Abstract / Summary
Background: Human papillomavirus (HPV) is the major cause of cervical cancer globally. Current vaccines have shown high efficacy at preventing HPV infection and other surrogate disease endpoints in randomised controlled trials (RCT). However, uptake is limited by accessibility and cost. Rapid clinical evaluation is critical for shortening time-to-market and encouraging the development and deployment of affordable vaccines. We aim to establish a validated correlate of protection that will support vaccine developers and decision makers with vaccine development, approval, dose schedules and deployment decisions. Methods: We utilised three existing Cochrane Systematic Reviews of RCTs of HPV vaccines and extended these to 1st January 2026. We identified and extracted data from trials where both efficacy and immunogenicity (HPV-specific antibodies) were reported. For HPV types where vaccine efficacy was reported but not immunogenicity (Types 31/33/45/52/58) we sourced antibody binding concentrations from separate immunogenicity studies. We linked antibody binding concentrations from each RCT to the corresponding vaccine efficacy data, matching by study, vaccine arm, number of doses, and HPV type. We use a Bayesian framework in a model-based meta-analysis to examine the association between antibody binding concentrations and vaccine efficacy. Findings: Combining data on both vaccine-targeted and non-targeted HPV types, we find an association between type-specific antibody binding concentrations and vaccine efficacy against the three HPV outcomes tested: incident infection, persistent infection and grade 2 or 3 cervical intraepithelial neoplasia (CIN2+) (p<0.0005, for all outcomes). We also compared this association with data from studies of natural infection, which had shown an association between antibody binding concentrations and risk of infection. This demonstrated that the association between vaccine-induced antibodies and vaccine efficacy was well aligned with the existing HPV literature on correlates of risk from natural exposure. Finally, we find evidence that the antibody concentrations associated with 50% (of the maximal) vaccine efficacy differ by type (p<0.0005, for example 16 sIU/ml for HPV58, but <0.5 IU/ml for HPV 16/18). Interpretations: HPV type-specific antibody binding concentrations are associated with protection against HPV related clinical outcomes, across different HPV types and vaccines, and after natural infection. The consistent observation that antibody binding concentrations correlate with protection provides support for the use of antibody binding for immunobridging, for example in non-inferiority studies when assessing similar vaccines. Funding: This work is supported by National Health and Medical Research Council of Australia and the University of New South Wales.