTE-A18, an updated meta-analysis will be conducted once mature survival results become available. This update will include longer follow-up data from ongoing phase III trials and adhere to a registered protocol to ensure transparency.

Despite the promising findings, several limitations must be acknowledged: Only two randomized controlled trials (RCTs) contributed pooled data for survival and response outcomes, limiting statistical power. Additionally, overall survival (OS) data remain immature across the included studies. Clinical heterogeneity was observed in terms of ICI type, treatment timing (concurrent vs. sequential), and radiotherapy regimens. Biomarker data (such as PD-L1 expression, tumor mutational burden [TMB], and tumor-infiltrating lymphocytes [TILs]) were inconsistently reported. Furthermore, the small number of studies precluded formal subgroup or sensitivity analyses.

This meta-analysis supports the integration of ICIs into CCRT for LACC, particularly in cases with immunogenic tumors or PD-L1 positivity. However, the long-term survival benefit remains to be confirmed as OS data matures. Future trials should focus on determining the optimal timing and duration of ICI administration (e.g., neoadjuvant, concurrent, or adjuvant), conducting biomarker-driven subgroup analyses, and assessing real-world safety, cost-effectiveness, and access across global populations. Additionally, integrating immune monitoring (such as TCR clonality and TILs) into clinical trial design will be essential to optimize treatment strategies.

The addition of ICIs to standard chemoradiotherapy is associated with a significant improvement in PFS and ORR in patients with LACC, without a significant increase in severe toxicity. Although OS data are still immature, the current evidence suggests a potential survival benefit. The integration of ICIs into definitive treatment requires further validation through biomarker-stratified, long-term, and real-world studies.