Evaluation of laboratory based molecular technologies for the detection of human papillomavirus

Abstract

Human papillomaviruses (HPV) are associated with over 95% of cervical cancer (CC) incidence. Cytology screening using the Papanicolaou test (Pap smear) to detect potentially cancerous cells is a current ‘gold standard’ for identifying CC risk. There is growing evidence that molecular assays detecting high-risk HPV genotypes offer a relatively more sensitive and less invasive CC screening alternative to cytology. However, some of these available HPV DNA testing platforms have little to no published validation data and others have high costs per test. More analytical evaluations of HPV tests and optimization of front-end processing variables, e.g., nucleic acid extraction, are required to establish feasible workflows for high-throughput processing. Thus, this study sought to compare the performance of three extraction protocols on residual liquid-based cytology (LBC) and to evaluate singleplex and duplex laboratory developed tests (LDT) to determine an optimal workflow for HPV detection. Spin column, heat lysis and magnetic bead-based extraction were performed on 46 residual LBC specimens and the extracted DNA for each was tested on singleplex HPV16 and HPV18 real-time PCR assays to determine the most suitable extraction method. After the best-performing extraction method was determined, it was applied to processing additional residual specimens that were tested on singleplex (n=61) and duplex PCR (n=175) assays targeting HPV genotypes 16, 18, 31, 35, 45, 52 and 58. The PCR results were compared to the BD Onclarity™ HPV assay. Automated magnetic bead-based extraction had higher average agreement (92.4%) compared to spin column extraction (90.3%) and heat lysis (90.2%) across both HPV16 and HPV18. Singleplex PCR had 70.0 – 95.0% overall agreement across targeted genotypes while the duplex PCR had 93.6% overall agreement when compared to the BD Onclarity™ HPV assay. The duplex assay had good intra- (%CV less than 3%) and inter- run performance (%CV less the 7%), comparable to commercial tests.