IJC International Journal of Cancer

Transcription

1 IJC International Journal of Cancer CADM1 and MAL methylation status in cervical scrapes is representative of the most severe underlying lesion in women with multiple cervical biopsies Romy van Baars 1, Jacolien van der Marel 1, Peter J.F. Snijders 2, Agata Rodriquez-Manfredi 3, Bram ter Harmsel 4, Henk A.M. van den Munckhof 1, Jaume Ordi 5, Marta del Pino 3, Miekel M. van de Sandt 1, N. Wentzensen 6, Chris J.L.M. Meijer 2 and Wim G.V. Quint 1 1 DDL Diagnostic Laboratory, Rijswijk, The Netherlands 2 Department of Pathology, VU University Medical Center, Amsterdam, The Netherlands 3 Department of Obstetrics and Gynaecology, Hospital Clınic, Barcelona, Spain 4 Roosevelt Kliniek, Leiden, The Netherlands 5 Department of Pathology, CRESIB (Centre De Recerca En Salut Internacional De Barcelona) Hospital Clinic, Faculty of Medicine-University of Barcelona, Barcelona, Spain 6 Division of Cancer Epidemiology and Genetics, National Cancer Institute, Bethesda, MD Recent studies have shown that CADM1/MAL methylation levels in cervical scrapes increase with severity and duration of the underlying cervical intraepithelial neoplasia (CIN) lesion. Multiple lesions of different histological grades and duration are frequently present on the cervix. To gain more insight into the possible epigenetic heterogeneity and its consequences for the methylation status in cervical scrapes, we performed an exploratory study of CADM1/MAL methylation in different grades of CIN lesions present in women with multiple cervical biopsies. CADM1-M18 and MAL-M1 methylation was assessed using a standardised, multiplex, quantitative methylation specific PCR on 178 biopsies with various grades of CIN in 65 women, and in their corresponding cervical scrapes. CADM1/MAL methylation positivity increased with disease severity, from 5.5% in normal biopsies to 63.3% and 100% in biopsies with CIN3 and cervical cancer, respectively. In the majority (8/9) of women where besides a CIN2/3 lesion a biopsy from normal cervical tissue was present, the CIN2/3 biopsy was CADM1/MAL methylation positive and the normal biopsy was CADM1/MAL methylation negative. A good concordance (78%) was found between CADM1/MAL methylation results on the scrapes and the biopsy with the worst diagnosis, particularly between samples of women with CIN3 and cervical cancer (92% and 100% concordance, respectively). Thus, in women with multiple cervical biopsies, CADM1/MAL methylation increases with severity of the lesion and is lesion-specific. CADM1/MAL methylation status in cervical scrapes appears to be representative of the worst underlying lesion, particularly for CIN3 and cervical cancer. Key words: DNA methylation, tumour suppressor genes, human papillomavirus, cervical intraepithelial neoplasia, cervical cancer Abbreviations: ASC-US: atypical squamous cells of undetermined significance; ASC-H: atypical squamous cells cannot exclude HSIL; CADM1: cell adhesion molecule 1; CIN: cervical intraepithelial neoplasia; Cq: quantification cycle; DEIA: DNA enzyme immuno assay; ECC: endocervical curettage; HG-CIN: high-grade cervical intraepithelial neoplasia; HPV: human papillomavirus; HrHPV: high-risk human papillomavirus; HSIL: high-grade squamous intraepithelial lesion; LBC: liquid-based cytology; LiPA: line probe assay; LSIL: lowgrade squamous intraepithelial lesion; MAL: T-lymphocyte maturation associated protein; NILM: negative for intraepithelial lesion or malignancy; NPV: negative predictive value; PCR: polymerase chain reaction; PPV: positive predictive value; qmsp: quantitative methylation specific PCR; SPF: short PCR fragment; TZ: transformation zone Additional Supporting Information may be found in the online version of this article. P.J.F.S. and C.J.L.M.M. are minority shareholders of Self-screen BV, a spin-off company of the VU University Medical Center. C.J.L.M.M. is a minority shareholder of Diassay. P.J.F.S. has been on speakers bureau of Abbott, Roche, Qiagen, Gen-probe and Seegene. C.J.L.M.M. has received speakers fee from GSK, Qiagen, Merck and Roche, served occasionally on the scientific advisory board/expert meeting of GSK, Qiagen, Merck, Roche and Genticel and has occasionally been consultant for Qiagen. W.G.V.Q. is a shareholder of LBP, the producer of the PreCursor-M test Grant sponsors: Stichting Pathologie Ontwikkeling en Onderzoek (SPOO) Foundation, The Netherlands and Instituto de Salud Carlos III (ICSIII)-Fondos de Investigacion Sanitaria, and ERDF one way to Europe; ; Grant numbers: PI12/01165 and PI12/01231 DOI: /ijc History: Received 2 Mar 2015; Accepted 19 June 2015; Online 28 July 2015 Correspondence to: Wim Quint, DDL Diagnostic Laboratory, Visseringlaan 25, 2288ER Rijswijk, The Netherlands, Tel.: 131-(88) , Fax: 131-(88) , wim.quint@ddl.nl Int. J. Cancer: 138, (2016) VC 2015 UICC

2 464 CADM1 and MAL methylation on lesion level What s new? When screening for cervical cancer, various characteristics can provide information. For instance, some results have shown that the degree of methylation in the CADM1/MAL genes can indicate how long an HPV infection has been present. The more methylated, the more severe the pre-cancerous lesion (CIN). But what if the patient has multiple lesions? The results of this study confirmed that more advanced lesions did show more methylation. The authors then went on to demonstrate that the methylation status detected in cervical scrapes seems to correspond with the most advanced lesions present on the cervix, especially cervical cancer or CIN3. Almost all cervical cancers are caused by a persistent infection with high-risk human papillomavirus (hrhpv) and preceded by premalignant stages referred to as cervical intraepithelial neoplasia (CIN). 1,2 Infections with hrhpv typically result in viral replication in differentiated cell layers of the squamous epithelium and release of progeny viral particles at the epithelial surface (productive infection). 3 Most infections are cleared by the immune system within 2 years. 4 Some infections may persist, and the viral gene expression may become disturbed. As a result, high expression levels of the viral oncogenes E6 and E7 in dividing epithelial cells trigger an oncogenic condition (a so-called transforming infection), which is often morphologically characterized as highgrade CIN (HG-CIN; CIN2/CIN3). 5 Progression of highgrade lesions to invasive cervical cancer requires the accumulation of additional epigenetic and genetic alterations, a process that may take 20 to 30 years. 6,7 This concept is supported by the fact that a longer duration of a preceding hrhpv infection has been associated with an increased number of genetic alterations in HG-CIN. 8 Hypermethylation of CpG islands in promoter regions of tumour suppressor genes has also been recognised as a molecular change from HG- CIN towards cervical cancer. 9,10 Recent studies have shown that methylation levels of two genes, i.e., CADM1 and MAL, in cervical scrapes are related to the duration of preceding HPV infection and the severity of the underlying CIN lesion Women with HG-CIN and hrhpv infections lasting >5 years had significantly higher methylation levels than those with infections of <5 years duration, and these levels were exceptionally high in cervical cancers. 11 Testing for CADM1 and MAL methylation in cervical scrapes can thus distinguish longstanding, more advanced HG-CIN lesions from early incident lesions. These findings suggest that CADM1 and MAL methylation analysis on cervical smears might serve as triage marker of hrhpv-positive women for colposcopy referral and potentially subsequent treatment. 14 However, as a result of multiple independent infections, multiple lesions or complex lesions of different histological grades can be present on the cervix. 15,16 With the aid of laser capture microdissection it has been shown that women with multiple HPV genotypes in the scrape can have distinct CIN lesions or components of CIN lesions containing one specific HPV genotype. 17 It is, however, yet unknown to what extent methylation events are lesion-specific in women with multiple cervical lesions of different histological grade. Moreover, the possible consequence of the presence of multiple underlying lesions for the methylation status of the corresponding cervical scrape is unknown. To address these topics, we first examined the variation in methylation results of CADM1 and MAL between different grades of CIN present in women with multiple cervical biopsies, and second correlated CADM1 and MAL methylation in cervical scrape samples with the corresponding biopsies of the same women. Material and Methods Study population The investigated study group was part of a prospective observational cohort of 610 women aged 17 years and older, referred for colposcopy to the Hospital Clınic, Barcelona, Spain, because of abnormal cervical cytology. 18 From this cohort we randomly selected 70 of 328 women with multiple cervical biopsies, such that different grades of CIN were sufficiently represented in this sample set. These comprised 14 women with CIN3, 16 women with CIN2, 14 with CIN1, and 5 women with cervical carcinoma as most severe lesion. As controls, 21 women that were histologically normal were selected. This study received approval from the medical ethical board of the local hospital (protocol number ). Written informed consent was obtained from each woman. Colposcopy procedure Colposcopic examination was performed and all findings were digitised using a digital colposcopy imaging system (Boundary Marking Tool) created by the National Cancer Institute (NCI) in collaboration with the National Library of Medicine (NLM), Bethesda. 19 After applying a 5% acetic acid solution for eliciting the acetowhite epithelial response, an image of the cervix was taken and all information was recorded in the Boundary Marking Tool. With this system the correlation between the clinical colposcopic image, histology and biomarkers can be studied. Colposcopists mark cervical lesions, indicate location of the biopsy and annotate their colposcopic impression per lesion. During colposcopy, up to four biopsies were collected from different abnormal areas or different regions in one large complex abnormal

3 van Baars et al. 465 area of the cervix. If less than four colposcopically directed biopsies were taken, a biopsy from normal appearing epithelium (visually negative tissue) was added. An endocervical curettage (ECC) was performed if the colposcopist regarded it clinically indicated. Pathological diagnosis and grading Biopsy specimens were fixed in 10% buffered formalin and paraffin-embedded. Haematoxylin and Eosin sections were examined by a local pathologist and classified as negative, CIN1, CIN2, CIN3 or cervical carcinoma. All biopsies were reviewed by a second independent gynaecological pathologist. In case of disagreement between the original and review diagnosis, a third pathologist reviewed the discordant cases independently and a consensus diagnosis was reached by agreement between two of three interpretations. All pathologists were blinded for HPV status and cytology result. The overall histological diagnosis per women was based on the worst diagnosis found in all biopsy specimens of each woman. HPV detection and genotyping A liquid-based cytology (LBC) sample was obtained prior to colposcopy. Cytological examination and classification was performed at the local laboratory according to the Bethesda 2001 classification. DNA isolation was performed using the QIAamp MinElute Virus Spin kit (Qiagen, Valencia, CA). HPV DNA detection was performed with the SPF 10 -PCR-DEIA-LiPA 25 system (SPF 10 HPV LiPA 25 version 1; Labo Bio-Medical Products, Rijswijk, The Netherlands) as described previously. 20,21 The LiPA 25 can identify 25 HPV genotypes (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, and 74) by reverse hybridisation on a line probe assay. DNA from histology specimens was isolated by a proteinase K procedure as described elsewhere 22 and HPV DNA detection and genotyping was likewise performed with the SPF 10 -PCR- DEIA-LiPA 25 method. 20,21 Each run contained negative and internal and external positive controls to monitor for efficiency of DNA isolation, PCR amplification, hybridisation, and genotyping procedures. Contamination or failure of analyses was not encountered. An in-house human reference gene, RNaseP, was used as a quality control for the amount of amplifiable DNA 23 and Phocine herpesvirus was used as an internal control for the absence of PCR inhibition in the isolated material. 24 PCR inhibition was not encountered. Women with insufficient DNA quality (i.e., negative RNaseP housekeeping gene PCR results) of their LBC specimen were excluded. Of the 70 selected patients, five women were excluded because of poor DNA quality of the LBC sample. Bisulfite treatment and quantitative methylation specific PCR After determination of the amount of amplifiable DNA, up to 250 ng of DNA was subjected to bisulfite conversion using the EZ DNA Methylation kit (Zymo Research, Orange, CA). If less than 100 ng of DNA was present, 3 ml carrier RNA (Qiagen) was added. The bisulfite conversion protocol of the cervical scrapes was adapted for the biopsies with an additional pretreatment incubation for 10 min at 958C and denaturation for both sample types was performed at 958C for 20 min. After bisulfite treatment, a standardised, multiplex quantitative methylation specific PCR (qmsp) (prototype PreCursor-M test, Self-screen B.V., Amsterdam, The Netherlands) was used to identify the methylation status of the promoter regions of CADM1-M18 and MAL-M1. In addition, a sample reference for the bisulfite-converted human reference gene ß-actin was included in this multiplex test to determine the total amount of converted human DNA present in the reaction. The multiplex format enables simultaneous amplification and detection of all targets within one reaction. The quantification cycle (Cq-value) was determined for each target and the ratio of the methylated marker to the total amplifiable bisulfite-treated DNA of each sample was calculated by the DDCq method. 25 This method shows the relative difference between each marker and ß-actin in a sample compared with the internal calibrator, a plasmid containing the bisulfite-converted sequence of all targets. Cervical scrape samples were scored CADM1/MAL methylation positive when either one or both markers had a target gene/ß-actin ratio above a predefined specific threshold value. These threshold values were exactly as described in the kit insert of the PreCursor-M kit, which had been set such that a CIN31 specificity of 70% was reached on a set of cervical scrape samples from a screening population using the method as described by Hesselink et al. 14 Histology samples were scored CADM1/MAL methylation positive when one or both markers had a target gene/ß-actin ratio above a fixed calculated threshold value. For each gene the threshold was the upper limit of the 99% confidence interval of the mean ratio of all histologically normal samples. Statistical analysis The frequencies of hrhpv, CADM1, MAL and CADM1 and/ or MAL (CADM1/MAL) methylation positive cervical biopsies in relation to their histological diagnosis and the frequency of hrhpv, CADM1, MAL and CADM1/MAL methylation positive cervical scrapes in relation to the worst underlying histological diagnosis were calculated. We assessed concordance in CADM1/MAL methylation status between cervical scrapes and biopsies of each patient and stratified for histological grade. A woman was considered CADM1/MAL methylation positive on biopsies if at least one of the collected biopsies was CADM1/MAL methylation positive, irrespective of histological grade. To avoid bias from the relatively large proportion of cancers in this study, sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of CADM1/MAL methylation testing on cervical scrapes and biopsies were only calculated for CIN2/3 and CIN3. Independent sample t-tests and linear logistic regression were used to examine the relation between age

4 466 CADM1 and MAL methylation on lesion level Table 1. Positivity rates for hrhpv, MAL, CADM1 and CADM1 and/or MAL methylation in cervical scrapes in relation to their most severe underlying histological diagnosis (on patient level) MAL and/or Worst histological Total hrhpv positive MAL positive CADM1 positive CADM1 positive diagnosis (n 5 65) n n % n % n % n % Negative CIN CIN CIN Carcinoma All six scrapes were hrhpv positive. 2 Both hrhpv negative scrapes were CADM1 and/or MAL methylation negative. and CADM1/MAL methylation status in women with CIN21. p Values below 0.05 were considered significant. Results Study population The median age of the 65 women included in the study was 33.0 years (range years). Five of them had invasive carcinoma (of whom four squamous cell carcinomas and one adenocarcinoma), 12 CIN3, 13 CIN2, and 14 CIN1 as worst diagnosis. Twenty one women only had histomorphologically normal biopsies. Of the total number of 178 biopsies taken, the histological diagnoses were invasive carcinoma (n 5 8), CIN3 (n 5 30), CIN2 (n 5 27), CIN1 (n 5 22) and normal (n 5 91). The cytology results were squamous cell carcinoma (n 5 3), adenocarcinoma (n 5 1), HSIL (n 5 28), LSIL (n 5 10), ASC-H (n 5 2), ASC-US (n 5 7), and negative for intraepithelial lesion or malignancy (NILM) (n 5 14). In Supporting Information Table 1 the relation between the cytology result and the most severe diagnosis per patient is shown. The hrhpv positivity rate in the LBC samples was 81.5% (53/65) (Table 1). Six (9.2%) women were HPV negative and six (9.2%) had an infection with a low-risk HPV type. CADM1/MAL methylation results on cervical scrape The proportion of women with a positive hrhpv test and CADM1, MAL and CADM1/MAL methylation test in the cervical scrapes in relation to the most severe underlying histological grade are shown in Table 1. Both hrhpv and CADM1/MAL methylation positivity rates increased in relation to the worst underlying histological diagnosis. A 100% hrhpv positivity rate was found for CIN2, CIN3 and carcinoma. The percentage of CADM1/MAL methylation positive women was similar in women with a worst diagnosis of normal/cin1 (i.e., 10/35; 28.6%) and CIN2 (3/13; 23.1%), but increased to 50% (6/12) of women with CIN3 and 100% (5/ 5) of women with carcinoma. Two women (14.3%) with CIN1 were hrhpv negative. They were also CADM1/MAL methylation negative. Eleven of 21 (52.4%) histologically negative women were hrhpv positive, six of whom (54.5%) were also CADM1/MAL methylation positive. All 12 hrhpv negative women with histologically normal biopsies had a CADM1/MAL methylation negative cytology sample. CADM1/MAL methylation marker results on cervical biopsies (on lesion level) Methylation patterns in biopsy specimens showed a strong relationship to histological diagnosis (Table 2). CADM1/ MAL methylation positivity increased with disease severity from 5.5% (5/91) in normal biopsies, 13.6% (3/22) in CIN1 biopsies, 29.6% (8/27) in CIN2 biopsies to 63.3% (19/30) in CIN3 biopsies and 100% (8/8) in cervical carcinomas. All biopsies with CIN3 and carcinoma were hrhpv positive. In addition, 25/27 (92.6%) CIN2 biopsies and 20/22 (90.9%) CIN1 biopsies were hrhpv positive. All hrhpv negative CIN1 and CIN2 biopsies were also CADM1/MAL methylation negative. The five histologically normal biopsies that were CADM1/MAL methylation positive were all hrhpv negative. Two of them had NILM cytology and none of them had a CADM1/MAL methylation positive cervical scrape sample. Also after restricting the analysis to the most severe underlying histological grade per woman, positivity for hrhpv and CADM1/MAL methylation in cervical biopsies increased proportional to lesion severity (Supporting Information Table 2). CADM1/MAL methylation in women with multiple biopsies of different histological grade In this study population, nine women with methylation positive CIN2 or three biopsies had at least one additional normal biopsy. Figure 1a shows an example of a woman with both methylation positive CIN3 biopsies and methylation negative normal biopsies. In eight (88.9%) of these nine women the CADM1/MAL positivity was confined to the biopsies containing the lesions. Five women with CIN3 as the worst diagnosis had no adjacent normal biopsy, but at least one additional biopsy with CIN2 or 3. Two of them had methylation positive CIN3 biopsies next to methylation negative CIN3 biopsies. Figure

5 van Baars et al. 467 Table 2. Positivity rates for hrhpv, MAL, CADM1 and CADM1 and/or MAL methylation in cervical biopsies (on lesion level) Total hrhpv positive MAL positive CADM1 positive MAL and/or CADM1 positive Diagnosis N n % n % n % n % Negative CIN CIN CIN Carcinoma All five biopsies were hrhpv negative. 2 Both hrhpv negative biopsies were CADM1 and / or MAL methylation negative. 1b shows a colposcopic image of a woman with two CIN3 biopsies and one CIN2 biopsy with different HPV types. One HPV33 positive CIN3 biopsy was methylation positive and another HPV16 positive CIN3 biopsy was methylation negative. Also the CIN2 lesion, in which both HPV16 and HPV33 were detected, was methylation negative. In one woman all CIN2/3 biopsies were methylation positive. In two women all CIN2/3 biopsies were methylation negative. Thus in women with multiple CIN2/3 biopsies methylation may differ per biopsy. Concordance in methylation status between cervical scrapes and biopsies, stratified for histological diagnosis The overall concordance in methylation status between the cervical scrape and corresponding biopsy representing the worst diagnosis was 78% (51/65) (Table 3). These involved 16 women with concordant methylation positive scrape/ biopsy results and 35 with concordant negative scrape/biopsy result. Six of 65 (9.2%) women were methylation positive on one of the biopsies, but not on the scrape. One of them had a CIN3 lesion and three a CIN2 lesion. Eight of 65 (12.3%) women were methylation positive on the cervical scrape, but not on corresponding biopsies; one of them had CIN2, three CIN1 and four had only histologically normal biopsies. In two women with only histologically negative biopsies, the transformation zone (TZ) was not visible during colposcopy, but no ECC was performed. In four women the amount of amplifiable DNA was relatively low (Cq-value ß-actin >30). Concordance increased with lesion grade: in CIN3 cases concordance was 92% (11/12, of which six were positive on both biopsy and scrape) and in carcinomas this was 100% (all five double positive on biopsy and scrape). In comparison, we found 98% (64/65) agreement in hrhpv detection. Methylation in relation to age Methylation positive women with CIN21 (n 5 14) were older than methylation negative women with CIN21 (n 5 16) (mean of 42 vs. 30 years; p ). In addition, in a linear regression analysis, age in women with CIN21 was associated with increased MAL (estimate higher MAL levels, 95% confidence interval [CI]: , p ) and CADM1 methylation levels (estimate higher CADM1 levels, 95% CI: , p ). Detection rate of CIN2/3 The sensitivity of CADM1/MAL methylation in cervical scrape samples for the detection of CIN2/3 was slightly lower than in biopsies, although this difference was not statistically significant (36% and 46%, respectively; p ) (Table 4). The specificity for the detection of CIN2/3 was 71% and 86%, for cervical and biopsy analysis, respectively (p ). Also, no significant difference in the sensitivity or specificity for the detection of CIN3 could be found between the cervical scrape samples and the biopsies (p and p ). The sensitivity for the detection of CIN3 in cervical scrape samples and biopsies was 50% and 58%, respectively, and the specificity was 73% and 79%, respectively. The PPV for both CIN2/3 and CIN3 was higher for the methylation test on biopsies than on scrapes, although not significantly (71% vs. 47%; p , and 41% vs. 32%; p , respectively). The NPV for CIN2/3 was 70% for biopsies and 61% for scrapes (p ) and for CIN3 88% for biopsies and 85% for scrapes (p ). Discussion We showed that at biopsy level methylation of CADM1/MAL genes was specific for each lesion in women with multiple lesions. In the majority of women with both a CIN2/3 lesion and a negative biopsy, CADM1/MAL methylation positivity was confined to the biopsy containing the lesion. CADM1/ MAL methylation positivity also diverged between different lesions on the cervix; it was increased proportional to the severity of the underlying histology. These findings support the concept that CADM1/MAL hypermethylation is a lesionspecific event and does not involve the lesion environment as a field effect. Moreover, irrespective of the presence of multiple lesions, CADM1/MAL methylation analysis of the cervical scrape was highly representative of that in the worst lesion, particularly if this was CIN3 or cervical cancer. It is of note that some of the differences found in this study lack

6 468 CADM1 and MAL methylation on lesion level Figure 1. The correlation between the clinical colposcopic image, cytology, histology, HPV genotype and MAL and CADM1 methylation on the cervical scrape and per biopsy. (a) Example 1; HPV: human papillomavirus. (b) Example 2; HPV: human papillomavirus, random biopsy: biopsy obtained from visual negative tissue. statistical significance. This likely reflects the rather small sample size studied. Combined positivity for the two markers CADM1 and MAL increased proportionally with the severity of the underlying CIN lesions as previously found. 11,14 We found 50% (6/ 12) of cervical scrapes of women with CIN3 to be methylated for CADM1 and/or MAL. Although this percentage tends to be a lower percentage than the 68% found by Overmeer et al. in a previous study, 13 caution is needed to compare results because of the small sample size in the current study. The sensitivity for CIN3 was not significantly higher by biopsy testing compared to testing of a cervical scrape (7/12 vs. 6/12 positive). Hence, the nondetection of methylation in cervical scrapes generally reflects the presence of methylation negative

7 van Baars et al. 469 Table 3. Concordance in CADM1/MAL methylation status between cervical scrapes and cervical biopsies (representing most severe diagnosis per woman) by histological grade Histology N Biopsy1/scrape1 Biopsy1/scrape2 Biopsy2/scrape1 Biopsy2/scrape2 Concordance Overall (51/65) 78% Negative (15/21) 71% CIN (11/14) 79% CIN (9/13) 69% CIN (11/12) 92% Carcinoma (5/5) 100% Table 4. Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of CADM1/MAL methylation testing on cervical scrapes and cervical biopsies for the detection of CIN2/3 and CIN3 (on patient level) Cervical scrape Cervical biopsies CIN2/3 CIN3 CIN2/3 CIN3 Sensitivity 36% (9/25) 50% (6/12) 48% (12/25) 58% (7/12) Specificity 71% (25/35) 73% (35/48) 86% (30/35) 79% (38/48) PPV 47% (9/19) 32% (6/19) 71% (12/17) 41% (7/17) NPV 61% (25/41) 85% (35/41) 70% (30/43) 88% (38/43) lesions, rather than a suboptimal detection of methylation in the cervical scrape. A recent study has revealed that methylation levels are particularly high in more advanced CIN3 lesions that display a longer duration of preceding HPV infection, which can be considered a proxy for lesion duration. 11 Thus methylation positive CIN3 lesions may represent advanced CIN3 lesions with a high short-term risk to cancer. 8,11 In line with these findings, CADM1/MAL methylation was present in CIN3 biopsies from women with carcinoma and in scrapes and biopsy specimens of all women with carcinoma cases in this study. Furthermore, methylation positive women with CIN21 were older than methylation negative women with CIN21 (42 vs. 30 years; p ). This may reflect the fact that CIN21 lesions in younger women represent more often early lesions with a high likelihood of spontaneous regression. 26 These findings suggest that CADM1/MAL methylation testing might distinguish women with longstanding CIN21 lesions with an increased short-term progression risk to carcinoma and therefore might be an attractive triage marker to reassure absence of cancer of longstanding CIN disease in hrhpv positive women. Of course, further studies are needed to investigate whether methylation negative CIN21 have indeed a less aggressive course than methylation positive CIN21. Although less well studied, also other human DNA methylation markers and HPV DNA methylation tests might be candidate triage markers for hrhpv positive women The difference in populations and testing methods may explain why the methylation frequency of certain markers vary between studies. 32 CIN2 and to lesser extent CIN3 are heterogeneous disease categories in terms of risk of progression to invasive cancer and number of chromosomal aberrations. 34,35 Here, we describe that by CADM1/MAL methylation analysis such a heterogeneity can also be found in cervical lesions in one woman. Methylation events present in severe lesions were generally well detectable in corresponding cervical scrapes. In cases showing a methylation negative scrape but one or more positive biopsies, poor scrape sampling seems a plausible reason. 36 On the other hand, presence of methylation positive scrapes but negative biopsies may point to presence of more severe lesions missed by biopsy (e.g., endocervical), a feature that might emphasize the need of performing an ECC. 37,38 Also, from some biopsies rather low amounts of DNA could be extracted. When omitting all cases with possible inadequate sampling (due to lower DNA quantity (ß-actin >30) or to endocervical location of the TZ), the concordance between methylation status of cervical scrapes and biopsies increased to 91% (51/56). Finally, methylation positivity rates depend on the thresholds used. Differences in cell composition between cervical scrape and biopsy samples result in different background methylation levels. 13,39 Accordingly, we adjusted the assay thresholds for positivity to the type of sample. In this study the thresholds for positivity in scrapes were determined with a standardised assay validated on the basis of a previous training set analysis, setting thresholds to give an optimal balance between detection of background methylation and lesion-specific hypermethylation with a specificity of 70%. Since for biopsies no such validated thresholds were available, we here used the upper limit of the 99% confidence intervals of the mean ratio of all

8 470 CADM1 and MAL methylation on lesion level histologically negative biopsies to score biopsy samples methylation positive or negative. Nevertheless, despite these different scoring systems, we found a substantial agreement in methylation status of the biopsies and the corresponding cervical scrape of the same women. Previous studies have shown that for some genes methylation as determined in the scrape is reflecting the methylation status as determined in the biopsy. 39,40 However, these studies have neither addressed CADM1 and MAL, the genes studied in this manuscript, nor did these provide details about the possible presence of multiple underlying lesions and their consequence, if any, for the methylation status of the corresponding cervical scrape. The main strength of our study is that we were able to relate methylation results on scrapes with those of multiple biopsies and various parameters, such as histology, HPV genotype, age and colposcopic image. The histological diagnosis of all biopsies was done by consensus diagnosis of expert pathologists and local pathologists who were blinded to HPV status and cytology result at the time of diagnosis. Moreover, if less than four colposcopydirected biopsies were obtained a biopsy from visually normal tissue was taken. Therefore, we were able to compare methylation results from CIN lesions and normal tissue from the same patient. In eight out of nine women who were methylation positive and from whom biopsies with HG-CIN and normal tissue were taken, the normal biopsy was consistently methylation negative. In women with only HG-CIN biopsies, methylation status could differ per biopsy. From this we can conclude that CADM1/MAL methylation does not involve a field effect, but is specifically lesion driven. In conclusion, we found in this study with multiple carefully located biopsies of different grades that CADM1/MAL methylation was lesion-specific and increased with the severity of the CIN lesion. Moreover, in women with multiple cervical lesions CADM1/MAL methylation status determined on the cervical scrape appears to be representative for the CADM1/MAL methylation status of the worst underlying cervical lesion, particularly in women with CIN31. Acknowledgements The authors thank Nadia Abu-Lhiga, Hospital Clınic for her clinical assistance. The authors thank Lorena Marimon, Hospital Clınic and Leonie van den Berg, DDL Diagnostic Laboratory for their technical assistance. The authors thank all colposcopists, laboratory personnel and the women who have participated in this study. The authors thank David Jenkins for his guidance on writing the manuscript. References 1. Bosch FX, Lorincz A, Munoz N, et al. The causal relation between human papillomavirus and cervical cancer. J Clin Pathol 2002; 55: Walboomers JM, Jacobs MV, Manos MM, et al. Human papillomavirus is a necessary cause of invasive cervical cancer worldwide. J Pathol 1999; 189: Doorbar J, Quint W, Banks L, et al. The biology and life-cycle of human papillomaviruses. Vaccine 2012; 30(Suppl 5):F55 F Ostor AG. Natural history of cervical intraepithelial neoplasia: a critical review. Int J Gynecol Pathol 1993; 12: Doorbar J. Molecular biology of human papillomavirus infection and cervical cancer. Clin Sci (Lond) 2006; 110: McCredie MR, Sharples KJ, Paul C, et al. Natural history of cervical neoplasia and risk of invasive cancer in women with cervical intraepithelial neoplasia 3: a retrospective cohort study. Lancet Oncol 2008; 9: Vink MA, Bogaards JA, van Kemenade FJ, et al. Clinical progression of high-grade cervical intraepithelial neoplasia: estimating the time to preclinical cervical cancer from doubly censored national registry data. Am J Epidemiol 2013; 178: Bierkens M, Wilting SM, van Wieringen WN, et al. Chromosomal profiles of high-grade cervical intraepithelial neoplasia relate to duration of preceding high-risk human papillomavirus infection. Int J Cancer 2012; 131:E579 E Herman JG, Baylin SB. Gene silencing in cancer in association with promoter hypermethylation. N Engl J Med 2003; 349: Jones PA, Gonzalgo ML. Altered DNA methylation and genome instability: a new pathway to cancer? Proc Natl Acad Sci USA 1997; 94: Bierkens M, Hesselink AT, Meijer CJ, et al. promoter methylation levels in hrhpv-positive cervical scrapes increase proportional to degree and duration of underlying cervical disease. Int J Cancer 2013; 133: Overmeer RM, Henken FE, Bierkens M, et al. Repression of MAL tumour suppressor activity by promoter methylation during cervical carcinogenesis. J Pathol 2009; 219: Overmeer RM, Louwers JA, Meijer CJ, et al. Combined CADM1 and MAL promoter methylation analysis to detect (pre-)malignant cervical lesions in high-risk HPV-positive women. Int J Cancer 2011; 129: Hesselink AT, Heideman DA, Steenbergen RD, et al. Combined promoter methylation analysis of CADM1 and MAL: an objective triage tool for high-risk human papillomavirus DNA-positive women. Clin Cancer Res 2011; 17: Bekkers RL, Melchers WJ, Bulten J, et al. Localized distribution of human papillomavirus genotypes in the uterine cervix. Eur J Gynaecol Oncol 2002; 23: Yang HP, Zuna RE, Schiffman M, et al. Clinical and pathological heterogeneity of cervical intraepithelial neoplasia grade 3. PLoS One 2012; 7: e 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: van der Marel J, van Baars R, Quint WG, et al. The impact of human papillomavirus genotype on colposcopic appearance: a cross-sectional analysis. BJOG 2014; 121: Jeronimo J, Long LR, Neve L, et al. Digital tools for collecting data from cervigrams for research and training in colposcopy. J Low Genit Tract Dis 2006; 10: 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 hybridization line probe assay for detection and identification of anogenital human papillomavirus. J Clin Microbiol 1999; 37: 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: Luo W, Yang H, Rathbun K, et al. Detection of human immunodeficiency virus type 1 DNA in dried blood spots by a duplex real-time PCR assay. J Clin Microbiol 2005; 43: van Doornum GJ, Guldemeester J, Osterhaus AD, et al. Diagnosing herpesvirus infections by realtime amplification and rapid culture. J Clin Microbiol 2003; 41: Livak KJ, Schmittgen TD. Analysis of relative gene expression data using real-time quantitative PCR and the 2(-delta delta C(T)) method. Methods 2001; 25: Moscicki AB, Ma Y, Wibbelsman C, et al. Rate of and risks for regression of cervical intraepithelial neoplasia 2 in adolescents and young women. Obstet Gynecol 2010; 116: Brentnall AR, Vasiljevic N, Scibior-Bentkowska D, et al. A DNA methylation classifier of cervical precancer based on human papillomavirus and human genes. Int J Cancer 2014; 135: Clarke MA, Wentzensen N, Mirabello L, et al. Human papillomavirus DNA methylation as a potential biomarker for cervical cancer. Cancer Epidemiol Biomarkers Prev 2012; 21:

9 van Baars et al Cuzick J, Bergeron C, von Knebel Doeberitz M, et al. A new technologies and procedures for cervical cancer screening. Vaccine 2012;30(Suppl 5): F107 F Eijsink JJ, Lendvai A, Deregowski V, et al. A four-gene methylation marker panel as triage test in high-risk human papillomavirus positive patients. Int J Cancer 2012; 130: Hansel A, Steinbach D, Greinke C, et al. A promising DNA methylation signature for the triage of high-risk human papillomavirus DNA-positive women. PLoS One 2014; 9:e Wentzensen N, Sherman ME, Schiffman M, et al. Utility of methylation markers in cervical cancer early detection: appraisal of the state-of-the-science. Gynecol Oncol 2009; 112: Wilting SM, van Boerdonk RA, Henken FE, et al. Methylation-mediated silencing and tumour suppressive function of hsa-mir-124 in cervical cancer. Mol Cancer 2010; 9: Wilting SM, Steenbergen RD, Tijssen M, et al. Chromosomal signatures of a subset of high-grade premalignant cervical lesions closely resemble invasive carcinomas. Cancer Res 2009; 69: Steenbergen RD, Snijders PJ, Heideman DA, et al. Clinical implications of (epi)genetic changes in HPV-induced cervical precancerous lesions. Nat Rev Cancer 2014; 14: Wentzensen N, Schiffman M, Dunn ST, et al. Grading the severity of cervical neoplasia based on combined histopathology, cytopathology, and HPV genotype distribution among 1,700 women referred to colposcopy in Oklahoma. Int J Cancer 2009; 124: Petry ku, Luyten A, Scherbring S. Accuracy of colposcopy management to detect CIN3 and invasive cancer in women with abnormal screening tests: results from a primary HPV screening project from 2006 to 2011 in Wolfsburg, Germany. Gynecol Oncol 2013; 128: Pretorius RG, Zhang WH, Belinson JL, et al. Colposcopically directed biopsy, random cervical biopsy, and endocervical curettage in the diagnosis of cervical intraepithelial neoplasia II or worse. Am J Obstet Gynecol 2004; 191: Feng Q, Balasubramanian A, Hawes SE, et al. Detection of hypermethylated genes in women with and without cervical neoplasia. J Natl Cancer Inst 2005; 97: Reesink-Peters N, Wisman GB, Jeronimo C, et al. Detecting cervical cancer by quantitative promoter hypermethylation assay on cervical scrapings: a feasibility study. Mol Cancer Res 2004; 2: