A population-based study of human papillomavirus genotype prevalence in the United States: Baseline measures prior to mass human papillomavirus vaccination

Authors

  • Cosette M. Wheeler,

    Corresponding author
    1. Department of Pathology and Department of Obstetrics and Gynecology, University of New Mexico Health Sciences Center, House of Prevention Epidemiology (HOPE), Albuquerque, NM
    • University of New Mexico Health Sciences Center, Department of Pathology, House of Prevention Epidemiology (HOPE), 1816 Sigma Chi Rd NE, MSC02—1670 Albuquerque, NM 87131, USA
    Search for more papers by this author
    • Tel.: [505-272-5785], Fax: +[505-277-0265]

  • William C. Hunt,

    1. Department of Pathology and Department of Obstetrics and Gynecology, University of New Mexico Health Sciences Center, House of Prevention Epidemiology (HOPE), Albuquerque, NM
    Search for more papers by this author
  • Jack Cuzick,

    1. Centre for Cancer Prevention, Wolfson Institute of Preventive Medicine, Queen Mary University of London, UK
    Search for more papers by this author
  • Erika Langsfeld,

    1. Department of Pathology and Department of Obstetrics and Gynecology, University of New Mexico Health Sciences Center, House of Prevention Epidemiology (HOPE), Albuquerque, NM
    Search for more papers by this author
  • Amanda Pearse,

    1. Department of Pathology and Department of Obstetrics and Gynecology, University of New Mexico Health Sciences Center, House of Prevention Epidemiology (HOPE), Albuquerque, NM
    Search for more papers by this author
  • George D. Montoya,

    1. Department of Pathology and Department of Obstetrics and Gynecology, University of New Mexico Health Sciences Center, House of Prevention Epidemiology (HOPE), Albuquerque, NM
    Search for more papers by this author
    • Deceased

  • Michael Robertson,

    1. Department of Pathology and Department of Obstetrics and Gynecology, University of New Mexico Health Sciences Center, House of Prevention Epidemiology (HOPE), Albuquerque, NM
    Search for more papers by this author
  • Catherine A. Shearman,

    1. Department of Pathology and Department of Obstetrics and Gynecology, University of New Mexico Health Sciences Center, House of Prevention Epidemiology (HOPE), Albuquerque, NM
    Search for more papers by this author
  • Philip E. Castle,

    1. American Society for Clinical Pathology Institute, Washington, DC
    Search for more papers by this author
  • For The New Mexico HPV Pap Registry Steering Committee

    Search for more papers by this author
    • Current members of the NMHPVPR Steering Committee are as follows: Nancy E. Joste, MD, Walter Kinney, MD, Cosette M. Wheeler, PhD., William C. Hunt, MS, Deborah Thompson, MD MSPH, Susan Baum, MD MPH, Linda Gorgos, MD MSc, Alan Waxman, MD MPH, David Espey MD, Jane McGrath MD, Steven Jenison, MD, Mark Schiffman, MD MPH, Philip Castle, PhD MPH, Vicki Benard, PhD, Debbie Saslow, PhD, Jane J.. Kim PhD, Mark H. Stoler MD and Jack Cuzick, PhD. This work was funded by a grant to C.M.W from the US National Cancer Institute (NCI) R01CA134779. Dr. Castle was previously supported by the Intramural Research Program of the NIH, NCI. HPV Linear Array reagents and equipment to automate HPV genotyping assays were provided by Roche Molecular Systems, Inc. C.M.W. has received funds through the University of New Mexico from GlaxoSmith Kline and Merck and Co., Inc., for HPV vaccine studies and equipment and reagents for HPV genotyping from Roche Molecular Systems. P.E.C. has received HPV tests and testing for research at a reduced or no cost from Roche and Qiagen, has a nondisclosure agreement with Roche, and serves on a Merck Data Safety and monitoring board.


Abstract

Currently, two prophylactic human papillomavirus (HPV) vaccines targeting HPV 16 and 18 have been shown to be highly efficacious for preventing precursor lesions although the effectiveness of these vaccines in real-world clinical settings must still be determined. Toward this end, an ongoing statewide surveillance program was established in New Mexico to assess all aspects of cervical cancer preventive care. Given that the reduction in cervical cancer incidence is expected to take several decades to manifest, a systematic population-based measurement of HPV type-specific prevalence employing an age- and cytology-stratified sample of 47,617 women attending for cervical screening was conducted prior to widespread HPV vaccination. A well-validated polymerase chain reaction (PCR) method for 37 HPV genotypes was used to test liquid-based cytology specimens. The prevalence for any of the 37 HPV types was 27.3% overall with a maximum of 52% at age of 20 years followed by a rapid decline at older ages. The HPV 16 prevalences in women aged ≤20 years, 21–29 years or ≥30 years were 9.6, 6.5 and 1.8%, respectively. The combined prevalences of HPV 16 and 18 in these age groups were 12.0, 8.3 and 2.4%, respectively. HPV 16 and/or HPV 18 were detected in 54.5% of high-grade squamous intraepithelial (cytologic) lesions (HSIL) and in 25.0% of those with low-grade SIL (LSIL). These baseline data enable estimates of maximum HPV vaccine impact across time and provide critical reference measurements important to assessing clinical benefits and potential harms of HPV vaccination including increases in nonvaccine HPV types (i.e., type replacement).

The discovery that persistent HPV infection causes cervical cancer1 and that HPV is a necessary cause of cervical cancer everywhere in the world2 has revolutionized cervical cancer prevention in the 21st century. DNA tests for the detection of HPV3–8 and prophylactic vaccines against HPV 6, 11, 16 and 18 infections and related cervical disease9–11 have been developed and approved by the US Food and Drug Administration and regulatory bodies throughout the world. Despite the role of the Pap test in reducing the cervical cancer burden in many western countries including the United States (US),12 the landscape of cervical cancer prevention will increasingly shift toward utilizing these HPV-targeted technologies because they are more efficacious and, if used rationally in an age-appropriate manner,13 are more cost-effective than Pap-based screening programs.14–16 Already, national HPV vaccination programs have been established in Australia, the United Kingdom and elsewhere,17, 18 and HPV-based screening has been introduced in the United States19 and is being considered in Europe.20 In the future, new HPV vaccines that target HPV 6, 11, 16, 18, 31, 33, 45, 52 and 58 may become available13 (http://clinicaltrials.gov/ct2/show/NCT00943722). Ultimately, cervical cancer prevention based on HPV vaccination in 11- to 12-year old girls and HPV testing in women 30 and older should replace Pap-based programs and could be the flagship for cancer prevention and cost-effective preventive programs globally.21

The implementation of any new technology requires surveillance to measure the population effectiveness. Randomized clinical trials are essential for establishing the efficacy of a new intervention but for logistic reasons are not capable of establishing overall efficacy in the general population in which an intervention will be implemented.22 Ultimately, surveillance in the general population is needed to assess the true clinical benefits and costs of implementing new technologies on the outcomes of interest.

To that end, a statewide registry, the New Mexico HPV Pap Registry (NMHPVPR), was established in 2006 to monitor the impact of HPV vaccine introduction and changes in cervical cancer screening behaviors over time. It includes records of all Pap and HPV tests, and all cervical, vaginal and vulvar biopsies. As a first goal, we wanted to describe the distribution of HPV genotypes in a large sample of the general New Mexico population of females participating in cervical screening prior to widespread HPV vaccination, thus providing a baseline for comparison with future vaccinated populations. Preventing cervical cancer is the primary goal of HPV vaccination but surrogate endpoints are required because reductions in cancer will take a decade or more to observe even in populations with high vaccine coverage.23 Among the available surrogate endpoints, changes in HPV type-specific prevalence will be a more sensitive, objective and earlier indicator of HPV vaccine impact24 versus precancer endpoints which are subjected to interobserver variation.25 Our study enables the estimations of HPV vaccine impact among women spanning a broad age range and with normal and abnormal cytology. In addition, these data provide a critical reference measurement important to assessing unanticipated outcomes such as increases in nonvaccine HPV types (i.e., HPV-type replacement).

Material and Methods

Registry

The New Mexico HPV Pap Registry (NMHPVPR) is located at the University of New Mexico and acts as a designee of the New Mexico Department of Health. The NMHPVPR operates under NMAC 7.4.3, which specifies the list of Notifiable Diseases and Conditions for the state of New Mexico. In 2006, with the intention of monitoring the impact of HPV vaccination, NMAC 7.4.3 specified that laboratories must report to the NMHPVPR all Pap cytology, cervical pathology and HPV tests performed on “individuals living in New Mexico/” vs New Mexico residents. Residents is a legal status and we do not know legal status only address. NMAC 7.4.3 was updated in 2009 to include vulvar and vaginal pathology (http://nmhealth.org/ERD/healthdata/documents/NotifiableDiseasesConditions022912final.pdf).

Study population and sample

During the 17-month period of December 2007 through April 2009 approximately 379,000 Pap tests were reported to the NMHPVPR by nine in-state and seven out-of-state clinical laboratories. We collected all available liquid cytology Pap specimens from seven out of the nine in-state laboratories, which accounted for 79% of all Pap tests done during this period. From each of these seven labs, we randomly selected specimens for HPV genotyping within four strata defined by the age of the woman (≤30 years versus >30 years) and by the cytologic result on the lab report (negative versus abnormal). Target sampling proportions varied by strata: 45% of negative specimens in women ≤30 years, 8% of negative specimens in women >30 years and 100% of abnormal specimens, regardless of the woman's age. Ultimately, a total of 59,644 specimens were retrieved and successfully genotyped for HPV. A schematic description of the sample design is shown in Supporting Information Figure 1. Prior to HPV genotyping, specimens were deidentified by the use of randomly assigned study-specific identifiers. The UNM Human Research Review Committee approved our study.

HPV genotyping

The LINEAR ARRAY HPV Genotyping Test (HPV LA; Roche Diagnostics, IN) is a qualitative test for 37 HPV genotypes incorporating selective PCR amplification with biotinylated PGMY 09/11 L1 region consensus primers and colorimetric detection of amplified products bound to immobilized HPV genotype-related oligonucleotide probes on a LINEAR ARRAY HPV genotyping strip. PGMY-based HPV genotyping with the HPV LA and a prototype Line Blot assay have been previously reported in detail.26–28 Following vigorous mixing of the original liquid cytology specimens, an aliquot of 500 μL of SurePath™ (Becton, Dickinson, NJ) or ThinPrep® (Hologic, MA) was transferred to 12 mm × 75 mm polypropylene tubes and DNA was purified using a Cobas X421 robot (Roche Molecular Systems [RMS], CA). The robot performed proteinase K digestion and inactivation with the final DNA eluate (150 μL) delivered into a 96-well QiaAmp plate (Qiagen, Valencia, CA). Fifty microliter (50 μL) of purified DNA was transferred to a tube with 50 μL of HPV LINEAR ARRAY mastermix, and the mixture was amplified by PCR using a Applied Biosystems Gold-plated 96-well GeneAmp PCR System 9700 as specified by the manufacturer. Controls for contamination and assay sensitivity were included in each 96-well assay.

Stability (storage temperature and time) and sensitivity studies were conducted prior to initiating the statewide sampling of Pap specimens (data not shown). Stability of β-globin and HPV DNA was consistent with the previous reports.29 HPV genotyping results were equivalent for SurePath™ samples held at 4°C following clinical collection whether processed and subjected to HPV genotyping at 7 days or up to 45 days postclinical collection. Therefore, SurePath™ samples were stored at 4°C at clinical labs and at the study lab and subjected to aliquoting, purification and HPV genotyping between 30 and 45 days following clinical collection. HPV genotyping results were equivalent for ThinPrep® samples stored at room temperature and subjected to HPV genotyping between 45 days and 6 months following clinical collection.

Using the Roche HPV LA detection kit, hybridizations were automated using Tecan ProfiBlot-48 robots (Tecan, Austria). The Roche HPV LA Genotyping Test detects 13 high- and 24 low-risk HPV types. HPV 52 is not determined directly by a type-specific probe but rather by a probe that cross hybridizes with HPV 33, 35, 52 and 58. The presence of HPV 52 was inferred only if the crossreactive probe was hybridized but there was no hybridization detected for the HPV 33, 35 and 58 type-specific probes. Notably, concurrent infections of type 52 with the three other types cannot be detected. Two independent readers interpreted the presence of HPV genotypes using a reference template provided by the manufacturer. Any discrepancies were identified by a custom computer application applied to the data input and were adjudicated by a third review.

HPV genotype groupings

In addition to HPV-type-specific prevalence, the combined prevalence of various groupings of HPV types was also computed: (i) groups based on the risk for cervical cancer: carcinogenic (HPV 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59 and 68) or low-risk (all other types including 6, 11, 26, 40, 42, 53, 54, 55, 61, 62, 64, 66, 67, 69, 70-73, 81-84, IS39 and 89); (ii) the five most common HPV types that have been detected in cervical cancer worldwide (HPV 16, 18, 31, 33 and 45)30, 31; (iii) groups based on types targeted by HPV vaccines: bivalent Cervarix™ (HPV 16 and 18), quadrivalent Gardasil™ (HPV 6, 11, 16 and 18) and the next generation nonavalent HPV vaccine (6, 11, 16, 18, 31, 33, 45, 52 and 58); (iv) groups based on phylogenetic species32 group: α 1 (HPV 42), α 3 (HPV 61, 62, 72, 81, 83, 84 and 89), α 5 (HPV 26, 51, 69, 82 and IS39), α 6 (HPV 53, 56 and 66), α 7 (HPV 18, 39, 45, 59, 68 and 70), α 8 (HPV 40), α 9 (HPV 16, 31, 33, 35, 52, 58 and 67), α 10 (HPV 6, 11 and 55), α 11 (HPV 64 and 73), α 13 (HPV 54) and α 15 (HPV 71).

Cytologic classifications

Cytologic results were classified according to the 2001 Bethesda System (TBS)33: high-grade squamous intraepithelial lesion (HSIL), atypical squamous cells cannot rule out HSIL (ASC-H), atypical glandular cells of undetermined significance (AGUS), low-grade squamous intraepithelial lesion (LSIL), atypical squamous cells of undetermined significance (ASC-US) and negative for intraepithelial lesion or malignancy. A small number of cytologic results were reported as LSIL cannot rule out HSIL (LSIL-H), a non-TBS categorization that has not been validated as a reproducible or clinically valuable cytologic category. We therefore grouped LSIL-H with ASC-H. Cytologic results reported as cervical intraepithelial neoplasia (CIN) grade 1 were classified as LSIL and results reported as CIN grade 2 (CIN2), CIN grade 3 (CIN3), carcinoma in situ or possible carcinoma were classified as HSIL. Cytologic results of “atypical squamous and glandular cells of undetermined significance” were classified as AGUS. Cytology was based on local readings and no attempt was made to review them centrally or undertake quality assurance activities.

Statistical methods

The study sample was selected prior to performing a woman-level linkage of the Pap test database and, as a consequence, multiple Pap specimens were collected and genotyped for some women. Subsequently, we performed a probabilistic linkage of all records in the NMHPVPR database using Link Plus (http://www.cdc.gov/cancer/npcr/tools/registryplus/lp_tech_info.htm), which allowed us to restrict the analysis to a single Pap specimen (the chronologically earliest) for women with more than one specimen in the study sample. This reduced the number of specimens in the analysis from 59,644 to 54,872. The linkage also gave us a Pap history for all women represented in the study sample extending back to at least January 2006.

All analyses were performed using sample survey techniques appropriate for a stratified random sample with unequal sampling fractions. The sample fractions are shown in Supporting Information Figure 1. Sample weights were computed as the inverse of the population sample fraction and variances were computed by the Taylor series linearization method. The SAS (v 9.2) procedure SURVEYFREQ was used to compute all proportions and SURVEYLOGIST was used to compute adjusted odds ratios. Smooth curves for the prevalence of HPV by age were computed using the LOESS procedure on the age-specific prevalence estimates from SURVEYFREQ. Confidence intervals were based on normal approximations.

Abbreviations

AGUS: atypical glandular cells of undetermined significance; ASC-H: atypical squamous cells cannot rule out HSIL; ASC-US: atypical squamous cells of undetermined significance; CIN: cervical intraepithelial neoplasia; CIN2: cervical intraepithelial neoplasia grade 2; CIN3: cervical intraepithelial neoplasia grade 3; HPV: human papillomavirus; HPV LA: HPV LINEAR ARRAY; HSIL: high-grade squamous intraepithelial lesions; LSIL: low-grade squamous intraepithelial lesions; LSIL-H: LSIL cannot rule out HSIL; NHANES, National Health and Nutrition Examination Surveys; NMHPVPR: New Mexico HPV Pap Registry; PCR: polymerase chain reaction; TBS: the 2001 Bethesda System

Results

Figure 1 shows the overall prevalence of carcinogenic HPV types by the number of months between the genotyped Pap specimen and the preceding Pap test for the woman. Detection of HPV was less common in Pap specimens taken 10 months or more (≥300 days) after the preceding Pap test (14.2% carcinogenic and 25.9% overall) versus those taken <10 months (29.9% carcinogenic [p < 0.0001] and 46.7% overall [p < 0.0001]). The prevalence of HPV was intermediate for specimens with no record of a preceding Pap test (19.2% carcinogenic and 32.6% overall). Based on these results, we used a working definition for a screening Pap test as one with no preceding Pap test in the previous 299 days, including those with no record of any preceding Pap test. This definition assumes that anyone returning in <10 months after the last screen was doing so for reasons other than routine annual screening. All estimates of HPV prevalence reported here have been restricted to these 47,617 screening Pap tests.

Figure 1.

Percent of cervical Pap specimens that are positive for any carcinogenic HPV type by the number of months since the previous Pap test. Specimens from women with no previous Pap test in the Registry are indicated as N. Pap specimens taken <10 months (<300 days) after the preceding Pap test (darker shading) are not classified as screens and were excluded from analyses.

The prevalence of each of the 13 carcinogenic or high-risk HPV types, individually and in various groupings, is summarized in Table 1. The prevalence of any HPV was 27.3% (95% confidence interval [CI]: 26.8–27.8%) and any carcinogenic HPV was 15.3% (95% CI: 14.9–15.7%). The prevalences of the five most common carcinogenic HPV genotypes were HPV 16 (3.5%), HPV 51 (2.3%), HPV 39 (2.1%), HPV 59 (2.1%) and HPV 52 (1.9%). The combined prevalence of the HPV 16 and HPV 18 was 4.5%, of the five most prevalent carcinogenic HPV types (HPV 16, 18, 31, 33 and 45) found in invasive cancer worldwide30 and in the United States31 was 7.2%, and of the seven carcinogenic types (HPV 16, 18, 31, 33, 45, 52 and 58) in the proposed nonavalent vaccine was 9.5%, the latter accounting for 62% of all carcinogenic HPV types detected. Low-risk HPV types were detected in 19.5% of specimens and 12.0% of specimens contained only low-risk HPV types. The five most common low-risk types were HPV 62 (3.0%), HPV 53 (2.9%), HPV 54 (2.3%), HPV 66 (2.3%) and HPV 61 (2.1%). The prevalence of each low-risk HPV type by age is summarized in Supporting Information Table 1. The most prevalent phylogenetic species group detected in women aged <30 years was the α-9 group (HPV 16, 31, 33, 35, 52, 58 and 67) followed by α-3 (HPV 61, 62, 72, 81, 83, 84 and 89). In contrast, for women aged ≥30 years the α-3 group was the most prevalent group detected followed by α-9. The prevalence of each phylogenetic species group by age is summarized in Supporting Information Table 2. Infection with multiple HPV types was common (data not shown). Among women who were positive for any type of HPV, multiple types were detected in 41.8% (95% CI: 40.8–42.8%) and were more common in specimens with abnormal cytology (62.5%, 95% CI: 61.6–63.5%) than in negative specimens (37.4%, 95% CI: 36.2–38.6%).

Table 1. The percent of cervical Pap specimens positive for carcinogenic HPV types by the age of the woman1
inline image

The overall prevalence of HPV rose sharply with increasing age in adolescent girls to a maximum of 52% at age 20 years and then rapidly declined after age 30 years, with a prevalence of 32% at age 30 and <20% by age 50 years (Fig. 2a). Low risk types were relatively more common at older ages. Thus, while low-risk and carcinogenic types were equally common at age 20 years (37% for both), in women ≥30 years of age low-risk HPV types were much more commonly detected than carcinogenic HPV types (14.9 versus 9.5%). Figure 2b shows the prevalence of carcinogenic HPV types included in the bivalent, quadrivalent and nonavalent vaccine by age. At age 20 years, the prevalence of HPV 16 and 18, the quadrivalent HPV vaccine types (HPV 6, 11, 16 and 18) and the nonavalent HPV vaccine types were 13, 17 and 27%, respectively. Figure 2c shows the reduction in prevalence of carcinogenic HPV as vaccine types are removed from the collection of carcinogenic types, modeling the impact of HPV vaccination now and in the future. This assumes 100% eradication of those types, which may be unachievable.

Figure 2.

The age-specific prevalence of HPV in cervical Pap specimens. (a) The percent of specimens positive for any HPV type, any carcinogenic HPV type, and any low-risk HPV type. (b) The percent positive for any bivalent HPV vaccine type (16 or 18), any quadrivalent HPV vaccine type (6, 11, 16 or 18) and any nonavalent HPV vaccine type (6, 11, 16, 18, 31, 33, 45, 52 or 58). (c) The percent positive for any carcinogenic HPV type, any carcinogenic type other than 16 or 18, any carcinogenic type other than 16, 18, 31, 33 and 45, and any carcinogenic type other than 16, 18, 31, 33, 45, 52 or 58. The specimen may be positive or negative for any of the excluded types.

As summarized in Table 2, the prevalence of carcinogenic HPV increased with the severity of the cytologic result (excluding AGUS): Negative (12.5%), ASC-US (41.7%), LSIL (74.4%) and HSIL (90.9%). The prevalence of HPV 16, 18, 31, 33 and 58 increased, whereas the prevalence of HPV 39, 45, 51, 56, 59 and 68 decreased. For low-risk types, the prevalence increased from negative (17.4%) to ASC-US (41.0%) to LSIL (69.9%), but then decreased in HSIL specimens (42.3%). This pattern held with few exceptions for all of the individual low-risk HPV types.

Table 2. The percent of cervical Pap specimens positive for carcinogenic HPV types by cytologic result1
inline image

The prevalence of each low-risk type and each phylogenetic group by cytologic result is summarized in Supporting Information Tables 3 and 4, respectively. The α-3 group was the most prevalent group detected in women with negative cytology results (9.9%). For all abnormal cytology groups, the α-9 group was the most prevalent group and the prevalence of this group increased from AGUS (16.4%) to ASC-US (27.3%) to LSIL (45.5%) to ASC-H (58.0%) to HSIL (73.8%). Notably, the α-7 group was the second most prevalent group detected in women with an AGUS test result (12.3%).

The age-adjusted odds ratios for high-grade disease (HSIL, ASC-H or AGUS) versus cytologic negative in specimens with a single type of HPV relative to HPV negative specimens are shown in Figure 3. Carcinogenic HPV types 33, 16, 18, 31 and 58 had odds ratios >10.0, HPV 35, 51, 52 and 45 had odds ratios between 5.0 and 10.0, and HPV 39, 56, 59 and 68 had odds ratios <5.0. The only low-risk HPV types with odds ratios >5 were HPV 82, IS39 and 67. Supporting Information Figure 2 shows the same odds ratios without restriction to single HPV infections. Although the range of the odds ratios is decreased, the ranking of the HPV types is very similar.

Figure 3.

Age-adjusted odds ratios and 95% CIs for cytologic result of High-Grade (HSIL, ASC-H or AGUS) versus Negative in specimens with HPV of specified type (single infection) compared to those without HPV infection. Numbers on right margin are the total number of single HPV infections of the specified type. No high-grade results found in specimens with HPV types 69 (n = 3), 64 (n = 5), 26 (n = 3) or 11 (n = 2). Reference lines for odds ratios of 1, 5 and 10 are added for readability.

Discussion

As part of our statewide New Mexico HPV prevention program, we measured the prevalence of 37 individual HPV types prior to widespread HPV vaccination. To our knowledge, this is the largest systematic population-based HPV prevalence study conducted to date. The effort was conducted specifically for the purpose of measuring HPV vaccine impact and integrating HPV vaccination and cervical screening programs. The HPV prevalence estimates reported here are intended for use as a baseline for comparison to future samplings of the population. The large sample size enables precise estimates of both increases and decreases in HPV type-specific prevalence and, through linkage to cervical pathology results in the NMHPVPR, detection of changes in the incidence of HPV type-specific disease outcomes including CIN2 and CIN3.

Women aged ≤30 years of age represent the most relevant group for detecting an early impact of HPV vaccination on the prevalence of HPV. The stratified random sampling strategy employed in our study allowed us to increase the precision of HPV prevalence estimates in this younger age group as well as in women with abnormal cytology. HPV prevalence within abnormal cytology categories is important for understanding the degree that coinfections with nonvaccine and vaccine HPV types will modify the clinical impact of HPV vaccines. Furthermore, both abnormal and normal cytology categories will be essential to monitor HPV prevalence in the whole population as multiple types and changes in the reading of cytology may impact on the prevalence in each group separately and therefore mask any potential HPV type replacement.

In assessing the impact of different types on HSIL, we have provided two analyses, one restricted to single infections (Fig. 3) and one including coinfections (Supporting Information Fig. 2), as it is impossible to determine which type was associated with the lesion when multiple types are present. Including coinfections would lead to a dilution of the true effects owing to misattribution. As there is no strong evidence for preference of coinfection with specific pairs of types, this is unlikely to lead to a serious bias, but further detailed analysis of coinfections is currently ongoing and will be reported separately. These analyses of coinfections will be required to carefully address any potential suggestion of type replacement.

Among carcinogenic HPV types, the most commonly detected types in the population were HPV 16 followed by HPV types 51, 39 and 59. HPV types 39 and 59 have been reported previously as prevalent HPV types detected in invasive cervical cancers from Central and South America.34 In line with this observation, the Hispanic and Native American ancestry of New Mexico's populations may represent historical founder effects that would be expected to influence the HPV-type distribution observed in our population sample. Consistent with the previously reported increased association of the α-7 group with adenocarcinoma30, 35, 36 (versus squamous cell carcinoma), this group was the second most prevalent group detected in women diagnosed with AGUS.

Both carcinogenic and low-risk HPV had peak prevalence around age 20 followed by a rapid decline. Low-risk HPV declined less rapidly with age than carcinogenic HPV so that, while carcinogenic and low-risk HPV were equally common in younger ages, by age 30 low-risk HPV, especially types of the α-3 phylogenetic group were detected more frequently than carcinogenic HPV. Little attention has been devoted to studying differences of type distribution at different ages. Similar to another US study that included large numbers of older women,37 we failed to observe any notable upturn in HPV prevalence at older ages.

Our data were consistent with the previously reported approximate threefold excess risk of CIN2+ associated with HPV 16 compared to all other carcinogenic HPV types.38 Potentially of importance is the observation that HPV 33 has the highest age-adjusted odds ratio for HSIL when examining women with single infections. Notably, when considering samples with coinfections, HPV 33 had the second highest age-adjusted odds ratio for HSIL, only surpassed by HPV 16. HPV 33 incident infections are at high risk of progression to CIN2/339, 40 and among HPV 33-positive women the risk of carcinoma in situ, adenocarcinoma in situ or invasive cervical cancer relative to HPV 16 has been previously reported to be essentially equal to that of HPV 16-positive women.31 It has also been suggested that HPV 33 may have a competitive advantage over several other oncogenic HPV types41 and it may be important to evaluate this following HPV 16/18 vaccination.

A considerably smaller US study of 4,150 females age 14–59 years derived from consecutive National Health and Nutrition Examination Surveys (NHANES; 2003–2006) was reported as a means for estimating HPV vaccine impact.42, 43 The NHANES study employed the same LA HPV genotyping assay used in our study. The most common HPV type detected in the NHANES study was not HPV 16 but rather low-risk HPV 62 (6.5%). The next most prevalent HPV types were HPV 53 (5.8%) and HPV 84 (4.8%). Overall, HPV prevalence for the same 37 HPV types was considerably higher in the NHANES study (42.5%, 95% CI: 40.3–44.7%) compared to our study (27.3%, 95% CI: 26.8–27.8%) with the greatest difference observed in the older age groups. The NHANES study used a cervicovaginal self-sampling method, which may have resulted in increased prevalence, consistent with the lower specificity of self-collection compared to clinician collection with HPV DNA testing, and probably led to an oversampling of vaginal HPV types including HPV 62 and 84.44, 45

Perhaps also because of cervicovaginal sampling the NHANES-based study reported a higher prevalence for HPV types 6 and 11 in girls 14–19 versus those 20–24 years of age or older. In contrast, the highest prevalence of HPV 16 and 18 and for low-risk and carcinogenic HPV-type groupings was seen in those 20–24 years of age.

Another study conducted to estimate the potential impact of HPV vaccines was reported from Denmark.46 Both our New Mexico and the Danish study analyzed samples from women participating in cervical screening. The Danish study excluded follow-up Pap tests after an abnormal cervical cytology/histology diagnosis within 12 months of the index sample. It is important to note that different age strata were reported for the NHANES, New Mexico and Danish studies and the two youngest age categories overlapped the peak of HPV prevalence at around age 20. This could at least partially explain differences in results although inherent differences in the underlying type-specific HPV prevalence, as well as age-related sexual behaviors of the populations may be contributing factors. For example, in the NHANES study, approximately 16% of individuals reported never having sexual intercourse, which would be expected to be less common in those participating in cervical screening.

One of the limitations of our study is that women would have to be undergoing screening to be included but this will be the case in future samplings where we will target the same laboratories and compare age distributions within- and between-laboratories for the two sampling periods. Unscreened women would presumably have higher prevalence of CIN2/3 and hence any sampling strategy targeting women who are screening will likely underestimates the prevalence of carcinogenic HPV genotypes.

Population-based HPV prevalence estimates represent both persistent and incident HPV infections and will be expected to differ for a variety of reasons including: the sampling strategy, the overall exposure risk profile of the population being sampled including the proportion of individuals who are not yet sexually active, the cervical sampling methods and devices and the sensitivity and specificity of the HPV detection assays. Differences will also be expected when assessing routine cervical screening tests versus samples which include repeat referral Pap tests wherein the later would be expected to yield higher HPV prevalence estimates. To address this, we assessed cervical screening histories using the NMHPVPR surveillance data through woman-level linkages. The resulting sample used for estimating HPV prevalence excluded specimens from women who had any record in the Registry of a Pap test within the previous 10 months. In support of using this latter selection criterion, we provide evidence of significantly greater HPV prevalence among Pap tests collected at intervals of <10 months since a woman's preceding Pap test. We highlight our working definition of a screening (versus a diagnostic) Pap test given the inherent challenges expected when conducting longitudinal evaluations and related HPV prevalence comparisons in the context of cervical screening. Evolving data applicable to the US and other populations support lengthening of cervical screening intervals and starting screening at an older age (e.g., age, 25 or 30 years), which may itself result in a corresponding reduction in the detection of at least some proportion of transient HPV infections and their associated abnormalities, which are known to be most common in younger women. We acknowledge that careful consideration of sampling strategies will be critical when proposing cross-sectional approaches to estimate HPV genotype prevalence as an indicator of HPV vaccine impact.

With these caveats in mind, reductions in the prevalence of certain HPV genotypes resulting from HPV vaccination will be expected to precede or be coincident with reductions in clinically relevant measures such as abnormal cervical cytology and histology. Comparisons of results from various populations would benefit from standardization of HPV prevalence estimates. Standardization procedures should consider establishing a specific overall age range (e.g., 20–59 years) and include adjustments for the population age distribution, HPV types included in the prevalence measures and HPV assay sensitivity referenced to Hybrid Capture 2 HPV Test (Qiagen, Silver Spring, MD) or to international HPV standards.47 Specific proposals for reporting HPV prevalence are being developed and will be reported separately.

The impact of HPV vaccines on cervical cancer incidence will be dependent on vaccine-related issues such as age of vaccination in relation to the age of sexual initiation for the population, vaccine coverage, efficacy achieved with one, two and three doses of vaccine, the level of crossprotection afforded against HPV types not included in the vaccine preparations, durability of vaccine efficacy as well as population characteristics including the baseline prevalence of HPV types for which HPV vaccination provides protection versus the prevalence of other HPV types that will not be prevented by current formulations, and future screening efficacy and coverage. Given these complexities, it will be difficult to directly translate measurable changes in HPV prevalence to expected reductions in clinically meaningful endpoints including abnormal cytology and CIN. Models have been proposed,23 but they all rely on assumptions that have not been verified. Yet, HPV is the necessary cause of cervical cancer and hence the reduction in HPV prevalence must therefore correlate with cancer protection.

Data from phase III clinical trials where over 90% of participants received three doses of HPV vaccine9–11, 48 provide us with the best current estimates of potential vaccine impact using approximated naïve populations as well as more general populations that included sexually active women with prior or current HPV infections. In the combined FUTURE I and II population that was naïve to 14 HPV types at baseline,47 the vaccine impact on Pap test abnormalities irrespective of causal HPV type at the end of the study (average follow-up, 3.6 years) resulted in an overall reduction of 17.1% for any abnormality and there was a general increase in reduction with increasing lesion severity: LSIL, 17.0% reduction, 95% CI (8.8–24.4%); HSIL, 44.5% reduction, 95% CI (4.3–68.6%).

In 2006, the quadrivalent HPV vaccine was licensed in the United States. Although New Mexico focused its initial vaccine implementation efforts on girls aged 11–14, we acknowledge that at least some proportion of the women who underwent Pap testing had been vaccinated. We are currently working to describe HPV vaccine implementation in New Mexico using statewide HPV vaccine administrative data. Exclusions of vaccinated women from HPV prevalence estimates can therefore be reconsidered in subsequent analyses. Through comprehensive HPV genotyping studies such as that reported here, through similar HPV genotyping studies in population-based CIN and through linkages between the NMHPVPR and the HPV vaccine administrative data, we will be able to contribute to more precisely understanding longitudinal changes in HPV prevalence and HPV-related disease incidence in New Mexico's populations in the vaccine era. Overall, these efforts will be important to establishing the rational integration of HPV vaccination and cervical screening as complementary cervical cancer prevention strategies and will serve to inform public health practices and policy.

Acknowledgements

The authors are grateful to the directors and staff of the New Mexico laboratories contributing specimens to this investigation and to Ray Apple PhD who has continuously extended efforts to enable the laboratory capacity required to perform a statewide effort of this magnitude. Special thanks to their students, Caitlin Hillygus, Christopher Romero and Anthony Pena working on this project and to the members of the New Mexico HPV Pap Registry (NMHPVPR) Steering Committee who supported the concept and directions of the NMHPVPR through their generous efforts over many years.

Ancillary