IJC International Journal of Cancer

Transcription

1 IJC International Journal of Cancer The complex relationship between human papillomavirus and cervical adenocarcinoma Anco Molijn 1, David Jenkins 1, Wen Chen 2, Xun Zhang 2, Edyta Pirog 3, Wu Enqi 4, Bin Liu 2, Johannes Schmidt 5, Jiangfeng Cui 2, Youlin Qiao 2, and Wim Quint 1 on behalf of Chinese HPV Typing Group 1 DDL Diagnostic Laboratory, Visseringlaan 25, 2288 ER, Rijswijk, The Netherlands 2 Cancer Hospital, Chinese Academy of Medical Sciences, Beijing, China 3 MINZU University of China, Haidian District, Beijing, China 4 Weill Medical College of Cornell University, 525 E 68th Street, New York, NY 5 GlaxoSmithKline Biologicals, Avenue Fleming 20, Wavre, Belgium Human Papillomavirus (HPV) is reported in % of cervical adenocarcinoma (CADC) globally. We investigated this relationship in a hospital-based survey in China. 718 CADC samples from nine Chinese regions were analysed. Expert pathologists reviewed cases with p16 and progesterone receptor immunostaining. Cases were tested for HPV using whole-tissue sections (WTS) and laser-capture microdissection. All cases were HPV-tested by L1 based broad-spectrum SPF 10 -DEIA-LiPA 25 PCR. Negative cases were tested for DNA adequacy and with E6 oncogene, type-specific HPV PCRs. Using WTS-PCR CADC showed overall 75% HPV-positivity (33 100% for different histological types). LCM-PCR showed that none of minimal deviation or serous CADC, and <10% of all clear cell and endometrioid CADC were HPV-positive in tumour cells. Usual and adenosquamous CADC showed a single HPV genotype in 60 and 78% cases. In some cases, HPV was found in adjacent cervix but not in tumour. HPV 16, 18 and 45 accounted for 90% of HPV in tumour cells. Patients with HPV-positive tumours were on average 6 years younger and presented at a lower clinicopathological stage as compared to patients with HPV-negative cancers. CADC is diverse pathologically and in HPV status. Special histopathological tumor subtypes may develop through different cellular and molecular pathways. Between 20 and 40% usual and adenosquamous types, in particular these diagnosed in older women and at advanced FIGO stages, are not driven by oncogenic HPV. In these cases HPV may not be involved in carcinogenisis or maybe lost during tumour progression. Many studies have addressed the role of Human Papillomavirus (HPV) in cervical cancer, leading to the view that HPV is a necessary cause of invasive cervical carcinoma (ICC). 1 3 Large population studies of cervical adenocarcinoma (CADC) have consistently found HPV in <80% of almost 2,000 cases in whole-tissue sections (WTS) with large variation between regions. 4 8 CADC accounts for 10 20% of ICC 4,9,10 and the incidence of adenocarcinoma relative to squamous carcinoma of the cervix, has increased in some countries, especially in younger women The projected worldwide incidence of CADC in 2015 is estimated at 56, China contributes about one-quarter of the global burden of ICC with about 50% of CADC occurring in South-East Asia where there is limited cytological screening. 9,10 This Key words: cervical cancer, cervical adenocarcinoma, HPV, human papilloma virus, SPF10, epidemiology, P16, china, laser-capture microdissection, vaccination Author Contribution Statement: All authors approved the final draft of the report. A.M., W.C., D.J., J.S., W.Q. and Y.Q. participated in the study conception and design, data collection and analysis, interpretation of the results, and writing the report. X.Z., W.E. and E.P. participated in data collection, interpretation of the results, and writing the report. B.L. and J.C. participated in data collection, analysis and interpretation of the results Chinese HPV typing group was involved in data collection, interpretation of the results and final approval of the report. Conflicts of interests: W.Q. and A.M. have received support for travel to meetings for the study from GlaxoSmithKline. D.J. provided consultancy and is a former employee of GlaxoSmithKline Biologicals. J.S. is an employee of GlaxoSmithKline Biologicals. Y.Q. has received support for travel to meetings for the study from GlaxoSmithKline. E.P. provided consultancy to DDL Diagnostic Laboratory. All investigators at study clinical sites were funded through their institutions to undertake the study protocol. Other authors did not report any conflict of interest A.M. and W.C. contributed equally to this work. Grant sponsor: GlaxoSmithKline Biologicals SA DOI: /ijc History: Received 7 Apr 2015; Accepted 17 June 2015; Online 4 Aug 2015 Correspondence to: A.C. Molijn, DDL Diagnostic Laboratory, Rijswijk, The Netherlands, Tel.: , anco.molijn@ddl.nl Int. J. Cancer: 138, (2016) VC 2015 UICC

2 410 Cross-sectional study to explore the relation of HPV to adenocarcinoma What s new? Cervical adenocarcinoma is generally considered a virally caused disease but human papilloma viruses (HPV) are not always found in tumor tissue. The authors found 75% of cervical adenocarcinoma HPV-positive in Chinese women. HPV negativity correlated with late stage adenocarcinoma in older women, and many were negative for the tumor suppressor p16, consistent with tumors not driven by HPV. This implicates that HPV might not be involved in the carcinogenesis of these tumors or is lost during tumor progression. study is a multicentre survey of 718 routinely diagnosed, confirmed, recent cases of CADC from nine regions in China. To address important methodological issues there was central expert pathology review with immunohistochemical (IHC) markers, including progesterone receptor (PR) to minimise misclassification of endometrial cancer. We used a very sensitive broad-spectrum HPV PCR for HPV L1 gene, and typespecific HPV PCR targeting the E6 oncogene of high-risk genotypes to identify HPV missing the L1 gene due to integration. Laser-capture microdissection-pcr (LCM-PCR) determined whether HPV DNA was located in tumour cells or elsewhere in cervical mucosa and evaluated multiple HPV infections. 1,4,5,11,15 19 HPV presence in tumour cells was then related to histological tumour type, age at presentation, tumour stage and presence of p16 INK4a or PR positivity. Material and Methods Case identification and specimen collection Institutional review board approval was obtained from the Ethics Committee, Cancer Institute, Chinese Academy of Medical Sciences, Beijing, China (CICAMS) and participating centers. Pathology records were searched from 2010 to 2005 for routine diagnoses of invasive CADC at nine regional Figure 1. Overview of study design. medical centers across China. Age at presentation and FIGO stage were recorded and the cases were anonymized. We identified 1,051 cases diagnosed and managed as CADC in proportion to regional populations. 170 cases were not available for release by the sites (Fig. 1). Slides and a representative block of formalin-fixed CADC were obtained from 881 cases and submitted to CICAMS for central review and further studies. Pathological Review, Sectioning and Immunohistochemistry Of 881 cases, 163 cases were excluded, 154 on central review and nine with inadequate DNA (Fig. 1). Formalin-fixed, paraffin blocks from 718 cases were confirmed as CADC and sectioned according to a sandwich procedure. 15 Immunostaining was performed on 4 lm sections using heat-induced epitope retrieval with citrate buffer and a primary mouse monoclonal antibody anti-p16 INK4a clone JC8 (Lab Vision Corporation, Neomarkers, Fremont CA) for p16ink4a and a rabbit monoclonal antibody clone 1E2 (Roche Corporation, Ventana Medical Systems, Arizona) for progesterone receptor (PR) followed by visualization using Dako EnVisionTM FLEX1 kit (Dako, Carpinteria, CA). Immunostaining results for p16 INK4a (nuclear and cytoplasmic) and PR (nuclear) were graded on a scale of 0 to 3 according to percentage of positive tumour cells: 0-no staining; 1-focal (1 25%); 2- patchy (26 75%); 3-diffuse (>75%). Consensus Histologic Review and Tumour Categorization An expert gynaecological pathologist at CICAMS (XZ) examined Hematoxylin and Eosin stain (H&E) and immunostained sections and made a diagnosis. The H&E section and IHC slides graded above 0 were scanned for digital microscopy (Aperio Technologies, Vista, CA) and sent to DDL Laboratories, Rijswijk, Netherlands. A second expert (ECP) made an independent diagnosis on these images. P16 INK4a and PR scores by each pathologist were averaged. A final average score of 1.5 or higher was considered positive. All cases were classified according to a simplified World Health Organisation (WHO) classification (Table 1). 5,17,20,21 Final diagnosis was established if there was complete agreement between the two expert reviews or agreement on differential diagnosis. For cases with inconsistency between the experts, the results of the combined IHC score were

3 Molijn et al. 411 Table 1. Pathology classification categories for cervical adenocarcinoma 20 Histology diagnosis code ADC-CX ASC ADC-NOS ADC-MIN ADC-CC ADC-ENDO ADC-SER NEC Description Adenocarcinoma of endocervix of usual mucinous type: includes intestinal type and villoglandular type Adenosquamous carcinoma of endocervix containing at least 5% of adenocarcinoma and squamous components: includes adenoid basal carcinoma, glassy cell carcinoma and clear cell adenosquamous carcinoma Adenocarcinoma of the cervix not otherwise specified: all cases not classified by the review algorithm Minimal deviation adenocarcinoma: includes gastric-type adenocarcinoma Clear cell adenocarcinoma and mesonephric adenocarcinoma of cervix Endometrioid adenocarcinoma present at the endocervix Serous adenocarcinoma of cervix Adenocarcinoma with neuroendocrine differentiation considered. For diagnosis of ADC-CX a p16 INK4a score of 2 or more with PR score less than 2 was required, and for diagnosis of ADC-ENDO the opposite scores were required. For ADC-MIN, a negative p16 INK4a and PR were considered as confirming the diagnosis. For ADC-SER, a negative PR result and variable p16 INK4a, and for ADC-CC, a negative PR result and a p16 INK4a score of at least 0.5 was considered diagnostic. DNA Isolation and HPV DNA Detection in WTS To achieve maximum sensitivity of HPV detection the testing combined L1-based broad-spectrum HPV SPF 10 version 1 assay, type-specific PCR for HPV 16 and 18 and a novel E6- based multiplex type-specific system (Fig. 1). Total DNA was isolated from paraffin embedded material by a proteinase K extraction procedure as described previously. 22 HPV DNA was first detected by PCR amplification and HPV genotyping using the HPV SPF 10 -DNA Enzyme Immunoassay (DEIA)- Line Probe Assay (LIPA) 25 system (version1; Labo Bio- Medical Products, Rijswijk, The Netherlands) as described previously. 23,24 The SPF 10 PCR primer set amplifies a 65 bp fragment from the L1 region of at least 54 HPV genotypes. Amplification products were detected using the HPV SPF 10 - DEIA and genotyped using reverse hybridization assay LiPA 25 containing probes for 25 HPV genotypes (HPV 6,11,16,18,31,33,34,35,39,40,42,43,44,45,51,52,53,54,56,58,59, 66,68/73,70,74). Cases negative for HPV by SPF 10 -DEIA system were diluted 1/10 and HPV PCR was repeated to confirm absence of PCR inhibitors. Cases that remained HPVnegative and all HPV-positive cases negative for HPV 16 and/or 18 were tested with type-specific PCR-DEIA assay, 25 to selectively amplify a 92 bp HPV 16 fragment (TS16) in the E6/7 region and 126 bp HPV 18 fragment (TS18) in the L1 region. Cases that were still negative were tested for the presence of HPV E6 genes using MPTS12 PCR reverse hybridization assay 26 that detects nine HPV genotypes (HPV 16,18,31,33,35,45,52,58,59). DNA Adequacy Testing All HPV-negative samples (n 5 183) and 101/535 (18.9%) positive specimens were tested using human RNaseP/PhHV qpcr. 27,28 A minimum inclusion threshold for adequate human DNA was established at 100 detectable copies of human DNA for WTS-PCR and two copies for LCM-PCR. These thresholds were set to allow detection of HPV DNA at the level of a single copy in tumour cells by LCM-PCR and in mixed samples by WTS-PCR. Laser-Capture Microdissection Laser-capture microdissection (LCM) of 170 selected cancer cases (Fig. 1) was performed as described previously to determine the presence of specific HPV genotypes in tumour cells. 15 Examples of tumour regions and adjacent mucosa selected for LCM are shown in Figures 2 and 3. In total >500 representative areas were marked using digital microscopy (Aperio Technologies Inc, Vista, CA) by an expert pathologist (DJ). The selected regions were excised with the Leica Laser Microdissection system from membrane-coated slides. HPV detection and DNA adequacy testing was performed as described above. Statistical Methods Population means and 95% confidence intervals were calculated using Vasserstats. Cases were compared using the Student s t test for means using SPSS 210 (IBM). The Cochran- Armitage trend test was used to assess the relationship between FIGO-stage and HPV-positivity and p16 expression and HPV-positivity, and v 2 test and Fisher s exact test were used to compare the HPV rates by histological type using SAS 92 (SAS Institute, Cary, NC). Probability levels <0.05 were considered as statistically significant. Results HPV detection in WTS in different histologic categories of cervical adenocarcinoma Of the collected 718 cases of CADC, ADC-CX was the most common histologic category (59%), followed by ASC (15%)

4 412 Cross-sectional study to explore the relation of HPV to adenocarcinoma Figure 2. Examples of HPV analysis of adenocarcinoma cases analyzed by lasercapture microdissection followed by HPV PCR. (a): Classical cervical adenocarcinoma. Multiple infection of HPV18 and 52 detected in whole tissue section. LCM-PCR detected HPV 18 in ADC-CX and HPV 52 in CIN lession. (b) Minimal deviation adenocarcinoma: HPV16 detected in whole tissue section. LCM-PCR detected HPV 16 squamous epithelium and HPV negative tumour. (c) Clear cell adenocarcinoma: HPV16 detected in whole tissue section. LCM-PCR detected HPV 16 in CIN lesion and HPV negative tumour. while other categories accounted individually for a small proportion of cases (Table 2). With WTS-PCR using L1 based SPF 10 -PCR there were 482 cases (67%) positive for HPV DNA. Additional testing using type-specific E6 PCR added 53 HPV-positive cases (7%). The highest HPV-positivity rate was in NEC (100%), followed by ASC (88%) and ADC-CX (82%). Other categories showed HPV-positivity rates of 33 51%. Of all HPV-positive cases, 89% showed a single HPV type and 11% multiple HPV types. HPV Genotype Distribution in WTS HPV 16,18,52,45 were the most common types detected as single infections in WTS of different histological categories (Table 3). HPV 16 and HPV 18 were found with similar frequencies and accounted for 89% of single infections in ADC-CX, ASC and NEC. For other pathological categories HPV 16 was much more common (57%) than HPV 18 (13%) with v and p values < Tumour Cell HPV-Positivity in LCM Samples LCM-PCR was performed on ADC-CX with multiple HPV infections (n 5 38, three cases excluded: one case with <2 copies human DNA, two cases with inadequate sampling), ADC- CX with single HPV infection (n 5 60, two cases excluded: one case with <2 copies human DNA and one case HPV untypable on WTS), ASC with multiple HPV infections (n 5 9) and as well as on cases from the low HPV prevalence categories (ADC-MIN, -CC, -ENDO, -SER and -NOS) that

5 Molijn et al. 413 Figure 3. Mean age of patients in different CADC categories and by different HPV status with 95% CI s. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.] were found to be HPV-positive using WTS (n 5 69). The rate of tumour cell HPV-positivity in LCM samples was lower than in WTS and varied by histological category (Table 4). This finding could not be ascribed to inadequate DNA in LCM samples, as the total amount of human DNA was similar for 140 HPV-positive LCM tumour samples [92 RNAseP copies/pcr (95% CI: )] compared to 221 HPV-negative LCM-PCR tumour samples [87 RNAseP copies/pcr (95% CI: )]. No HPV DNA was identified in any of ADC-MIN and ADC-SER, and HPV was found in only 13% of ADC- ENDO and ADC-CC, respectively. HPV Genotyping in LCM Samples Of the total 102 cases of ADC-CX and ASC analysed by LCM and comprising cases with single and multiple HPVs detected by WTS, 63 were positive for HPV by LCM-PCR. In all 63 cases only a single HPV type was detected in tumour cells. No multiple infections were identified. In addition, a limited range of HPV types was detected compared to WTS-PCR. HPV 16 (N 5 26, 41%), 18 (N 5 24, 38%) and 45 (N 5 7, 11%) accounted for 90% of HPV types detected with only single cases of HPV 59, 31 and 58 (N 5 3, 5%) and three cases with untypable HPV. Table 2. WTS HPV-positivity in different histological tumour subtypes Histopathological diagnosis (% of total cases) HPV-positive cases (% in category) Single HPV types (% HPV-positive) Multiple HPV types (% HPV-positive) (N 5 718) (N 5 535) (N 5 475) (N 5 60) ADC-CX 426 (59%) 350 (82%) 312 (89%) 38 (11%) ASC 110 (15%) 97 (88%) 88 (91%) 9 (9%) ADC-NOS 55 (8%) 22 (40%) 19 (86%) 3 (14%) ADC-MIN 43 (6%) 22 (51%) 19 (86%) 3 (14%) ADC-CC 30 (4%) 15 (50%) 10 (67%) 5 (33%) ADC-ENDO 24 (3%) 8 (33%) 7 (88%) 1 (13%) ADC-SER 18 (3%) 9 (50%) 8 (89%) 1 (11%) NEC 12 (2%) 12 (100%) 12 (100%) 0 (0%) TOTAL 718 (100%) 535 (75%) 475 (89%) 60 (11%)

6 414 Cross-sectional study to explore the relation of HPV to adenocarcinoma Table 3. Human papillomavirus genotyping in single infected cases grouped by high and low prevalence histological categories HPV 16 HPV 18 HPV 52 HPV 45 Other HPV types 1 N N Rate N Rate N Rate N Rate N Rate ADC-CX, ASC, NEC % % 13 3% 5 1% 29 7% ADC-NOS, -MIN, -ENDO, -SER % 8 13% 2 3% 4 6% 13 21% All ADC % % 15 3% 9 2% 42 42% 1 Other HPV types is a summary of HPV type 6, 11, 31, 33, 34, 35, 39, 40, 42, 44, 51, 53, 54, 56, 58, 59, 66, 68 or 73 and six HPV-positive but untypable cases. A few exploratory cases which were found to be HPVpositive in WTS-PCR but HPV-negative in the tumour cells by LCM-PCR were sampled more extensively. In some instances it was possible to locate HPV in the adjacent cervical mucosa. Figure 2a shows a case in which HPV 52 and HPV 18 were detected by WTS-PCR. LCM-PCR showed that HPV 52 was present in CIN2 and CIN3, and HPV 18 was found in ADC-CX tumor cells. For some cases it was possible to locate HPV in other cervical sites although the tumour cells were HPV-negative by LCM-PCR. Figures 2b and 2c show two cases (ADC-MIN and ADC-CC) that were positive for HPV on WTS-PCR. Both were negative by LCM-PCR in tumour cells for HPV DNA but HPV DNA was located in normal squamous cervical epithelium and an area of CIN2 respectively. Mean Age in HPV-Positive and HPV-Negative ADC-CX The mean age of patients for all 718 cases (Fig. 3) was 46.0 years (95% CI: ). The mean age of patients with tumours with high HPV prevalence (ADC-CX, ASC and NEC) was 44.8 years (95% CI: ), and these patients were significantly younger (p < ) than patients with tumours with low HPV prevalence (ADC-MIN, -CC, - ENDO, -SER and -NOS) whose average was 49.8 years (95% CI: ). The mean ages at diagnosis of women with ADC-CX by HPV status as determined by WTS-PCR were further stratified. For HPV-positive cases (N 5 350, 82%) the mean age Table 4. HPV-positivity in tumour cell samples by LCM-PCR Histological Category Cases positive for HPV by WTS PCR further analysed by LCM-PCR (N) LCM-PCR HPVpositive cases (N) (%) ADC-CX % ASC % ADC-NOS % ADC-MIN % ADC-CC % ADC-ENDO % ADC-SER 9 0 0% was 44.0 years (95% CI: ) and for HPV-negative cases (N 5 76, 18%) 50.5 years (95% CI: ). The trend between age and HPV status was consistent in 93 ADC-CX cases in which tumour cell HPV status was determined by LCM-PCR. For HPV-positive cases (N 5 56, 60%) the mean age was 43.2 years (95% CI: ) and for HPV-negative cases (N 5 38, 38%) 48.5 years (95% CI: ). FIGO Stage in WTS-PCR And LCM-PCR HPV-Positive and HPV-Negative ADC-CX The relationship between FIGO stage, age at diagnosis and HPV-positivity by WTS-PCR and LCM-PCR is shown in Table 5. There is a statistically significant trend of decreasing HPV-positivity by WTS-PCR (Z , p ) and LCM-PCR (Z , p ) in ADC-CX with increasing tumor stage. The FIGO stage was also related to age and more advanced tumours occurred at an older mean age. To exclude over-representation of endometrial tumours at higher FIGO stage we showed 97% of ADC-CX were PR negative and calculated the frequency of PR positivity in relationship to FIGO stage. For FIGO stage 1 (N 5 216), four cases were PR positive whereas for FIGO Stage 2 4 (N 5 125), one case was PR positive. p16ink4a expression rates in HPV-positive and negative ADC-CX HR-HPV was present in 93% of p16 positive cases (n 5 350: p < 0.001) based on WTS analysis and in 89% of the studied LCM cases (N 5 56: p < 0.001) compared to only 54% in both HPV-negative WTS cases (N 5 76) and HPV-negative LCM cases (N 5 37) (Table 6). Discussion It is a widely held belief that every cervical cancer is driven by HPV DNA. HPV is routinely detected in as much as 90% of squamous cell carcinomas using PCR with consensus primers. Re-testing of HPV-negative tumours with PCR using type specific primers and excluding the cases with inadequate DNA yielded HPV positivity of 99.7%. 1 The detection of HPV in adenocarcinomas was always lower than in squamous cell carcinomas: recent CADC studies 4 6,32 using standardized methodologies showed HPV positivities of 62, 76 and 90%, respectively. In situ hybridization has been previously

7 Molijn et al. 415 Table 5. WTS-PCR and LCM-PCR HPV-positivity by FIGO stage and by age in ADC-CX WTS-PCR FIGO stage (N 5 340) Stage 1 (N 5 216) Stage 2 (N 5 90) Stage 3/4 (N 5 34) Mean age (95% CI) 42.8 (95% CI: ) 47.1 (95% CI: ) 51.9 (95% CI: ) HPV-positive (n 5 284) (% by grade) 191 (88.4%) 71 (78.9%) 22 (62.8%) LCM-PCR FIGO stage (N 5 76) Stage 1 (N 5 43) Stage 2 (N 5 23) Stage 3/4 (N 5 10) Mean age (95% CI) 40.5 (95% CI: ) 50.2 (95% CI: ) 48.8 (95% CI: ) HPV-positive (n 5 46) (% by grade) 30 (71.4%) 13 (56.5%) 2 (20.0%) used to study the complexitiy of association of HPV with adenocarcinoma. These studies were limited by the variable sensitivity of the technology but already indicated a limited involvement of HPV in some cervical adenocarcinomas. 33 The lower prevalence of HPV in CADC as compared to squamous cell carcinoma has been interpreted as an artefact of the technical methods for HPV detection in formalin-fixed tissue. Our study suggests that there may be other explanations for these results. Our study was the largest study designed specifically to address issues related to adenocarcinomas. We wanted to provide high assay sensitivity and ascertain accurate localization of HPV within tumor cells. To that end we performed exhaustive HPV testing beginning with highly sensitive SPF10 PCR with consensus primers targeting the L1 gene followed by type specific PCR targeting the E6 gene. After detecting HPV in whole tissue section we proceeded to lasercapture miscrodissection of tumors followed by HPV PCR. Another issue of HPV testing in adenocarcinomas is presence of histologic tumor subtypes, some of which were previously analyzed in small studies and reported as HPV negative. To ascertain accurate classification in our study, central expert histopathologic review of cases was conducted aided with p16 INK4a and progesterone receptor immunostains, however pathological classification of these complex tumours is difficult and subject to interobserver variation. Usual-type adenocarcinoma, adenosquamous and neuroendocrine carcinomas (76% of CADC) showed a clear relationship to HPV infection with restricted genotypes, mainly HPV 16, 18 and 45 as shown by positivity in whole tissue Table 6. Expression of p16 in HPV-positive and negative ADC-CX by WTS-PCR and LCM-PCR WTS-PCR HPV-positive HPV-negative Total no. cases p16 positive 326 (93%) 41 (54%) p16 negative 24 (7%) 35 (46%) LCM-PCR HPV-positive HPV-negative Total no. cases p16 positive 50 (89%) 20 (54%) p16 negative 6 (11%) 17 (46%) sections and in tumor cells. Other CADC categories showed overall 45% HPV positivity in WTS but when LCM-PCR was performed there was no HPV in tumor cells of minimal deviation and serous adenocarcinoma as defined by our diagnostic algorithm, and infrequent HPV-positivity in clear cell, endometrioid and unspecified adenocarcinoma. In some cases, when HPV was not found in the tumor cells we could localize it to CIN or latent infection of normal cervical mucosa. The relation of HPV to clear-cell, minimal deviation and serous CADC has been previously questioned 5 and other likely genetic mechanisms were proposed. The proportion of these tumors as defined is similar in China (16%) to that reported in recent European 6 and global studies. 4,5 The 426 cases of usual-type adenocarcinoma allowed the exploration of clinicopathological and biomarker differences between women with HPV-positive and negative tumors. HPVpositive cases occurred at a younger average age than HPVnegative tumors, had an earlier clinical stage and higher p16 positivity. HPV types detected by LCM-PCR were very restricted, with HPV 16, 18 and 45 accounting for 90% of cases. Only single HPV types were detected in tumor cells by LCM even in cases in which multiple types were found in WTS. The finding confirms one virus one lesion theory of HPV carcinogenesis. These findings indicate that only a few HPV types are important in the pathogenesis of CADC and that although most early stage usual CADC are positive for HPV in tumour cells,hpvmaybelostinlatestagesofthistumor. The finding of true HPV-negative tumors categorized as usual histological subtype, particularly in late stages of spread, suggests that mechanisms other than continuous HPV drive are involved in the pathogenesis of these tumors, unlike in squamous cervical cancer. The high frequency of HPV infection in cervical adenocarcinoma in situ 16 is consistent with HPV role as an initiating factor. It has been reported that epigenetic changes such as methylation, chromosome abnormalities and mutations in TP53 and other genes occur in cervical cancers and differ between squamous cell and adenocarcinomas of the cervix Based on results of our study we propose a novel hypothesis for cervical carcinogenesis; our findings suggest a subset of CADC develop by epigenetic and/or somatic genetic changes in the absence of HPV or loss of drive from HPV oncogenes as CADC evolves. This hypothesis is supported by the finding of lower p16 INK4a

8 416 Cross-sectional study to explore the relation of HPV to adenocarcinoma positivity in HPV-negative usual CADC indicating the pathway of E7 expression/prb sequestration/p16 upregulation is not active. 17,34 Our findings confirm the dominant role of HPV in relation to CADC but show how difficult it is to define precisely the burden of CADC due to HPV infection from crosssectional data and with respect to the limitations of histological diagnosis. Current vaccines aimed at HPV 16 and 18 with cross-protection against HPV 45 should provide good protection against most HPV-driven CADC, however, the current study suggests that 3 5% of overall ICC is not driven by continued action of HPV DNA. Acknowledgements We thank the staff members of the Chinese sites that participated in this study and Hans Berkhof for statistical support. GlaxoSmithKline Biologicals SA had the opportunity to review a preliminary version of this manuscript for factual accuracy, but the authors were solely responsible for final content and interpretation. Chinese HPV typing group: Yu XH and Li LY, Jiangxi Women and Children s Hospital, Jiangxi; Wang J and Zeng L, Hunan Cancer Hospital, Hunan; Liu JH and Zhang Y, Sun Yat-sen University Cancer Hospital; Wang CY and Liu D, Liaoning Cancer Hospital, Liaoning; Cheng JX and Shen GQ, Xinjiang People s Hospital, Xinjiang; Wang P and Yuan Y, Shaanxi Cancer Hospital, Shaanxi; Zhou Q and Tang Y, Chongqing Cancer Hospital, Chongqing; and Xu M and Liao GD, Chongqing Medical College First Affiliated Hospital, Chongqing. References 1. Walboomers JM, Jacobs MV, Manos MM, et al. Human papillomavirus is a necessary cause of invasive cervical cancer worldwide. J Pathol 1999; 189: Franco EL, Rohan TE, Villa LL. Epidemiologic evidence and human papillomavirus infection as a necessary cause of cervical cancer. J Natl Cancer Inst 1999;91: Bosch FX, Manos MM, Munoz N, et al. Prevalence of human papillomavirus in cervical cancer: a worldwide perspective. J Natl Cancer Inst 1995; 87: de Sanjose S, Quint WGV, Alemany L, et al. Human papillomavirus genotype attribution in invasive cervical cancer: a retrospective crosssectional worldwide study. Lancet Oncol 2010;11: Pirog EC, Lloveras B, Molijn A, et al. HPV prevalence and genotypes in different histological subtypes of cervical adenocarcinoma, a worldwide analysis of 760 cases. Mod Pathol 2014;12: Tjalma WA, Fiander A, Reich O, et al. Differences in human papillomavirus type distribution in high-grade cervical intraepithelial neoplasia and invasive cervical cancer in Europe. Int J Cancer 2013;132: Adegoke O, Kulasingam S, Virnig B. Cervical cancer trends in the United States: a 35-year population-based analysis. J Women s Hlth 2012;21: Bray F, Carstensen B, Moller H, et al. Incidence trends of adenocarcinoma of the cervix in 13 European countries. Cancer Epidemiol Biomarkers Prev 2005;14: Ferlay J, Soerjomataram I, Ervik M, et al. GLO- BOCAN 2012 v1.0, Cancer incidence and mortality worldwide: IARC CancerBase No. 11. Lyon, France: International Agency for Research on Cancer; Available at: (Accessed on June 25, 2014). 10. Pimenta JM, Galindo C, Jenkins D, et al. Estimate of the global burden of cervical adenocarcinoma and potential impact of prophylactic human papillomavirus vaccination. BMC Cancer 2013;13: Castellsague X, Diaz M, de Sanjose S, et al. Worldwide human papillomavirus etiology of cervical adenocarcinoma and its cofactors: implications for screening and prevention. J Natl Cancer Inst 2006;98: Seoud M, Tjalma WAA, Ronsse V. Cervical adenocarcinoma: moving towards better prevention. Vaccine 2011;29: Smith HO, Tiffany MF, Qualls CR, et al. The rising incidence of adenocarcinoma relative to squamous cell carcinoma of the uterine cervix in the United States a 24-year population-based study. Gynecol Oncol 2000;78: Sasieni P, Castanon A, Cuzick J. Screening and adenocarcinoma of the cervix. Int J Cancer 2009; 125: Quint W, Jenkins D, Molijn A, et al. One virus, one lesion individual components of CIN lesions contain a specific HPV type. J Pathol 2012; 227: Pirog EC, Kleter B, Olgac S, et al. Prevalence of human papillomavirus DNA in different histological subtypes of cervical adenocarcinoma. Am J Pathol 2000;157: McCluggage WG. New developments in endocervical glandular lesions. Histopathology 2013;62: Park KJ, Kiyokawa T, Soslow RA, et al. Unusual endocervical adenocarcinomas: an immunohistochemical analysis with molecular detection of human papillomavirus. Am J Surg Pathol 2011;35: An HJ, Kim KR, Kim IS, et al. Prevalence of human papillomavirus DNA in various histological subtypes of cervical adenocarcinoma: a population-based study. Mod Pathol 2005;18: Tavassoli FA, Devilee P. Pathology and Genetics of Tumours of the Breast and Female Genital Organs. IARC Press: Lyon, Alfsen GC, Reed W, Abeler VM. Reproducibility of classification in non-squamous cell carcinomas of the uterine cervix. Gynecol Oncol 2003;90: Quint WG, Scholte G, van Doorn LJ, et al. Comparative analysis of human papillomavirus infections in cervical scrapes and biopsy specimens by general SPF(10) PCR and HPV genotyping. J Pathol 2001;194: Kleter B, van Doorn LJ, ter Schegget J, et al. Novel short-fragment PCR assay for highly sensitive broad-spectrum detection of anogenital human papillomaviruses. Am J Pathol 1998;153: Kleter B, van Doorn LJ, Schrauwen L, et al. Development and clinical evaluation of a highly sensitive PCR-reverse hybridisation line probe assay for detection and identification of anogenital human papillomavirus. J Clin Microbiol 1999; 37: van Doorn LJ, Molijn A, Kleter B, et al. Highly effective detection of human papillomavirus 16 and 18 DNA by a testing algorithm combining broad-spectrum and type-specific PCR. J Clin Microbiol 2006;44: van Alewijk D, Kleter B, Vent M, et al. A human papillomavirus testing algorithm comprising a combination of the L1 broadspectrum SPF10 PCR assay and a novel E6 high-risk multiplex type-specific genotyping PCR assay. J Clin Microbiol 2013;51: Ou CY, Yang H, Balinandi S, et al. Identification of HIV-1 infected infants and young children using real-time RT-PCR and dried blood spots from Uganda and Cameroon. J Virol Methods 2007;144: van Doornum GJJ, Guldemeester J., Osterhaus ADME, et al. Diagnosing herpesvirus infections by real-time amplification and rapid culture. J Clin Microbiol 2003;41: Wright AA, Howitt BE, Myers AP, et al. Oncogenic mutations in cervical cancer: genomic differences between adenocarcinomas and squamous cell carcinomas of the cervix. Cancer 2013;119: Tornesello ML, Buonaguro L, Buonaguro FM. Mutations of the TP53 gene in adenocarcinoma and squamous cell carcinoma of the cervix: a systematic review. Gynecol Oncol 2013;128: Thomas LK, Bermejo JL, Vinokurova S, et al. Chromosomal gains and losses in human papillomavirus-associated neoplasia of the lower genital tract a systematic review and meta-analysis. Eur J Cancer 2014;50: Park JS, Kim YT, Lee A, et al. Prevalence and type distribution of human papillomavirus in cervical adenocarcinoma in Korean women. Gynecol Oncol 2013;130: Cooper K, Herrington CS, Lo ESF, et al. Integration of human papillomavirus types 16 and 18 in cervical adenocarcinoma. J Clin Pathol 1992;45: Steinestel J, Cronauer MV, Muller J, et al. Overexpression of p16 INK4a in urothelial carcinoma In Situ is a marker for MAPK-mediated epithelial-mesenchymal transition but is not related to human papillomavirus infection. PloS ONE 2013;8:e65189.