Cross-protective efficacy of two human papillomavirus vaccines: a systematic review and meta-analysis

Background The extent of cross-protection is a key element in the choice of human papillomavirus (HPV) vaccine to use in vaccination programmes. We compared the cross-protective efficacy of the bivalent vaccine (HPV 16 and 18; Cervarix, GlaxoSmithKline Biologicals, Rixensart, Belgium) and quadrivalent vaccine (HPV 6,11,16, and 18; Gardasil, Merck, Whitehouse Station, NJ, USA) against non-vaccine type HPVs. Methods We searched Medline and Embase databases, conference abstracts, and manufacturers' websites for randomised clinical trials assessing the efficacy of bivalent and quadrivalent vaccines against persistent infections (lasting >= 16 months) and cervical intraepithelial neoplasia (CIN) associated with the non-vaccine type HPVs (types 31, 33,45,52, and 58). We included studies of participants who were HPV DNA negative before vaccination for all HPV types assessed. We assessed heterogeneity in vaccine efficacy estimates between trials with I-2 and chi(2) statistics. Findings We identified two clinical trials (Females United to Unilaterally Reduce Endo/Ectocervical Disease [FUTURE] I and II) of the quadrivalent vaccine and three (Papilloma Trial Against Cancer In Young Adults [PATRICIA], HPV007, and HPV-023) of the bivalent vaccine. Analysis of the most comparable populations (pooled FUTURE I/II data vs PATRICIA) suggested that cross-protective vaccine efficacy estimates against infections and lesions associated with HPV 31,33, and 45 were usually higher for the bivalent vaccine than the quadrivalent vaccine. Vaccine efficacy in the bivalent trial was higher than it was in the quadrivalent trial against persistent infections with HPV 31 (77.1% [95% CI 67.2 to 84.4] for bivalent vaccine vs 46.2% [15.3 to 66.4] for quadrivalent vaccine; p=0.003) and HPV 45 (79.0% [61.3 to 89.4] vs 7.8% [-67.0 to 49.3]; p=0.0003), and against CIN grade 2 or worse associated with HPV 33 (82.3% [53.4 to 94.7] vs 24.0% [-71.2 to 67.2]; p=0.02) and HPV 45 (100% [41.7 to 1001 vs -51.9% [-1717.8 to 82.6]; p=0.04). We noted substantial heterogeneity between vaccine efficacy in bivalent trials against persistent infections with HPV 31 (I-2=69%, p=0.04) and HPV 45(I-2=70%, p=0.04), with apparent reductions in cross-protective efficacy with increased follow-up. Interpretation The bivalent vaccine seems more efficacious against non-vaccine HPV types 31,33, and 45 than the quadrivalent vaccine, but the differences were not all significant and might be attributable to differences in trial design. Efficacy against persistent infections with types 31 and 45 seemed to decrease in bivalent trials with increased follow-up, suggesting a waning of cross-protection; more data are needed to establish duration of cross-protection.