Differences in human papillomavirus type distribution in high-grade cervical intraepithelial neoplasia and invasive cervical cancer in Europe

Authors


  • Conflicts of Interest: W.A. Tjalma has received support for travel to meetings for the study or other purposes from GSK, Merck and Sanofi Pasteur. A. Fiander has been on advisory boards for both GSK and Sanofi Pasteur MSD and received research grant funding and support to attend HPV-related conferences from both companies. N. Powell's laboratory has received funding from GSK, including for the work in support of this study; he has received honoraria from GSK and Sanofi Pasteur to speak/advise on HPV and has also received support for travel to study-related meetings. A.M. Nowakowski and his institution have been paid by GSK for participation in the study, and he has received support for travel to study-related meetings; he has also been paid by GSK and MSD for lectures. B. Kirschner has received support from GSK for travel and is a paid consultant to GSK's Advisory Board on vaccine-related issues. J. O'Leary, V. Damaskou, M. Repanti and their institutions received funding from GSK to complete the work disclosed in this publication. E.A. Joura and his institution have received grants from GSK and Merck for HPV vaccine studies; he is also a paid member of the board of Merck (HPV vaccines), has received travel expenses from Merck, Sanofi Pasteur MSD and GSK and has also been paid by GSK, Merck and Sanofi Pasteur MSD for lectures. M. Rosenlund was employed by GSK Biologicals at the time of study; B. Colau, K. Holl and D. Rosillon are employed by GSK Biologicals; B. Colau and M. Rosenlund also have stock or stock options in GSK Biologicals. D. Schledermann has received support from GSK for congress registration and travel. A. Savicheva and E. Shipitsyna have received consulting fees or honoraria for their participation in the study and travel expenses and have previously participated in a HPV vaccine trial for GSK. The employer of A. Molijn and W. Quint, DDL Diagnostic Laboratories, has received support from GSK for travel to meetings for the study or other purposes, fees for participation in review activities such as data monitoring boards, statistical analysis and end-point committees and also grants and fees for consultancy outside the submitted work. The employer of S. Collas De Souza and A. Raillard, 4Clinics, has received consulting fees from GSK for statistical analysis. D. Jenkins was, at the time of this study, initially an employee and then consultant in HPV to GSK Biologicals. O. Reich, R. Koiss, K. Kukk, R. Vladareanu, L. Kolomiets and J. Ordi declare no conflicts of interest. Role of the funding source: The costs related to the collection of tissue and the analysis of data from both studies, together with the development of this article, were funded and coordinated by GlaxoSmithKline Biologicals.

Abstract

Knowledge of differences in human papillomavirus (HPV)-type prevalence between high-grade cervical intraepithelial neoplasia (HG-CIN) and invasive cervical cancer (ICC) is crucial for understanding the natural history of HPV-infected cervical lesions and the potential impact of HPV vaccination on cervical cancer prevention. More than 6,000 women diagnosed with HG-CIN or ICC from 17 European countries were enrolled in two parallel cross-sectional studies (108288/108290). Centralised histopathology review and standardised HPV-DNA typing were applied to formalin-fixed paraffin-embedded cervical specimens dated 2001–2008. The pooled prevalence of individual HPV types was estimated using meta-analytic methods. A total of 3,103 women were diagnosed with HG-CIN and a total of 3,162 with ICC (median ages: 34 and 49 years, respectively), of which 98.5 and 91.8% were HPV-positive, respectively. The most common HPV types in women with HG-CIN were HPV16/33/31 (59.9/10.5/9.0%) and in ICC were HPV16/18/45 (63.3/15.2/5.3%). In squamous cell carcinomas, HPV16/18/33 were most frequent (66.2/10.8/5.3%), and in adenocarcinomas, HPV16/18/45 (54.2/40.4/8.3%). The prevalence of HPV16/18/45 was 1.1/3.5/2.5 times higher in ICC than in HG-CIN. The difference in age at diagnosis between CIN3 and squamous cervical cancer for HPV18 (9 years) was significantly less compared to HPV31/33/‘other’ (23/20/17 years), and for HPV45 (1 year) than HPV16/31/33/‘other’ (15/23/20/17 years). In Europe, HPV16 predominates in both HG-CIN and ICC, whereas HPV18/45 are associated with a low median age of ICC. HPV18/45 are more frequent in ICC than HG-CIN and associated with a high median age of HG-CIN, with a narrow age interval between HG-CIN and ICC detection. These findings support the need for primary prevention of HPV16/18/45-related cervical lesions.

In Europe, there are an estimated 61,000 new cases and 28,000 deaths from invasive cervical cancer (ICC) annually. ICC is the second most frequent cancer among women aged between 15 and 44 years.1, 2 ICC and/or its precursor lesions remain a major public health and socioeconomic problem in countries with and without screening programmes.3

Persistent infection with high-risk (HR) human papillomavirus (HPV) types is necessary for the transformation of normal cervical epithelium to high-grade cervical intraepithelial neoplasia (HG-CIN) and progression to ICC.4–6 To date, more than 40 anogenital HPV types have been identified, of which at least 14 types are associated with ICC.7–9 The necessary role of HR-HPV infection in the development of cervical neoplasia provides an opportunity to reduce the disease burden through prevention programmes including prophylactic HPV vaccination.10–12

Type-specific distributions of HPV in HG-CIN and ICC have been reported in Europe2, 8, 13 and suggest that the frequency of specific HR-HPV types differs between HG-CIN and ICC. Prospective follow-up has shown a long-term increase in the risk of HG-CIN in women with persistent HPV16 and HPV18 infections.14 Studies also indicate an under-representation of HPV18 and HPV45 in HG-CIN compared to ICC and the relatively late diagnosis of CIN3 associated with these types.15, 16 However, there is still a lack of observational data across European countries reporting the prevalence of HPV types in ICC versus HG-CIN. Therefore, the knowledge on HPV-type distribution in ICC and HG-CIN is still largely incomplete in many European countries.

The primary objectives of these studies were to assess the HPV-type distribution detected in HG-CIN [CIN2, CIN2/3, CIN3 and adenocarcinoma in situ (AIS)] and/or ICC [squamous cervical cancer (SCC), adenocarcinoma (ADC), adenosquamous carcinoma (ASC) and ‘other’ diagnoses] diagnosed in women aged ≥18 years in 17 European countries. Prevalence ratios of specific HPV types in ICC versus HG-CIN (ICC/HG-CIN prevalence ratios) and the median age of HG-CIN and ICC diagnoses for different HPV types were estimated. Centralised expert histopathological review and standardised HPV DNA typing for 14 HR-HPV types were applied to more than 6,000 parallel samples of HG-CIN and ICC. The pooled prevalence of individual HPV types was estimated using meta-analytic methods to control for heterogeneity observed among the countries.

Abbreviations

ADC: adenocarcinoma; AIS: adenocarcinoma in situ; ASC: adenosquamous carcinoma; CI: confidence interval; HG-CIN: high-grade cervical intraepithelial neoplasia; HPV: human papillomavirus; HR: high risk; ICC: invasive cervical cancer; LR: low risk; SCC: squamous cell carcinoma

Material and Methods

Study design and population

HERACLES (HPV Epidemiology Research Applied to Cervical Lesions: a European Study) and SCALE (Study on Cervical Cancer Lesions in Europe) were parallel, cross-sectional, multicentre studies of HPV-type distribution in women diagnosed with HG-CIN and ICC, respectively. Thirteen European countries participated in HERACLES (Austria, Czech Republic, Denmark, Estonia, Greece, Hungary, Ireland, Norway, Poland, Portugal, Romania, Russia and Spain), and 12 European countries in SCALE (Belgium, Czech Republic, Denmark, Germany, Greece, Hungary, Norway, Poland, Portugal, Romania and Scotland and Wales, both in the United Kingdom). Eight countries were involved in both studies (Denmark, Greece, Portugal, Norway, Hungary, Czech Republic, Poland and Romania).

In each participating country, only study sites that maintained an archive of cervical excision specimens were selected. From each participating country, 290 consecutive archived formalin-fixed paraffin-embedded cervical specimens of HG-CIN and/or ICC diagnosed between 2001 and 2008 were collected (Fig. 1). The collection procedures were standardised. If several specimens were available for a subject, the most recent paraffin block containing the area with the highest grade of CIN or the primary ICC obtained prior to chemo/radiotherapy, was selected. Information about age at specimen collection, year of specimen collection and original histological diagnosis were obtained.

Figure 1.

Study cohorts and exclusion criteria for (a) HERACLES and (b) SCALE.

To be included in the studies (Total enrolled cohort), specimens had to come from a woman aged ≥18 years at the time of specimen collection, be the specimen on which the relevant diagnosis was made, be of appropriate size (≤2 cm diameter and ≥2 mm thickness) and be adequately preserved. The Histologically eligible cohort included only specimens for which histological examination on sections flanking those taken for DNA analysis were concordant for grade and type of abnormality and for which the histological diagnosis was eligible for study inclusion. If the number of specimens withdrawn in one country exceeded 5%, additional specimens were requested from that country. Once the histological diagnosis was deemed eligible, each specimen was tested for the presence of HPV DNA, and HPV-positive cases (i.e., HPV+) were included in the HPV+ cohort (Fig. 1).

Each study was approved by the appropriate Institutional Review Board and/or Independent Ethics Committee in that participating country and was conducted in accordance with the Declaration of Helsinki and Good Clinical Practice guidelines. Where required, written informed consent was obtained from participating subjects prior to any study procedure on clinical specimens.

Laboratory procedures

Following anonymisation, specimens included in the Total enrolled cohort were shipped to a central laboratory (DDL Diagnostic Laboratory, Rijswijk, The Netherlands) for centralised histopathology review in order to standardise the diagnosis of HG-CIN and ICC across the countries. To confirm the presence of HG-CIN or worse in the sections used for DNA analysis, 4-μm sections for haematoxylin and eosin (H&E) staining were taken immediately before and after the sections used for HPV DNA analysis. Two sets of 3 μm × 8 μm sections were taken for whole-section HPV DNA analysis. All cutting procedures were performed in a histology unit specialised in molecular analysis, with strict and comprehensive measures to prevent cross contamination. H&E sections were examined by an experienced gynaecological histopathologist, blinded to the initial diagnosis. If lesions of different grades were present in a single section, the worst grade of lesion diagnosed represented the study clinical diagnosis.

Samples with an eligible standardised histopathological diagnosis were further tested for HPV using the SPF10-DEIA/LiPA25-polymerase chain reaction (PCR) system (SPF10-LiPA25; version 1, Labo Biomedical Products, Rijswijk, The Netherlands, based on licensed Innogenetics technology).17, 18 Total DNA was isolated and extracted using a proteinase K lysis procedure. HPV DNA was amplified and detected by SPF10-PCR DEIA. The HPV genotype was identified in all positive samples by reverse hybridisation probe assay (SPF10-LiPA25), which detected 14 HR (16/18/31/33/35/39/45/51/52/56/58/59/66/68/73) and 11 low-risk (LR) HPV types (6/11/34/40/42/43/44/53/54/70/74).17

Each run contained positive and negative controls to monitor DNA isolation, PCR amplification, HPV detection and genotyping procedures. The DNA of HPV-negative samples was diluted tenfold and the testing was repeated. As a quality control measure, the pathological diagnosis of all HPV-negative CIN2, CIN2/3 and CIN3, all AIS and ADC and a random selection of ‘other’ diagnoses of both HG-CIN and ICC were reviewed by up to three independent expert pathologists who were blinded for HPV status. Cases were classified according to the guidelines of the 2003 WHO classification of tumours of the female genital tract.19 In case of disagreement, a majority decision was accepted; if all pathologists disagreed, the case was rejected.

The aim of this process was to confirm the presence of a lesion meeting study criteria and to obtain consistent histological classification of HG-CIN and ICC from a wide range of sources, rather to investigate agreement between the histological diagnoses assessed by the central laboratory (DDL) and the routine diagnoses of local country laboratories.

Statistical analysis

The primary population for analysis was the HPV+ cohort comprising women with an eligible standardised diagnosis of HG-CIN or ICC and HPV-positive (HPV+) PCR (Fig. 1). Sample size calculations demonstrated that with 210 HPV+ women in each country, the precision of the 95% confidence interval (CI) of percentages for each HPV type would not exceed 6% when percentages were lower than 20%, and 5% when percentages were lower than 10%. Considering that the sensitivity of the HPV testing method was 90% for HG-CIN samples and 95% for ICC samples and that the validation rate was 80% for HG-CIN samples and 85% for ICC samples, 290 women diagnosed with either HG-CIN or ICC were enrolled in HERACLES and/or SCALE studies in each country, respectively.

Proportions of HPV types in women diagnosed with HG-CIN (CIN2, CIN2/3, CIN3 and AIS) and/or ICC (SCC, ADC, ASC and ‘other’ diagnoses) by country were computed with their two-sided 95% CIs (Clopper-Pearson method).20 For each individual HPV type, Cochran Q test assessed the null hypothesis of homogeneity of the prevalence of the HPV type across the countries, at 5% significance level.

For all countries combined, the pooled prevalence of individual HPV types in women infected with a single HPV type and diagnosed with HG-CIN or ICC was estimated using meta-analytic methods. For each individual HPV type, the proportion in each country was first transformed (Freeman-Tukey double-arcsine variance-stabilising transformation).21 The transformed proportions were then weighted by the inverse of the sum of their within-country variance and their between-countries variance.22 The mean of the weighted transformed proportions was back-transformed to obtain the pooled proportion of the given HPV type. The overall prevalence of individual HPV types in this method was computed with its 95% CI.

The ICC/HG-CIN prevalence ratio, with its 95% CI,23 was computed for the different HPV types. Comparison of age difference between CIN3 and SCC and between AIS and ADC for the different HPV types was performed using contrasts in a two-way ANOVA analysis with histological diagnosis and HPV type as the main factors and an interaction term.

All statistical analyses were performed using SAS software (version 9.1).

Results

A total of 3,979 (65.5%) of the 6,080 selected women and 3,626 (80.8%) of the 4,486 selected women were enrolled in the Total enrolled cohorts of HERACLES and SCALE studies, respectively (Fig. 1).

The Histologically eligible cohorts included 3,103 (78.0%) women with the diagnosis of HG-CIN and 3,162 (87.2%) women with the diagnosis of ICC. Among women diagnosed with HG-CIN: CIN3, CIN2, CIN2/3, AIS and ‘other’ accounted for 73.6, 15.3, 9.4, 0.7 and 0.9% of cases, respectively. Among women diagnosed with ICC: SCC, ADC and ‘other’ accounted for 77.7, 13.4 and 8.9%, respectively (Table 1). The median age (years) (range) at the time of specimen collection was 34 years (18–86) for HG-CIN [34 years (18–86) for CIN3, 33 years (19–84) for CIN2, 36 years (19–77) for CIN2/3 and 35 years (22–55) for AIS] and 49 years (19–99) for ICC [50 years (19–99) for SCC and 45 years (22–90) for ADC] (Table 1).

Table 1. HERACLES (HG-CIN) and SCALE (ICC) cohort demographics
inline image

Forty-six (1.5%) and 259 (8.2%) women diagnosed with HG-CIN and ICC, respectively, had cervical specimens that were HPV-negative. Women diagnosed with HPV-negative ICC tended to be older, with the highest proportion of these cases being diagnosed in women aged ≥61 years (45.6%; Fig. 2).

Figure 2.

Age-specific HPV-type distribution in women diagnosed with HG-CIN or ICC.

Multiple HPV infection occurred in 17.2% of HG-CIN and 4.4% of ICC diagnoses. Multiple HPV infection was more frequently observed in younger women diagnosed with HG-CIN, decreasing in frequency with increasing age (Fig. 2).

The HPV+ cohorts included 3,057 (98.5%) women with HG-CIN and 2,903 (91.8%) women with ICC (Fig. 1). A total of 2,445 (80%) HPV+ women with HG-CIN and 2,715 (93.5%) HPV+ women with ICC were infected with a single HPV type. The median age at the time of specimen collection was 34 years (range: 18–86 years) for HPV + HG-CIN and 48 years (19–99 years) for HPV + ICC (Table 1).

In women infected with a single HPV type and diagnosed with HG-CIN, the most common types were HPV16 (59.9%), HPV33 (10.5%), HPV31 (9.0%), HPV52 (3.9%) and HPV18 (3.6%). In CIN2, HPV16/31/33/52/18 accounted for 42.5/13.3/10.2/5.2/3.3%; and in CIN3, 64.5/8.4/10.9/3.5/2.8%. HPV16 and HPV18 accounted for all AIS (47.7 and 52.3%, respectively; Table 2).

Table 2. Pooled estimates of HR-HPV-type prevalence among women with HG-CIN or ICC and infected with a single HPV type, overall and by histological diagnosis, in 17 European countries between 2001 and 2008
inline image

In women infected with a single HPV type and diagnosed with ICC lesions, the most common types were HPV16 (63.3%), HPV18 (15.2%), HPV45 (5.3%), HPV33 (4.6%) and HPV31 (3.7%). In SCC, these types accounted for 66.2, 10.8, 5.0, 5.3 and 4.1% of cases; and in ADC, 54.2, 40.4, 8.3, 1.5 and 1.4% of cases (Table 2).

In women infected with a single HPV type, the pooled prevalence of LR-HPV types was 0.65% (95% CI = 0.00–3.63) in HG-CIN and 0.73% (95% CI = 0.00–4.30) in ICC.

In each country studied, HPV16 was the most common type to be associated with both HG-CIN and ICC in women infected with a single HPV type. The prevalence varied from 47.1% in Norway to 71.9% in Estonia for HG-CIN and from 54.4% in Norway to 72.8% in Poland for ICC. Similar variations across the countries were observed for HPV18/31/33/45 and ‘other’ types (Table 3).

Table 3. Country-specific HPV16/18/31/33/45 and ‘other’ type distribution in women infected with a single HPV type and diagnosed with HG-CIN or ICC in 17 European countries between 2001 and 2008
inline image

Patterns of age-specific HPV prevalence differed between HG-CIN and ICC. In women infected with a single HPV type and diagnosed with HG-CIN, the most frequent age group was 26–30 years (21.3%), with decreasing numbers of women in the older age groups. The proportion of women infected with a single HPV type and diagnosed with ICC increased with increasing age, with the most frequent age group being ≥61 years (23.5%). This pattern was also observed for women with single HPV16/31/33-related HG-CIN and ICC diagnoses [HPV16 (largest group): 26–30 years (22.3%) for HG-CIN, ≥61 years (22.7%) for ICC; HPV31: 26–30 years (25.9%) for HG-CIN, ≥61 years (35.6%) for ICC and HPV33: 31–35 years (24.1%) for HG-CIN; ≥61 years (41.0%) for ICC]. In contrast, the highest proportion of women with single HPV18-related HG-CIN and ICC diagnoses was observed in the 31- to 35-year age group (24.4%) and in the 51- to 60-year age groups (19.3%), respectively, suggesting that there is less time between diagnoses of HPV18-related HG-CIN and ICC than for those related to HPV16/31/33. A similar pattern to that of HPV18-related HG-CIN and ICC was observed for women diagnosed with HPV45-related HG-CIN and ICC [26–30 years (22.2%) for HG-CIN; 51–60 years (20.4%) for ICC].

In the 534 women infected with multiple HPV types and diagnosed with HG-CIN, the most frequent HPV types were HPV16 (59.9%; 55.6–64.1), HPV31 (26.0%; 22.4–30.0), HPV52 (23.4%; 19.9–27.2), HPV33 (18.0%; 14.8–21.5), HPV51 (15.0%; 12.1–18.3) and HPV18 (14.2%; 11.4–17.5).

In the 138 women infected with multiple HPV types and diagnosed with ICC, the most frequent HPV types were HPV16 (52.9%; 44.4–61.4), HPV18 (26.8%; 19.6–35.0), HPV52 (21.0%; 14.5–28.8), HPV31 (19.6%; 13.3–27.2), HPV45 (18.8%; 12.7–26.4) and HPV33 (17.4%; 11.5–24.8). In women diagnosed with SCC, these HPV types were detected in 52.9, 25.0, 24.0, 23.1, 16.3 and 19.2% of cases; and in ADC in 41.2, 29.4, 11.8, 5.9, 35.3 and 11.8% of cases.

In the eight countries participating in both the HERACLES and the SCALE studies, the total number of women diagnosed with HG-CIN and ICC and included in the Histologically eligible cohorts was 1,923 and 2,138, respectively. The demographic and clinical characteristics of these two populations were similar to those in the overall population in the respective studies (Table 1). The frequencies of HPV types in HG-CIN and ICC were also similar to those seen in the overall population (data not shown).

In women infected with a single HPV type, the ICC/HG-CIN prevalence ratio varied for individual HPV types. Overall, HPV39/18/45 were 4.7/3.5/2.5 times more prevalent in ICC than HG-CIN. HPV16 was common in both ICC and HG-CIN with an ICC/HG-CIN prevalence ratio of 1.1, whereas HPV31/33/35/52 were 0.4/0.4/0.5/0.5 less frequent in ICC than HG-CIN (Table 4).

Table 4. ICC/HG-CIN prevalence ratios for the different HR-HPV types, overall and by age class
inline image

The median age at diagnosis varied according to HPV type, with CIN3 being diagnosed earlier in women infected with HPV16 (34 years), HPV31 (32.5 years) and HPV33 (35 years) and at a later median age in women with HPV18 (38 years) and HPV45 (42 years). In contrast, SCC tended to be diagnosed earlier in women infected with HPV16 (49 years), HPV18 (47 years) or HPV45 (43 years) and at a later age in women infected with HPV31, HPV33 and ‘other’ HPV types (Fig. 3). The difference in age at specimen collection between CIN3 and SCC for HPV18 (9 years) was statistically significantly narrower compared to the difference in age at specimen collection for HPV31 (23 years), HPV33 (20 years) and ‘other’ (17 years; p-values of <0.001, 0.001 and 0.011, respectively), and the difference in age at specimen collection between CIN3 and SCC for HPV45 (1 year) was statistically significantly narrower compared to the difference in age at specimen collection for HPV16 (15 years), HPV31 (23 years), HPV33 (20 years) and ‘other’ (17 years; p-values of 0.005, <0.001, <0.001 and 0.001, respectively). The difference, though not statistically significant, in age at specimen collection between AIS and ADC for HPV18 (6 years) was lower than for HPV16 (13 years; p = 0.162; Fig. 3).

Figure 3.

Median age of women from 17 European countries infected with a single HPV type and diagnosed with CIN3, SCC, AIS or ADC.

Discussion

These are the first parallel studies to estimate HPV-type prevalence in more than 6,000 women diagnosed with either HG-CIN or ICC across 17 European countries, using standardised and validated methods for specimen preparation, with centralised pathology review and HPV DNA testing. The pooled prevalence of individual HPV types was estimated using meta-analytic methods to control for heterogeneity observed among the countries. A comparison of age at diagnosis for different HPV types was undertaken to investigate type-specific age distributions of HPV-associated HG-CIN and ICC.

Consistent with other studies, we found that HPV16 was the most frequent type detected across Europe in both HG-CIN and ICC. Overall, HPV16/18/45/39/59/68 were more frequently detected in women diagnosed with ICC than HG-CIN, whereas HPV31/33/35/51/52/58/66 were more frequently detected in women diagnosed with HG-CIN than ICC. This is consistent with the results of other studies of type-specific HPV prevalence across the spectrum of HPV-related cervical diagnoses.7, 8, 13, 14

In each country studied, HPV16 was the most prevalent type associated with both HG-CIN and ICC; however, its absolute prevalence in both HG-CIN and ICC varied across the different countries. Such variations may not only reflect differences in the prevalence of these types in the general populations of different countries24 but are also likely to result from differences in screening practice and coverage.

Our studies also aimed to determine the ICC/HG-CIN prevalence ratios of specific HPV types in Europe. In women infected with a single HPV type, the ICC/HG-CIN prevalence ratios of HPV39/18/45 were greater than 1. These types all belong to the HPV α-7 clade. This is consistent with other studies showing the relative infrequency of these viruses (especially HPV18) in HG-CIN compared to ICC, despite their important role in the genesis of ICC. This is especially the case for ADC, which is more likely to be missed by screening, and for which the precursor of AIS is infrequent. HPV16 was equally frequent in both ICC and HG-CIN, which reflects its overall importance in both HG-CIN and ICC. Overall, the ICC/HG-CIN prevalence ratios of HPV31/33/35/52 were below or equal to 1, reflecting their importance in HG-CIN and a lesser role in ICC.

In addition to the overall prevalence of specific HPV types in ICC versus HG-CIN, the age-specific prevalence in ICC and HG-CIN reflects another distinguishing aspect of HPV16/18/45-related cervical lesions compared to lesions infected with other HPV types. ICC associated with HPV16/18/45 tended to be diagnosed at an earlier age than ICC related to ‘other’ HPV types. This, together with the above findings, reinforces the hypothesis that HPV16/18/45 may have a greater potential for rapid neoplastic transformation than other types as found in other studies.7, 15, 25

The results confirm the later presentation of HPV18- and HPV45-related CIN3 and the narrow age range between the detection of HPV18- and HPV45-related CIN3 and HPV18- and HPV45-related ICC.15, 26 It was also notable that the study identified no cases of HPV45-related AIS despite identifying several cases of HPV45-related ADC. These observations are consistent with data on HPV-type-specific DNA integration, suggesting that HPV18 and HPV45 may promote a higher degree of chromosomal instability than other types.25, 27

Our observations on the ICC/HG-CIN prevalence ratios and differences in median age at diagnosis for HG-CIN and ICC associated with HPV18/45/39, which are all of the same α-7 clade, also suggest that this group of HPV types may have a distinct natural history different from that of HPV16 or other HPV types. HPV16 predominates in both HG-CIN and ICC, especially in SCC. HPV types of the α-7 clade (18/39/45) are much more frequent in ICC than in HG-CIN, being relatively infrequent in the latter. In comparison, HPV31 and HPV33 have a large median age difference between HG-CIN and ICC and are more prevalent in HG-CIN than in ICC.28

Another finding of our study was that although the fraction of HPV16-related CIN3 was comparable to the fraction of HPV16-related ICC (and to that reported in the literature7, 13), the fraction of HPV18- and HPV45-related CIN3 was five times less than the fraction of HPV18- and HPV45-related ICC. This difference suggests that cytological screening and other European cervical cancer prevention practices may not be effective in detecting HG-CIN lesions related to HPV18 and HPV45. These and the above observations support the argument for prophylactic HPV vaccination and HPV-type-specific testing as an adjunct to the current cytological screening programmes.29

The higher prevalence of HPV18 and HPV45 in ADC compared to SCC is consistent with other recent data7, 28–30 and suggests that these HPV types may differ from HPV16 in their target cell specificity. Hence, HPV16 infection results in predominantly squamous cervical neoplasia, whereas HPV18 and HPV45 have a greater tendency to induce glandular cervical neoplasia.28, 29

HPV DNA was identified in 98.5% of HG-CIN samples, reflecting the high sensitivity of the PCR detection method used. HPV was found in 91.8% of ICC samples. Although this rate is higher than that measured in another similar and recent study,7 the lower rate of HPV positivity in ICC samples (particularly ADC samples) compared to HG-CIN may reflect poorer preservation of DNA in the larger ICC samples, a high frequency of integration of HPV into the host genome of ICC with loss of the L1 sequences amplified by the SPF10 primers, a lower HPV copy number in ICC relative to fewer episomal infections or varying sensitivities of the SPF10 probe for different HPV types, especially in the presence of multiple infections.31, 32 A small fraction of ICC without the presence of HPV may be truly HPV-negative.33 Some types of ADC are considered to be HPV-negative cancers.34 Indeed, in our study, ADC types possibly not associated with HPV composed 1.9% of the ICC cases.

The strengths of our studies include the large number of subjects and the diversity of selected European countries with and without screening programmes to which a standardised study protocol was applied. The selection criteria and the recent time period (2001–2008) give an up-to-date picture of the most common HPV types associated with cervical neoplasia in Europe. The standardised specimen collection, central histopathological review, use of a well-validated highly sensitive method for the detection of HPV in formalin-fixed paraffin-embedded tissue and further investigation of HPV-negative cases35 reduced the potential for pathological misclassification and increased both sensitivity and specificity of HPV detection. SPF10-LiPA25 recognises 25 HPV types and has the potential to amplify at least 54 individual HPV types.36 It has proven clinical significance and demonstrated high performance compared to other tests in terms of sensitivity, reproducibility and coverage of HPV types.17, 32, 36–38 Pooled data analysis allowed the inclusion of a large sample size and a wide geographical coverage across Europe while controlling for possible discrepancies and biases between the countries and study populations (e.g., differing geographical HPV-type distribution and variations in cervical cancer screening policies and methods). For comparison of ICC versus HG-CIN, restriction to the eight countries common to both studies also increased the robustness of the analysis.

The study design did not include a normal control population and was cross-sectional so that any consideration of actual progression from HG-CIN to ICC for individual HPV types remains hypothetical. However, our findings provide important information regarding differences in the ICC/HG-CIN prevalence ratios of individual HPV types and differences in median age of diagnosis of lesions associated with individual HR-HPV types. Estimates of the prevalence of infrequent HPV types and results on rarer histological diagnoses (e.g., AIS) are limited by number of cases and resulting wide CIs. A higher proportion (22%) of HG-CIN specimens were eliminated compared to the proportion (12.8%) of ICC specimens. This resulted from consistent, expert central pathology review being able to exclude cases of CIN1 that had been erroneously submitted. We do not anticipate that this would introduce systematic bias in HPV type results.

In conclusion, these robust and high-quality data indicate that HPV16 predominates both in HG-CIN and ICC across Europe. HPV types of the α-7 clade (18/39/45) are much more prevalent in ICC than in HG-CIN, being relatively infrequent in the latter. Furthermore, women with HPV16/18/45-associated ICC were diagnosed at an earlier median age than women diagnosed with ICC associated with other HR-HPV types, whereas HPV18- and HPV45-associated HG-CIN were detected at a later median age. These variations in HPV prevalence observed with age and type of lesion suggest that HPV18- and HPV45-related CIN3 may progress more rapidly to SCC than do other HR-HPV types related to CIN3, suggesting important differences between individual types with respect to their processes of infection and rates of transformation. Finally, our data on HPV-type distribution in both HG-CIN and ICC have implications for both HPV screening and prophylactic vaccination programme considerations in Europe.

Contribution to Authorship

All authors served as principal investigators and/or were involved in the original idea and design of the study, wrote the protocol and report or analysed and interpreted data. All authors reviewed and approved the article. W.A.T., A.F., O.R., N.P., A.M.N., B.K., R.K., J.O'L., E.A.J., M.R., B.C., S.C. De S., D.J. and K.H. analysed and/or interpreted the data and comprised the writing committee; D.J. and B.C. contributed to trial design; W.A.T., A.F., O.R., N.P., A.M.N., B.K., R.K., J.O'L., E.A.J., D.S., K.K., V.D., M.Re., R.V., L.K., A.S., E.S. and J.O. recruited subjects and conducted the research. A.R. and D.R. contributed to data analysis and interpretation. A.M. and W.Q. analysed and interpreted laboratory results.

Acknowledgements

The authors thank all study participants and their families. They gratefully acknowledge the work of the central and local study co-ordinators and the staff members of the sites who participated in this study. Contribution to statistical support was provided by Aurélie Le Plain and Marie-Cecile Bozonnat (4Clinics, Paris, France, Contract Research Organization on behalf of GlaxoSmithKline Biologicals). Writing support services were provided by Catherine Streeton (Streeton Associates, Melbourne, Australia); editing and publication co-ordinating services were provided by Veronique Delpire and Mandy Payne (Words and Science, Brussels, Belgium).

Appendix

HERACLES/SCALE Study Group Collaborators

Austria: R. Horvat (Department of Pathology, Medical University of Vienna, Austria); L. Six (Department of Gynaecology and Obstetrics, Medical University of Vienna, Austria) and M. Spacek (GlaxoSmithKline, Vienna, Austria). Belgium: F. Krikelda (University Hospital of Liège, Liège, Belgium); M. Praet (Ghent University Hospital, Ghent, Belgium); P. Simon (Erasme Hospital ULB, Brussels, Belgium) and B. De Vos (GlaxoSmithKline, Genval, Belgium). Czech Republic: R. Kodet and L. Rob (University Hospital Motol, Prague, Czech Republic); P. Hrdlickova-Stachova, R. Nenutil and P. Rotterova (Masaryk Memorial Cancer Institute, Brno, Czech Republic); L. Sevcik and J. Dvorackova (University Hospital Ostrava, Ostrava-Poruba, Czech Republic); M. Pluta and H. Robova (University Hospital Motol, Prague 5, Czech Republic) and E. Kaliskova (GlaxoSmithKline, Prague, Czech Republic). Denmark: J. Junge (Hvidovre University Hospital, Copenhagen, Denmark) and M. Herslov (GlaxoSmithKline, Broendby, Denmark). Estonia: I. Vaasna (Tartu University Hospital, Tartu, Estonia) and S. Oro (GlaxoSmithKline, Tallinn, Estonia). Germany: K. Diedrich (Universitätsklinikum Schleswig-Holstein, Klinikum für Frauenheilkunde und Geburtshilfe, Lübeck, Germany); F. Gieseking and L. Woelber (Universitätsklinikum, Hamburg, Germany); B. Schmalfeldt (Frauenklinik des Klinikums Rechts der Isar der Techn. Universität München, München, Germany); M.W. Beckmann and H. Mehlhorn (Friedrich-Alexander Universität, Erlangen, Germany); C. Hanusch (Frauenklinik, Rotkreuzklinikum München, Germany); W.-D. Höpker (Praxis Nordhausen, Germany); T. Kirchner, F. Bergauer, C. Dannecker and D. Mayr (Pathologisches Institut der LMU München, München, Germany); H. Kölbl, B. Euteneuer and K. Nilges (Klinik und Poliklinik für Geburtshilfe, Hamburg, Germany); K.U. Petry and C. Liebrich (Klinikum der Stadt Wolfsburg Klinik für Frauenheilkunde, Geburtshilfe und Gynäkologische Onkologie, Wolfsburg, Germany); H. Nenning (Institut für Pathologie am Elsapark, Leipzig, Germany) and C. Eisenhofer-Wolff (GlaxoSmithKline, Munich, Germany). Greece: N. Apostolikas (Department of Histopathology, General Hospital of Patras, Patras, Greece); S. Markaki (General Hospital of Athens ‘Alexandra’, Athens, Greece); T. Zaraboukas (Medical Cross-Balkanic Centre of Thessaloniki, Pylaia, Greece); A. Pantzaki (Ippokratio General Hospital, Thessaloniki, Greece); I. Papaspirou (General Hospital of Athens ‘Alexandra’, Athens, Greece); E. Zairi (Medical Cross-Balkanic Centre of Thessaloniki, Pylaia, Greece) and E. Pournara (GlaxoSmithKline, Athens, Greece). Hungary: A. Mosonyi (Hetényi Géza Hospital, Szolnok, Hungary); T. Barna (Hospital Kútvölgyi, Budapest) and M. Pal (GlaxoSmithKline, Budapest, Hungary). Ireland: C. Martin and O. Sheils (CERVIVA research consortium, funded by the Health Research Board Ireland, based at the Department of Pathology, The Coombe Women and Infants University Hospital and Trinity College, Dublin, Ireland) and G. Power (GlaxoSmithKline, Rathfarnham, Ireland). Norway: B. Hagen (St. Olav's Hospital, Trondheim, Norway); J. Lømo (Oslo University Hospital, Oslo, Norway); O.E. Iversen, B. Bertelsen, H. Haugland and O.K. Vintermyr (Haukeland University Hospital, Bergen, Norway) and S. Thoresen (GlaxoSmithKline, Oslo, Norway). Poland: A. Basta, R. Jach and R. Tomaszewska (Jagiellonian University, Krakow, Poland); J. Kotarski and B. Barczynski (Medical University of Lublin, Poland); R. Lyszcz and A. Maj (Jan BozySPSzW Hospital, Lublin, Poland); B. Pawlowska-Wakowicz (deceased) (SPSK1 Hospital, Poland) and A. Ksiazek (GlaxoSmithKline, Warsaw, Poland). Portugal: I. Pinto (Instituto Português de Oncologia Francisco Gentil—EPE, Oporto, Portugal); T. Simões (Hospitais da Universidade de Coimbra, Coimbra, Portugal); A.I. Belo (Maternidade Dr. Alfredo da Costa, Lisbon, Portuga), M.J. Brito (Hospital Garcia de Orta, Almada, Portugal); L.D.P. Gonçalves (Hospital Amadora Sintra, Lisbon, Portugal) and M. Carichas (GlaxoSmithKline, Lisbon, Portugal). Romania: F. Nicula (Oncology Institute ‘Prof. Dr. Ion Chiricuţă’, Cluj, Romania); M. Onofriescu (Obstetrics and Gynaecology Clinic Hospital ‘Cuza Vodă’, Iaşi, Romania); R. Mociulschi (deceased) (Alfred Rusescu Institute for Mother and Childcare, Bucharest, Romania) and D. Constantinescu (GlaxoSmithKline, Bucharest, Romania). Russia: G. Minkina (Moscow University of Medicine and Dentistry, Moscow, Russia) and M. Scherbakov (GlaxoSmithKline, Moscow, Russia). Spain: M. Castro Sánchez and L. Cortés Lambea (Hospital de Móstoles, Madrid, Spain); A. Torné Blade (Hospital Clinic-IDIBAPS, University of Barcelona, Barcelona, Spain) and R. Cambronero Martínez (GlaxoSmithKline, Tres Cantos, Spain). United Kingdom: K. Cuschieri (Royal Infirmary of Edinburgh, Edinburgh, Scotland, UK) and S. Blaikie (GlaxoSmithKline, Uxbridge, UK). D. Luyts (GlaxoSmithKline Biologicals, Rixensart, Belgium) and A. Crochard (employee of GlaxoSmithKline Biologicals, Rixensart, Belgium, at the time of trial design) contributed to study protocol and study conduct. E.C. Pirog (Cornell University, NY), G. McCluggage (Queens University, Belfast, UK) and M. Wells (Sheffield University, Sheffield, UK) were consulted as independent experts to review and validate the histopathological diagnoses of selected samples performed at DDL's central laboratory.

Ancillary