Materials and Methods
Multicenter Clinical Trial Design
A total of 5,415 eligible women were enrolled with informed consent at 25 collection sites in the United States between September 2010 and February 2012. The selection criteria were as follows: age younger than 35 years; any age with high-risk status, defined as having been previously diagnosed with an abnormal Papanicolaou or positive HPV test; or not being screened in the previous 5 years. Two cervical samples were taken from each patient in accordance with the manufacturers' recommendations: a liquid-based cytology (LBC) sample was collected using the BD SurePath collection device and medium (BD Diagnostics, Burlington, NC) and a second sample was collected using the Digene HC2 DNA collection device and placed in STM (Qiagen). An LBC result was obtained from each patient and patients with either abnormal cytology or a positive HC2 result were referred to colposcopy, in which a four-quadrant biopsy specimen was taken via endocervical curettage. All testing was performed centrally in three independent laboratories. Abnormal cytology and all histology were adjudicated by an expert pathology panel and the consensus result was considered to be the final study result.
Retrospective Cohort Study Design
A total of 541 (10%) residual BD SurePath specimens were selected from the multicenter clinical trial samples for testing with the BD Onclarity HPV Assay. These included all samples with biopsy-confirmed CIN 2 (n = 104), 79 samples with abnormal cytology but lower than CIN 2 on cervical biopsy, and 358 samples that were negative for intraepithelial lesion or malignancy (NILM) on cytology and lower than CIN 2 on cervical biopsy. Residual samples were stored at room temperature and shipped on cold packs from the trial sites to BD within 60 days of collection where they were immediately stored at −20°C for 8 to 24 months before testing. All samples were tested in a blinded fashion using the BD HPV assay. The results were subsequently compared with the HC2 results obtained in the multicenter study from the fresh STM specimens.
HPV Testing
HC2 testing was performed at the time of collection in accordance with the manufacturer's recommended protocol using the STM specimens collected in the multicenter trial. BD Onclarity HPV Assay was used on retrospective BD SurePath residual vial specimens using the fully automated Viper LT system (BD Diagnostics, Sparks, MD). The details of the BD Viper LT system will be reported elsewhere. Briefly, residual (pregradient) SurePath samples were brought to room temperature, vortexed briefly, and 0.5 mL of specimen was transferred to 1.7 mL of a proprietary BD HPV LBC diluent. The resultant mixture was heated at 120°C for approximately 30 minutes to lyse and homogenize the specimen and then cooled to room temperature. A volume of 0.8 mL of the resultant lysate was then extracted using BD Viper ferric oxide particle DNA binding and magnetic extraction and eluted in 400 μL of elution and neutralization buffer. The BD Onclarity HPV Assay design selection criteria included the ability to detect 100 copies of one HPV type in the presence of one million copies of competing HPV targets, which ensures that mixed infections can be reliably detected. It is a three-well, four-channel real-time PCR test that can provide individual genotyping information for six HPV types (HPV 16, 18, 31, 45, 51, and 52) while simultaneously screening for all 14 high-risk viruses. The rest of the 14 high-risk types are reported in groups of two or three viruses (HPV 33/58, HPV 56/59/66, and HPV 35/39/68). Each of three tubes contains fluorescent real-time PCR probes for four separate optical channels including one channel for human β-globin sequence, which acts as a sample adequacy and sample processing control Table 1 . Aliquots (50 μL) of the DNA eluate were dispensed into each of three PCR tubes (G1, G2, and G3) to rehydrate the dried-down master mix and hot start Taq polymerase. PCR was performed using the following parameters: 95°C enzyme activation step for 15 minutes followed by 40 cycles at 95°C for 30 seconds and 55°C for 60 seconds. HPV-positive specimens were identified using a previously defined cycle threshold algorithm method that was derived using receiver operating characteristic curves.
The BD Onclarity HPV Assay participated in the WHO HPV Laboratory Network (LabNet) HPV DNA Proficiency Study, 2013. The assay correctly identified each of the 46 blinded panel samples and was therefore found to be "proficient for detection of HPV 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, 68a, and 68b" (personal communication, Joakim Dillner, MD, WHO HPV LabNet, International HPV Reference Center, Stockholm, Sweden).
Denaturing High-Performance Liquid Chromatography (DHPLC) Double-Stranded Sequencing Discordant hrHPV Test Method
Proprietary HPV type-specific primers (Transgenomic Inc, Omaha, NE) were designed to detect 14 high-risk strains (HPV 16, 18, 31, 33, 35, 39 45, 51, 52, 56, 58, 59, 66, and 68) and 11 low-risk genotypes (HPV 6, 11, 26, 30, 53, 67, 69, 70, 82, 85, and 97) in the L1 region of the virus. Prototype designs were validated empirically using human and HPV type-specific DNA. Residual SurePath samples (blinded to the original HPV results) were extracted using a commercially available extraction kit and the purified DNA was used for PCR amplification using gene-specific primer sets. Each PCR reaction also contained a primer pair specific for the human KRAS gene that acted as an internal (sample adequacy/processing) control. Each experimental run was performed using a positive control containing 8,000 copies of the relevant plasmid DNA from each of the 25 HPV genotypes in the assay and 60 ng of human genomic DNA. Negative run controls were also included to ensure system integrity. Analysis of the strain-specific amplifications was performed on the WAVE HS System (Transgenomic Inc) Figure 1. Each amplification reaction was routinely run on a triethylammonium acetate/acetonitrile gradient under nondenaturing conditions and compared using plasmid control amplifications. Any amplification that showed a peak at the expected retention time for a given genotype was bidirectionally sequenced using standard Sanger sequencing. Approximately 100 bases of the resulting sequence for any predicted strain was then confirmed using nucleotide sequencing alignment and the nucleotide database of the National Library of Medicine (Basic Local Alignment Search Tool). Primer designs were optimized to maximize the distance between individual genotype peaks so that even mixed infections would be resolved. Mixed infections that resulted in fully or partially overlapping peaks were not typically observed. However, the DHPLC method can be adjusted to resolve these samples by rerunning the DNA mixture under partially denaturing conditions on the column. The peak-to-peak separation is then increased, allowing the different DNA species to be separated before they are collected in individual fractions for sequencing.
(Enlarge Image)
Figure 1.
Ability of the denaturing high-performance liquid chromatography (DHPLC) technique to resolve mixed human papillomavirus (HPV) infections. The WAVE HS System DHPLC chromatogram shows the analysis of the HPV type-specific L1 polymerase chain reaction amplifications. Each amplification reaction contains HPV-specific primers and human KRAS (internal control) primers. The top trace shows the results of a UV detector scan of eluted DNA fragments from a mixture of nine different HPV types (indicated on top of figure) eluted from the DHPLC column and illustrates how different mixed HPV genotypes are separated. Each individual peak corresponds to the designated HPV type. The lower traces show the results from the corresponding HPV types run individually on the same column. The dotted vertical lines demonstrate that the genotype-specific peaks had the same retention time on the column whether present as a mixture or individually. In patients with mixed infections, the different genotypes are separated on the column, eluted as distinct fractions, and then sequenced individually using double-stranded Sanger sequencing chemistry.
Statistical Analysis
Test performance estimates were calculated with exact 95% confidence intervals (CIs) using the Pearson-Klopper method in the binom R package. McNemar exact test comparisons were calculated using the exact 2 × 2 R package. Differences between the HC2 and BD HPV assays were calculated using Fisher exact test in Minitab 16 statistical software (Minitab, State College, PA). The risk for disease among patients with a given genotype relative to HPV-negative patients was calculated by unconditional maximum likelihood estimation using the epitools R package.