The approval of immunotherapy combinations has fundamentally changed treatment for persistent, recurrent, or metastatic cervical cancer. Multiple randomized controlled trials (RCTs) have now established an anti-programmed cell death protein 1 (PD-1) antibody (pembrolizumab), an anti–PD-1 ligand 1 (PD-L1) antibody (atezolizumab), and a bispecific antibody targeting both PD-1 and cytotoxic T lymphocyte-associated protein 4 (CTLA-4) (cadonilimab), each in combination with chemotherapy ± bevacizumab, as active regimens in this setting. With several options now available, the absence of comparative studies presents a challenge for optimal, individualized treatment selection.
To address this, Liang and colleagues conducted a systematic review, which identified 3 phase III RCTs comparing immune checkpoint inhibitor regimens with chemotherapy-based control arms (BEATcc [n = 410], KEYNOTE-826 [n = 617], and COMPASSION-16 [n = 445]), and a fourth trial (GOG 240) that served as an external control to account for differences in bevacizumab inclusion in the control arms of the first 3 trials. These 4 trials enrolled 1,924 patients with persistent, recurrent, or metastatic cervical cancer. The analysis examined overall survival (OS), progression-free survival (PFS), objective response rate, and the incidence of grade ≥3 treatment-related adverse events (AEs).
Using validated statistical analysis for cross-trial insights (frequentist network meta-analysis and Bucher’s indirect comparison method), the authors did not find significant differences in efficacy between the 3 regimens in the all-comers populations regardless of whether bevacizumab was included in the chemotherapy backbone. There were PD-L1 expression data available for KEYNOTE-826 and COMPASSION-16. In subset analyses, patients with PD-L1 Combined Positive Score (CPS) ≥1 derived significant benefit from immunotherapy combinations in both studies. However, among patients with CPS <1, neither pembrolizumab nor cadonilimab significantly improved OS or PFS vs chemotherapy, although the authors reported that cadonilimab showed a trend toward PFS benefit therein (hazard ratio [HR] 0.65, 95% CI: 0.42–1.03; P = .06). In the subset of patients with metastatic cervical cancer, the authors reported no significant benefit for pembrolizumab and a trend for improved OS (HR 0.73, 95% CI: 0.52–1.02; P ≈ .05) and a statistically significant PFS benefit (HR 0.70, 95% CI: 0.54–0.92; P <.05) with the addition of cadonilimab to chemotherapy. The relevance of this is unclear, as a heterogeneity analysis found no significant difference between these outcomes for cadonilimab– vs pembrolizumab-treated patients.
Regarding safety, all 3 immunotherapy regimens were associated with increased rates of grade ≥3 AEs compared with control arms, but the authors pointed out that while KEYNOTE-826 showed a statistically significant increase in severe toxicities vs control, the increases observed with BEATcc and COMPASSION-16 did not reach statistical significance. However, KEYNOTE-826 had more statistical power than the other 2 studies, and statistical significance could be an artifact of this difference. The authors further characterized the rate of grade 3–4 immune-related AEs with cadonilimab plus chemotherapy as numerically lower than pembrolizumab-based combinations in KEYNOTE-826 (10% vs 12%), noting that further confirmation is warranted.
Further cross-trial comparisons were made to gain insight into the efficacy of immune checkpoint inhibitors in cervical cancer in Asian populations vs the rest of the world. The analyses found numeric but not statistically significant differences.
The insights reported from these analyses raise interesting questions for further study. However, they must be interpreted with caution until the observations can be evaluated prospectively. Although this publication used a robust statistical model, limitations of the cross-trial comparisons and statistical power considerations must be acknowledged. It remains unclear whether there are inherent differences between therapies directed at PD-1, PD-L1, and dual PD-1/CTLA-4. Although differences were reported, and some reached the P <.05 threshold for significance, the trends reported had no adjustment for multiple comparisons in this post hoc data analysis.
High Level
This network meta-analysis lends support to the growing body of evidence supporting immunotherapy combinations as initial treatment for recurrent or metastatic cervical cancer, while also highlighting important gaps that will require further investigation before the field can draw definitive conclusions about differential efficacy and safety across regimens. In the bevacizumab combination setting, the indirect comparison between pembrolizumab and atezolizumab suggests broadly comparable outcomes, though the confidence intervals are wide and head-to-head data are lacking; therefore, institutional access, formulary considerations, and patient-level safety profiles remain reasonable guides for selection between the 2. The numeric trends favoring cadonilimab in the non-bevacizumab setting and in PD-L1–negative patients are intriguing, but these analyses were not powered for direct comparison and should be interpreted cautiously. Whether these differences represent a clinically meaningful signal or reflect the inherent limitations of indirect comparison methodology remains an open question. Academic institutions are well positioned to address these uncertainties to establish more reliable evidence on which to base treatment differentiation, particularly for the CPS <1 population, where meaningful options remain limited.
Ground Level
For community oncologists treating patients with persistent, recurrent, or metastatic cervical cancer, this analysis may offer practical guidance for navigating an increasingly complex treatment landscape. The core takeaway is that first-line immunotherapy combinations are now standard of care in this setting, with optimal regimen selection dependent on 2 key factors: PD-L1 status (CPS) and bevacizumab eligibility. In patients with CPS ≥1 who are bevacizumab eligible, pembrolizumab- or atezolizumab-based regimens offer well-established and comparable survival benefits, and either can be selected on the basis of access, tolerability history, and patient preference. For patients who are not candidates for bevacizumab and/or are PD-L1 negative, cadonilimab plus chemotherapy may represent a compelling option for further investigation according to guidance from academic centers. However, it is unclear whether this agent will become available in the US and/or whether the same efficacy can be achieved by combining anti–PD-(L)1 and anti–CTLA-4 antibodies in patients who are not eligible for bevacizumab and/or have PD-L1–negative tumors.