The Clinical Performance of Primary HPV Screening, Primary HPV Screening Plus Cytology Cotesting, and Cytology Alone at a Tertiary Care Hospital

The Clinical Performance of Primary HPV Screening, Primary HPV Screening Plus Cytology Cotesting, and Cytology Alone at a Tertiary Care Hospital Jung-Woo Choi MD, PhD; Younghye Kim MD, PhD; Ju-Han Lee MD, PhD; and Young-Sik Kim MD, PhD BACKGROUND: Algorithms for primary human papillomavirus (HPV) screening, primary HPV screening plus cytology cotesting, and cytology alone were evaluated previously in large cohort trials for cervical cancer detection, although the quality of cytology in those studies was controversial. To investigate whether these 3 algorithms would be applicable in routine practice at a tertiary care hospital, the authors compared their clinical performance. In addition, the prevalence of HPV genotypes was determined. METHODS: Cervical cytology samples (n 5 1000) were tested using liquid-based cytology (LBC), a nucleic acid hybridization assay, real-time polymerase chain reaction analysis, and direct HPV DNA sequencing. The clinical performance of the 3 algorithms was compared among women in different age groups (age range, 17-86 years; median age, 44.7 years). RESULTS: For cervical intraepithelial neoplasia grade 2 or worse (CIN 21), the sensitivity of primary HPV screening alone, cotesting, and LBC alone was 71.7%, 72.5%, and 63.8%, respectively; whereas the specificity was 87.5%, 96.5%, and 97.4%, respectively. Cotesting and LBC alone had slightly higher positive predictive values for CIN 2 1 (97.8% and 98.9%, respectively) than primary HPV screening alone (91%), whereas primary HPV screening alone and cotesting demonstrated higher negative predictive values (63.6% and 62.5%, respectively) than LBC alone (43.2%). High-risk HPV types were detected in 24.3% of individuals. The most common type was HPV type 16 (HPV-16) followed by multiple HPV infections and HPV-58, HPV-52, HPV-31, HPV-35, HPV-51, HPV-39, HPV-56, HPV-33, HPV-18, HPV-59, and HPV-45. CONCLUSIONS: Primary HPV screening alone in a tertiary care hospital demonstrated a performance that was equivalent to that of cotesting for CIN 21, irrespective of patient age. With regard to the distribution of HPV genotypes, the nonavalent HPV vaccine would prevent approximately 60% of high-risk HPV. Cancer (Cancer Cytopathol) 2016;124:144-52. VC 2015 American Cancer Society. KEY WORDS: cervical intraepithelial neoplasia; genotype; human papillomavirus (HPV); real-time polymerase chain reaction (PCR); screening. INTRODUCTION Cytology has been the conventional approach for cervical cancer screening; however, the use of cytology as a screening test has been restricted because of its relatively low sensitivity for high-grade cervical dysplastic lesions as well as its complex and subjective diagnostic categories. 1 Moreover, cytology identifies women who have cervical intraepithelial neoplasia 2 or worse (CIN 21) but not those who have a potential risk for developing CIN 21. In 2000, these limitations led to the advent of various high-risk (HR) human papillomavirus (HPV) assays, such as Hybrid Capture 2 (HC2) (QIAGEN, Hilden, Germany), as cytology adjuncts for cervical cancer screening. HPV testing has superior sensitivity compared with cytology for detecting cervical precancerous lesions and permits the early detection and prevention of cervical cancers, thus reassuring women who have negative screening Corresponding author: Young-Sik Kim, MD, PhD, Department of Pathology, Korea University Ansan Hospital, 123, Jeokgeum-Ro, Danwon-Gu, Ansan-Si, Gyeonggi-Do, 425-707, Republic of Korea; Fax: (011) 8231-412-5324; apysk@korea.ac.kr Department of Pathology, Korea University Ansan Hospital, Ansan, Korea We thank Drs. Yong-Suk Nam and Sung-Won Hong for their technical assistance. Received: August 16, 2015; Revised: September 21, 2015; Accepted: September 22, 2015 Published online October 12, 2015 in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/cncy.21632, wileyonlinelibrary.com 144 Cancer Cytopathology February 2016

Screening Algorithms for Cervical Cancer/Choi et al results. 2 HPV has been the primary risk factor for cervical cancer, 3 and12hrhpvgenotypes(hpvtype16[hpv- 16], HPV-18, HPV-31, HPV-33, HPV-35, HPV-39, HPV-45, HPV-51, HPV-52, HPV-56, HPV-58, and HPV-59) belong to group I carcinogens according to the International Agency for Research on Cancer. 4 Among these, HPV-16 and HPV-18 progress to CIN 3 1 more frequently than the other HR HPV types, accounting for 68% to 82% of cervical cancers worldwide. 5,6 In the United States, either HPV screening plus cytology cotesting every 5 years or primary cytology with HPV tests every 3 years for the triage of patients who have equivocal results, such as atypical squamous cells of undetermined significance (ASC-US), has been recommended as primary cervical cancer screening for women ages 30 to 65 years. 7 In April 2014, based on results from the Addressing the Need for Advanced HPV Diagnostics trials, 8 the US Food and Drug Administration approved primary HPV screening for women aged 25 years, followed by HPV-16/HPV-18 genotyping and/or reflex cytology. Several large trials reported that primary HPV screening was superior to cytology alone and was at least as effective as cotesting at the same screening intervals. 8 10 Compared with cytology-based screening, the primary HPV screening algorithm is considered a simple and economically efficient strategy for screening patients in clinical practices and has advantages for the detection of adenocarcinoma. 11 However, the clinical performance of primary HPV screening and cotesting with liquid-based cytology (LBC) has not been compared in routine clinical practice. Moreover, cytology results in the trial cohorts were neither highly qualified nor consistent. The objective of this study was to compare the effectiveness of primary HPV screening, primary HPV screening plus LBC cotesting, and LBC alone and determine whether the new algorithms would be applicable in routine practice at a tertiary care hospital. To accomplish this goal, we developed an in-house, real-time HPV genotyping polymerase chain reaction (PCR) system, which was comparable to direct HPV DNA sequencing. The performances of the 3 algorithms were assessed for their ability to detect high-grade cervical dysplasia or worse in cervical cytology and/or biopsy samples. In addition, we determined the prevalence of HPV genotypes using our validated real-time HPV PCR system to identify which HR HPV types could be protected against by the newly licensed nonavalent HPV vaccine. MATERIALS AND METHODS Clinical Samples One thousand consecutive cytology specimens were obtained from women who visited the gynecology clinics at the Korea University Ansan Hospital for routine screening or follow-up between November 2011 and May 2012. This study was approved by the institutional review board (IRB) of Korea University Ansan Hospital (IRB no. AS12071). For each patient, cervical samples were taken using a cytobrush. The acquired cells were prepared for BD SurePath liquid-based Papanicolaou (Pap) tests (Becton, Dickinson and Company, Franklin Lakes, NJ) and were collected in Cervical Sampler solution (QIAGEN) for HR HPV detection and/or genotyping by real-time PCR, HC2, and direct HPV DNA sequencing. The mean age of the study population was 44.7 years (range, 17-86 years). Cytology results were available from 987 women based on specimen adequacy criteria according to the 2004 Bethesda System. A colposcopic biopsy was ordered for 197 women with ASC-US or worse and/or HPV-positive results. HC2 Assay Collectively, the HC2 test detects 13 oncogenic HPV types (HPV-16, HPV-18, HPV-31, HPV-33, HPV-35, HPV- 39, HPV-45, HPV-51, HPV-52, HPV-56, HPV-58, HPV-59, and HPV-68), and we performed the test according to the manufacturer s instructions. Briefly, HPV DNA was denatured and then hybridized with a specific HPV RNA probe cocktail, resulting in the formation of RNA-DNA hybrids. The hybrids were captured by antibodies specific to the RNA-DNA hybrids. Immobilized hybrids were reacted with alkaline phosphatase-conjugated antibodies and detected with a chemiluminescent substrate. The light signals were measured using a luminometer and are reported in relative light units/cutoff; samples that had 1.0 relative light units/cutoff were regarded as positive. Real-Time HPV Genotyping PCR DNA was extracted from specimens using the QIAmp DNA mini kit (QIAGEN) and was eluted into 200 ll Tris-ethylene diamine tetraacetic acid buffer. The primers and probes for the multiplex PCR assay were designed using the software program Primer Express (ThermoFisher Scientific, Waltham, Mass), and target genes of the HPV Cancer Cytopathology February 2016 145

TABLE 1. Primers and Probes for Human Papillomavirus Genotyping by Multiplex Real-Time Polymerase Chain Reaction HPV Type Primer and Probe Sequences Target Gene Amplicon Size, bp HPV-16 F: 5 0 -TTC GGT TGT GCG TAC AAA GC-3 0 E7 62 R: 5 0 -GCC CAT TAA CAG GTC TTC CAA A-3 0 P: 5 0 -FAM-CAC ACG TAG ACA TTC G-MGB-3 0 HPV-18 F: 5 0 -CGC GCT TTG AGG ATC CAA-3 0 E6 58 R: 5 0 -AGT TCC GTG CAC AGA TCA GGT A-3 0 P: 5 0 -VIC-ACG GCG ACC CTA CAA-MGB-3 0 HPV-31 F: 5 0 -ATG TTC AAA AAT CCT GCA GAA AG-3 0 E6 65 R: 5 0 -GGT ATT TCC AAT GCC GAG CTT-3 0 P: 5 0 -VIC-CTC GGA AAT TGC ATG AA-MGB-3 0 HPV-33 F: 5 0 -AAA AAC CAC GAA CAT TGC ATG A-3 0 E6 72 R: 5 0 -CGC ACT GTA GTT CAA TGT TGT GTA TAG-3 0 P: 5 0 -FAM-TTG TGC CAA GCA TTG-MGB-3 0 HPV-35 F: 5 0 -TTA CAT GTC AAA AAC CGC TGT GT-3 0 E6 75 R: 5 0 -CGA TGT TAT GGA ATC GTT TTT TTT C-3 0 P: 5 0 -VIC-CAG TTG AAA AGC AAA GAC AT-MGB-3 0 HPV-39 F: 5 0 -TGA TAT AGT ACT GTA TAT GTA TGT GCA TTG TGC-3 0 E5 94 R: 5 0 -GGT GGG AAA CCA TGT TTA TTA GTA ACA A-3 0 P: 5 0 -FAM-AAC TAC TGT ACA TAG CTT T-MGB-3 0 HPV-45 F: 5 0 -ATG GAG TTA GTC ATG CAC AAC T-3 0 E7 85 R: 5 0 -CCG TCA CAC TTA CAA CAT ACA CAC A-3 0 P: 5 0 -ROX-CCA GCC CGA CGA GCC GAA CC-BHQ3-3 0 HPV-51 F: 5 0 -GGG CCT GAA GAA AAG CAA AA-3 0 E6 59 R: 5 0 -CCC GCT ATT TCA TGG AAC CTT-3 0 P: 5 0 -FAM-TTG GTG GAC GAA AAA-MGB-3 0 HPV-52 F: 5 0 -AAG AAA AAG AAA GAC ATG TTA ATG CAA A-3 0 E6 81 R: 5 0 -AAC ACT CTG AAC AGC GCC CT-3 0 P: 5 0 -ROX-TCC AAC GAC CCA TAA TAT TAT GAA ATC-BHQ3-3 0 HPV-56 F: 5 0 -AAG CAA TTG CAT TGT GAC AGA AA-3 0 E6 65 R: 5 0 -CAT GAC CCG GTC CAA CCA-3 0 P: 5 0 -VIC-AGA CGA TTT CAT CTA ATA GC-MGB-3 0 HPV-58 F: 5 0 -CAC GGA CAT TGC ATG ATT TGT G-3 0 E6 91 R: 5 0 -CGC TGC AAA GTC TTT TTG CAT T-3 0 P: 5 0 -ROX-CAG GCG TTG GAG ACA TCT GTG CAT GA-BHQ3-3 0 HPV-59 F: 5 0 -CTT GTG TGC TAC GAG CAA TTA CCT-3 0 E7 91 R: 5 0 -GTC TAG CTA GTA GCA AAG GAT GAT TAA CTC-3 0 P: 5 0 -ROX-CTC CGA CTC CGA GAA TGA AAA AGA TGA ACC-BHQ3-3 0 Control F: 5 0 -CTG GCA CCC AGC ACA ATG-3 0 b-actin 69 R: 5 0 -GCC GAT CCA CAC GGA GTA CG-3 0 P: 5 0 -Cy5-ATC AAG ATC ATT GCT CCT CCT GAG CGC-BHQ2-3 0 Abbreviations: bp, base pairs; HPV, human papillomavirus; F, forward; HPV, human papillomavirus; P, probe; R, reverse. genome were from the E5, E6, and E7 regions (Table 1). All sequences of the primers and probes were verified by the Basic Local Alignment Search Tool (BLAST) analysis (National Library of Medicine, National Institutes of Health, Bethesda, Md). The melting temperature (Tm) of each primer and probe and the amplicon size were designed to be similar to each other to ensure optimum multiplexing. The Tm of the primers was between 588C and 608C, whereas the Tm of the probes was between 688C and 708C. The amplicon lengths were between 58 and 94 base pairs. b-actinwasusedasaninternalcontrol for checking the quality of DNA in the samples. The multiplex real-time PCR assay was performed in 20 ll ofa reaction mixture containing 4 ll of DNA, 8 ll of 2 3 PowerAmp real-time PCR master mixes (Thermo- Fisher Scientific), and 4 ll of each primer and probe mixture. The primer and probe mixtures contained 500 nm of each primer, 100 nm of each probe, and an internal positive control. The PCR reactions were carried out in the ABI 7500 real-time PCR system (Applied Biosystems, Waltham, Mass) under the following conditions: 2 minutes at 508C, 10 minutes at 948C, and 35 cycles consisting of 15 seconds at 958Cand1minuteat608C. Negative controls without template DNA were included in each run. Fluorescent data were collected and analyzed (Kogene Biotech, Seoul, Korea). Cycle threshold (CT)-values were plotted against the log 10 value of target DNA. A correlation coefficient > 0.98 for each HPV type was obtained. The assay cutoff level for CT values was set between 30.0 and 35.0 according to HR HPV genotype. 146 Cancer Cytopathology February 2016

Screening Algorithms for Cervical Cancer/Choi et al TABLE 2. Primers for Human Papillomavirus Genotyping by Direct Sequencing HPV Type Primer Sequences Target Gene Amplicon Size, bp HPV-16 F: 5 0 -GTT GGA CAT CCC TAT TTT CCT AT-3 0 L1 284 R: 5 0 -GCA TTT GCT GCA TAA GCA C-3 0 HPV-18 F: 5 0 -ATA TTT TAG GGT TCC TGC AGG-3 0 L1 290 R: 5 0 -CCC TAA CGT CCT CAG AAA CA-3 0 HPV-31 F: 5 0 -GCA GTG CTA GGC TGC TTA CAG TA-3 0 L1 221 R: 5 0 -CCC GCG ACC TAC CTC TAA A-3 0 HPV-33 F: 5 0 -TAT TTT TCT ATT AAA AAT CCT ACT AAC GC-3 0 L1 202 R: 5 0 -CGC CTA ATG GCT GCC CT-3 0 HPV-35 F: 5 0 -ACC CAT ACT ATG CTA TTA AAA AAC AAG-3 0 L1 292 R: 5 0 -CTG TTA TCT GTA CCA GAG TTA CCA A-3 0 HPV-39 F: 5 0 -TAA AGT GGG TAT GAA TGG TGG-3 0 L1 275 R: 5 0 -TGT CCT TAT TGG TGG TTG ATG-3 0 HPV-45 F: 5 0 -GGT TGT ACC TAA TGG TGC AGG-3 0 L1 289 R: 5 0 -ATT ATC CCT AAC ATC CTG CGT-3 0 HPV-51 F: 5 0 -CCC TAT TTT CCA ATA CCT AAA ACC T-3 0 L1 316 R: 5 0 -TCT GTT TGT TGT CAA CAG ATG TG-3 0 HPV-52 F: 5 0 -TAA AAA CAC CAG TAG TGG TAA TGG-3 0 L1 287 R: 5 0 -CCC TAT TAT CTA TAC CAG GTT TAC CA-3 0 HPV-56 F: 5 0 -CTA AGG ACA ATA CCA AAA CAA ACA-3 0 L1 288 R: 5 0 -CAA CTG ATA TAT TGT CCC TAC TAT CTT C-3 0 HPV-58 F: 5 0 -GGC AAT CCA TAT TTT TCC ATC-3 0 L1 311 R: 5 0 -GAT AAG CAT TCC CTG TTA TCA GAC-3 0 HPV-59 F: 5 0 -TTA AAG TAC CTA AAG GTG GTA ATG G-3 0 L1 297 R: 5 0 -CAG ATA CAT TAT CAC GTG TAT CTT TG-3 0 Abbreviations: bp, base pairs; F, forward; HPV, human papillomavirus; R reverse. Direct HPV DNA Sequencing Target genes from the L1 region of the HR HPV genome were amplified by PCR using genotype-specific primer sets (Table 2). PCR products were purified and sequenced using the BigDye Terminator Cycle Sequencing kit (Applied Biosystems). The reaction products were analyzed with an ABI Prism 310 Genetic Analyzer (Applied Biosystems), and DNA sequences were aligned with the BLAST database on the National Center for Biotechnology Information website to confirm specific genotypes. Three Cervical Cancer Screening Algorithms For women in different age groups, we retrospectively calculated and compared the clinical performance of the 3 screening strategies (Fig. 1). One strategy was a primary HPV screening in which women who were positive for HPV-16/HPV-18 and those who were positive for 12 other HR HPV types who had abnormal cytology results were referred to colposcopy. For the second strategy, we used cotesting with both LBC and the HR HPV test, which is currently the preferred method according to US screening guidelines. With cotesting, it is recommended that women who are positive for both tests, those who are HPV-negative but have abnormal cytology results (highgrade squamous intraepithelial lesion [HSIL], atypical squamous cells cannot rule out HSIL [ASC-H], or cancer), and HPV-16/HPV-18 positive women who have negative cytology results undergo colposcopy. The third strategy was LBC testing, and women who had LBC results of either >ASC-US or ASC-US with HR HPV positivity were referred for colposcopic biopsy. In all 3 screening algorithms, HR HPV positivity and genotyping for HR HPV types were determined using multiplex, real-time PCR. Statistical Analysis The chi-square test was used to compare the proportions of HR HPV-positive results between real-time PCR, HC2, and direct HPV DNA sequencing. Overall concordant rates and Cohen j values for HC2 and direct HPV DNA sequencing were calculated using the real-time PCR result as the reference. The sensitivities and specificities of HPV tests were determined based on the results of cytology (HSIL or worse [HSIL1]) or biopsy (CIN 21). The sensitivities, specificities, positive predictive values, and negative predictive values for all 3 screening algorithms were calculated using either CIN 2 1 or CIN 3 1 as the gold standard. Statistical significance was defined as P <.05. All statistical analyses were performed using the software package SPSS for Windows 10.0 (SPSS Inc, Chicago, Ill). Cancer Cytopathology February 2016 147

RESULTS Validation of Real-Time HPV Genotyping PCR Figure 1. The 3 screening algorithms for cervical cancer detection are illustrated: primary human papillomavirus (HPV) screening alone, liquid-based cytology (LBC)/HPV cotesting, and LBC alone. ASC-US indicates atypical squamous cells of undetermined significance; ASC-H, atypical squamous cells cannot rule out high-grade squamous intraepithelial lesion; HPV2, HPV-negative; HPV1, HPV-positive; HR, high-risk; HSIL, high-grade squamous intraepithelial lesion; LSIL, low-grade squamous intraepithelial lesion. Among the 1000 samples, 243 (24.3%), 267 (26.7%), and 252 (25.2%) were positive for HR HPV types according to the results from real-time HPV genotyping PCR, HC2, and direct HPV DNA sequencing, respectively. Real-time HPV genotyping PCR demonstrated a concordant rate of 88.8% with HC2 (j 5 0.705) and 98.9% with direct HPV DNA sequencing (j 5 0.970) (Table 3). For detecting HSIL1, the performance of real-time HPV genotyping PCR was equivalent to that of direct HPV DNA sequencing but had reduced sensitivity and a superior specificity compared with HC2. HSIL 1 was detectable by real-time HPV genotyping PCR, HC2, and direct HPV DNA sequencing in 92 (91.1%), 96 (95%), and 93 (92.1%) of the 101 samples, respectively. Of the 97 samples with CIN 21, 89 (91.8%), 93 (95.9%), and 90 (92.8%) were positive according to the results from realtime HPV genotyping PCR, HC2, and direct HPV DNA sequencing, respectively. These results suggest that, like other HPV assays, the newly designed real-time HPV genotyping PCR approach could be used as an effective and validated screening test. Comparison of the 3 Screening Algorithms Of the 1000 women, 922 women between ages 25 and 65 years were selected for the current analysis, because women within this age range are examined for HPV according to current US screening and management guidelines. Among allagegroups,womenbetweenages25and29yearshad the highest prevalence of HR HPV (33 of 70 women; 47.1%) and frequently had abnormal cytology and biopsy results (Table 4). The infected genotypes were primarily HPV-16/HPV-18 (15 of 33 women; 45.5%). The sensitivity of primary HPV screening, cotesting, and LBC alone for the detection of CIN 2 1 was 71.7%, 72.5%, and 63.8%, respectively. Sensitivity for detecting CIN 3 1 was highest with primary HPV screening (62.8%), followed by TABLE 3. Comparison of Human Papillomavirus Detection by Real-Time Polymerase Chain Reaction, Hybrid Capture 2 and Direct Sequencing a Real-Time PCR Results HPV Assay Result Positive, n 5 243 Negative, n 5 757 Total No. Concordance Rate, % j Coefficient b HC2 Positive 199 68 267 88.8 0.705 Negative 44 689 733 Direct sequencing Positive 242 10 252 98.9 0.97 Negative 1 747 748 Abbreviations: HC2, Hybrid Capture 2 (QIAGEN, Hilden, Germany); HPV, human papillomavirus; PCR, polymerase chain reaction. a Chi-square tests were used to determine the associations (real-time PCR vs HC2, P <.001; real-time PCR vs direct sequencing, P <.001). b P <.001. 148 Cancer Cytopathology February 2016

Screening Algorithms for Cervical Cancer/Choi et al TABLE 4. Human Papillomavirus Status, Abnormal Cytology, and Biopsy Results According to the Age of Women at Screening Age Group: No. of Patients (%) Status 20 y 21-24 y 25-29 y 30-39 y 40-49 y 50-65 y 66 y Total No. of Patients. 5 16 70 232 382 238 57 1000 (100) HR HPV1 3 (60) 4 (25) 33 (47.1) 64 (27.6) 73 (19.1) 50 (21) 16 (28.1) 243 (24.3) HPV-16/HPV-181 1 (33.3) 1 (25) 15 (45.5) 18 (28.1) 19 (26) 11 (22) 5 (31.3) 70 (28.8) LBC ASC-US 2 (40) 5 (31.3) 22 (31.4) 72 (31) 92 (24.1) 51 (21.4) 17 (29.8) 261 (26.1) CIN 21 1 (20) 1 (6.3) 11 (15.7) 30 (12.9) 31 (8.1) 16 (6.7) 8 (14) 98 (9.8) CIN 31 1 (20) 1 (6.3) 10 (14.3) 27 (11.6) 25 (6.5) 10 (4.2) 6 (10.5) 80 (8) Abbreviations: (1), positive; ASC-US, atypical squamous cells of undetermined significance; CIN, cervical intraepithelial lesion; HPV, human papillomavirus; HR, high risk; LBC, liquid-based cytology. cotesting (60%), and LBC (52.2%). The LBC (97.4%) and cotesting (96.5%) specificities for detecting CIN 2 1 were similarly high, whereas primary HPV screening recorded the lowest specificity (87.5%). All 3 algorithms displayedhighspecificityforcin31.cotestingandlbc alone had comparably higher positive predictive values for CIN 2 1 (97.8% and 98.9%, respectively) compared with primary HPV screening (91%), whereas primary HPV screening and cotesting had higher negative predictive values for CIN 2 1 (63.6% and 62.5%, respectively) compared with LBC (43.2%). Regardless of the results from subanalyses of screening in different age groups, cotesting for the detection of CIN 2 1 revealed minimally higher sensitivities and specificities than those of primary HPV screening. In contrast, the sensitivity of primary HPV screening for detecting CIN 3 1 was slightly higher than that of cotesting in all subgroup analyses by age. LBC had the highest specificity and the lowest sensitivity among the 3 screening algorithms, whereas primary HPV screening had the highest negative predictive value and the lowest specificity (Table 5). Prevalence of HPV Genotypes The mean age of the study population was 44.7 years (age range, 17-86 years). Cytology results were available for 987 patients based on 2004 Bethesda System specimen adequacy criteria. According to the cytology and/or HPV results, 197 patients underwent colposcopic biopsies. Realtime HPV genotyping PCR revealed that 243 patients (24.3%) were positive for HR HPV types. The most common type was HPV-16, followed by multiple HPV infections and HPV-58, HPV-52, HPV-31, HPV-35, HPV- 51, HPV-39, HPV-56, HPV-33, HPV-18, HPV-59, and HPV-45, in descending order (Fig. 2). Three HR HPV types (HPV-16, HPV-58, and HPV-52) comprised almost 50% of the total infected specimens. HPV-18 was identified in 6 patients (2.4%). Multiple HR HPV infections were identified in 41 patients (16.9%), among which HPV-52, HPV-58, HPV-16, and HPV-39 were the predominant types identified (Table 6). Among patients who had CIN 21, HPV-16 (29.3%) and multiple infections (18.2%) were frequently discovered. DISCUSSION In this study, we used validated HPV genotyping and qualified LBC results to demonstrate that the clinical performance of primary HPV screening in a tertiary hospital practice was comparable to that of cotesting for CIN 2 1 or CIN 3 1 irrespective of patient age. This supports the finding that the current use of cotesting is preferred for cervical cancer screening, and primary HPV screening could replace cotesting in some clinical situations. In addition, the prevalence of HR HPV was 24.3%, which included HPV-16; multiple HPV infections; and HPV- 58, HPV-52, HPV-31, HPV-35, HPV-51, HPV-39, HPV-56, HPV-33, HPV-18, HPV-59, and HPV-45, in order of decreasing prevalence. Consistent with the results from several observational trials, 8 10 our results indicate that primary HPV screening in our clinical setting had a clinical performance comparable to that of cotesting for CIN 2 1 and CIN 31. Thus, the primary HPV screening algorithm could be introduced into the current screening programs instead of cotesting. However, the currently approved primary HPV screening algorithm needs to be improved, because it displayed lower specificity despite its excellent sensitivity. Indeed, the lower specificity of primary HPV screening may have been caused by the unselective detection of early HR HPVinfectedlesionsthatdidnotmanifestthehistologicfeatures Cancer Cytopathology February 2016 149

TABLE 5. Clinical Performance of 3 Screening Algorithms for Cervical Intraepithelial Neoplasia Grade 2 or Greater Percentage (No./Total No. of Patients) Algorithm Performance CIN 21 CIN 31 Ages 21-65 y Primary HPV Prevalence 49.5 (90/182) 39.6 (72/182) Sensitivity 71.3 61.7 Specificity 88.1 98.5 PPV 91.1 98.6 NPV 64.1 60 Cotesting Prevalence 49.5 (90/182) 39.6 (72/182) Sensitivity 72.1 59 Specificity 96.7 100 PPV 97.8 100 NPV 63 54.5 LBC Prevalence 49.5 (90/182) 39.6 (72/182) Sensitivity 63.6 51.4 Specificity 97.6 100 PPV 98.9 100 NPV 44.6 38.2 Ages 25-65 y Primary HPV Prevalence 50.3 (89/177) 40.7 (72/177) Sensitivity 71.7 62.8 Specificity 87.5 98.4 PPV 91 98.6 NPV 63.6 60 Cotesting Prevalence 50.3 (89/177) 40.7 (72/177) Sensitivity 72.5 60 Specificity 96.5 100 PPV 97.8 100 NPV 62.5 54.3 LBC Prevalence 50.3 (89/177) 40.7 (72/177) Sensitivity 63.8 52.2 Specificity 97.4 100 PPV 98.9 100 NPV 43.2 37.1 Ages 30-65 y Primary HPV Prevalence 48.8 (78/160) 38.8 (62/160) Sensitivity 71.4 62.2 Specificity 87.1 98.4 PPV 89.7 98.4 NPV 65.9 62.2 Cotesting Prevalence 48.8 (78/160) 38.8 (62/160) Sensitivity 72.4 59 Specificity 96.4 100 PPV 97.4 100 NPV 64.6 56.1 LBC Prevalence 48.8 (78/160) 38.8 (62/160) Sensitivity 62.6 50.4 Specificity 97.3 100 PPV 98.7 100 NPV 43.9 37.8 Ages 40-65 y Primary HPV Prevalence 45.3 (48/106) 33 (35/106) Sensitivity 70.7 60.3 Specificity 85.4 100 PPV 85.4 100 NPV 70.7 67.6 Cotesting Prevalence 45.3 (48/106) 33 (35/106) Sensitivity 71.9 54.7 Specificity 95.2 100 PPV 95.8 100 NPV 69 59.2 LBC Prevalence 45.3 (48/106) 33 (35/106) Sensitivity 64.4 47.9 Specificity 97 100 TABLE 5. Continued Percentage (No./Total No. of Patients) Algorithm Performance CIN 21 CIN 31 PPV 97.9 100 NPV 55.2 46.5 Abbreviations: CIN, cervical intraepithelial neoplasia; HPV, human papillomavirus; LBC, liquid-based cytology; NPV, negative predictive value; PPV, positive predictive value; prevalence, proportion of detected cases. Figure 2. The distribution of high-risk (HR) human papillomavirus (HPV) genotypes is illustrated according to results from multiplex, real-time polymerase chain reaction analysis. CIN II 1 indicates cervical intraepithelial neoplasia grade 2 or greater. of CIN 21, which comprised 28.8% of the HR HPVpositive patients who underwent biopsy in this study. LBC displayed the lowest sensitivity and the highest specificity among the 3 screening algorithms, as expected. With the recent approval of the HPV test as a primary cervical cancer screening method, there is great concern about whether primary HPV screening can sufficiently replace the prevailing cytology-based screening or cotesting. Unlike previous trials, we performed this study using high-quality LBC results, which were interpreted by experienced cytopathologists and, with an efficient multiplex, real-time PCR assay, were validated by direct HPV DNA sequencing as a robust gold standard. For the reliable detection of HR HPV types, we designed real-time PCR primers targeting the E5, E6, and E7 regions, because the L1 and L2 regions of the HPV genome can be lost during viral integration into the host genome, leading to a false-negative result. 12 Compared with cytology or cotesting methodology, primary HPV screening has a simple, intuitive algorithm, which could help both clinicians and patients to comply 150 Cancer Cytopathology February 2016

Screening Algorithms for Cervical Cancer/Choi et al TABLE 6. High-Risk Human Papillomavirus Genotypes in Multiple Human Papillomavirus-Infected Cervical Lesions Histology No. of Samples HPV Types (No. of Samples) Chronic cervicitis 2 35/39 (1), 39/56/59 (1) CIN 1 8 18/39 (1), 18/52 (1), 33/39 (1), 33/52 (2), 35/39 (1), 51/52 (1), 51/52/58 (1) CIN 2 4 39/56 (1), 52/56 (1), 52/58 (1), 16/33/39/45/52 (1) CIN 3 13 16/39 (1), 16/51 (1), 16/56 (1), 31/35 (1), 33/51 (1), 33/56 (1), 35/52 (1), 45/58 (1), 52/58 (1), 58/59 (1),16/31/33 (1), 16/51/58 (1), 39/51/56 (1) SCC 1 33/52 (1) Unknown 13 16/18/39/58 (1), 16/31 (1), 16/51 (1), 16/52 (1), 16/58 (1), 18/52 (1), 33/56/58 (1), 39/52 (1), 51/56 (2), 51/58 (1), 52/58/59 (1), 56/58 (1) Total 41 Abbreviations: CIN, cervical intraepithelial lesion; HPV, human papillomavirus; SCC, squamous cell carcinoma. with testing. Furthermore, primary HPV screening can be performed economically. In addition, it has been reported that the reassurance of HPV-negative women with primary HPV screening every 3 years was superior to cytology every 3 years and was nearly equivalent to cotesting every 5 years. 13 However, despite its great advantages, the clinical application of primary HPV screening has considerable obstacles. Primary HPV screening has lower specificity than the other methods, leading to increased referral to colposcopy, which is an insufficient triaging strategy for HPVpositive women, and there are uncertainties about the screening intervals for HPV-negative women. 14 In addition, the presence of cervical cancers, especially adenocarcinoma, unrelated to HPV (range, 6%-18%) should always be considered when the primary HPV screening strategy is used. 15 This study included too few invasive carcinomas to reach a significant conclusion on the advantage of 1 screening algorithm compared with the others. Previous retrospective studies regarding HPV detection in specimens before the diagnosis of invasive cancer have indicated that the sensitivity of HPV testing was lower than would have been expected based on its sensitivity for precancerous lesions. 16,17 In the current study, of 1000 samples, only 5 diagnoses of squamous cell carcinoma and 2 diagnoses of adenocarcinoma were included. Overall, the HR HPV-negative rate for squamous cell carcinoma was 20% (1 of 5 samples). The 1 HR HPV-negative sample was consistently negative in all HPV assays, including real-time PCR, HC2, and direct HPV DNA sequencing. In contrast, the 2 patients with adenocarcinoma had positive HR HPV test results. The primary HPV screening algorithm could not detect 1 patient with squamous cell carcinoma, whereas screening programs with LBC or cotesting could successfully identify all patients who had carcinoma. Therefore, it is important to recognize that primary HPV screening rarely missed the diagnosis of HPV-negative cervical cancer. We observed that HPV-16, HPV-58, HPV-52, and HPV-31 were the most prevalent types in our study population. Because these types are detected dominantly in developed countries, 18 it is noteworthy that HPV-18 was relatively uncommon in this study. Instead, there were significant increases in HPV-35, HPV-39, HPV-51, and HPV-56, 3 of which (HPV-35, HPV-39, and HPV-56) were excluded from the targeting genotypes of the nonavalent HPV vaccine and were rare in previous Korean reports. 19,20 Since the development of commercial HPV vaccines, such as the quadrivalent vaccine (HPV-6, HPV-11, HPV-16, and HPV-18; Gardasil; Merck & Company, Inc, Kenilworth, NJ) and the bivalent vaccine (HPV-16 and HPV-18; Cervarix; GlaxoSmithKline, London, UK), the prevention of cervical high-grade dysplastic lesions has been successful. 21,22 Recently, the nonavalent vaccine was approved by the US Food and Drug Administration and now confers protection against 5 additional HR HPV types (HPV-31, HPV-33, HPV-45, HPV-52, and HPV-58) compared with Gardasil. The nonavalent vaccine is expected to offer an overall protection rate against cervical cancer of up to 90%. 23 The prevalence of HR HPV subtypes, including multiple HPV infections that are not covered by the nonavalent vaccine, was 39.9% in our study population and 28% in the patients who had CIN 21. Thus, vaccination strategies that cover the emerging and uncovered HR HPV types need to be further improved. In conclusion, the clinical performance of primary HPV screening was comparable to that of cotesting for Cancer Cytopathology February 2016 151

CIN 2 1 or CIN 3 1 irrespective of patient age. Considering the prevalence distribution of HPV genotypes, the new nonavalent HPV vaccine would prevent approximately 60% of HR HPV-positive cases in the population. FUNDING SUPPORT This study was supported by grants from Korea University (grant K1421351) and by the Basic Science Research Program through the National Research Foundation of Korea funded by the Ministry of Science, Information and Communications Technologies, and Future Planning (grant 2013R1A1A1058146). CONFLICT OF INTEREST DISCLOSURES The authors made no disclosures. REFERENCES 1. Denny LA, Wright TC Jr. Human papillomavirus testing and screening. Best Pract Res Clin Obstet Gynaecol. 2005;19:501-515. 2. Ronco G, Giorgi-Rossi P, Carozzi F, et al. Efficacy of human papillomavirus testing for the detection of invasive cervical cancers and cervical intraepithelial neoplasia: a randomised controlled trial. Lancet Oncol. 2010;11:249-257. 3. Walboomers JM, Jacobs MV, Manos MM, et al. Human papillomavirus is a necessary cause of invasive cervical cancer worldwide. J Pathol. 1999;189:12-19. 4. Bouvard V, Baan R, Straif K, et al. A review of human carcinogens part B: biological agents. Lancet Oncol. 2009;10:321-322. 5. Khan MJ, Castle PE, Lorincz AT, et al. The elevated 10-year risk of cervical precancer and cancer in women with human papillomavirus (HPV) type 16 or 18 and the possible utility of type-specific HPV testing in clinical practice. J Natl Cancer Inst. 2005;97:1072-1079. 6. Li N, Franceschi S, Howell-Jones R, Snijders PJ, Clifford GM. Human papillomavirus type distribution in 30,848 invasive cervical cancers worldwide: variation by geographical region, histological type and year of publication. Int J Cancer. 2011;128:927-935. 7. Moyer VA, US Preventive Services Task Force. Screening for cervical cancer: US Preventive Services Task Force recommendation statement. Ann Intern Med. 2012;156:880-891, W312. 8. Wright TC, Stoler MH, Behrens CM, Sharma A, Zhang G, Wright TL. Primary cervical cancer screening with human papillomavirus: end of study results from the ATHENA study using HPV as the first-line screening test. Gynecol Oncol. 2015;136:189-197. 9. Dillner J, Rebolj M, Birembaut P, et al. Long term predictive values of cytology and human papillomavirus testing in cervical cancer screening: joint European cohort study [serial online]. BMJ. 2008; 337:a1754. 10. Katki HA, Kinney WK, Fetterman B, et al. Cervical cancer risk for women undergoing concurrent testing for human papillomavirus and cervical cytology: a population-based study in routine clinical practice. Lancet Oncol. 2011;12:663-672. 11. Ronco G, Dillner J, Elfstrom KM, et al. Efficacy of HPV-based screening for prevention of invasive cervical cancer: follow-up of four European randomised controlled trials. Lancet. 2014;383:524-532. 12. Noffsinger AE, Suzuk L, Hui YZ, Gal AA, Fenoglio-Preiser CM. Differential sensitivities of E6 type-specific and L1 consensus primers in the detection of human papillomavirus in anal carcinoma. Mod Pathol. 1995;8:509-514. 13. Gage JC, Schiffman M, Katki HA, et al. Reassurance against future risk of precancer and cancer conferred by a negative human papillomavirus test [serial online]. J Natl Cancer Inst. 2014;106:dju153. 14. Huh WK, Ault KA, Chelmow D, et al. Use of primary high-risk human papillomavirus testing for cervical cancer screening: interim clinical guidance. Gynecol Oncol. 2015;136:178-182. 15. de Sanjose S, Quint WG, Alemany L, et al. Human papillomavirus genotype attribution in invasive cervical cancer: a retrospective cross-sectional worldwide study. Lancet Oncol. 2010;11:1048-1056. 16. Tao X, Griffith CC, Zhou X, et al. History of high-risk HPV and Pap test results in a large cohort of patients with invasive cervical carcinoma: experience from the largest women s hospital in China. Cancer Cytopathol. 2015;123:421-427. 17. Zhao C, Li Z, Nayar R, et al. Prior high-risk human papillomavirus testing and Papanicolaou test results of 70 invasive cervical carcinomas diagnosed in 2012: results of a retrospective multicenter study. Arch Pathol Lab Med. 2015;139:184-188. 18. Vinodhini K, Shanmughapriya S, Das BC, Natarajaseenivasan K. Prevalence and risk factors of HPV infection among women from various provinces of the world. Arch Gynecol Obstet. 2012;285:771-777. 19. Hwang Y, Lee M. Comparison of the AdvanSure human papillomavirus screening real-time PCR, the Abbott RealTime High Risk human papillomavirus test, and the Hybrid Capture human papillomavirus DNA test for the detection of human papillomavirus. Ann Lab Med. 2012;32:201-205. 20. Park Y, Lee E, Choi J, Jeong S, Kim HS. Comparison of the Abbott Real Time High-Risk Human Papillomavirus (HPV), Roche Cobas HPV, and Hybrid Capture 2 assays to direct sequencing and genotyping of HPV DNA. J Clin Microbiol. 2012; 50:2359-2365. 21. Gertig DM, Brotherton JM, Budd AC, Drennan K, Chappell G, Saville AM. Impact of a population-based HPV vaccination program on cervical abnormalities: a data linkage study [serial online]. BMC Med. 2013;11:227. 22. Paavonen J, Jenkins D, Bosch FX, et al. Efficacy of a prophylactic adjuvanted bivalent L1 virus-like-particle vaccine against infection with human papillomavirus types 16 and 18 in young women: an interim analysis of a phase III double-blind, randomised controlled trial. Lancet. 2007;369:2161-2170. 23. Serrano B, Alemany L, Tous S, Bruni L, Clifford GM, Weiss T, et al. Potential impact of a nine-valent vaccine in human papillomavirus related cervical disease. Infect Agent Cancer. 2012;7:38. 152 Cancer Cytopathology February 2016