Evaluation of triage strategies for high-risk human papilloma virus testing as primary screening for cervical cancer / 肿瘤

ABSTRACT

Objective: To evaluate the performance and application value of several triage strategies for high-risk human papillomavirus (HR-HPV) DNA testing as primary screening of human cervical cancer. Methods: The data from cross-sectional population-based studies for screening human cervical cancer carried out in Xiangyuan county of Shanxi province, China, 2003 were reanalyzed. Totally 1788 women were included in this study. All women were screened by liquid-based cytology (LBC), HR-HPV DNA testing-hybrid capture 2 (HPV-HC2) (for primary screening and analysis of high- and low-level viral loads), and visual inspection with acetic acid (VIA), then all positive women in any test were referred for colposcopy. If the colposcopy result was positive, the woman was referred for biopsy for the final pathological result. This study simulated the strategy in which HPV-HC2 testing was used as primary screening of cervical cancer, then LBC, VIA, HPV-HC2 high- and low-level viral load tests were used as the triage for HPV-positive women. The above strategies were compared by sensitivity, specificity, colposcopy referral rate and other indexes for high-grade squamous intraepithelial lesion (HSIL) and invasive cervical cancer. Moreover, the application value of the above strategies was analyzed by receiver operator characteristic (ROC) curve and the area under curve (AUC) (αcorrection = 0.0125). Results: The colposcopy referral rate, sensitivity, specificity and positive predictive value (PPV) of HPV primary screening with HPV-HC2 for detecting HSIL and invasive cervical cancer were 18.1%, 95.7%, 85.0% and 20.4%, respectively. After HPV primary screening, the referral rates of three strategies of triage with LBC, VIA and HR-HPV-HC2 low viral load were decreased to 8.7%, 4.5% and 9.8%, respectively; the specificity values were increased to 94.3%, 97.5% and 93.1%, respectively; the positive predictive values were increased to 37.2%, 46.9% and 32.0%, respectively; the sensitivity values were decreased to 84.1%, 55.1% and 81.2%, respectively. The difference of AUC between HPV primary screening and its combination with the triage of LBC or HPV-HC2 low viral load was not statistically significant (both P > 0.012 5), but the difference of AUC between HPV primary screening and its combination with the triage of VIA or HPV-HC2 high viral load was statistically significant (both P < 0.001). Conclusion: Considering the economy and adaptability, increasing the cut-off of HPV DNA testing (cut-off ≥10 pg/mL) to improve the screening efficiency of human cervical cancer is feasible and effective in low-resource and less-developed areas. HR-HPV primary screening with the triage of LBC can be introduced widespreadly in high-developed areas. HR-HPV primary screening with the triage of VIA is feasible in rural areas, but the skills of the health care providers responsible for testing should be trained and improved.