Uncovering the HPV types causing cervical cancer

About 50 years ago, the causality of cervical cancer was attributed to human papillomavirus (HPV) by the German virologist Harald zur Hausen. Since then, HPV has been shown to cause cancer in several anatomical sites and hundreds of HPV types have been discovered, prompting further research into the carcinogenic potential of HPVs. This virus-induced carcinogenicity has evolved and co-diverged with humans over time. For instance, most carcinogenic HPV genotypes belong to the same evolutionary clade in the α genus — these and related HPV genotypes are considered as potentially carcinogenic. From an epidemiological perspective, understanding the carcinogenicity of HPV genotypes in invasive cervical cancer (ICC) is key to predicting the effectiveness of HPV vaccination and screening programmes. Towards this goal, in a recent study published in The Lancet, Wei and colleagues conducted a comprehensive systematic review to evaluate the proportion of ICC in the population that is attributable to individual HPV genotypes (known as the attributable fraction, or AF) at global and regional levels. We selected this paper because it highlights the wide diversity of HPV types with attributable causality of ICC, and serves as a stepping stone towards harnessing this knowledge for improved prevention strategies.

The authors analysed 1,174 studies that performed HPV genotyping on cervical swabs or biopsies, including more than 110,000 HPV-positive ICC cases and 2,750,000 individuals with normal cervical cytology from 121 countries. Genotype-specific prevalences were compared between ICC cases and people with normal cytology and used to calculate odds ratios (ORs), showing whether one group is more or less likely to be associated with individual HPV types. HPV genotypes with an OR significantly higher than one were deemed causal of ICC. To estimate the regional AFs, attributable risks were derived from the ORs and multiplied by the regional HPV genotype-specific prevalence. Global AFs were estimated by weighting the regional AFs by the number of ICC cases in each region. The authors then used a Bayesian model to generate AF estimates, considering both HPV prevalence in ICC and ORs, and providing the 95% credible intervals (a Bayesian analogue to confidence intervals).