Quantitative Analysis of Methylation Status of the PAX1 Gene for Detection of Cervical Cancer

Transcription

1 ORIGINAL ARTICLE Quantitative Analysis of Methylation Status of the PAX1 Gene for Detection of Cervical Cancer Tien-Hung Huang, MS,*Þ Hung-Chen Lai, MD, PhD,þ Hwan-Wun Liu, MD, PhD, Cuei Jyuan Lin, MS,*Þ Kai-Hung Wang, MS,*Þ Dah-Ching Ding, MD, PhD, # and Tang-Yuan Chu, MD, PhD* # Objective: Although aided by high-risk human papillomavirus (HPV) DNA test, early detection of cervical cancer is still a challenge. Hypermethylation of the paired boxed gene 1(PAX1) was recently reported as a characteristic of cervical cancer. This study designed a quantitative measure of PAX1 methylation and compared its efficacy to the currently available Hybrid Capture 2 (HC2) HPV test in detection of cervical cancer. Methods: Using real-time quantitative methylation-specific polymerase chain reaction, we measured the percentage of PAX1 methylation in cervical scrapings obtained from a hospital-based cohort of women with cervical neoplasia of different severities and compared the efficacy of diagnosis of cervical cancer to that of the HC2 HPV test. Results: From 73 cervical scrapings, with diagnoses of normal (n = 17), cervical intraepithelial neoplasm 1 (CIN1; n = 10), CIN2 (n = 18), CIN3 (n = 14), and invasive cancer (n = 14), the percentage of PAX1 methylation was determined. The percent of methylated reference of invasive cancer (mean [SE], 56.7 [7.1]) was significantly higher than CIN3 (6.5 [2.3]) and the other milder lesions (1.0 [0.3]; P G ). At a cutoff percent of methylated reference value of 4.5, PAX1 methylation was found in 100% of invasive cancer tissue as compared with 0% of normal tissue, 10% of CIN1, 11% of CIN2, and 43% of CIN3 (P G ). As a comparison, the HC2 HPV test result was positive in 5.9% of normal tissue, 70% of CIN1, 55.6% of CIN2, 71.4% of CIN3, and 100% of invasive cancer. In addition to cancer tissue, methylation of PAX1 was also found in normal tissue adjacent to the cancer lesion (9/11, 82%) but much less in the remote normal tissues (2/5, 40%), indicating a field methylation. Conclusions: In this hospital-based study, quantitative measurement of PAX1 hypermethylation in cervical scrapings is highly sensitive and is more specific than HC2 in detection of cervical cancer. Key Words: PAX1, DNA methylation, Cervical cancer, HPV test, Field methylation Received September 4, 2009, and in revised form October 26, Accepted for publication October 27, (Int J Gynecol Cancer 2010;20: 513Y519) *Department of Research, Center for Cervical Cancer Prevention, Buddhist Tzu Chi General Hospital, and School of Medicine, Graduate Institute of Medical Science, Tzu Chi University, Hualien; Department of Obstetrics and Gynecology, Tri-Service General Hospital; Laboratory of Epigenetics, National Defense Medical Center, Taipei; and Department of Occupational Medicine, Copyright * 2010 by IGCS and ESGO ISSN: X DOI: /IGC.0b013e3181c7fe6e Buddhist Tzu Chi General Hospital, Graduate Institute of Clinical Medicine, and #Department of Obstetrics and Gynecology, Buddhist Tzu Chi General Hospital, Tzu Chi University, Hualien, Taiwan, Republic of China. Address correspondence and reprint requests to Tang-Yuan Chu, MD, PhD, Department of Obstetrics and Gynecology, Buddhist Tzu Chi General Hospital, 707, Section 3, Chung-Yang Rd, Hualien 970, Taiwan, Republic of China. tychu@mail2000.com.tw. International Journal of Gynecological Cancer & Volume 20, Number 4, May

2 Huang et al International Journal of Gynecological Cancer & Volume 20, Number 4, May 2010 nfection of high-risk human papillomavirus (HPV) is a Inecessary cause of cervical carcinogensis. 1 Early infection of HPV usually induces transient cervical intraepithelial neoplasia (CIN1), whereas persistent infection may result in premalignant cervical intraepithelial neoplasia (CIN2 or CIN3). Testing for HPV DNA in cervical scrapings is performed as a triage for unequivocal Papanicolaou test results 2 and for primary screening for cervical cancer, typically with high negative predictive value. 3,4 However, because most infections are subclinical, transient, and without cancerous or precancerous lesions. Human papillomavirus testing has been used to facilitate the screening of cervical cancer but is insufficient in the diagnosis of cervical cancer. Recently, transcriptional silencing of tumor suppressor genes by hypermethylation of CpG sites of promoters was reported as an early and crucial event in cancer development. 5,6 In cervical carcinogenesis, a variety of tumor suppressor genes including tumor suppressor in lung cancer 1 (TSLC1), zinc finger, MYND-type containing 10 (BLU), and Ras association domain family member 1 (RASSF1A) were shown methylation silenced. 7Y10 We have recently reported the paired boxed gene 1 (PAX1) as a novel methylationsilenced gene in cervical cancer development, and methylation was specifically present in the carcinoma in situ (CIS) to invasive cancer stages. 11 The PAX1 gene belongs to a highly conserved family of developmentally controlled genes that encode transcription factors and play a role in pattern formation during embryogenesis in vertebrates. 12 Specifically, PAX1 is expressed during the development of the skeleton, thymus, and parathyroid glands 13,14 and is frequently methylation silenced in cervical and ovarian cancers. 11,15 Other than the role of frequent methylation silencing of the tumor suppressor PAX1 in cervical cancer cells, the functional role of PAX1 in carcinogenesis remains unknown. The methylation-specific polymerase chain reaction (PCR; MSP) test 16 for DNA methylation uses visual detection of ethidium bromideystained PCR products in agarose gel, which is subjective in result reading and insufficient to assess the wide range of methylation percentage in cervical scrapings. In the present study, we performed quantitative measurement of PAX1 gene methylation in cervical scrapings obtained from the full spectrum of cervical neoplastic diseases from normal tissue, CIN1, CIN2, and CIN3 to invasive cancer and compared the results with HPV DNA testing on the same scrapings. MATERIALS AND METHODS Study Subjects and Diagnosis During the period between January 2006 and April 2009, women referred for colposcopic examination or surgical interventions because of abnormal Papanicolaou test results or known cervical cancer at the Buddhist Tzu-Chi General Hospital, Hualien, Taiwan, were enrolled for this study. Pelvic examination and colposcopy were performed by the same colposcopist (C.T.Y.). Healthy women who visited the hospital for routine Papanicolaou test during the same period were randomly invited as healthy controls. A standard guideline issued by the National Health Research Institute, Taiwan, for screening, diagnosis, and management of cervical neoplasia ( file/tcog/gog_b.pdf) was followed in all subjects. The exclusion criteria included previous diagnosis of HPV-related genital lesions or neoplasia, current pregnancy, absence of the uterine cervix, history of cancer at other sites, and history of immune compromise diseases. A total of 73 women, including 17 healthy controls (normal tissue), 10 with CIN1, 18 with CIN2, 14 with CIN3 (including 7 CISs), and 14 with invasive cancers were engaged in this study. All the CINs and invasive cancers were confirmed by histologic examination. The age range of the 73 women at enrollment were approximately 22 to 85 years (mean [SD], 48.5 [1.7] years; Table 1). For the characterization of the field methylation in cancer lesions and normal tissues, cervical cancer patients underwent radical hysterectomy, noncancer women underwent total hysterectomy on account of benign uterine tumors, and patients with gynecologic cancers of a noncervical origin were recruited for tissue procurement. The study was approved by the institutional review board of the Tzu Chi General Hospital, and informed consent was obtained from each subject before the study. TABLE 1. The PMR values of PAX1 methylation in the cervical scrapings of healthy controls and biopsy-confirmed cervical neoplasia n Age PMR of PAX1 Gene Mean (SE) Range Mean (SE) Normal (2.8)* 0Y (0.3) 1.0 (0.3) CIN (4.9) 0Y (1.2) CIN (2.5) 0Y (0.5) CIN (3.1) 0Y (2.3) 31.6 (6.1) Invasive cancer (3.9) 14.1Y (7.1) *P = 0.15, comparing the healthy and other case groups. P G , comparing the CIN3 and invasive cancer groups with the healthy, CIN1, and CIN2 groups. P G , comparing the invasive cancer with noncancer groups. 514 * 2010 IGCS and ESGO

3 International Journal of Gynecological Cancer & Volume 20, Number 4, May 2010 PAX1 Methylation Versus HPV Test Clinical Specimens Cervical scrapings were collected by a cytobrush before colposcopy, cervical biopsy, or other cervical procedures. The swabs were transferred into a universal tube containing 3 ml of phosphate-buffered saline. After thorough agitation, they were then dispensed, snap frozen, and stored at j80-c. For the procurement of tissues from topologically different sites, surgical tissues were carefully procured in respect to their anatomic locations and nature of the tumor lesions. Primary and metastatic cancers, as well as normal cervix adjacent to a cancer lesion (NNC), and normal tissues remote to the cervical cancer lesion (NCx), such as those from the vagina and the endometrium, were procured during the radical hysterectomy. Tissue specimens were collected in multiple tubes and snap frozen in liquid nitrogen. This procedure was typically completed within 20 minutes after tissue resection. Overall, the studied tissues included 23 invasive cancers, 4 metastatic lesions, 11 normal tissues from noncancer patients, 11 NNCs, and 5 NCx s. All specimens were procured under an institutional review boardy approved project of the tissue banking system of Tzu-Chi General Hospital. Human Papillomavirus Test Infection of high-risk HPV in the specimens was detected by the Hybrid Capture 2 (HC2) test (Digene, Silver Spring, Md). In brief, the DNA in a specimen was denatured and hybridized with a cocktail of RNA probes directed against a panel of 13 high-risk HPVs. The RNA-DNA hybrids were then captured by hybrid-specific antibodies and detected by alkaline phosphataseylinked second antibody and chemiluminescence. The coefficients of variation of the triplicate tests of both the positive and negative controls were approximately 3% to 7%, and the ratio of the positive to negative mean was approximately 3.2:4.9. DNA Preparation, MSP, and Quantitative MSP Assay DNA of cervical scrapings was extracted using the QIAmp tissue kit (Qiagen, Hilden, Germany). For detection of DNA methylation, we used the EZ DNA Methylation-Gold Kit (Zymo Research, California) to convert cytocines to uracils, whereas methylated cytosines remained unmodified. For gel-based MSP of the PAX1 gene, bisulfite-converted DNA was PCR-amplified with methylation-specific primers (forward, 5 TATTTTGGGTTTGGGGTCGC3 ; reverse, 5 CCCGAAAACCGAAAACCG3 ) and nonymethylationspecific primers (forward: 5 GTTTATTTTGGGTTTG GGGTTGTG3 ; reverse: 5 CACCCAAAAACCAAAAA CCAC3 ) in a PCR master mix reagent Kit (Roche Diagnostic, Mannheim, Germany). The PCR program included an initiation at 94-C for 10 minutes, 40 cycles of annealing at 64-C for 30 seconds, extension at 72-C for 30 seconds, denaturing at 94-C for 30 seconds, and a final extension at 72-C for 7 minutes. The MSP products were resolved in 2.0% agarose gel containing ethidium bromide and visualized under UV light. For quantitative PCR of PAX1 methylation, primers and Taqman probe (TIB MOLBIOL, Berlin, Germany) specific to previously defined methylated CpG sites in cervical cancer 11 were designed. The forward primer (5 CGGTTAGACGAATTTTTTTTAATCGGATGA3 ) recognized 3 methylated CpG sites, the reverse primer (5 CCCGCGACCCCAAAC3 ) recognized 2 methylated CpG sites, and the Taqman probe (FAM-5 CGCCCGCTC CAAAACCTA3 -TAMARA) recognized 2 methylated CpG sites. The beta-actin gene (ACTB) was used as internal reference, with the sequences of forward primer, reverse primer, and Taqman probe as 5 TGGTGATGGAGGAGGTTTAG TAAGT3, 5 ACCACCACCCAACACACAATAACAAACA CA3, and FAM-5 ACCAATAAAACCTACTCCTCCCT TAA3 -TAMARA, respectively. Quantitative MSP (QMSP) analyses of PAX1 and ACTB were each performed by using the Roche LightCycler (Roche Diagnostics, Mannheim, Germany) and the Taqman kit (Roche Diagnostics) with the primers and the Taqman probes described previously. They were carried out with the same protocol consisting of a denaturing at 95-C for 10 minutes, followed by 50 cycles of denaturing at 95-C for 10 seconds and annealing/elongation at 60-C for 30 seconds. Genomic DNA of human peripheral blood leukocytes (white blood corpuscles [WBC]) treated with Sss-I methyltransferase (swbc) served as the fully methylated control and was used for calibration of PCR efficiency in serial dilutions. The efficiencies of QMSP for methylated PAX1 (E PAX1 ) and ACTB (E ACTB ) were and 2.053, respectively, indicating a high efficiency and good lineality of DNA amplification. Differences of quantities of methylated PAX1 and ACTB were measured in both test samples and Sss-1Ytreated WBC by the difference of the PCR cycle number needed to reach the crossing point or $ CP. The percentage of methylated reference (PMR) 17 of the PAX1 gene was calculated by the equation proposed by Pfaffl 18 : PMR = (E PAX1 ) $CP (swbcjsample)pax1 /(E ACTB ) $CP (swbcjsample)actb 100. FIGURE 1. Comparison of PAX1 hypermethylation in malignant and nonmalignant neoplasias. The PMR values obtained by QMSP were transformed as a function of ln(pmr + 1) and plotted according to clinical status. A horizontal bar indicating the mean (SE) of each group. The dotted line is the cutoff at a transformed value of 1.7, corresponding to an untransformed value (PMR = 4.5). * 2010 IGCS and ESGO 515

4 Huang et al International Journal of Gynecological Cancer & Volume 20, Number 4, May 2010 TABLE 2. The status of PAX1 methylation and HPV infection in noncancer conditions and invasive cervical cancer n PAX1 QMSP* P High-Risk HPV P Invasive cervical cancer % (14/14) G % (14/14) Noncancer conditions % (9/59) 47.5% (28/59) Healthy 17 0% (0/17) % (1/17) CIN % (1/10) 70% (7/10) CIN % (2/18) 55.6% (10/18) CIN % (6/14) 71.4% (10/14) *Determined by QMSP with a natural logarithm-transformed cut-off-value of 1.7. Determined by HC2 test. Statistical Analysis GraphPad Prism (version 5.0a) software was used for statistical analyses and data plotting. The PMR values obtained by QMSP were transformed as a function of ln(pmr + 1) and plotted according to clinical status to get the cutoff point that could best discriminate disease versus nondisease. The Mann-Whitney U test was used to analyze the distribution of age of the study subjects and the PMR values of the PAX1 gene. The Fisher exact and W 2 tests for trend were used to analyze the status of PAX1 methylation or HPV infection in the different groups. RESULTS Cervical Scrapings of Cancer and CIN3 Patients Had High PMR Values of PAX1 The proportion of methylation of the PAX1 gene was determined by QMSP in cervical scrapings obtained from the patients with various degrees of cervical neoplasia and from the healthy controls. The PMR value of PAX1 varied widely, depending on the extent of cervical dysplasia, and displayed a tendency closely related to the progression of cervical carcinogenesis (Table 1). Specifically, invasive cervical cancers and CIN3 had significant higher PMR values than the groups with less severe disease (P G ). A Cutoff Value of RMR Readily Distinguish Malignant From Nonmalignant Cervical Neoplasia Based on the pooled data of PAX1 methylation rates in the subjects with different severities of cervical neoplasia, the cutoff value that best distinguishes cancerous (CIS and invasive cancer) from noncancerous (CIN and normal tissue) conditions of the uterine cervix was determined. At a PMR cutoff value of 4.5, 100% of the invasive cancer and 43% of the CIN3 (including CIS) specimens were positive for PAX1 methylation compared with 0% for the normal, 10% for CIN1, and 11% for CIN2 specimens (Fig. 1, Table 2). At this cutoff value, specimens positive for invasive cancer were more significant than the noncancerous specimens (P G ). As a comparison, the HC2 HPV test result was positive in 100% of the invasive cancer and 47.5% of noncancerous specimens. Noteworthily, HPV was present in 70% and 5.9% of the benign conditions of CIN1 and the normal 516 tissues, respectively, but PAX1 methylation was present in only 10% and 0%, respectively (Table 2). PAX1 Methylation Test Was as Sensitive and Much More Specific Than the HC2 HPV Test in Detection of Cancer in the Studied Cohort We compared the performance of QMSP for PAX1 to the HC2 HPV test on the cervical scraping of the same cohort of cervical neoplasia. In the detection of invasive cervical cancer, both tests had 100% sensitivity and negative predictive value, but PAX1 methylation was more specific (84.7% vs 52.5%) and had a higher positive predictive value (60.9% vs 33.3%) than the HC2 test. In the detection of CIN3 and invasive cervical cancer, the sensitivity, the specificity, and the positive and negative predictive values of PAX1 methylation and HPV tests were respectively 71.4% vs 85.7%, 93.3% vs 60.0%, 87.0% vs 57.1%, and 84.0% vs 87.1% (Table 3). A Field Effect of PAX1 Methylation in Cervical Cancer and Adjacent Normal Tissues may Facilitate the Detection of Cancer-Associated Methylation From Cervical Scrapings Primary and metastatic cancer tissues and normal tissues adjacent to or remote from the primary cancer lesion TABLE 3. Performances of the PAX1 QMSP and HC2 tests in detection of CIN3 or invasive cervical cancer Sensitivity, % Specificity, % PPV, % NPV, % Detection of invasive cervical cancer PAX1 QMSP HPV test Detection of CIN3 or invasive cervical cancer PAX1 QMSP HPV test PPV, positive predictive value; NPV, negative predictive value. * 2010 IGCS and ESGO

5 International Journal of Gynecological Cancer & Volume 20, Number 4, May 2010 PAX1 Methylation Versus HPV Test TABLE 4. Topological controls used to test the field effect of PAX1 methylation Identification No. Primary Cancer Topological Control Specimens Diagnosis Site Methylation Procured Site Type Methylation SCC of cervix, stage IVB Cervix Positive Vagina MC Positive Adenocarcinoma of the cervix, stage IIA Cervix Positive Pelvic LN MC Positive SCC of the cervix, stage IIIA at least Cervix Positive Vagina MC Positive Recurrent SCC of the cervix, with Cervix N/A Common iliac LN MC Positive lymph node metastasis SCC of the cervix, stage IIA Cervix Positive Cervix NNC Positive SCC of the cervix, stage IIA Cervix Positive Cervix NNC Positive SCC of the cervix, stage 2a Cervix Positive Cervix NNC Positive SCC of the cervix, stage IIA Cervix Positive Cervix NNC Positive SCC of the cervix, stage IB Cervix Positive Cervix NNC Positive SCC of the cervix, stage IB 2 Cervix Positive Cervix NNC Positive Stroma under SCC NNC Positive SCC of the cervix, stage IIA Cervix Positive Cervix NNC Positive Adenocarcinoma of the cervix, stage IB Cervix Positive Cervix NNC Negative Poorly differentiated adenosquamous Cervix Positive Vagina NNC Positive carcinoma of the cervix, stage IB SCC of the cervix, stage IA Cervix N/A Cervix NNC Negative SCC of the cervix, stage IIA Cervix Positive Endometrium NCx Positive SCC of the cervix, stage IB 2 Cervix Positive Ovary NCx Negative Pelvic LN NCx Positive Adenocarcinoma of the cervix, stage IB 2 Cervix Positive Common iliac LN NCx Negative Neuroendocrine-differentiated carcinoma of the cervix, stage IVB Cervix Negative Vagina NCx Negative LN, lymph node; MC, metastatic cancer; N/A: not available; SCC, squamous cell carcinoma. were procured from patients who underwent radical hysterectomy for invasive cervical carcinoma and tested for PAX1 methylation by MSP (Table 4). As shown in Table 5, methylation of PAX1 was found in 1 of 11 normal cervical tissues from noncancer patients (normal next to tumor), 2 of 5 normal tissues remote to cervical cancer (NCx), 9 of 11 normal cervical tissues adjacent to cervical cancer (NNC), 21 of 23 primary cervical cancer tissues, and 4 of 4 metastatic cervical cancer tissues. DISCUSSION In the development of cervical cancer, the PAX1 gene is silenced by hypermethylation. 11 Using a methylation-specific PCR test of cervical scrapings, we have previously shown that PAX1 methylation was specifically found in most invasive cervical cancer specimens, in half of CIN2 or CIN3, and rarely in CIN1 and normal specimens in a hospital-based cohort in North Taiwan. 11 In this study, QMSP was conducted to examine the extent of methylation in another cohort in East Taiwan. We found a progressive increase in the level of PAX1 methylation in cervical scrapings in the spectrum of cervical neoplasia and a marked increase of methylation levels in CIN3 and invasive cancer. The methylation silencing of the PAX1 gene during cervical cancinogenesis suggests a tumor suppressor role of this paired boxed (PAX) gene. The PAX gene family has been classified into 4 classes. 19Y21 Members of the class 1 and 3 PAX genes are proposed to play an essential role in maintaining tissue-specific stem cells and have been found to be overexpressed in the development of several cancers. 19Y21 Members of classes 1 and 4, where PAX1, PAX4, PAX6, and TABLE 5. Methylation of PAX1 in cancer lesions and normal tissues of different topological relationship to cancer Tissue Specimen n Methylation % P NN NCx NNC Invasive cancer Metastatic cancer E-06 NN, normal tissue from noncancer patient. * 2010 IGCS and ESGO 517

6 Huang et al International Journal of Gynecological Cancer & Volume 20, Number 4, May 2010 PAX9 belong to, are much less overexpressed in cancer. 22 In fact, PAX6 and PAX9 have been reported to be associated with favorable clinical outcomes, 23,24 and PAX6 has a tumorsuppressive function in glioblastoma cells. 25 Although the structural differences between these 2 groups of PAX genes may explain the contradictory functions in tumorigenesis, 22 we hypothesize that members of classes 1 and 4 of the PAX gene family function as tumor suppressors in organs in which they do not normally have a developmental role. Recently, PAX4, known to be important in the development of gastrointestinal system, was reported to function as a tumor suppressor in human melanoma. 26 Our study suggested a potential tumor suppressor role for PAX1 in cervical carcinogenesis, given its normal function in the development of the skeleton, thymus, and parathyroid glands. The quantitative measurement of methylated gene in reference with a fully methylated standard by using real-time PCR is merited by its high sensitivity and consistency. 17,27 The specificity of the CpG site detection can be further assured by the Taqman probe. In our design, the forward and reverse primers recognized 3 and 2 methylated CpG sites, respectively, and 2 additional sites were recognized by the Taqman probe. Furthermore, measurement of fluorescence-labeled amplicons allows for accurate, sensitive, and quantitative measurement of the amplicons. In this studied hospital-based cases and controls, at a selected cutoff value for the QMSP test, all the normal and most of the CIN1 and CIN2 specimens were negative for PAX1 gene methylation, whereas all the invasive cervical cancer specimens were positive. Cervical scraping has long been a target for testing for cervical neoplasia and has been used here to detect cancerspecific gene methylation. The low representation of cancer cells in the cervical scrapings may not be of concern because the epigenetic change in cancer may act in a field. 28,29 Indeed, as we have shown here and previously, 27 methylation of PCDH10 and PAX1 genes acted in a field in cervical cancer and in the grossly normal-appearing adjacent tissues but not in the remote tissues. The field methylation may be a reflection of carcinogenic environment such as inflammation or hormone exposure or the stromal reaction that precedes the invasion of cancer. 30 The field methylation of genes related to cancer invasion helps to a sensitive detection of it in cervical scrapings. It also provides a possibility of testing of cervicovaginal secretions self-sampled by women, as that has been done for HPV test. The feasibility of this deserves further investigation. In current practice, the Papanicolaou test is used to screen for cervical cancer and the HPV test can facilitate the diagnosis when performed on the scrapings remaining after the Papanicolaou test. With the aid of the HPV test, which is highly sensitive and has a high negative predictive rate, screening with Papanicolaou test can be more sensitive and greatly increase its screening interval. 31 Testing for HPV is also helpful in prediction and guiding the management of equivocal or low-grade Papanicolaou test results. However, owing to its low specificity and high false positive rate, the HPV test cannot be used alone as a screening or diagnostic tool. As shown in this study, the methylation test of the PAX1 gene was both sensitive and specific in detecting cervical 518 cancer from cervical scrapings and thus can be a good candidate for in vitro testing of cervical cancer. In summary, in this hospital-based study, the PAX1 QMSP test can be as sensitive as but much more specific than the HPV test in detecting cervical cancer. Further studies in larger cohorts and in screening/postscreening populations are warranted to prove its efficacy in diagnosis and screening. REFERENCES 1. Parkin DM. The global health burden of infection-associated cancers in the year Int J Cancer. 2006;118:3030Y Levi AW, Kelly DP, Rosenthal DL, et al. Atypical squamous cells of undetermined significance in liquid-based cytologic specimens: results of reflex human papillomavirus testing and histologic follow-up in routine practice with comparison of interpretive and probabilistic reporting methods. Cancer. 2003;99:191Y Huang YK, You SL, Yuan CC, et al. Long-term outcomes of high-risk human papillomavirus infection support a long interval of cervical cancer screening. Br J Cancer. 2008;98: 863Y Dillner J, Rebolj M, Birembaut P, et al. Long term predictive values of cytology and human papillomavirus testing in cervical cancer screening: joint European cohort study. BMJ. 2008;337:a Jones P, Laird P. Cancer epigenetics comes of age. Nat Genet. 1999;21:163Y Jones PA, Baylin SB. The epigenomics of cancer. Cell. 2007; 128:683Y Steenbergen RD, Kramer D, Braakhuis BJ, et al. TSLC1 gene silencing in cervical cancer cell lines and cervical neoplasia. J Natl Cancer Inst. 2004;96:294Y Feng Q, Balasubramanian A, Hawes SE, et al. Detection of hypermethylated genes in women with and without cervical neoplasia. J Natl Cancer Inst. 2005;97:273Y Lai HC, Lin YW, Chang CC, et al. Hypermethylation of two consecutive tumor suppressor genes, BLU and RASSF1A, located at 3p21.3 in cervical neoplasias. Gynecol Oncol. 2007;104:629Y Kahn SL, Ronnett BM, Gravitt PE, et al. Quantitative methylation-specific PCR for the detection of aberrant DNA methylation in liquid-based Pap tests. Cancer. 2008;114: 57Y Lai H, Lin Y, Huang T, et al. Identification of novel DNA methylation markers in cervical cancer. Int J Cancer. 2008;123:161Y McGaughran JM, Oates A, Donnai D, et al. Mutations in PAX1 may be associated with Klippel-Feil syndrome. Eur J Hum Genet. 2003;11:468Y Wallin J, Eibel H, Neubuser A, et al. Pax1 is expressed during development of the thymus epithelium and is required for normal T-cell maturation. Development. 1996;122:23Y Schnittger S, Rao V, Deutsch U, et al. Pax1, a member of the paired box-containing class of developmental control genes, is mapped to human chromosome 20p11.2 by in situ hybridization (ISH and FISH). Genomics. 1992;14:740Y Su HY, Lai HC, Lin YW, et al. An epigenetic marker panel for screening and prognostic prediction of ovarian cancer. Int J Cancer. 2009;124:387Y Eads C, Danenberg K, Kawakami K, et al. MethyLight: a high-throughput assay to measure DNA methylation. Nucleic Acids Res. 2000;28:E32. * 2010 IGCS and ESGO

7 International Journal of Gynecological Cancer & Volume 20, Number 4, May 2010 PAX1 Methylation Versus HPV Test 17. Ogino S, Kawasaki T, Brahmandam M, et al. Precision and performance characteristics of bisulfite conversion and real-time PCR (MethyLight) for quantitative DNA methylation analysis. J Mol Diagn. 2006;8:209Y Pfaffl MW. A new mathematical model for relative quantification in real-time RT-PCR. Nucleic Acids Res. 2001;29:e Lang D, Powell SK, Plummer RS, et al. PAX genes: roles in development, pathophysiology, and cancer. Biochem Pharmacol. 2007;73:1Y Galindo RL, Allport JA, Olson EN. A Drosophila model of the rhabdomyosarcoma initiator PAX7-FKHR. Proc Natl Acad Sci U S A. 2006;103:13439Y Souabni A, Jochum W, Busslinger M. Oncogenic role of Pax5 in the T-lymphoid lineage upon ectopic expression from the immunoglobulin heavy-chain locus. Blood. 2007;109:281Y Robson EJ, He SJ, Eccles MR. A PANorama of PAX genes in cancer and development. Nat Rev Cancer. 2006;6:52Y Zhou YH, Tan F, Hess KR, et al. The expression of PAX6, PTEN, vascular endothelial growth factor, and epidermal growth factor receptor in gliomas: relationship to tumor grade and survival. Clin Cancer Res. 2003;9:3369Y Gerber JK, Richter T, Kremmer E, et al. Progressive loss of PAX9 expression correlates with increasing malignancy of dysplastic and cancerous epithelium of the human oesophagus. J Pathol. 2002;197:293Y Zhou YH, Wu X, Tan F, et al. PAX6 suppresses growth of human glioblastoma cells. J Neurooncol. 2005;71: 223Y Hata S, Hamada J, Maeda K, et al. PAX4 has the potential to function as a tumor suppressor in human melanoma. Int J Oncol. 2008;33:1065Y Coleman WB, Rivenbark AG. Quantitative DNA methylation analysis: the promise of high-throughput epigenomic diagnostic testing in human neoplastic disease. J Mol Diagn. 2006;8:152Y Wang KH, Liu HW, Lin SR, et al. Field methylation silencing of the protocadherin 10 gene in cervical carcinogenesis as a potential specific diagnostic test from cervical scrapings. Cancer Sci. 2009;100:2175Y Hanson JA, Gillespie JW, Grover A, et al. Gene promoter methylation in prostate tumoryassociated stromal cells. J Natl Cancer Inst. 2006;98:255Y Jones PA, Baylin SB. The epigenomics of cancer. Cell. 2007;128:683Y Huang Y, You S, Yuan C, et al. Long-term outcomes of high-risk human papillomavirus infection support a long interval of cervical cancer screening. Br J Cancer. 2008;98: 863Y869. * 2010 IGCS and ESGO 519