The underlying risk of cervical precancer in women with atypical squamous cells of undetermined significance (ASC-US) or low-grade squamous lesions (LSIL) should determine management. Genotyping for the most important carcinogenic HPV types (HPV16 and 18) could identify those women at highest risk requiring colposcopy or more intensive follow-up.
A meta-analysis was performed to assess the diagnostic accuracy of genotyping for HPV16, HPV16 & HPV18 (HPV1618) and for high-risk types (hrHPV) to detect prevalent cervical intraepithelial neoplasia, grade 2 or 3 or cancer (CIN2+, CIN3+) in women with ASC-US or LSIL cervical cytology. A literature search was performed in three electronic databases to identify eligible studies. Authors were contacted to request additional non-published data. Data pooling was performed using a bivariate normal model designed for diagnostic test accuracy, taking the intrinsic negative correlation between sensitivity and specificity into account. The clinical utility of the triage strategies were illustrated by pretest-posttest probability (PPP) plots.
Twenty-three studies evaluating 15 different HPV assays met criteria for inclusion. The pooled sensitivity of HPV16 genotyping to detect CIN2+ was 54% (95%CI [CI]: 50-58%) and 50% (CI: 47-54%) in ASC-US and LSIL, respectively. The pooled specificity was 86% (CI: 83-89%) and 83% (CI: 80-85%), in ASC-US and LSIL, respectively. The sensitivity was 10-14% higher and the specificity was 3-4% lower for the detection of CIN3+. Adding HPV18 increased sensitivity for CIN2+ compared to testing only for HPV16: ratios of 1.10 (CI: 1.03-1.19) in ASCUS and 1.11 (CI: 1.04-1.19) in LSIL. Adding HPV18 significantly decreased the specificity: ratio of 0.96 (CI: 0.94-0.97) for ASC-US and 0.93 (CI: 0.91-95) for LSIL. The gain in sensitivity and the loss in specificity were similar for detection of CIN3+.
The triage performance of HPV genotyping tests was demonstrated in PPP plots with predefined local decision thresholds. A post-test risk above the referral decision threshold suggests immediate referral to colposcopy while a post-test risk under the threshold suggests increased surveillance or release to routine screening. The average risk of underlying CIN3+ (PPV) was 17% and 19% in HPV1618-positive women with ASC-US or LSIL, respectively. The risk of CIN3+ among women not carrying HPV16 or 18 was 2% for ASC-US and 4% for LSIL. Being HPV1618 negative but positive for other hrHPV types is associated with a risk of CIN3+ of 4% and 5% for ASC-US and LSIL, respectively. Whether this risk is sufficiently low to avoid colposcopy and to recommend delayed retesting depends on local guidelines.