HPV prevalence and accuracy of HPV testing to detect high‐grade cervical intraepithelial neoplasia

Abstract

Concern was raised on using testing for high-risk (HR) human papillomavirus (HPV) in cervical cancer screening in populations where HPV prevalence is high. The impact of HR HPV prevalence on the efficiency of HPV test-based screening has never been directly evaluated. A meta-regression of the relationship between HR HPV prevalence and the specificity and positive predictive value (PPV) of HPV DNA testing for the presence of cervical intraepithelial neoplasia grade 2 or worse (CIN2+) was performed. Only studies that used Hybrid Capture 2 (HC2) were included. Country income (low-medium vs. high) was used as a proxy of previous screening. Twenty-six populations from 20 studies were included. For a 10% increase in HR HPV prevalence, HC2 specificity decreased by 8.41% [95% confidence interval (CI): 8.02-8.81], whereas PPV increased by 4.74% (95% CI: 2.45-7.03). HR HPV prevalence explained 98% of the variability in HC2 specificity and 38% of the variability in PPV. Country income did not affect specificity, but low-medium income was associated with higher PPV (3.81%; 95% CI: 1.53-6.10) after adjustment for HR HPV prevalence. When HR HPV prevalence is high, the specificity of HPV testing for CIN2+ decreases, but PPV does not decrease and it is high in inadequately screened populations. The number of HPV-positive women needing further assessment or treatment per CIN2+ case detected will therefore decrease and screening efficiency will improve. This is explained by the fact that HR HPV causes CIN2+: an increase in HR HPV prevalence is inevitably accompanied by an increase in CIN2+.