00019606-201106000-00006ArticleDiagnostic Molecular PathologyDiagnostic Molecular Pathology© 2011 by Lippincott Williams & Wilkins.20June 2011
p 101–104Impact of HPV Assay on Observed Population PrevalenceOriginal ArticlesUnger, Elizabeth R. PhD, MD; Steinau, Martin PhD; Lin, Jin-Mann S. PhD; Patel, Sonya S. BS; Swan, David C. PhDCenters for Disease Control and Prevention, National Center for Emerging and Zoonotic, Infectious Diseases (proposed), Chronic Viral Diseases Branch Atlanta, GADisclaimer: The findings and conclusions in this report are hose of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention.Reprints: Elizabeth R. Unger PhD, MD, Centers for Disease Control and Prevention, MS G41, Atlanta, GA (e-mail:
[email protected]).AbstractType-specific surveillance of human papillomavirus (HPV) has been proposed as an early indicator of vaccine impact. Longitudinal comparison of HPV typing results requires stable assays with high type-specific reproducibility. Assays are evolving and the impact of even minor changes in the assay format may be difficult to anticipate. We initiated a population-based study of HPV with the prototype line blot (PLB) assay. These reagents were replaced by the research use only Linear Array (LA) HPV Genotyping kit. The assays are similar in principle and earlier comparisons found increased sensitivity and detection of more types per sample with LA; however, in samples from women with cervical abnormalities, the overall concordance was good. Slight changes in sensitivity may be more significant in samples from a general population with lower viral loads in the samples. Residual extracts from 3001 self-collected vaginal swabs from women in the general US population originally tested with PLB were retested with LA. With LA, all the samples were hybridized. PLB hybridization was restricted to samples with probable amplicon in gel electrophoresis. For HPV detection, the agreement between the 2 assays was 78.6% (κ=0.55) with a positive concordance of 52.8%. However, this masks the observation that repeat testing with LA led to the detection of HPV in nearly twice as many samples. Agreement improves if comparison was restricted to the samples hybridized. These results emphasize that assay comparisons should consider the clinical-epidemiologic context of sample collection. Studies designed to examine temporal trends in type-specific prevalence should archive residual material to permit retesting if assays change.Monitoring the impact of the introduction of human papillomavirus (HPV) prophylactic vaccines presents unique challenge. Although cervical cancer is the outcome of greatest interest, its incidence cannot be expected to changes for decades after widespread vaccine uptake because of the recognized lag phase between the infection and the rare long-term sequelae of cancer. Type-specific surveillance of HPV in populations or sentinel sites has been proposed as an early indicator of vaccine impact, allowing a decrease in vaccine-targeted types to be observed and detecting possible replacement of vaccine-targeted types with other HPV types. This requires longitudinal comparison of HPV typing results, and requires that the type-specific reproducibility of the assays be extremely reliable.The technology of HPV detection and typing is continually evolving and the impact of even minor changes in the assay format may be difficult to anticipate. We initiated a population-based study of HPV with the prototype line blot (PLB) assay using reagents from Roche Molecular Diagnostics, Pleasanton, CA. These reagents were discontinued at the time that the manufacturer released the research use only Linear Array (LA) HPV Genotyping kit. The assays are similar in principle (compared in Table 1), but LA incorporated features to improve standardization. To ensure comparability of longitudinal results, residual extracts from samples originally tested with PLB were retested with LA. This report describes the differences in results from these 2 assays and emphasizes the importance of archiving biologic samples from studies designed for baseline and postvaccine determinations of type-specific HPV prevalence.JOURNAL/dimp/04.03/00019606-201106000-00006/table1-6/v/2021-02-17T200033Z/r/image-tiffComparison of AssaysMATERIALS AND METHODSSpecimen Collection and ProcessingSelf-collected vaginal swabs were obtained between 2002 and 2004 from a population-based sample of US women in the general population and processed as described earlier.1 In brief, DNA was extracted within 1 month of sample collection using slight modifications to the QIAmp Mini Kit (Qiagen, Valencia, CA). The extracts were further purified and concentrated with spin columns to yield a final 100 μL extract in 10 mM Tris-HCl, 0.1 mM ethylene diaminetetra-acetic acid. For every batch of 40 samples, 1 water blank was processed through all the extraction steps as a negative control. The extracts were tested and then archived at −20°C.HPV Genotyping TestsAll samples were first tested with the Roche PLB (reagents provided by Roche Molecular Systems Inc) either the first generation 27 probe format (HPV types 6, 11, 16, 18, 26, 31, 33, 35, 39, 40, 42, 45, 51, 52, 53, 54, 55, 56, 57, 58, 59, 66, 68, 73, 82, 83, and 84), or the second generation 37 probe format (added types 61, 62, 64, 67, 69, 70, 71, 72, 81, 89, IS39, and omitted HPV57).2,3 As described earlier, 5 μL of the extract was used in the 100-μL polymerase chain reaction (PCR) and samples with HPV band detected on ethidium bromide gel electrophoresis were hybridized to the typing strips.1 LA was performed on the extracts archived on an average of 3 years (range: 26.7 to 60.1 mo) by technologists blinded to the original results. The manufacturer's protocol for LA was modified to use the 5 μL extract in the 100-μL PCR, but was otherwise unchanged. All the samples were hybridized to the typing strip. Samples positive for the XR probe on the LA HPV strip, which were also positive for HPV 33, 35, or 58, required further evaluation to confirm or exclude the presence of HPV 52. The quantitative assay described below, with a threshold of 50 copies was used to determine the status of HPV 52 in these cases. For each platform, the typing results were recorded in a Microsoft Office Access database linked to study ID. Samples failing to detect HPV and the β-globin control were considered inadequate for evaluation.Quantitative PCRAll samples that were positive for HPV 16 or HVP 52 in either platform, and that had sufficient residual extract were tested by quantitative real-time PCR (qPCR) to determine the HPV copy number. HPV 16 and 52 copies were amplified with type-specific primers and a FAM-labeled TaqMan probe for the respective E6/E7 region together with β-globin-specific primers and probe using an ABI 7900 HT Sequence Detection System (Applied Biosystems, Foster City, CA). Primer and probe sequences and PCR conditions are detailed elsewhere4 (Onyekwuluje et al, manuscript in preparation).Data AnalysisPercent agreement and κ statistics were used to compare HPV detection in these specimens with LA and PLB. Concordance and the prevalence rates of positive detection were also calculated. We further performed a linear regression on the prevalence rates of LA versus PLB for the 26 HPV types included in both the PLB formats.RESULTSA total of 3260 samples originally evaluated with the PLB assay (∼40% 27-probe format, 60% 37-probe format) had residual extracts available for retesting with LA, and 3001 samples were adequate in both the testing platforms. For the samples adequate in both assays, the overall concordance for HPV detection (any HPV) was 78.6%, κ=0.55±0.01 (Table 2). The number of specimens testing positive by LA was nearly twice the number testing positive by PLB; 1350 (45%) by LA and 734 (24.5%) by PLB. Of the PLB-negative LA-positive samples, only 47 were attributable to the detection of types not included in the 27-type PLB format. Positive agreement, that is concordance restricted to those with HPV detected by at least 1 platform, was 52.8%. Prescreening PLB amplicons by gel electrophoresis contributed to decreased detection of HPV as the assay concordance improved if the comparison was restricted to the 1243 samples hybridized to the PLB typing strips (87.9%, κ=0.74±0.02, positive concordance 82.6%).JOURNAL/dimp/04.03/00019606-201106000-00006/table2-6/v/2021-02-17T200033Z/r/image-tiffOverall Concordance (78.6%) of the 2 Assay Formats (n=3001)Determination of type-specific agreement was restricted to the 26 types included in both the formats of PLB. Type-specific concordance, as measured by κ values, varied from 0.72 (95% confidence interval: 0.62, 0.83) for HPV 58 to 0.4 (0.23, 0.57) for HPV 67 (Table 3). The linear regression of the scatter-plot of type-specific prevalence in PLB versus LA indicates the change in sensitivity was similar for most types (R2=0.86), but HPV types 84 and 89 were relative outliers (higher change in detection by LA as compared with PLB). On an average, PLB detected 1.9±0.05 and LA 2.6±0.07 types per sample.JOURNAL/dimp/04.03/00019606-201106000-00006/table3-6/v/2021-02-17T200033Z/r/image-tiffFrequency of Type-specific Detection in Linear Array (LA) and Prototype Line Blot (PLB) Assays and Corresponding κ Values With Lower and Upper Limits of the 95% Confidence IntervalFor the samples positive for HPV 16 (n=130) or HPV 52 (n=125) with either assay for which sufficient residual extract was available, quantitation of the HPV copy number was determined. As shown in Table 4, samples with concordant detection had higher copy numbers than those that were positive in only 1 assay (P values for nonparametric tests were <0.0001 for the differences in the copy numbers of HPV 16 and HPV52). HPV 16 median copy number was 239-folds higher in the samples tested positive by LA and PLB as compared with those only positive by LA, and for HPV52 the difference was 453-folds.JOURNAL/dimp/04.03/00019606-201106000-00006/table4-6/v/2021-02-17T200033Z/r/image-tiffMedian and Mean Copy Numbers (per μL) of HPV 16 and 52 Determined by qPCR Stratified by Assay ConcordanceDISCUSSIONThe overall concordance in HPV detection between these 2 very similar assay platforms was 78.5%, but this masks the observation that repeat testing with LA-detected HPV in nearly twice as many samples. Such differences do not significantly impact the risk profile for HPV, nor the general patterns of HPV natural history, but would complicate the interpretation of HPV prevalence in studies when the testing platform changes.Our study results show a more pronounced difference between PLB and LA than that reported in earlier studies. The HPV prevalence was nearly doubled with the LA assay, despite an average of 3 years of storage. Coutlee et al5 reported an overall agreement of 93.8% and positive agreement of 92.5% for 528 anogenital samples from an HIV-positive cohort. Castle et al6 reported an overall agreement of 88% and positive agreement of 80% in 3335 samples from a population with equivocal and low-grade cytologic abnormalities. One explanation for the greater level of agreement in these studies might be that their samples were collected in populations more likely to have higher viral loads, a higher proportion of samples would therefore fall above the detection limit of the less sensitive PLB assay. HIV infection has been associated with higher HPV viral loads and HPV viral load is increased in women with cervical abnormalities.7,8 Our samples were collected from the general US population in which cervical abnormalities would be uncommon and therefore lower copy numbers of HPV would be encountered more frequently. In addition, the cervical component in a vaginal self-collected sample could be expected to be diluted with the vaginal pool, also contributing to the reduced viral load. Consistent with this suggestion, qPCR showed that viral DNA was significantly lower in samples that were only positive by LA when compared with those that were positive in both the assays.In addition, 2 technical aspects might also have contributed to the greater detection disagreement in our study. For the PLB assay, we restricted hybridization to the typing strips to those samples with HPV amplicons visualized by gel electrophoresis, whereas for the LA assay all the samples were hybridized. In preliminary studies with the PLB format using both gel electrophoresis and hybridization, we found few samples (in the order of ∼1%) that were gel negative but hybridization positive. The limited sensitivity of gel visualization did add to the number of samples falsely classified as negative by PLB.These results indicate the significant impact that changes in the assay, even those based on the same primer systems, can have on the observed type-specific prevalence. In addition, the difficulties in extrapolating assay differences when studying different populations and different sample collections are clearly shown. Finally, the importance of archiving samples from studies designed to examine temporal trends in type-specific prevalence is emphasized. These archived samples allowed retesting so that HPV data on the population-based samples collected before the introduction of the HPV vaccine could be directly compared with those collected in ongoing and future studies.REFERENCES1. Dunne EF, Unger ER, Sternberg M, et al. Prevalence of HPV infection among females in the United States. JAMA. 2006;297:813–819[Context Link][Full Text][Medline Link]2. Gravitt PE, Peyton CL, Apple RJ, et al. Genotyping of 27 human papillomavirus types by using L1 consensus products by a single-hybridization, reverse line blot detection method. J Clin Microbiol. 1998;36:3020–3027[Context Link][CrossRef][Medline Link]3. Gravitt PE, Peyton CL, Alessi TQ, et al. Improved amplification of genital human papillomaviruses. J Clin Microbiol. 2000;38:357–361[Context Link][CrossRef][Medline Link]4. Swan DC, Tucker RA, Holloway BP, et al. A sensitive, type-specific, fluorogenic probe Assay for detection of human papillomavirus DNA. J Clin Microbiol. 1997;35:886–891[Context Link][CrossRef][Medline Link]5. Coutlee F, Rouleau D, Petignat P, et al. Enhanced detection and typing of human papillomavirus (HPV) DNA in anogenital samples with PGMY primers and the linear array HPV genotyping test. J Clin Microbiol. 2006;44:1998–2006[Context Link][CrossRef][Medline Link]6. Castle PE, Gravitt PE, Solomon D, et al. Comparison of linear array and line blot assay for detection of human papillomavirus and diagnosis of cervical precancer and cancer in the atypical squamous cell of undetermined significance and low-grade squamous intraepithelial lesion triage study. J Clin Microbiol. 2008;46:109–117[Context Link][CrossRef][Medline Link]7. Huang Y, Huang MN, Li N, et al. Association between human papillomavirus DNA load and development of cervical intraepithelial neoplasia and cervical cancer. Int J Gynecol Cancer. 2008;18:755–760[Context Link][Full Text][CrossRef][Medline Link]8. Riva E, Serraino D, Pierangeli A, et al. Markers of human papillomavirus infection and their correlation with cervical dysplasia in human immunodeficiency virus-positive women. Clin Microbiol Infect. 2007;13:94–97[Context Link][Full Text][CrossRef][Medline Link]human papillomavirus; epidemiology; genotyping assays00019606-201106000-0000600005407_2007_297_813_dunne_prevalence_|00019606-201106000-00006#xpointer(id(citation_FROM_JRF_ID_d2865e349_citationRF_FLOATING))|11065404||ovftdb|SL00005407200729781311065404citation_FROM_JRF_ID_d2865e349_citationRF_FLOATING[Full Text]00005407-200702280-0002700019606-201106000-0000600005407_2007_297_813_dunne_prevalence_|00019606-201106000-00006#xpointer(id(citation_FROM_JRF_ID_d2865e349_citationRF_FLOATING))|11065405||ovftdb|SL00005407200729781311065405citation_FROM_JRF_ID_d2865e349_citationRF_FLOATING[Medline Link]1732752300019606-201106000-0000600004688_1998_36_3020_gravitt_papillomavirus_|00019606-201106000-00006#xpointer(id(citation_FROM_JRF_ID_d2865e384_citationRF_FLOATING))|11065213||ovftdb|SL00004688199836302011065213citation_FROM_JRF_ID_d2865e384_citationRF_FLOATING[CrossRef]10.1128%2FJCM.36.10.3020-3027.199800019606-201106000-0000600004688_1998_36_3020_gravitt_papillomavirus_|00019606-201106000-00006#xpointer(id(citation_FROM_JRF_ID_d2865e384_citationRF_FLOATING))|11065405||ovftdb|SL00004688199836302011065405citation_FROM_JRF_ID_d2865e384_citationRF_FLOATING[Medline Link]973806000019606-201106000-0000600004688_2000_38_357_gravitt_papillomaviruses_|00019606-201106000-00006#xpointer(id(citation_FROM_JRF_ID_d2865e419_citationRF_FLOATING))|11065213||ovftdb|SL0000468820003835711065213citation_FROM_JRF_ID_d2865e419_citationRF_FLOATING[CrossRef]10.1128%2FJCM.38.1.357-361.200000019606-201106000-0000600004688_2000_38_357_gravitt_papillomaviruses_|00019606-201106000-00006#xpointer(id(citation_FROM_JRF_ID_d2865e419_citationRF_FLOATING))|11065405||ovftdb|SL0000468820003835711065405citation_FROM_JRF_ID_d2865e419_citationRF_FLOATING[Medline Link]1061811600019606-201106000-0000600004688_1997_35_886_swan_papillomavirus_|00019606-201106000-00006#xpointer(id(citation_FROM_JRF_ID_d2865e454_citationRF_FLOATING))|11065213||ovftdb|SL0000468819973588611065213citation_FROM_JRF_ID_d2865e454_citationRF_FLOATING[CrossRef]10.1128%2Fjcm.35.4.886-891.199700019606-201106000-0000600004688_1997_35_886_swan_papillomavirus_|00019606-201106000-00006#xpointer(id(citation_FROM_JRF_ID_d2865e454_citationRF_FLOATING))|11065405||ovftdb|SL0000468819973588611065405citation_FROM_JRF_ID_d2865e454_citationRF_FLOATING[Medline Link]915714800019606-201106000-0000600004688_2006_44_1998_coutlee_papillomavirus_|00019606-201106000-00006#xpointer(id(citation_FROM_JRF_ID_d2865e490_citationRF_FLOATING))|11065213||ovftdb|SL00004688200644199811065213citation_FROM_JRF_ID_d2865e490_citationRF_FLOATING[CrossRef]10.1128%2FJCM.00104-0600019606-201106000-0000600004688_2006_44_1998_coutlee_papillomavirus_|00019606-201106000-00006#xpointer(id(citation_FROM_JRF_ID_d2865e490_citationRF_FLOATING))|11065405||ovftdb|SL00004688200644199811065405citation_FROM_JRF_ID_d2865e490_citationRF_FLOATING[Medline Link]1675759000019606-201106000-0000600004688_2008_46_109_castle_intraepithelial_|00019606-201106000-00006#xpointer(id(citation_FROM_JRF_ID_d2865e525_citationRF_FLOATING))|11065213||ovftdb|SL0000468820084610911065213citation_FROM_JRF_ID_d2865e525_citationRF_FLOATING[CrossRef]10.1128%2FJCM.01667-0700019606-201106000-0000600004688_2008_46_109_castle_intraepithelial_|00019606-201106000-00006#xpointer(id(citation_FROM_JRF_ID_d2865e525_citationRF_FLOATING))|11065405||ovftdb|SL0000468820084610911065405citation_FROM_JRF_ID_d2865e525_citationRF_FLOATING[Medline Link]1798919400019606-201106000-0000600009577_2008_18_755_huang_intraepithelial_|00019606-201106000-00006#xpointer(id(citation_FROM_JRF_ID_d2865e560_citationRF_FLOATING))|11065404||ovftdb|SL0000957720081875511065404citation_FROM_JRF_ID_d2865e560_citationRF_FLOATING[Full Text]00009577-200807000-0002200019606-201106000-0000600009577_2008_18_755_huang_intraepithelial_|00019606-201106000-00006#xpointer(id(citation_FROM_JRF_ID_d2865e560_citationRF_FLOATING))|11065213||ovftdb|SL0000957720081875511065213citation_FROM_JRF_ID_d2865e560_citationRF_FLOATING[CrossRef]10.1111%2Fj.1525-1438.2007.01092.x00019606-201106000-0000600009577_2008_18_755_huang_intraepithelial_|00019606-201106000-00006#xpointer(id(citation_FROM_JRF_ID_d2865e560_citationRF_FLOATING))|11065405||ovftdb|SL0000957720081875511065405citation_FROM_JRF_ID_d2865e560_citationRF_FLOATING[Medline Link]1802122200019606-201106000-0000600125357_2007_13_94_riva_immunodeficiency_|00019606-201106000-00006#xpointer(id(citation_FROM_JRF_ID_d2865e595_citationRF_FLOATING))|11065404||ovftdb|SL001253572007139411065404citation_FROM_JRF_ID_d2865e595_citationRF_FLOATING[Full Text]00125357-200701000-0001500019606-201106000-0000600125357_2007_13_94_riva_immunodeficiency_|00019606-201106000-00006#xpointer(id(citation_FROM_JRF_ID_d2865e595_citationRF_FLOATING))|11065213||ovftdb|SL001253572007139411065213citation_FROM_JRF_ID_d2865e595_citationRF_FLOATING[CrossRef]10.1111%2Fj.1469-0691.2006.01568.x00019606-201106000-0000600125357_2007_13_94_riva_immunodeficiency_|00019606-201106000-00006#xpointer(id(citation_FROM_JRF_ID_d2865e595_citationRF_FLOATING))|11065405||ovftdb|SL001253572007139411065405citation_FROM_JRF_ID_d2865e595_citationRF_FLOATING[Medline Link]17184295Comparison of AssaysOverall Concordance (78.6%) of the 2 Assay Formats (n=3001)Frequency of Type-specific Detection in Linear Array (LA) and Prototype Line Blot (PLB) Assays and Corresponding κ Values With Lower and Upper Limits of the 95% Confidence IntervalMedian and Mean Copy Numbers (per μL) of HPV 16 and 52 Determined by qPCR Stratified by Assay ConcordanceImpact of HPV Assay on Observed Population PrevalenceUnger Elizabeth R. PhD MD; Steinau, Martin PhD; Lin, Jin-Mann S. PhD; Patel, Sonya S. BS; Swan, David C. PhDOriginal ArticlesOriginal Articles220p 101-104
