DETECTION AND PROGNOSIS OF CERVICAL CANCER

FIELD OF THE INVENTION

The present invention relates to the area of cancer diagnostics and therapeutics. In 5 particular, it relates to methods and kits for identifying, diagnosing, prognosing, and monitoring cervical cancer. These methods include determining the methylation status or the expression levels of particular genes, or a combination thereof.

BACKGROUND TO THE INVENTION

Cervical cancer is the fifth most deadly cancer in women. Worldwide, approximately 10 500,000 cases of cervical cancer are diagnosed and about 250,000 women die from this disease annually (www.who.int/mediacentre/factsheets).

Most (80-90%) invasive cervical cancer develops in flat, scaly surface cells that line the cervix (called squamous cell carcinomas, SCC). Approximately 10-15% of cases develop in glandular surface cells (called adenocarcinomas, AdC). Less commonly, 15 cervical cancers have features of both SCC and AdC. These are called adenosquamous carcinomas or mixed carcinomas (www.cancer.org).

During the process of cervical cancer development, normal cervical cells gradually develop pre-cancerous changes that turn into cancer. Cervical cancer evolves from pre-existing noninvasive premalignant lesions referred to as cervical intraepithelial 20 neoplasias (CINs), ranging from CIN I (mild dysplasia) to CIN II (moderate dysplasia) to CIN III (severe dysplasia/carcinoma in situ). This process usually takes several years but sometimes can happen in less than a year. For most women, pre-cancerous cells will remain unchanged and disappear without any treatment.

Screening for malignant and premalignant disorders of the cervix is usually performed

25 according to the Papanicolaou (PAP) system. The cervical smears are examined by light microscopy and the specimens containing morphologically abnormal cells are classified into PAP I to V, at a scale of increasing severity of the lesion. But, present PAP test has some limitations and is not completely ideal for screening as it suffers from suboptimal single-test sensitivity, limited reproducibility, and many equivocal.

30 There is a strong association between certain subtypes of the Human Papillomavirus

(HPV) and cervical cancer. Studies have shown that only high-risk HPV types are

involved in the progression from cyto logical normal cervix cells to high grade squamous intraepithelial lesions. Around 15 high-risk (cancer-causing) HPV types have been identified. Although it has been suggested that high-risk HPV testing may improve cervical cancer screening, the specificity for high grade cervical neoplasia of high risk HPV testing is relatively low. This low specificity of HPV testing leads to a higher number of unnecessarily follow-up diagnostic workups (e.g. colposcopy) and unnecessarily treatment with cryotherapy or loop electro surgical excision procedure, which permanently alters the cervix and have unknown consequences on fertility and pregnancy.

To improve early detection, the combination of HPV and PAP tests is now approved by the FDA for screening women 30 years of age and older. However, co-testing substantially increases the cost of screening.

In the meanwhile, vaccines for preventing cervical cancer have been developed and one has already been approved by the FDA. But, immunization will only protect against HPV types that are targeted by the vaccine; protection will not be absolute and its longevity is uncertain; as yet, the possibility of genotype replacement cannot be excluded; and older women not covered by vaccination programs will continue to be at risk. Therefore, cervical screening will still be required for control.

Cancer bio markers have been described in literature and aberrant methylation of genes has been linked to cervical cancer (Virmani et al, 2001). In addition, methylation markers may serve for predictive purposes as they often reflect the sensitivity to therapy or duration of patient survival.

DNA methylation is a chemical modification of DNA performed by enzymes called methyltransferases, in which a methyl group (m) is added to certain cytosines (C) of DNA. This non-mutational (epigenetic) process (mC) is a critical factor in gene expression regulation. (See J.G. Herman, Seminars in Cancer Biology, 9: 359-67, 1999).

An early diagnosis is critical for the successful treatment of many types of cancer, including cervical cancer. If the exact methylation profiles of cervical tumors are available and drugs targeting the specific genes are obtainable, then the treatment of cervical cancer could be more focused and rational. Therefore, the detection and mapping of novel methylation markers is an essential step towards improvement of cervical cancer prevention, screening, and treatment. Thus, there is a continuing need

in the art to identify methylation markers that can be used for improved assessment of cervical cancer.

SUMMARY OF THE INVENTION

The present invention is based on the finding that several genes are identified as being differentially methylated in cervical cancers. This information is useful for cervical cancer screening, risk-assessment, prognosis, disease identification, disease staging, and identification of therapeutic targets. The identification of new genes that are methylated in cervical cancer allows accurate and effective early diagnostic assays, methylation profiling using multiple genes and identification of new targets for therapeutic intervention.

Accordingly, in a first aspect, the invention provides a method for identifying cervical cancer or its precursor, or predisposition to cervical cancer. Epigenetic modification of at least one gene selected from the group consisting of genes according to Table 1, is detected in a test sample containing cervical cells or nucleic acids from cervical cells. The test sample is identified as containing cells that are neoplastic, precursor to neoplastic, or predisposed to neoplasia, or as containing nucleic acids from cells that are neoplastic, precursor to neoplastic, or predisposed to neoplasia.

Preferably, the at least one gene is selected from a group of genes consisting of JAM3, LMXlA, CDOl, NID2, ALX3, ALX4, AR, ARID4A, ATM, AURKA, B4GALT1, BMP2, BMP6, BNIP3, C13orfl8, C16orf48, C9orfl9, CALCA, CAMK4, CCNAl, CCND2, CDHl, CDH4, CDK6, CDKNlB, CDKN2B, CLSTN2, CLU, COLlAl, CPTlC, CTDSPL, CYCLIND2, DAPKl, DBCl, DDXl 9B, DKK2, EGFR, EGR4, EPB41L3, FOS, FOXEl, GADD45A, GATA4, GDAPlLl, GNB4, GPNMB, GREMl, Gst-Pi, HHIP, HINl, H00K2, HOXAl, HOXAl 1, H0XA7, HOXDl, IGSF4, ISYNAl, JPH3, KNDCl, KRAS, LAMAl, LOC285016, LOX, LTB4R, MAL,

MTAP, MY018B, NDRG2, N0L4, NPTXl, NPTX2, 0GF0D2, PAK3, PAXl, PDCD4, PHACTR3, POMC, PRKCE, RAD23B, RALY, RARA, RASSFlA, RBP4, RECK, RPRM, SALL4, SEMA3F, SLC5A8, SLITl, SLIT2, SLIT3, SMPDl, SOCSl, SOXl, S0X17, SPARC, SPN, SST, TACl, TERT, TFPI-2, TLLl, TNFAIPl, TRMTl, TWISTl, UGTlAl, WIFl, WITl, WTl, XRCC3, and ZGPAT.

In one embodiment of the present invention, the detection of epigenetic modification comprises detection of methylation of a CpG dinucleotide motif in the gene and/or promoter region of the gene; and/or detection of expression of mRNA of the gene.

The invention also relates to a kit for assessing cervical cancer or its precursor, or predisposition to cervical cancer in a test sample containing cervical cells or nucleic acids from cervical cells. The kit comprises in a package: a reagent that (a) modifies methylated cytosine residues but not non-methylated cytosine residues, or that (b) 5 modifies non-methylated cytosine residues but not methylated cytosine residues; and at least one pair of oligonucleotide primers that specifically hybridizes under amplification conditions to a region of a gene selected from the group consisting of genes according to Table 1 and/or the aforementioned group of genes. The region is preferably within aboutlO kbp of said gene's transcription start site.

0 In a further aspect, the invention provides for oligonucleotide primers and/or probes and their sequences for use in the methods and assays of the invention.

The invention also relates to screening protocols for the screening of woman for cervical cancer and the precursors thereof. Such method for cervical cancer screening combines hr-HPV testing and methylation testing, or combines PAP tests with 5 methylation testing. Methylation testing in such screening method preferably detects the epigenetic modification of at least one gene selected from the group consisting of JAM3, LMXlA, CDOl, NID2, ALX3, ALX4, AR, ARID4A, ATM, AURKA, B4GALT1, BMP2, BMP6, BNIP3, C13orfl8, C16orf48, C9orfl9, CALCA, CAMK4, CCNAl, CCND2, CDHl, CDH4, CDK6, CDKNlB, CDKN2B, CLSTN2, CLU, 0 COLlAl, CPTlC, CTDSPL, CYCLIND2, DAPKl, DBCl, DDXl 9B, DKK2, EGFR, EGR4, EPB41L3, FOS, FOXEl, GADD45A, GATA4, GDAPlLl, GNB4, GPNMB, GREMl, Gst-Pi, HHIP, HINl, H00K2, HOXAl, HOXAI l, H0XA7, HOXDl, IGSF4, ISYNAl, JPH3, KNDCl, KRAS, LAMAl, LOC285016, LOX, LTB4R, MAL, MTAP, MY018B, NDRG2, N0L4, NPTXl, NPTX2, 0GF0D2, PAK3, PAXl, 5 PDCD4, PHACTR3, POMC, PRKCE, RAD23B, RALY, RARA, RASSFlA, RBP4,

RECK, RPRM, SALL4, SEMA3F, SLC5A8, SLITl, SLIT2, SLIT3, SMPDl, SOCSl, SOXl, SOX17, SPARC, SPN, SST, TACl, TERT, TFPI-2, TLLl, TNFAIPl, TRMTl, TWISTl, UGTlAl, WIFl, WITl, WTl, XRCC3, and ZGPAT. Dependent on the outcome, the women screened for cervical cancer is referred for colposcopy, or O referred for hr-HPV and/or PAP testing and/or methylation testing on a more regular basis.

Epigenetic loss of gene function can be rescued by the use of DNA demethylating agents and/or DNA methy transferase inhibitors. Accordingly, the invention also provides for a method for predicting the likelihood of successful treatment or 5 resistance to treatment of cancer with such agent. If the gene is methylated, the

likelihood of successful treatment is higher than if the gene is unmethylated, or methylated to a lesser degree. Conversely, if the gene is unmethylated, or methylated to a lesser degree, the likelihood of resistance to treatment is higher than if the gene is methylated.

In a related aspect, epigenetic loss of gene function(s) can identify the stage of the disease and from that the need of treatment. Accordingly, the invention provides for a method for predicting suitable treatment comprising determining the methylation status of a gene or a combination of genes. If the gene is methylated, the need of cervical resection is identified; if the gene is unmethylated or methylated to a lesser degree, it is decided that there is no need for cervical resection.

SUMMARY OF THE FIGURES

Figure 1: The number of probes (w) that is retrieved using parameters x (number of P- calls in primary cancers for probe), y (number of P-calls in untreated cell- lines for probe) and z (number of P-calls in treated cell-lines for probe).

Figure 2: Step-plot to determine optimal number of probes for further analysis. Step- plot of the number of retrieved known markers as a function of the position after relaxation ranking (this is the number of selected probes after ranking). The step plot shows the actual (observed) number of markers. If the markers were randomly distributed, one would expect the profile, marked with 'expected' (details in the text). The trend of the observed markers versus the number of selected probes is indicated with dashed lines.

Figure 3: (Hyper) methylation analysis of the promoter region (-430 to -5 of TSS) of the CCNAl gene by COBRA and sequence analysis.

A: schematic representation of the restriction enzyme sites (B: BstUI and T: Taql) in the virtual hypermethylated BSP nucleotide sequence after bisulfite treatment.

Vertical bars represent CG site, arrow represents TSS (retrieved from Ensembl). B: Result of COBRA analysis of the BSP products of 10 tumor samples (Tl-TlO), in vitro methylated DNA as a positive control (IV) and leukocyte DNA as a negative (unmethylated) control (L). C: Schematic representation of the sequencing results. From each tumor, the BSP- products were cloned into TOPO-pCR4 (Invitrogen) and sequencing (BaseClear) was performed on M13-PCR products of 7-9 independent clones. Circles represent CG dinucleo tides: the darker, the more clones at this site were methylated.

Figure 4: Representative COBRA on 3 gene promoters (SST, AUTS2 and SYCP3). A: schematic representation of the restriction enzyme sites in the virtual hypermethylated BSP nucleotide sequence after bisulfite treatment. (B: BstUI, T: Taql and H: Hinfl). Bars represent CG site and arrow is TSS (retrieved from Ensembl). B: Result of COBRA analysis of BSP products of tumor samples (Tl-TlO) and 5 normal cervices (N1-N5), in vitro methylated DNA as a positive control (IV) and leukocyte DNA as a negative (unmethylated) control (L); lane B is water blank.

Figure 5:

A: Position of the different primers relative to the TSS (transcription start site). Multiple primer designs are displayed by blue boxes and red boxes (=fmal primer pairs retained for the assays). The exon of ALX4 is indicated in green. The number of CpG count is spotted in blue over a region of 20Kb.

B: List of sequences for the different primer sets, converted and unconverted amplicon sequences used in Figure 5 A. Figure 6: Ranked methylation table from the Lightcycler platform. 27 methylation profiles from cervical cancer samples (left) are compared against 20 normal tissue samples (right). Samples are shown along the X-axis where each vertical column represents the methylation profile of one individual sample across the 63 different assays (Y-axis). Assays demonstrating the best methylation discriminators between the 2 groups are displayed at the top, with discrimination effect decreasing towards the bottom. The black boxes indicate the methylated results; grey boxes indicate the unmethylated results; white boxes indicate invalid results. (NA: not applicable; NT: not tested)

Figure 7: Amplification plot for the standard curve for TAC1_56187 Figure 8: Amplification plot for standard curve and samples for TAC1_56187 Figure 9: Linear regression of standard curve for TAC1_56187 Figure 10: Decision tree for ratio determination

Figure 11: Performance of the individual markers on cervical tissue samples using qMSP.

DETAILED DESCRIPTION OF THE INVENTION

We describe a new sorting methodology to enrich for genes which are silenced by promoter methylation in human cervical cancer. The pharmacological unmasking expression microarray approach is an elegant method to enrich for genes that are

silenced and re-expressed during functional reversal of DNA methylation upon treatment with demethylating agents. However, such experiments are performed in in vitro (cancer) cell lines mostly with poor relevance when extrapolating to primary cancers. To overcome this problem, we incorporated data from primary cancer samples in the experimental design. A pharmacological unmasking microarray approach was combined with microarray expression data of primary cancer samples. For the integration of data from both cell lines and primary cancers, we developed a novel ranking strategy, which combines reactivation in cell lines and no expression in primary cancer tissue.

We also used a Genome-wide Promoter Alignment approach with the capacity to define a further substantial fraction of the cancer gene promoter CpG island DNA methylome. Markers clustering with known methylation markers might indicate towards common mechanisms underlying the methylation event and thus identify novel genes that are more methylation-prone.

Studies of the genes defined by the different approaches will contribute to understanding the molecular pathways driving tumorigenesis, provide useful new DNA methylation biomarkers to monitor cancer risk assessment, early diagnosis, and prognosis, and permit better monitoring of gene re-expression during cancer prevention and/or therapy strategies.

Using the aforementioned techniques, we have identified cytosines within CpG dinucleotides of DNA from particular genes isolated from a test sample, which are differentially methylated in human cervical cancer tissue samples and normal cervical tissue control samples. The cancer tissues samples are hypermethylated or hypomethylated with respect to the normal samples (collectively termed epigenetic modification). The differential methylation has been found in genomic DNA of at least one gene selected from the group consisting of JAM3, LMXlA, CDOl, NID2, ALX3, ALX4, AR, ARID4A, ATM, AURKA, B4GALT1, BMP2, BMP6, BNIP3, C13orfl8, C16orf48, C9orfl9, CALCA, CAMK4, CCNAl, CCND2, CDHl, CDH4, CDK6, CDKNlB, CDKN2B, CLSTN2, CLU, COLlAl, CPTlC, CTDSPL, CYCLIND2, DAPKl, DBCl, DDX19B, DKK2, EGFR, EGR4, EPB41L3, FOS,

FOXEl, GADD45A, GAT A4, GDAPlLl, GNB4, GPNMB, GREMl, Gst-Pi, HHIP, HINl, H00K2, HOXAl, HOXAI l, H0XA7, HOXDl, IGSF4, ISYNAl, JPH3, KNDCl, KRAS, LAMAl, LOC285016, LOX, LTB4R, MAL, MTAP, MYO 18B, NDRG2, N0L4, NPTXl, NPTX2, OGFOD2, PAK3, PAXl, PDCD4, PHACTR3, POMC, PRKCE, RAD23B, RALY, RARA, RASSFlA, RBP4, RECK, RPRM,

SALL4, SEMA3F, SLC5A8, SLITl, SLIT2, SLIT3, SMPDl, SOCSl, SOXl, SOX17, SPARC, SPN, SST, TACl, TERT, TFPI-2, TLLl, TNFAIPl, TRMTl, TWISTl, UGTlAl, WIFl, WITl, WTl, XRCC3, and ZGPAT.

Accordingly, in a first aspect, the invention provides a method for identifying cervical 5 cancer or its precursor, or predisposition to cervical cancer. Epigenetic modification of at least one gene selected from the group consisting of genes according to Table 1, is detected in a test sample containing cervical cells or nucleic acids from cervical cells. The test sample is identified as containing cells that are neoplastic, precursor to neoplastic, or predisposed to neoplasia, or as containing nucleic acids from cells that 0 are neoplastic, precursor to neoplastic, or predisposed to neoplasia.

Preferably, the at least one gene is selected from a group of genes consisting of JAM3, LMXlA, CDOl, NID2, ALX3, ALX4, AR, ARID4A, ATM, AURKA, B4GALT1, BMP2, BMP6, BNIP3, C13orfl8, C16orf48, C9orfl9, CALCA, CAMK4, CCNAl, CCND2, CDHl, CDH4, CDK6, CDKNlB, CDKN2B, CLSTN2, CLU, COLlAl,5 CPTlC, CTDSPL, CYCLIND2, DAPKl, DBCl, DDXl 9B, DKK2, EGFR, EGR4, EPB41L3, FOS, FOXEl, GADD45A, GATA4, GDAPlLl, GNB4, GPNMB, GREMl, Gst-Pi, HHIP, HINl, H00K2, HOXAl, HOXAl 1, H0XA7, HOXDl, IGSF4, ISYNAl, JPH3, KNDCl, KRAS, LAMAl, LOC285016, LOX, LTB4R, MAL, MTAP, MY018B, NDRG2, N0L4, NPTXl, NPTX2, 0GF0D2, PAK3, PAXl, O PDCD4, PHACTR3, POMC, PRKCE, RAD23B, RALY, RARA, RASSFlA, RBP4, RECK, RPRM, SALL4, SEMA3F, SLC5A8, SLITl, SLIT2, SLIT3, SMPDl, SOCSl, SOXl, S0X17, SPARC, SPN, SST, TACl, TERT, TFPI-2, TLLl, TNFAIPl, TRMTl, TWISTl, UGTlAl, WIFl, WITl, WTl, XRCC3, and ZGPAT.

Preferably, at least one gene is selected from the group consisting of JAM3, LMXlA, 5 CDOl, NID2, CCNAl, HOXAI l, GREMl and TACl. Preferably, epigenetic silencing of a gene combination is detected and preferably selected from the group of gene combinations consisting of:

• NID2 and HOXAI l;

• JAM3, CDOl, HOXAI l, and CCNAl; 0 • JAM3 and HOXAI l;

• JAM3 , HOXAl 1 and GREM 1 ;

• JAM3, NID2, HOXAl 1 and CDOl ;

• JAM3, TACl, HOXAI l, and CDOl;

• JAM3, HOXAI l, and CDOl; 5 • JAM3 and CDOl;

• JAM3 and NID2;

• NID2 and CDOl;

• JAM3 and LMXlA

• NID2 and LMXlA, and 5 • JAM3, CDOl and NID2

"Identifying" a disease or predisposition of disease is defined herein to include detecting by way of routine examination, screening for a disease or pre-stadia of a disease, monitoring staging and the state and/or progression of the disease, checking for recurrence of disease following treatment and monitoring the success of a 0 particular treatment. The identification can also have prognostic value, and the prognostic value of the tests can be used as a marker of potential susceptibility to cancer.

The term "Epigenetic modification" can be described as a stable alteration in gene expression potential that takes place during development and cell proliferation, 5 mediated by mechanisms other than alterations in the primary nucleotide sequence of a gene. Three related mechanisms that cause alteration in gene expression are recognized: DNA methylation, histone code changes and RNA interference.

Epigenetic modification of a gene can be determined by any method known in the art. One method is to determine that a gene which is expressed in normal cells or other 0 control cells is less expressed or not expressed in tumor cells. Diminished gene expression can be assessed in terms of DNA methylation status or in terms of expression levels as determined by their methylation status, generally manifested as hypermethylation. Conversely, a gene can be more highly expressed in tumor cells than in control cells in the case of hypomethylation. This method does not, on its own, 5 however, indicate that the silencing or activation is epigenetic, as the mechanism of the silencing or activation could be genetic, for example, by somatic mutation. One method to determine that silencing is epigenetic is to treat with a reagent, such as DAC (5'-deazacytidine), or with a reagent which changes the histone acetylation status of cellular DNA or any other treatment affecting epigenetic mechanisms present in cells, O and observe that the silencing is reversed, i.e., that the expression of the gene is reactivated or restored.

Another means to determine epigenetic modification is to determine the presence of methylated CpG dinucleotide motifs in the silenced gene or the absence of methylation CpG dinucleotide motifs in the activated gene. In one embodiment, epigenetic

modification of a CpG dinucleotide motif in the promoter region of the at least one gene selected from a group of genes according to Table 1 is determined. Methylation of a CpG island at a promoter usually prevents expression of the gene. The islands can surround the 5' region of the coding region of the gene as well as the 3' region of the coding region. Thus, CpG islands can be found in multiple regions of a nucleic acid sequence. The term "region" when used in reference to a gene includes sequences upstream of coding sequences in a regulatory region including a promoter region, in the coding regions (e.g., exons), downstream of coding regions in, for example, enhancer regions, and in introns. All of these regions can be assessed to determine their methylation status. When the CpG distribution in the promoter region is rather scarce, levels of methylation are assessed in the intron and/or exon regions. The region of assessment can be a region that comprises both intron and exon sequences and thus overlaps both regions. Typically these reside near the transcription start site (TSS), for example, within about 10 kbp, within about 5 kbp, within about 3 kbp, within about 1 kbp, within about 750 bp, within about 500 bp, within 200 bp or within 100 bp. Once a gene has been identified as the target of epigenetic modification in tumor cells, determination of reduced or enhanced expression can be used as an indicator of epigenetic modification.

Expression of a gene can be assessed using any means known in the art. Typically expression is assessed and compared in test samples and control samples which may be normal, non-malignant cells. Either mRNA or protein can be measured. Methods employing hybridization to nucleic acid probes can be employed for measuring specific mRNAs. Such methods include using nucleic acid probe arrays (e.g. microarray technology, in situ hybridization, Northern blots). Messenger RNA can also be assessed using amplification techniques, such as RT-PCR. Sequencing-based methods are an alternative; these methods started with the use of expressed sequence tags (ESTs), and now include methods based on short tags, such as serial analysis of gene expression (SAGE) and massively parallel signature sequencing (MPSS). Differential display techniques provide another means of analyzing gene expression; this family of techniques is based on random amplification of cDNA fragments generated by restriction digestion, and bands that differ between two tissues identify cDNAs of interest. Specific proteins can be assessed using any convenient method including immunoassays and immuno-cytochemistry but are not limited to that. Most such methods will employ antibodies, or engineered equivalents thereof, which are specific for the particular protein or protein fragments. The sequences of the mRNA

(cDNA) and proteins of the markers of the present invention are known in the art and publicly available.

Alternatively, methylation-sensitive restriction endonucleases can be used to detect methylated CpG dinucleotide motifs. Such endonucleases may either preferentially cleave methylated recognition sites relative to non-methylated recognition sites or preferentially cleave non-methylated relative to methylated recognition sites. Non limiting examples of the former are Aat II, Ace III, Ad I, AcI I, Age I, AIu I, Asc I, Ase 1, AsiS I, Ban I, Bbe I, BsaA I, BsaH I, BsiE I, BsiW I, BsrV I, BssK 1, BstB I, BstN I, Bs I, CIa I, Eae I, Eag I, Fau I, Fse I, Hha I, mPl I, HinC II, Hpa 11, Npy99 I, HpyCAIV, Kas I, Mbo I, MIu I, MapA 1 1. Msp I, Nae I, Nar I, Not 1, PmI I, Pst I, Pvu I, Rsr II, Sac II, Sap I, Sau3A I, SfI I, Sfo I, SgrA I, Sma I SnaB I, Tsc I, Xma I, and Zra I. Non limiting examples of the latter are Ace II, Ava I, BssH II, BstU I, Hpa II, Not I, and Mho I.

Alternatively, chemical reagents can be used that selectively modify either the methylated or non-methylated form of CpG dinucleotide motifs. Modified products can be detected directly, or after a further reaction which creates products that are easily distinguishable. Means which detect altered size and/or charge can be used to detect modified products, including but not limited to electrophoresis, chromatography, and mass spectrometry. Examples of such chemical reagents for selective modification include hydrazine and bisulfite ions. Hydrazine-modified DNA can be treated with piperidine to cleave it. Bisulfite ion-treated DNA can be treated with alkali. Other means for detection that are reliant on specific sequences can be used, including but not limited to hybridization, amplification, sequencing, and ligase chain reaction. Combinations of such techniques can be used as is desired.

The principle behind electrophoresis is the separation of nucleic acids via their size and charge. Many assays exist for detecting methylation and most rely on determining the presence or absence of a specific nucleic acid product. Gel electrophoresis is commonly used in a laboratory for this purpose.

One may use MALDI mass spectrometry in combination with a methylation detection assay to observe the size of a nucleic acid product. The principle behind mass spectrometry is the ionizing of nucleic acids and separating them according to their mass to charge ratio. Similar to electrophoresis, one can use mass spectrometry to detect a specific nucleic acid that was created in an experiment to determine methylation (Tost, J. et al. 2003).

One form of chromatography, high performance liquid chromatography, is used to separate components of a mixture based on a variety of chemical interactions between a substance being analyzed and a chromatography column. DNA is first treated with sodium bisulfite, which converts an unmethylated cytosine to uracil, while methylated cytosine residues remain unaffected. One may amplify the region containing potential methylation sites via PCR and separate the products via denaturing high performance liquid chromatography (DHPLC). DHPLC has the resolution capabilities to distinguish between methylated (containing cytosine) and unmethylated (containing uracil) DNA sequences. Deng, D. et al. describes simultaneous detection of CpG methylation and single nucleotide polymorphism by denaturing high performance liquid chromatography.

Hybridization is a technique for detecting specific nucleic acid sequences that is based on the annealing of two complementary nucleic acid strands to form a double-stranded molecule. One example of the use of hybridization is a microarray assay to determine the methylation status of DNA. After sodium bisulfite treatment of DNA, which converts an unmethylated cytosine to uracil while methylated cytosine residues remain unaffected, oligonucleotides complementary to potential methylation sites can hybridize to the bisulfite-treated DNA. The oligonucleotides are designed to be complimentary to either sequence containing uracil (thymine) or sequence containing cytosine, representing unmethylated and methylated DNA, respectively. Computer- based microarray technology can determine which oligonucleotides hybridize with the DNA sequence and one can deduce the methylation status of the DNA. Similarly primers can be designed to be complimentary to either sequence containing uracil (thymine) or sequence containing cytosine. Primers and probes that recognize the converted methylated form of DNA are dubbed methylation- specific primers or probes

(MSP).

An additional method of determining the results after sodium bisulfite treatment involves sequencing the DNA to directly observe any bisulfite-modifications.

Pyro sequencing technology is a method of sequencing-by- synthesis in real time. It is based on an indirect bioluminometric assay of the pyrophosphate (PPi) that is released from each deoxynucleotide (dNTP) upon DNA-chain elongation. This method presents a DNA template-primer complex with a dNTP in the presence of an exonuclease-deficient Klenow DNA polymerase. The four nucleotides are sequentially added to the reaction mix in a predetermined order. If the nucleotide is complementary to the template base and thus incorporated, PPi is released. The PPi

and other reagents are used as a substrate in a lucif erase reaction producing visible light that is detected by either a luminometer or a charge-coupled device. The light produced is proportional to the number of nucleotides added to the DNA primer and results in a peak indicating the number and type of nucleotide present in the form of a pyrogram. Pyro sequencing can exploit the sequence differences that arise following sodium bisulfϊte-conversion of DNA.

A variety of amplification techniques may be used in a reaction for creating distinguishable products. Some of these techniques employ PCR. Other suitable amplification methods include the ligase chain reaction (LCR) (Barringer et al, 1990), transcription amplification (Kwoh et al. 1989; WO88/10315), selective amplification of target polynucleotide sequences (US Patent No. 6,410,276), consensus sequence primed polymerase chain reaction (US Patent No 4,437,975), arbitrarily primed polymerase chain reaction (WO90/06995), nucleic acid based sequence amplification (NASBA) (US Patent Nos. 5,409,818; 5,554,517; 6,063,603), micro satellite length polymorphism (MLP), and nick displacement amplification (WO2004/067726).

Sequence variation that reflects the methylation status at CpG dinucleotides in the original genomic DNA offers two approaches to PCR primer design. In the first approach, the primers do not themselves cover or hybridize to any potential sites of DNA methylation; sequence variation at sites of differential methylation are located between the two primers. Such primers are used in bisulfite genomic sequencing, COBRA, Ms-SNuPE. In the second approach, the primers are designed to anneal specifically with either the methylated or unmethylated version of the converted sequence. If there is a sufficient region of complementarity, e.g., 12, 15, 18, or 20 nucleotides, to the target, then the primer may also contain additional nucleotide residues that do not interfere with hybridization but may be useful for other manipulations. Exemplary of such other residues may be sites for restriction endonuclease cleavage, for ligand binding or for factor binding or linkers or repeats. The oligonucleotide primers may or may not be such that they are specific for modified methylated residues.

One way to distinguish between modified and unmodified DNA is to hybridize oligonucleotide primers which specifically bind to one form or the other of the DNA. After primer hybridization, an amplification reaction can be performed. The presence of an amplification product indicates that a sample hybridized to the primer. The specificity of the primer indicates whether the DNA had been modified or not, which in turn indicates whether the DNA had been methylated or not. For example, bisulfite

ions convert non-methylated cytosine bases to uracil bases. Uracil bases hybridize to adenine bases under hybridization conditions. Thus an oligonucleotide primer which comprises adenine bases in place of guanine bases would hybridize to the bisulfite- modified DNA, whereas an oligonucleotide primer containing the guanine bases would hybridize to the non-converted (initial methylated) cytosine residues in the modified DNA. Amplification using a DNA polymerase and a second primer yield amplification products which can be readily observed. This method is known as MSP (Methylation Specific PCR; Patent Nos 5,786,146; 6,017,704; 6,200,756). Primers are designed to anneal specifically with the converted sequence representing either the methylated or the unmethylated version of the DNA. Preferred primers and primer sets for assessing the methylation status of the concerned gene by way of MSP will specifically hybridize to a converted sequence provided in Table 2, or to its complement sequence. Most preferred primers and primer sets are provided in Table 1 and are represented by SEQ ID NO. 1 to 264. Sense primers comprise or consist essentially of SEQ ID NO. 1 to 132, antisense primers consist essentially of SEQ ID NO. 133 to 264. The amplification products can be optionally hybridized to specific oligonucleotide probes which may also be specific for certain products. Alternatively, oligonucleotide probes can be used which will hybridize to amplification products from both modified and non-modified DNA.

Thus, present invention provides for a method for identifying cervical cancer or its precursor, or predisposition to cervical cancer in a test sample containing cervical cells or nucleic acids from cervical cells comprising: contacting a methylated CpG- containing nucleic acid of at least one gene selected from the group consisting of genes according to Table 1 with bisulfite to convert unmethylated cytosine to uracil; detecting the methylated CpGs in the nucleic acid by contacting the converted nucleic acid with oligonucleotide primers whose sequence discriminates between the bisulfite- treated methylated and unmethylated version of the converted nucleic acid; and identifying the test sample as containing cells that are neoplastic, precursor to neoplastic, or predisposed to neoplasia, or as containing nucleic acids from cells that are neoplastic, precursor to neoplastic, or predisposed to neoplasia.

Modified and non-modified DNA can be distinguished with use of oligonucleotide probes which may also be specific for certain products. Such probes can be hybridized directly to modified DNA or to amplification products of modified DNA. Probes for assessing the methylation status of the concerned gene will specifically hybridize to the converted sequence but not to the corresponding non converted sequence. Probes

are designed to anneal specifically with the converted sequence representing either the methylated or unmethylated version of the DNA. Preferred converted sequences are provided in Table 2. Preferred probes anneal specifically with the converted sequence representing the methylated version of the DNA, or to the complement sequence thereof. Oligonucleotide probes can be labeled using detection systems known in the art. These include but are not limited to fluorescent moieties, radioisotope labeled moieties, bioluminescent moieties, luminescent moieties, chemiluminescent moieties, enzymes, substrates, receptors, or ligands.

Another way for the identification of methylated CpG dinucleotides utilizes the ability of the MBD domain of the McCP2 protein to selectively bind to methylated DNA sequences (Cross et al, 1994; Shiraishi et al, 1999). Restriction endonuclease digested genomic DNA is loaded onto expressed His-tagged methyl-CpG binding domain that is immobilized to a solid matrix and used for preparative column chromatography to isolate highly methylated DNA sequences. Variants of this method have been described and may be used in present methods of the invention.

Real time chemistry allows for the detection of PCR amplification during the early phases of the reactions, and makes quantitation of DNA and RNA easier and more precise. A few variants of real-time PCR are well known. They include Taqman® (Roche Molecular Systems), Molecular Beacons®, Amplifluor® (Chemicon International) and Scorpion® DzyNA®, Plexor™ (Promega) etc. The TaqMan® system and Molecular Beacon® system have separate probes labeled with a fluorophore and a fuorescence quencher. In the Scorpion® system the labeled probe in the form of a hairpin structure is linked to the primer.

Quantitation in real time format may be on an absolute basis, or it may be relative to a methylated DNA standard or relative to an unmethylated DNA standard. The absolute copy number of the methylated marker gene can be determined; or the methylation status may be determined by using the ratio between the signal of the marker under investigation and the signal of a reference gene with a known methylation (e.g. β- actin), or by using the ratio between the methylated marker and the sum of the methylated and the non-methylated marker.

Real-Time PCR detects the accumulation of amplicon during the reaction, but alternatively end-point PCR fluorescence detection techniques may be used. Confirming the presence of target DNA at the end point stage may indeed be sufficient and it can use the same approaches as widely used for real time PCR.

DNA methylation analysis has been performed successfully with a number of techniques which are also applicable in present methods of the invention. These include the MALDI-TOFF, MassARRAY (Ehrich, M. et al. 2005), MethyLight (Trinh B. et al. 2001), Quantitative Analysis of Methylated Alleles (Zeschnigk M. et al. 5 2004), Enzymatic Regional Methylation Assay (GaIm et al., 2002), HeavyMethyl (Cottrell, SE et al, 2004), QBSUPT, MS-SNuPE (Gonzalgo and Jones, 1997), MethylQuant (Thomassin H. et al. 2004), Quantitative PCR sequencing, and Oligonucleotide-based microarray systems (Gitan RS et al. ,2006).

The number of genes whose modification is tested and/or detected can vary: one, two, 10 three, four, five, six, seven, eight, nine or more genes according to Table 1 can be tested and/or detected. Detection of epigenetic modification of at least one, two, three, four, five, six, seven, eight, nine or more genes according to Table 1 can be used as an indication of cancer or pre-cancer or risk of developing cancer. The genes are preferably selected from the group of JAM3, LMXlA, CDOl, NID2, CCNAl, 15 HOXAI l, GREMl and TACl. Preferred gene combinations include

• NID2 and HOXAI l;

• JAM3, CDOl, HOXAI l, and CCNAl;

• JAM3 and HOXAI l;

• JAM3 , HOXAl 1 and GREM 1 ;

20 • JAM3, NID2, HOXAl 1 and CDOl ;

• JAM3, TACl, HOXAI l, and CDOl;

• JAM3, HOXAI l, and CDOl;

• JAM3 and CDOl;

• JAM3 and NID2;

25 • NID2 and CDOl;

• JAM3 and LMXlA

• NID2 and LMXlA, and

• JAM3, CDOl and NID2.

The accession numbers corresponding to the listed genes can be found at

30 http://www.ncbi.nlm.nih.gov. Of course, as appropriate, the skilled person would appreciate that functionally relevant variants of each of the gene sequences may also be detected according to the methods of the invention. For example, the methylation status of a number of splice variants may be determined according to the methods of the invention. Variant sequences preferably have at least 90%, at least 91%, at least 35 92%, at least 93%, at least 94%, at least 95%, at least 96%, at least 97%, at least 98%,

or at least 99% nucleotide sequence identity with the nucleotide sequences in the database entries. Computer programs for determining percentage nucleotide sequence identity are available in the art, including the Basic Local Alignment Search Tool (BLAST) available from the National Center for Biotechnology Information.

It is possible for the methods of the invention to be used in order to detect more than one gene of interest in the same reaction. Through the use of several specific sets of primers, amplification of several nucleic acid targets can be performed in the same reaction mixture. This may be termed "multiplexing". Multiplexing can also be utilized in the context of detecting both the gene of interest and a reference gene in the same reaction.

The term "test sample" refers to biological material obtained from a mammalian subject, preferably a human subject, and may be any tissue sample, body fluid, body fluid precipitate, or lavage specimen. Test samples for diagnostic, prognostic, or personalized medicine uses can be obtained from cyto logical samples, from surgical samples, such as biopsies, cervical conization or hysterectomy, from (formalin fixed) paraffin embedded cervix or other organ tissues, from frozen tumor tissue samples, from fresh tumor tissue samples, from a fresh or frozen body fluid such as blood, serum, lymph, or from cervical scrapings, cervical smears, cervical washings and vaginal excretions. Such sources are not meant to be exhaustive, but rather exemplary. A test sample obtainable from such specimens or fluids includes detached tumor cells and/or free nucleic acids that are released from dead or damaged tumor cells. Nucleic acids include RNA, genomic DNA, mitochondrial DNA, single or double stranded, and protein-associated nucleic acids. Any nucleic acid specimen in purified or non- purified form obtained from such specimen cell can be utilized as the starting nucleic acid or acids. The test samples may contain cancer cells or pre-cancer cells or nucleic acids from them. Preferably, the test sample contains squamous cell carcinomas cells or nucleic acids from squamous cell carcinomas, adenocarcinoma cells or nucleic acids of adenocarcinoma cells, adenosquamous carcinoma cells or nucleic acids thereof. Samples may contain mixtures of different types and stages of cervical cancer cells.

Present invention also relates to screening protocols for the screening of woman for cervical cancer and the precursors thereof. Traditionally the Pap Smear has been the primary screening method for the detection of abnormality of the cervix, but its performance is suboptimal. Human Papillomavirus has been associated with the development of cervical cancer. Five high-risk types, 16, 18, 31, 45, and 58, and in particular HPV types 16 and 18 account for approximately 70% of all cervical

carcinomas. A small percentage of women showing persistent infection progress from Low-grade to High-grade lesions. The introduction of methylation markers now adds a new dimension to the screening for and treatment of cervical lesions. Method for cervical cancer screening may combine high-risk human papillomavirus (hr-HPV) testing and methylation testing; or cytological evaluation and methylation testing; or hr-HPV testing and cytological evaluation and methylation testing.

Thus, a further embodiment of the present invention relates to a method for cervical cancer detection or screening comprising the steps of: a) providing a test sample comprising cervical cells or nucleic acids from cervical cells; b) assaying the test sample of step a) for high-risk human papillomavirus (hr-HPV); c) if b) is positive for the presence of hr-HPV, assaying the methylation status of at least one gene selected from the group consisting of genes according to Table 1; d) if the gene of c) is methylated, refer the woman for colposcopy; e) if the gene of c) is unmethylated, refer the woman to a more regular screening for the presence of hr-HPV.

The present invention relates further to a method for cervical cancer detection or screening comprising the steps of: a) providing a test sample comprising cervical cells or nucleic acids from cervical cells; b) assaying the test sample of step a) for hr-HPV ; c) if b) is positive for the presence of hr-HPV, assaying the methylation status of at least one gene selected from the group consisting of genes according to Table 1, and/or typing the hr-HPV for the presence of HPV16 and/or HPV18; d) if the gene of c) is methylated, and/or HPV16 and/or HPV18 positive, refer the woman for colposcopy; e) if the gene of c) is unmethylated, refer the woman to a more regular screening for the presence of hr-HPV.

In a related embodiment, the invention provides for a method for cervical cancer detection or screening comprising the steps of: a) performing cytology evaluation on a test sample comprising cervical cells or nucleic acids from cervical cells; b) if a) is positive, assaying the methylation status of at least one gene selected from the group consisting of genes according to Table 1; c) if the at least one gene of b) is methylated, refer the woman for colposcopy;

d) if the at least one gene of b) is unmethylated, refer the woman to cytology testing on a more regular basis.

In a related embodiment, the invention provides for a method for cervical cancer detection or screening comprising the steps of: a) assaying the methylation status of at least one gene selected from the group consisting of genes according to Table 1; b) if the at least one gene of b) is methylated, perform cytology testing; c) if b) is tested positive, refer the woman for colposcopy; d) if b) is negative, refer the woman to methylation testing on a more regular basis.

In all aspects of the invention, the test sample is preferably a cervical, cervicovaginal or vaginal sample of a woman.

The phrase "cervical cancer screening" refers to organized periodic procedures performed on groups of people for the purpose of detecting cervical cancer.

The phrase "assaying for hr-HPV" refers to testing for the presence of hr-HPV. There are various PCR based assays commercially available to measure hr-HPV copy number or viral load in clinical samples. Many testing methods have been used to detect the presence of HPV in cervicovaginal specimens, including viral load quantification, Southern blot, polymerase chain reaction (PCR), ViraPap (Life Technologies, Gaithersburg, MD), Hybrid Capture tube testing, Hybrid Capture microtiter plate assays, and CISH. For instance, assaying for hr-HPV may be performed with the FDA approved Hybrid Capture II assay (Digene Corp., Silver Spring, Md.) with a probe cocktail for 13 carcinogenic types.

The so-called "high risk" HPV types are those strains of HPV more likely to lead to the development of cancer, while "low-risk" viruses rarely develop into cancer. The list of strains considered high risk is being adapted with the time and the increase in epidemiological knowledge. As such, those hr-HPV types comprise, without being limited to, strains 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 68, and 69. Preferred "high risk" HPV types are HPV 16 and HPVl 8.

The phrase "HPV16 testing" refers to testing for the presence of hr-HPV type 16. Similarly, "HPV 18 testing" refers to testing for the presence of hr-HPV type 18. The various methods allowing type-specific HPV testing are well known to the person skilled in the art and are applicable in the methods of present invention. For instance,

testing for the presence of hr-HPV-16 may be accomplished by PCR amplification using primers specific for HPV type 16, which are known by the skilled in the art.

The phrase "performing cytological evaluation" refers to the cytomorphological assessment of cervical samples, which is usually performed by especially trained medical staff. The various methods allowing cytological testing are well known to the person skilled in the art and are applicable in the methods of present invention. Cytological evaluation may be performed with the known Papanicolaou (PAP) smear test. Alternative means for cytological evaluation include liquid based cytology with for example the ThinPrep technique (Cytyc Corporation, Marlborough, Massachusetts, USA).

The term "triaging" refers to sorting out or classifying patients in order to establish priority of treatment's necessity, priority of proper place of treatment, or any other priority in terms of patient management.

The test sample will most of the time be obtained from a subject suspected of being tumorigenic or from a subject undergoing routine examination and not necessarily being suspected of having a disease. Alternatively the sample is obtained from a subject undergoing treatment, or from patients being checked for recurrence of disease.

Testing can be performed diagnostically or in conjunction with a therapeutic regimen. Testing can be used to monitor efficacy of a therapeutic regimen, whether a chemotherapeutic agent or a biological agent, such as a polynucleotide. Epigenetic loss of function of at least one gene selected from the group consisting of genes according to Table 1 can be rescued by the use of DNA demethylating agents and/or DNA methyltransferase inhibitors. Testing can also be used to determine what therapeutic or preventive regimen to employ on a patient. Moreover, testing can be used to stratify patients into groups for testing agents and determining their efficacy on various groups of patients.

Demethylating agents can be contacted with cells in vitro or in vivo for the purpose of restoring normal gene expression to the cell. Suitable demethylating agents include, but are not limited to 5-aza-2'-deoxycytidine, 5-aza-cytidine, Zebularine, procaine, and L-ethionine. This reaction may be used for diagnosis, for determining predisposition, and for determining suitable therapeutic regimes. Accordingly, the invention also provides for a method for predicting the likelihood of successful treatment or resistance to treatment of cancer with such agent. If the gene is methylated, the

likelihood of successful treatment is higher than if the gene is unmethylated, or methylated to a lesser degree. Conversely, if the gene is unmethylated, or methylated to a lesser degree, the likelihood of resistance to treatment is higher than if the gene is methylated.

In a related aspect, epigenetic loss of gene function(s) can identify the stage of the disease and from that the need of treatment. Accordingly, the invention provides for a method for predicting suitable treatment comprising determining the methylation status of a gene or a combination of genes. If the gene is methylated, the need of cervical resection is identified; if the gene is unmethylated or methylated to a lesser degree, it is decided that there is no need for cervical resection. In cases of early stage (CIN) and carcinoma in situ, abnormal tissue is removed by cryosurgery, laser surgery, conization, or simple hysterectomy (removal of the uterus). Invasive cervical cancer is treated with radical hysterectomy (removal of the uterus, fallopian tubes, ovaries, adjacent lymph nodes, and part of the vagina).

To attain high rates of tumor detection, it may be necessary to combine the methods of the invention with established methods and/or markers for cervical cancer identification (Malinowski D, 2007), such as morphology-based detection methods, HPV methylation testing (Badal et al. 2004, Kalantari et al. 2004), KRAS and BRAF mutation detection (Kang et al. 2007), chromosomal amplification (Rao et al. 2004), protein expression (Keating et al. 2001)and HPV detection methods (Brink et al.

2007): several HPV detection kits are known in the art and commercially available, for example kits such as Digene® HPV Test (Qiagen), AMPLICOR HPV Test (Roche), HPV High-Risk Molecular Assay (Third Wave Technologies), LINEAR ARRAY HPV Genotyping Test (Roche), INNO-LiPA HPV Genotyping (Innogenetics), PapilloCheck (Greiner Bio-One GmbH), PreTect HPV-Proofer (Norchip), NucliSENS

EasyQ HPV (BioMerieux), F-HPV typing™ (molGENTIX, S.L.) may be utilized. Such examples are not meant to be exhaustive, but rather exemplary.

Another aspect of the invention is a kit for assessing methylation in a test sample. Kits according to the present invention are assemblages of reagents for testing methylation. They are typically in a package which contains all elements, optionally including instructions. The package may be divided so that components are not mixed until desired. Components may be in different physical states. For example, some components may be lyophilized and some in aqueous solution. Some may be frozen. Individual components may be separately packaged within the kit. The kit may contain reagents, as described above for differentially modifying methylated and non-

methylated cytosine residues. Desirably the kit will contain oligonucleotide primers which specifically hybridize to regions within about 10 kbp, within about 5 kbp, within about 3 kbp, within about 1 kbp, within about 750 bp, within about 500 bp, within 200 bp or within 100 bp kb of the transcription start sites of the genes/markers listed in Table 1.

Typically the kit will contain both a forward and a reverse primer for a single gene or marker. If there is a sufficient region of complementarity, e.g., 12, 15, 18, or 20 nucleotides, then the primer may also contain additional nucleotide residues that do not interfere with hybridization but may be useful for other manipulations. Exemplary of such other residues may be sites for restriction endonuclease cleavage, for ligand binding or for factor binding or linkers or repeats. The oligonucleotide primers may or may not be such that they are specific for modified methylated residues. The kit may optionally contain oligonucleotide probes. The probes may be specific for sequences containing modified methylated residues or for sequences containing non-methylated residues. The kit may optionally contain reagents for modifying methylated cytosine residues. The kit may also contain components for performing amplification, such as a DNA polymerase and deoxyribonucleotides. Means of detection may also be provided in the kit, including detectable labels on primers or probes. Kits may also contain reagents for detecting gene expression for one of the markers of the present invention. Such reagents may include probes, primers, or antibodies, for example. In the case of enzymes or ligands, substrates or binding partners may be sued to assess the presence of the marker. Kits may contain 1, 2, 3, 4, or more of the primers or primer pairs of the invention. Kits that contain probes may have them as separate molecules or covalently linked to a primer for amplifying the region to which the probes hybridize. Other useful tools for performing the methods of the invention or associated testing, therapy, or calibration may also be included in the kits, including buffers, enzymes, gels, plates, detectable labels, vessels, etc.

According to a further aspect, the invention also employs or relies upon or utilizes oligonucleotide primers and/or probes to determine the methylation status of at least one gene selected from a group of genes consisting of JAM3, LMXlA, CDOl, NID2,

ALX3, ALX4, AR, ARID4A, ATM, AURKA, B4GALT1, BMP2, BMP6, BNIP3, C13orfl8, C16orf48, C9orfl9, CALCA, CAMK4, CCNAl, CCND2, CDHl, CDH4, CDK6, CDKNlB, CDKN2B, CLSTN2, CLU, COLlAl, CPTlC, CTDSPL, CYCLIND2, DAPKl, DBCl, DDX19B, DKK2, EGFR, EGR4, EPB41L3, FOS, FOXEl, GADD45A, GAT A4, GDAPlLl, GNB4, GPNMB, GREMl, Gst-Pi, HHIP,

HINl, HOOK2, HOXAl, HOXAI l, HOXA7, HOXDl, IGSF4, ISYNAl, JPH3, KNDCl, KRAS, LAMAl, LOC285016, LOX, LTB4R, MAL, MTAP, MYO 18B, NDRG2, N0L4, NPTXl, NPTX2, 0GF0D2, PAK3, PAXl, PDCD4, PHACTR3, POMC, PRKCE, RAD23B, RALY, RARA, RASSFlA, RBP4, RECK, RPRM, SALL4, SEMA3F, SLC5A8, SLITl, SLIT2, SLIT3, SMPDl, SOCSl, SOXl, S0X17, SPARC, SPN, SST, TACl, TERT, TFPI-2, TLLl, TNFAIPl, TRMTl, TWISTl, UGTlAl, WIFl, WITl, WTl, XRCC3, and ZGPAT. Preferred probes and their sequences bind to at least one of the polynucleotide sequences listed in Table 2, Figure 5B or to the complement sequence thereof. Preferred primers and probes are selected from the primers and probes comprising or consisting essentially of the nucleotide sequences set forth in Table 1. Related to this, the invention also provides for an isolated polynucleotide which consists of a nucleotide sequence listed in Table 1, Table 2 and Figure 5B.

The above disclosure generally describes the present invention. All references disclosed herein are expressly incorporated by reference. A more complete understanding can be obtained by reference to the following specific examples which are provided herein for purposes of illustration only, and are not intended to limit the scope of the invention.

Table 1: MSP assays and primer design

Table 2: Amplicon details (converted sequences issuing from the methylated version of the DNA)

Table 3: qMSP Molecular Beacon sequences

EXAMPLES

Example 1: Discovery of methylation markers in cervical cancer, using relaxation ranking

To identify genes that are downregulated in cervical cancer due to promoter hypermethylation and to enrich for those genes that are most frequently involved in cervical cancer, a multistep approach was used combining:

• Affymetrix expression microarray analysis on a panel of frozen tissue samples from 39 human primary cervical cancers to identify cancer- specific down-regulated genes.

• Affymetrix expression microarray analysis on a panel of 4 different cervical cancer cell lines in which the expression of (hyper)methylated genes was re-activated upon treatment with 5-aza-2'deoxycytidine (DAC) (blocking DNA methylation), and/or trichostatin A (TSA) (inhibiting histone deacetylase - HDAC).

Data from both approaches were combined, and a novel non-parametrical ranking and selection method was applied to identify and rank candidate genes. Using in silico promoter analysis we restricted the analysis to those candidate genes that carry CpG- islands. The new approach resulted in a significant enrichment of hypermethylated genes: we compared the first 3000 high-ranking candidate probes with lists of imprinted genes, X-chromosome located genes and known methylation markers. In addition, we determined the hypermethylation status of the 10 highest ranking candidate genes in both cervical cancers and normal cervices using COBRA (COmbined Bisulfite Restriction Analysis).

MATERIAL AND METHODS

Primary cervical tissue samples:

For the expression microarray analysis, tissues from 39 early stage frozen cervical cancer samples were used from a collection of primary tumors surgically removed between 1993 and 2003 (University Medical Center Groningen, Groningen, The Netherlands). All cervical cancer patients underwent gynecological examination for staging in accordance with the International Federation of Gynecology and Obstetrics (FIGO) criteria (Finan et al., 1996). Tumor samples were collected after surgery and stored at -80 ⁰ C. The stage of cervical cancer patients included 33 FIGO stage IB

(85%) and 6 FIGO stage HA (15%). The median age of the cervical cancer patients was 46 years (IQ range 35 - 52 yr.).

For COBRA and BSP (Bisulfite Sequencing PCR), 10 (of the 39) primary cervical cancers and 5 controls (normal cervix) were used. The age-matched normal cervical controls were women without a history of abnormal PAP smears or any form of cancer and planned to undergo a hysterectomy for benign reasons during the same period. 5 Normal cervices were collected after surgery and histologically confirmed.

Informed consent was obtained from all patients participating in this study.

Cervical cancer cell lines:

Four cervical carcinoma cell lines were used: HeLa (cervical adenocarcinoma,

HPVl 8), SiHa (cervical squamous cell carcinoma, HPV 16), CSCC-7 (non-keratinizing

10 large cell cervical squamous cell carcinoma, HPV 16) and CC-8 (cervical adenosquamous carcinoma, HPV45). HeLa and SiHa were obtained from the American Tissue Type Collection. CSCC-7 and CC-8 (Koopman et al, 1999) were a kind gift of Prof. GJ Fleuren (Leiden University Medical Center, Leiden, The Netherlands). All cell lines were cultured in DMEM/Ham's F12 supplemented with

15 10% fetal calf serum.

Cell lines were treated for 3 days with low to high dose (200 nM, 1 μM or 5 μM) 5- aza-2'deoxycytidine (DAC), 200 nM DAC with 300 nM trichostatin A (TSA) after 48 hours, or left untreated. Cells were split to low density 24 hours before treatment. Every 24 hours DAC was refreshed. After 72 hours cells were collected for RNA 20 isolation.

RNA and DNA isolation:

From the frozen biopsies, four 10-μm-thick sections were cut and used for standard RNA and DNA isolation. After cutting, a 3-μm-thick section was stained with haematoxylin/eosin for histological examination and only tissues with >80% tumor

25 cells were included. Macrodissection was performed to enrich for epithelial cells in all normal cervices.

For DNA isolation, cells and tissue sections were dissolved in lysis buffer and incubated overnight at 55°C. DNA was extracted using standard salt-chloroform extraction and ethanol precipitation for high molecular DNA and dissolved in 250 μl

30 TE-4 buffer (10 mM Tris; 1 mM EDTA (pH 8.0)). For quality control, genomic DNA was amplified in a multiplex PCR containing a control gene primer set resulting in

products of 100, 200, 300, 400 and 600 bp according to the BIOMED-2 protocol (van Dongen et al, 2003).

RNA was isolated with TRIzol reagent (Invitrogen, Breda, The Netherlands) according to manufacturer's protocol. RNA was treated with DNAse and purified using the RNeasy mini-kit (Qiagen, Westburg, Leusden, The Netherlands). The quality and quantity of the RNA was determined by Agilent Lab-on-Chip analysis (ServiceXS, Leiden, The Netherlands, www.serviceXS.com).

Expression data:

Gene expression for 39 primary cancers and 20 cell line samples was performed using the Affymetrix HGU 133 Plus 2.0 array with 54,675 probes for analysis of over 47,000 human transcripts. The labeling of the RNA, the quality control, the microarray hybridization and scanning were performed by ServiceXS according to Affymetrix standards. For labeling, ten microgram of total RNA was amplified by in vitro transcription using T7 RNA polymerase.

Quality of the microarray data was checked using histograms, boxplots and a RNA degradation plot. One cell line sample was omitted because of poor quality. Using BioConductor (Gentleman et al., 2004), present (P), absent (A) or marginal (M) calls were determined with the MAS5 algorithm. MAS5 uses a non-parametric statistical test (Wilcoxon signed rank test) that assesses whether significantly more perfect matches show more hybridization signal than their corresponding mismatches to produce the detection call for each probe set (Liu et al., 2002). The relaxation ranking approach only relied on P-calls. Some samples were analyzed in duplicate, and the profile of P-calls is highly similar (93-95 % of the probesets have an identical P/M/A call).

Relaxation ranking algorithm:

In order to identify the most promising markers that are methylated in cervical cancer, we assumed that such markers should be silenced in cancer cells and upregulated upon re-activation after DAC/TSA treatment; therefore, the best methylation markers will be genes represented by probes with:

• no expression in primary cervical cancers: P-calls=0 out of 39 cancers

• no expression in (untreated) cervical cancer cell lines: P-calls=0 out of 4 cell lines

• expression in cervical cancer cell lines treated with DAC (or DAC in combination with TSA): P-calls=15 out of 15 treated cell lines

To select for those gene probes that would be the best candidate hypermethylated genes in cervical cancer, we present the relaxation ranking algorithm. Probe sets were ranked, not primarily based on the number of P-calls and thus explicitly setting thresholds, but primarily driven by the number of probe sets that would be picked up, based on selection criteria (the number of P-calls in primary cancers, untreated and treated cell lines). The stricter (e.g. P-calls: 0 - 0 - 15) these selection criteria, the lower the number of probes that meet with these criteria; while if the conditions become more and more relaxed (higher number of P-calls in primary cancers and untreated cell lines, and lower number of P-calls in treated cell lines), the more probes will comply. In the end, using P-calls: 39 - 4 - 0 as criteria, all probe sets were returned. This way, there was no need to define a 'prior' threshold for the number of P-calls.

The following sorting method was applied:

(1) All possible conditions were generated and the number of probes that were picked up under these conditions was calculated: a. the number of samples with expression (P) of a certain probe in i. primary cervical cancer samples is called x _S am _P ie ii. cervical cancer cell lines is called y _sa mpie iii. treated cervical cancer cell lines is called z _samp i _e b. all combinations of x, y and z are made i. x (the number of P-calls in primary cancers) varies from 0 to 39 ii. y (the number of P-calls in untreated cell lines) from 0 to 4 iii. z (the number of P-calls in treated cell lines) from 0 to 15 iv. In total, 3200 combinations of x, y and z can be made c. a probeset was found under each of these generated conditions x, y and z if: i- Xsampie ≤ x (number of P-calls for probe in primary cancers smaller or equal compared to condition) AND ϋ. ysampie ≤ y (number of P-calls for probe in untreated cell lines smaller or equal compared to condition) AND iii. z _samp ie > z (number of P-calls for probe in treated cell lines larger or equal compared to condition)

d. under very strict conditions (x=0, y=0, z=15) no probes were found, while under the most relaxed conditions (x=39, y=4, z=0) all probes were returned. For all combinations of x, y and z, the number of probes that complied (w), was stored

(2) The data was sorted with w as primary criterion (ascending), followed by x (ascending), y (ascending) and z (descending)

(3) This sorted dataset was analyzed row per row. In row i, the W ₁ probes retrieved with criteria X ₁ V ₁ Z ₁ were compared with the list of probes, already picked up in rows 1 to i-1. If a probe did not occur in this list, it was added to the list

(4) This process continued until there were m (user-defined) probes in the list

DNA methylation analysis using COBRA and bisulphate sequencing:

To validate the (hyper)methylated status of candidate gene probes, DNA extracted from 10 cervical cancers and 5 normal cervices were analyzed using BSP and COBRA. Bisulfite modification of genomic DNA was performed using the EZ DNA methylation kit (Zymogen, BaseClear, Leiden, The Netherlands). The 5' promoter region of the tested gene was amplified using bisulfite treated DNA. PCR primers for amplification of specific targets sequences are listed in Table 4. COBRA was performed directly on the BSP products as described by Xiong et al. (Xiong and Laird, 1997) using digestions with BstUI, Taql and/or Hinfl according the manufacture's protocol (New England Biolabs Inc., Beverly, MA). For sequence analysis, the BSP products were purified (Qiagen) and subjected to direct sequencing (BaseClear, Leiden, The Netherlands). Leukocyte DNA collected from anonymous healthy volunteers and in vitro CpG methylated DNA with Sssl (CpG) methyltransferase (New England Biolabs Inc.) were used as negative and positive control, respectively.

Table 4: list of primers used for BSP ( ^l : +1 is transcription start site (TSS); ^2' Several primer pairs were tested, however, none worked)

l oiwϋi'i :>■>«!<.•! si. 3 \

DAZL TTTGGGGGTGATGTGTGTGTTT TCTCCCTCAACTCACCATAATA 54 -161 312 NM_001351

ADARBl ² NM_015834

SYCP3 AAAATTTAAAAATTGGAAGGTATT ACCTCACTAATCAAAAACAACCTCT 54 -208 +186 NM_153694

AGG

AUTS2 TTTTAAAAGTGATAAAGTTGGTTA CCCTTTTCTTTCTCCTCTCTTTCT 56 +300 -184 NM_Cl!>a/ϋ

TGG T

NNAT GGTTAGGGATTGGGGAGAA GCTAAACTTACCTACAACAACAC 54 -271 210 NM_005386

SST GGGGTATGTGGAATTGTGTG AAA TCT CCT TAC CTA CTT CCC C 54 -185 +276 NM_001048

HTRA3 GTYGGTTTTGTYGTTATGTAGGY AAC TTC ACT TCC TCC CTA ACC 57 +190 +622 NM_053044

ZFP42 AGTAGGTGTTTGTTGAAGATAG ACT CAT AAC ACA CAT AAC CAT C 60 +308 +580 NM_174900

NPTXl GGTAGTGGGGGTTTGATAG AAATAATCTCCTTCTACTACAACAC 54 -2 +372 NM_002522

GDA TATAGAAGGTGGAGGAAGTTGA CACCTCCATAAAACAAATCCAAA 54 -239 +194 NM_004293

CCNAl TATAGTTGGAGTTGGAGGGT AAACAACTAACAAATACACTAAAA 54 -279 +146 NM_153694

RESULTS

To identify novel markers that are methylated in cervical cancer, we applied a multistep approach that combines re-expression of silenced hypermethylated genes in cervical cancer cell lines (using DAC and DAC/TSA), downregulated expression in 39 cervical cancers expression, and selection of candidate markers using a relaxing ranking algorithm. The best profile of a candidate marker would be: no expression in any of the 39 cervical primary cancers and 4 untreated cancer cell lines, but re- activation of expression after demethylation and/or blocking of histone deacetylation in all 15 cell lines treated with various combinations of DAC/TSA (P-calls: 0 - 0 - 15). However, none of the probe sets showed this ideal profile. To generate a list of candidate genes, a relaxation ranking algorithm was applied.

The only variable used in the relaxation ranking is the number of probes we would like to retrieve. As shown in Figure 1, the number of probes retrieved (w) with parameters x, y and z (the number of P-calls in respectively primary tumor samples, untreated and treated cell lines) follows a complex profile which consists not only of additive elements, but also interactions between the parameters. In general, the number of P- calls in primary cancer samples (x) has the largest influence on w. The sorting methodology has the advantage that no cut-off values have to be chosen for x, y and z, and therefore there is no need to implicitly link a relative weight factor to the parameters.

To calculate the most optimal number of potentially hypermethylated candidate markers for further analysis, we estimated this number based on known (i.e. described in literature) methylation markers in cervical cancer. Forty- five known methylation markers were found using text-mining using GeneCards (Rebhan et ah, 1997) for aliases/symbols to query PubMed through NCBI E-Utils. The position of the markers after ranking ("observed") was determined as shown in the step plot in Figure 2. If the markers would be randomly distributed in the ranking, the profile would be similar to the curve, marked 'expected'. This 'expected' curve is not a straight line, but is calculated based on whether a probe could be assigned with a gene symbol and taking probes into account that are associated with a gene that is already associated with an earlier selected probe. The number of observed methylation markers has in general the same slope as expected. However, until about 3000 probes, the slope of the number observed markers versus the number of selected probes (in dashed lines) cannot be explained if the markers would be randomly distributed as its steepness is much higher. When selecting more than 3000 probes, the slope suddenly decreases to a level that is close to random distribution. This enrichment can also statistically be proven. Therefore, we selected the first 3000 probes, referred to as TOP3000, in the ranking for further analysis. In this TOP3000 list, 2135 probes are associated with a gene symbol, of which 1904 are unique.

Validation of the 10 highest-ranking candidate genes using COBRA:

In order to validate whether the highest ranking genes represent markers that are functionally hypermethylated in cervical cancer, we performed COBRA on bisulfite- treated DNA of 10 cervical cancers and 5 normal cervices. For this analysis we focused on those first 10 genes from the highest ranking probe-list (Table 5) that:

• represent a known gene (i.e. gene symbol)

• contain a CpG-island surrounding the TSS

• are located on any chromosome except chromosome X

• are expressed in less than 15 carcinomas

BSP was used to amplify the CpG-islands of these candidate genes using bisulfite- treated DNA and COBRA to determine the methylation status. CCNAl (at position

49) was included as a positive control for the highest listed, reported cervical cancer specific methylation gene promoter. BSP/COBRA of CCNAl revealed that 6 of 10 carcinomas are methylated at the restriction enzyme sites (Tl, T3, T5, T7, T9 and TlO in Figure 3). Sequence analysis of the BSP-products (on average 7-9 independent

clones for each carcinoma) of these 10 carcinomas revealed that in 6 carcinomas the promoter is hypermethylated in good agreement with the COBRA results (Figure 3C).

Table 5: Methylation status using COBRA of the 10 highest ranking gene promoters. Gene selected for further validation after applying additional criteria. Included is CCNAl on position 47 (original position 241) as the highest ranking cervical-cancer- associated hypermethylated gene. Methylation status was determined by BSP/COBRA (see Figure 3 and Figure 4).

Gene Chromosomal Methylation in Methylation in Rank symbol location cancer normal

1 DAZL 3p24.3 9/9 5/5

2 ADARBl 21q22.3 Nd Nd

3 SYCP3 12q 9/9 5/5

4 AUTS2 7ql l.22 0/9 0/5

5 NNAT 20ql 1.2 9/9 5/5

6 SST 3q28 7/9 0/5

7 HTRA3 4pl6.1 1/9 0/5

8 ZFP42 4q35.2 9/9 5/5

9 NPTXl 17q25.1 5/10 0/5

10 GDA 9q21.13 0/9 0/5

47 CCNAl 6/10 0/5

Table 5 summarizes the methylation status of the 10 highest ranking genes in 10 cervical cancer and 5 normal cervices using COBRA. One gene (ADARBl at rank 2) could not be analyzed for methylation as no specific BSP products could be amplified using several combinations of primer pairs. Interestingly, using the BSP products of the other 9 listed genes, 7 (78%) showed methylation in carcinomas (Table 5). Four genes are hypermethylated in all 9 tested cancers, while for SST (7 of 9 carcinomas), HTRA3 (1 of 9 carcinomas) and NPTXl (5 of 10 carcinomas) not all tested carcinomas are hypermethylated. Figure 4 shows representative methylation analysis of 3 genes using COBRA. Three (NNAT, SST and NPTXl) of the 7 hypermethylated gene promoters have been reported to be methylated in tumors previously. Taken

these data together, these findings showed that the relaxation ranking algorithm resulted in a very significant enrichment for genes with a positive methylation status.

A cervical-cancer-specific hypermethylated marker is only of relevance for the diagnosis of (pre-) malignant disease in case normal cervical epithelium is not methylated. COBRA analysis of 5 normal cervices for all 9 genes revealed that 4 genes (DAZL, SYCP3, ZFP42 and NNAT) are hypermethylated in all 5 samples (Table 5). On the other hand, of the 7 genes hypermethylated in cervical cancer specimens, 3 genes (SST, HTRA3 and NPTXl) did not show DNA methylation in any of the normal cervices of 5 independent individuals. We observed the same methylation profile for CCNAl that was reported previously as a cervical cancer specific gene (Kitkumthorn et al., 2006) with hypermethylation in only 6 of 10 tumors but none of the 5 normals (Table 5; Figure 3).

Example 2: BROAD analysis: Genome- wide Promoter Alignment.

The "Database of Transcription Start Sites" (DBTSS) (Suzuki et al., 2004) mapped each transcript sequence on the human draft genome sequence to identify its transcriptional start site, providing more detailed information on distribution patterns of transcriptional start sites and adjacent regulatory regions. The promoters of the above identified TOP3000 genes were separately mapped on the genome-wide alignment of all promoter associated CpG islands. All the promoter sequences were subsequently aligned by clustalW algorithm (Li 2003; Thompson et al., 1994). Treeillustrator (Trooskens et al., 2005) was used to visualize the large guide tree in addition to indicating the location of the known markers. Some regions on the "circle" are denser in known markers than others, indicating that there might be a sequence mechanism located in the small region around the TSS which makes certain genes more methylation-prone. The genes were selected as candidates to be methylated if they were located in a cluster, i.e. less than 9 nodes (distance to the closest neighboring marker) away from a marker already described in the literature. These genes were assigned a score, calculated as follows: if the gene is a known literature marker, score +10, if a known marker is one node away, score +9, if there are markers two nodes away: addition to score = number of markers * 8, etc. The genes were ranked according to this score.

A final gene selection was made based on the ranking, the opportunity to design primers, genes to be known as tumor suppressor genes and expert knowledge on their

function, history and mutation status in other cancer types. Also known genes from literature and previous research were included for confirmation.

A final selection of markers resulting from the above set out approaches, were tested on tissue using the Base5 methylation profiling platform (Straub et al. 2007). Differential methylation of the particular genes was assessed using Base5 methylation profiling platform as follows: DNA was extracted from cervical samples, bisulfite converted, and selected regions of the particular genes were amplified using primers whose sequence represented converted or non-converted DNA sequences. Amplification was monitored in real-time set up using SYBRgreen. Data analyses designed to cope with inherent variance {i.e., noise) in measured Ct and Tm values were applied to withhold 112 different assays for detecting differential methylation of ALX3, ALX4, AR, ARID4A, ATM, AURKA, B4GALT1, BMP2, BMP6, BNIP3, C13orfl8, C16orf48, C9orfl9, CALCA, CAMK4, CCNAl, CCND2, CDHl, CDH4, CDK6, CDKNlB, CDKN2B, CLSTN2, CLU, COLlAl, CPTlC, CTDSPL, CYCLIND2, DAPKl, DBCl, DDX19B, DKK2, EGFR, EGR4, EPB41L3, FOS, FOXEl, GADD45A, GAT A4, GDAPlLl, GNB4, Gst-Pi, HHIP, H00K2, HOXAl, HOXAI l, H0XA7, IGSF4, ISYNAl, JAM3, JPH3, KNDCl, KRAS, LMXlA, LOC285016, LOX, MTAP, MY018B, N0L4, NPTXl, OGFOD2, PAK3, PAXl, PDCD4, PHACTR3, POMC, PRKCE, RAD23B, RALY, RARA, RBP4, RECK, RPRM, SEMA3F, SLC5A8, SLITl, SLIT2, SLIT3, SMPDl, SOCSl, SOXl, SPARC, SPN, SST, TERT, TFPI-2, TLLl, TNFAIPl, TRMTl, TWISTl, UGTlAl, WIFl, WITl, WTl, XRCC3, and ZGPAT in cervical cancer tissue samples.

MATERIAL AND METHODS

Samples:

A total of 201 frozen tissue samples (87 cervical cancer samples, the majority derived from squamous cell carcinomas; and 114 normal tissues) were collected by UMC Groningen. If the tissue contained more than 20% stromal cells, the samples were macro-dissected to enrich for tumor cells.

DNA isolation and bisulphite modification:

DNA was isolated using proteinase K digestion and phenol/chloroform extraction.

DNA concentration was measured using NanoDrop Spectrophotometer. From each sample, up to 2 μg of genomic DNA was converted using a bisulphite based protocol (EZ DNA Methylation Kit™, ZYMO Research, Orange, CA.).

Detection of hypermethylation:

Methylation specific PCR (MSP) primers were designed for each of the genes assessed for (hyper)methylation. An example on primer design spanning a large region of the promoter is provided in Figure 5A and 5B for ALX4.

5 For some genes more primer pairs were designed giving a total of 424 different assays. These assays were applied on 8 sub-arrays of 2 OpenArray™ plates by BioTrove Inc. The beta-actin assay was applied on each sub-array as an internal control. Quality control was performed using an in vitro methylated DNA sample and a negative control sample. The selectivity and the reproducibility were checked. After DNA 10 conversion and purification, beta-actin copy number was determined by qMSP. The equivalent of 1500 beta-actin copies per sample was applied per sub-array of an OpenArray™ plate on a real-time qPCR system (BioTrove Inc.) using the DNA double strand-specific dye SYBRgreen for signal detection.

The cycling conditions were: 90 ⁰ C-IO seconds, (43°C 18 seconds, 49°C 60 seconds, 15 77°C 22 seconds, 72°C 70seconds, 95°C 28 seconds) for 40 cycles, 70 ⁰ C for 200 seconds, 45°C for 5 seconds. A melting curve was generated in a temperature range between 45°C and 94°C.

Analysis of methylation:

For each combination of assays and samples two parameters were collected using an 20 algorithm which is part of the standard data analysis package offered by the supplier. The parameters were the Ct value (threshold cycle number) of the assessed amplicon and the melting temperature of the assessed amplicon. The following data analysis workflow was applied to the results created by the software which came with the system OpenArray™ system: Data was collected for each combination of assays and

25 samples in the two sets of samples used. Results were filtered using the following approach. Read outs from not loaded reaction spaces were removed from analysis. Technical Control assays were removed from the data set. Assays known to not work for other than biological reasons were removed from the analysis. Per sub-array, signals were only interpreted if there was a positive beta-Actin call. Ct values > 0 for

30 each gene were normalized using the Ct values collected for the gene beta-Actin. This resulted in two files containing the results for each set of sample. 201 samples were tested of which 6 gave invalid results. In total 79,170 reactions were performed of which 74,110 were valid. For the data analysis, 2 boundaries were defined: an upper

bound on beta-Actin-normalized-Ct (banCt) and a lower bound on Melting Temperature (Tm). Samples below the banCt boundary and above the Tm boundary are considered to be "methylated", others (including all samples with no signal, i.e. Ct>40) are classified as "unmethylated". In both dimensions the set of candidate boundaries consists of all values in between 2 measurements, plus infinity (the equivalent of no boundary). The set of candidate models for "methylated" then consists of all combinations of candidate Tm lower bound and a banCt upperbound. A score is computed for each of these candidate models, as follows. Count: cancers inside boundaries = true positives (TP), cancers outside boundaries = false negatives (FN), normals inside boundaries = false positives (FP), normals outside boundaries = true negatives (TN). A binomial test was applied to find out how unusual it is to have at least TP successes in (TP+FP) trials where the probability of success is (TP+FN). The lower this probability value is the better. Then quality control data were taken into account to determine the most robust boundaries. Using the standard deviations (StDevQC) observed in the QC, a series of increasingly "noisy" datasets were generated. The measurements are replaced by a value randomly selected from a normal distribution with average equal to the observed measurement and standard deviation equal to StDevQC multiplied by a value that gradually (10 noise levels) increases from 0 to 2. Each time the score of the candidate model is computed by applying the 2 steps above (i.e., count and binomial test). All these scores (11 in total: 1 for "no noise" and 10 for noise levels 0.2, 0.4, ..., 2) are added up to obtain the ultimate accumulated score. The candidate model with the best (i.e. lowest) accumulated score is retained. This same score of the best candidate model for each marker is also used for ranking the markers.

RESULTS

A high throughput, real-time methylation specific detection platform was applied on two groups of samples isolated from cervical cancer tissue and from corresponding normal cervical tissue. In this study it was shown that a number of genes are differentially methylated in cervical cancer. We identified 112 different assays for detecting 96 different genes being differentially methylated in human cervical cancer tissue and normal cervical tissue control samples. The genes identified are ALX3, ALX4, AR, ARID4A, ATM, AURKA, B4GALT1, BMP2, BMP6, BNIP3, C13orfl8, C16orf48, C9orfl9, CALCA, CAMK4, CCNAl, CCND2, CDHl, CDH4, CDK6, CDKNlB, CDKN2B, CLSTN2, CLU, COLlAl, CPTlC, CTDSPL, CYCLIND2, DAPKl, DBCl, DDX19B, DKK2, EGFR, EGR4, EPB41L3, FOS, FOXEl,

GADD45A, GATA4, GDAPlLl, GNB4, Gst-Pi, HHIP, HOOK2, HOXAl, HOXAI l, HOXA7, IGSF4, ISYNAl, JAM3, JPH3, KNDCl, KRAS, LMXlA, LOC285016, LOX, MTAP, MY018B, N0L4, NPTXl, 0GF0D2, PAK3, PAXl, PDCD4, PHACTR3, POMC, PRKCE, RAD23B, RALY, RARA, RBP4, RECK, RPRM, 5 SEMA3F, SLC5A8, SLITl, SLIT2, SLIT3, SMPDl, SOCSl, SOXl, SPARC, SPN, SST, TERT, TFPI-2, TLLl, TNFAIPl, TRMTl, TWISTl, UGTlAl, WIFl, WITl, WTl, XRCC3, and ZGPAT

The resulting assays have the assay details provided in Table 1, Table 2, and Figure 5B. 0

Example 3: Further Assay selection: Base 5 - Lightcycler platform

Of the different assays listed in Table 1 previously identified using the Base5 methylation platform, the top 63 ranked assays plus β-actin (ACTB) were transferred to the Lightcycler platform in order to further fine-tune the selection of the best 5 cervical cancer methylation markers. This platform allows the assessment of markers in a system which is closer to, and provides information valuable for the subsequent development of, a final, scaled up MSP assay. The 64 assays (Table 6) were applied on a 384 well plate by Sigma. Six repeats of the assay set fitted on a 384 well plate. The samples were randomized per plate. O The sample set selected for the Lightcycler analysis was also previously used in the Base 5 analysis in order to make a compared analysis: a total of 27 cervical tumor samples and 20 controls (frozen tissue) were collected by UMC Groningen.

Table 6: The 64 selected assays which were applied on the Lightcycler platform

Tissue slides were deparaffinized using 100% xylene followed by 100% ethanol. Pellet was resuspended in a buffer containing SDS-proteinase K, and DNA was extracted with phenol-chloroform followed by ethanol precipitation. DNA concentration was measured using NanoDrop Spectrophotometer. From each sample, up to 3 μg of genomic DNA was converted using a bisulphite based protocol (EZ DNA Methylation

Kit™, ZYMO Research). After DNA conversion and purification, equivalent of 20 ng of gDNA was used per reaction. All the samples were tested on Lightcycler using

Sybergreen as detector and the amplicon size was determined by capillary electrophoresis.

Quality control was performed using in vitro methylated DNA sample, unmethylated DNA sample (Chemicon International, CA, USA; Cat.# S7821 and Cat.# S7822) and no template control sample (H ₂ O). From the Lightcycler platform, the Ct values (cycle number at which the amplification curves cross the threshold value, set automatically by the software) and melting curves (Tm) were generated. From the capillary electrophoresis platform, size of the amplicon and intensity of the signal detected were generated. For each assay, Tm and amplicon size parameters were determined in in vitro methylated DNA sample, unmethylated DNA sample and no template control sample. The measured Tm and amplicon size values were compared to the calculated values. If the Tm or amplicon size values were out of the range of the calculated ones, the assay was considered as non specific and disqualified. All the 64 assays were specific.

A sample is considered methylated if Ct is under 40 and if Tm and amplicon size are within the boundaries of Tm +/- 2 degrees and amplicon size +/- 10 bp. The intensity of the band detected by capillary electrophoresis had to be higher than 20. Those evaluation criteria have been developed based on concordance with existing Molecular Beacon based qMSP assays, to ensure that the conclusions drawn from these data would be predictive of MSP assays developed subsequently.

DNA methylation calls were compared between cervical cancer and control patients. An assay ranking with the set of samples was generated and the results are summarized in the methylation table of Figure 6. A one-tailed Fisher's exact test was used as a scoring function to rank the candidate markers. The calculation of Fisher's exact test

was based on a formula as described by Haseeb Ahmad Khan in "A visual basic software for computing Fisher's exact probability" {Journal of Statistical Software, vol. 08, issue i21, 2003).

A comparison between the results coming from the Base 5 (Biotrove) and the Lightcycler platforms has been performed. Most of the interesting assays discovered on the Base 5 platform were confirmed on the Lightcycler platform.

Example 4: QMSP

Seventeen assays (ALX3, C13ORF18, DBCl, EPB41L3, GATA4, HOXAI l, JAM3, JPH3, LMXlA, NOL4, PAK3, SLIT2 23672, SLIT2 23676, SOXl, TERT, TFPI2 and TWIST 1 3) were further selected based on their performance on the Biotrove and

Lightcycler platforms and on complementarity analysis to maximize discriminatory power. For these assays, qMSPs using Molecular Beacon as detection system were designed (3 designs, if possible, were evaluated per assay) and tested on control samples. For this selection, assays were judged on several criteria, including background fluorescence, dynamic of the curve, and level of fluorescence generated.

PCR material was used for generating standard curves for quantification of the results.

Five assays did not meet the desired specifications (EPB41L3, SOXl, SLIT2 23672, DBCl, and SLIT2 23676) and may be redesigned in a later phase or can be used on another detection platform. The remaining 12 assays were further tested on converted DNA of cervix cancer cell lines.

All these results were taken into account to decide which assays should be further verified on cervical tissue samples collected by UIg (normal PE tissue samples #13, cancer PE tissue samples #17) and/or UMCG (normal frozen tissue samples #20, cancer frozen tissue samples #27).

Seventeen (CCNAl, CDO1 55928, CDO1 55929, GREMl, GPNMB, HINl, HOXDl, LAMAl, LTB4R, MAL, NDRG2, NID2, NPTX2, RASSFlA, SALL4, SOX17, and TACl) additional good performing assays were also selected for further verification on the cervix tissue samples. These candidates were taken from other in- house cancer projects, and were not tested on the Biotrove/Lightcycler platform as described above.

DNA was isolated from the cervix tissue samples using a phenol-chloroform procedure, quantified using the picogreen method and 1.5 μg of DNA was bisulphite treated using the ZYMO kit.

qMSPs were carried out in a total volume of 12 μl in 384 well plates in an ABI PRISM 7900HT instrument (Applied Biosystems). The final reaction mixture consisted of in- house qMSP buffer (including 80.4 nmol of MgC12), 60 nmol of each dNTP, 0.5 U of Jump Start Taq polymerase (SIGMA), 72 ng of forward primer, 216 ng of reverse primer, 1.92 pmol of Molecular Beacon detection probe, 6.0 pmol of ROX (passive reference dye) and 72 ng of bisulphite converted genomic DNA. Thermal cycling was initiated with an incubation step of 5 minutes at 95°C, followed by 45 cycles (95°C for 30 seconds, 57°C for 30 seconds, 72°C for 30 seconds). A finalizing step was performed at 72°C for 5 minutes to conclude cycling. These conditions were similar for all the test genes as well as for ACTB. Cell lines [in vitro methylated DNA sample and unmethylated DNA sample (Chemicon International, CA, USA; Cat.# S7821 and Cat.# S7822)] were included in each run as positive and negative controls, and entered the procedure at the DNA extraction step. Primers and molecular beacon sequences used for the different qMSPs are summarized in Table 1 and Table 3. Corresponding amplicons are summarized in Table 2.

Ct values were determined using the SDS software (version 2.2.2.) supplied by Applied Biosystems with automatic baseline settings and threshold. The slopes and R ² values for the different standard curves were determined after exporting data into MS Excel. As an example, Figure 7 shows the amplification plot obtained for the standard curve for TAC1_56187 (960000 copies to 9.6 copies of the gene) and Figure 8 shows the amplification plot obtained for the standard curve and for all samples for TAC1_56187. The Ct values plotted against the Log Copies of TAC1_56187 (Figure 9) give a R ² of 0.9995 and the efficiency of the reaction is 99.35%. In addition to the test genes, the independent reference gene β-actin (ACTB) was also measured. The ratios between the test genes and ACTB were calculated to generate the test result. The samples were classified as methylated, unmethylated, or invalid based on the decision tree shown in Figure 10.

A provisional cut-off was defined for each gene, chosen based on the greater of either the highest value seen among the controls or a value 3 times the standard deviation of the values from control samples.

The one-tailed Fisher's exact test as described above was used as a scoring function to rank the candidate markers (Journal of Statistical Software, vol. 08, issue i21, 2003).

Table 7 summarizes the results obtained for TACl 56187. Table 8 summarizes the results obtained for all the tested markers on tissue samples. The individual

performances of the assays are shown in Figure 11 and the assays are ranked according their p-value (Fisher's exact test). The best performing markers were further tested on clinical samples (scrapings).

Table 7: Summary of the test results for TAC1_56187 on cervix tissue samples. In column "methylation status", the black boxes indicate the methylated results; white boxes indicate the unmethylated results.

N Normals 38 |

Table 8: Summary of the performance results of all the tested markers on tissue samples.

Example 5: Best performing markers tested on clinical cervical scraping samples

Cervical scraping samples were collected under the Cervical Cancer Clinical Collaborative Research Agreement study of ONCO with the Gynecology Department of the UMCG hospital. The scraping samples were taken from patients who were referred to the hospital with an abnormal PAP smear or because they were suspected for cervical carcinoma. Gynecological examination under general anesthesia was performed in all cervical cancer patients for staging in accordance with the International Federation of Gynecology and Obstetrics (FIGO) criteria. Control scraping samples were taken from women who visited the hospital for a non-malignant condition, e.g. fibroids, prolaps uteri or hypermenorrhea, and who were scheduled to undergo a hysterectomy. While the patient was under general anesthesia, the cervix was scraped with an Ayres spatula and brush. The scraped cells were suspended in 5- ml PBS. Cytospins for cytomorphological assessment were made (1/5 volume). Cytospins were Papanicolaou stained and routinely classified according to a modified Papanicolaou system (Hanselaar AG. Kwaliteit van cytopathologisch onderzoek in het herziene bevolkingsonderzoek naar baarmoederhalskanker. Nederlands Tijdschrift voor Obstetrie en Gynaecologie 1996; 109:207-210) without knowledge of the clinical data. The remaining 4-ml of the scraped cells was centrifuged, washed, aliquoted, snap-frozen in liquid nitrogen and stored at -80 ⁰ C. DNA was extracted using standard salt-chloroform extraction and ethanol precipitation. DNA of the pellet was used for qMSP of a panel of good performing markers for cervical cancer and also for HPV typing. DNA was extracted from the scraped cells using standard salt-chloroform extraction and ethanol precipitation for high molecular DNA, dissolved in 250 μL TE-4 buffer (10 mM Tris; 1 mM EDTA, pH 8.0) and kept at -2O ⁰ C until tested.

Presence of high risk HPV was analyzed by PCR using HPV 16 and HPV 18 specific primers on DNA of the scraping samples. On all HPV 16- or HPV 18- negative cases, general primer-mediated PCR was performed using two HPV consensus primer sets,

CPI/CPIIG and GP5+/6+, with subsequent nucleotide sequence analysis, as described previously [by Wisman et al Int j cancer 2006]. qMSP was performed after bisulphite treatment on denatured genomic DNA. The assays were carried out as described above. The samples were classified as

5 methylated, unmethylated, or invalid as described above. The results obtained for all the tested markers on scraping samples from cervical cancer patients and from control patients were ranked according their p-value (Fisher's exact test) (Table 9). Some markers have a higher sensitivity for squamous cell carcinoma than for adenocarcinoma (NID2, JPH3, CCNAl) and some markers have a higher sensitivity

10 for adenocarcinoma than for squamous cell carcinoma (JAM3, CDOl, HOXAl 1).

Various combinations of markers were evaluated to see if such a combination could increase the sensitivity while still maintaining a high level of specificity. In all cases, if any marker of a combination panel was positive, the sample was classified as methylated. Examples of the performance of combination of markers are summarized 15 in Table 10. It can be seen that several combinations provided a sensitivity and specificity greater than 90%.

Table 9: Summary of the results obtained for all the tested markers on scraping samples from cervical cancer patients and from control patients (Sens: sensitivity; SCC: squamous cell carcinoma; Ade: adenocarcinoma; cncr: cancer; Ctrl: control).

20

Table 10: Examples of the performance of combination of markers on scraping samples from cervical cancer patients and from control patients (Sens: sensitivity; SCC: squamous cell carcinoma; Ade: adenocarcinoma; cncr: cancer; Ctrl: control).

HPV testing will certainly continue to occupy a significant position in the diagnosis of cervical cancer. With this in mind, the best performing methylation markers were tested on scraping samples from patients who were referred to the hospital with an abnormal Pap smear and these samples were also tested for hr HPV and HPV16. The provisional cut off as defined above was reduced in order to obtain the highest possible sensitivity and specificity compared to the performance of hrHPV. The results of these tests are shown in Table 11. For these testing, the classification of pre-cancerous (CIN) conditions were used. Sensitivity was calculated for samples indicating cancer, CIN 2 and CIN 3, while specificity was calculated for those samples from controls, and those indicating CIN 1 or CIN 0 after cytological examination. Overall the specificity of the methylation markers was higher compared to hr-HPV or HPVl 6 testing but with a lower sensitivity. Combinations of methylation markers (where at least one of the markers scores positive) showed a comparable sensitivity and specificity for cancers and controls, but a much higher specificity for CINO and CINl. The sensitivity for CIN3 and CIN2 is however somewhat lower. In order to increase the sensitivity for CIN3 and CIN2 detection, an analysis was made of combining the

results of methylation markers and HPV16 (Table 12). The sensitivity as well as the specificity increased if HPV 16 was combined with the methylation markers.

Table 11 : Overall summary of the methylation marker(s) results on scraping samples from patients who were referred to the hospital with an abnormal Pap smear, and from cervical cancer and control patients. (Sens: sensitivity; Spec: specificity; CINO, CINl, CIN2, CIN3: cervical intraepithelial neoplasia grade 0, 1, 2, and 3; cncr: cancer; Ctrl: control, NA: not applicable).

Table 12: Overall summary results of methylation marker(s) in combination with HPV 16 on scraping samples from patients who were referred to the hospital with an abnormal Pap smear, and from cervical cancer and control patients. (Sens: sensitivity; Spec: specificity; CINO, CINl, CIN2, CIN3: cervical intraepithelial neoplasia grade 0, 1, 2, and 3; cncr: cancer; Ctrl: control).

As cytology is currently been used and hr-HPV testing has been suggested as primary screening tool in population-based cervical screening, we simulated the effect on the performances of the methylation tests if only cytology (Table 13) or hr-HPV (Table 14) positive patients were selected. The triage simulations were based on the performance results obtained in Table 11 and Table 12. The performance of cytology and hr-HPV testing were based on data from literature.

The performances of the triage tests showed much higher specificity resulting in fewer referrals for colposcopy than did cytology or hr-HPV testing alone but were less sensitive. Testing for hr-HPV types has a higher sensitivity for detecting CIN2+ than cytology. The NPV is close to 100% thus allowing for less frequent screening and longer screening intervals without jeopardizing patients' safety. But, the enthusiasm for using HPV testing in primary screening has been tempered by its somewhat poorer PPV (19%) in comparison with cytological analysis (27%). Using methylation as triage test, the PPVs were much higher.

Taking the limitations of cytology and the decreased disease prevalence due to the introduction of HPV vaccination programs into account, it is proposed to use a highly sensitive and objective screening test such as HPV DNA testing to identify the rare cases of cancer precursors and to combine it, when positive, with another test which has a high degree of specificity, such as methylation testing. Moreover, methylation is measuring changes in the host cells, as precursor of cervix cancer, while HPV is detecting the causative agent. This is an ideal methodology for a screening and a triage assay because they should measure different but complementary biological signals. Table 13: The simulation of the performance of Cytology test as a first-line screening test on 70000 women and the methylation marker test(s) in- or excluding HPV 16 as triage test. (Sens: sensitivity; Spec: specificity; CINO/1, CIN2+: cervical intraepithelial neoplasia grade 0 and 1, and grade 2 and 3 and cancers; PPV: positive predictive value; NPV: negative predictive value).

Table 14: The simulation of the performance of hr-HPV test as a first-line screening test on 70000 women and the methylation marker test(s) in- or excluding HPV 16 as triage test. (Sens: sensitivity; Spec: specificity; CINO/1, CIN2+: cervical intraepithelial neoplasia grade 0 and 1, and grade 2 and 3 and cancers; PPV: positive predictive value; NPV: negative predictive value).

REFERENCES

The disclosure of each reference cited in this disclosure is expressly incorporated herein.

Badal et al, The human papillomavirus- 18 genome is efficiently targeted by cellular DNA methylation, Virology 324 (2004) 483- 492.

Barringer KJ, Orgel L, Wahl G, Gingeras TR._Gene. 1990 Apr 30; 89(1): 117-22.

Brink et al, HPV detection methods, Disease Markers 23 (2007) 273-281.

Cottrell, SE et al. A real-time PCR assay for DNA-methylation using methylation- specific blockers. Nuc Acids Res 2004, 32, 1.

Cross, SH et al. Nature Genetics 1994, 6, 236-244.

Deng, D. et al. Simultaneous detection of CpG methylation and single nucleotide polymorphism by denaturing high performance liquid chromatography. 2002 Nuc. Acid Res, 30, 3.

Ehrich, M. et al. 2005. Introduction to DNA methylation analysis using the MassARRAY system. SEQUENOM™ product preview note.

Finan, M.A. et al. (1996) Radical hysterectomy for stage IBl vs IB2 carcinoma of the cervix: does the new staging system predict morbidity and survival? Gynecol Oncol, 2, 139-147.

GaIm et al. Enzymatic Regional Methylation Assay: A Novel Method to Quantify Regional CpG Methylation Density. Genome Research, Vol. 12, Issue 1, 153-157, January 2002.

Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, et al. (2004) Bioconductor: open software development for computational biology and bioinformatics. Genome Biol 5: R80.

Gitan RS et al. Methylation-specifϊc oligonucleotide microarray: a new potential for high-throughput methylation analysis. 2006 Genome Research 12:158-164.

Gonzalgo and Jones, Rapid quantitation of methylation differences at specific sites using methylation-sensitive single nuclear primer extension. 1997 Nuc Acid Res, 25, 12.

Kalantari et al, Conserved Methylation Patterns of Human Papillomavirus Type 16 DNA in Asymptomatic Infection and Cervical Neoplasia, Journal of virology, Dec. 2004, p. 12762-12772.

Kang et al, Inverse correlation between RASSFlA hypermethylation, KRAS and BRAF mutations in cervical adenocarcinoma, Gynecol Oncol 2007 Jun; 105(3):662-6. Epub 2007 Mar 13.

Keating et al, Ki-67, Cyclin E and pl6 (INK4A) are complementary surrogate biomarkers for human papillomavirus-related cervical neoplasia. Am J Surg Pathol 25, 884-891 (2001).

Kitkumthorn, N. et al. (2006) Cyclin Al promoter hypermethylation in human papillomavirus-associated cervical cancer. Bmc Cancer, 6.

Koopman,L.A. et al. (1999) Recurrent integration of human papillomaviruses 16, 45, and 67 near translocation breakpoints in new cervical cancer cell lines. Cancer Research, 59, 5615-5624.

Kwoh DY, Davis GR, Whitfield KM, Chappelle HL, DiMichele LJ, Gingeras TR. Transcription-based amplification system and detection of amplified human immunodeficiency virus type 1 with a bead-based sandwich hybridization format. Proc Natl Acad Sci U S A. 1989 Feb; 86(4): 1173-7.

Li KB. ClustalW-MPI: ClustalW analysis using distributed and parallel computing.

Bioinformatics 2003;19(12): 1585-6.

Liu, G. et al. (2003) NetAffx: Affymetrix probesets and annotations. Nucleic Acids Res., 31, 82-86.

Malinowski D. (2007) Multiple biomarkers in molecular oncology. Molecular diagnostics application in cervical cancer detection. Expert Rev. MoI. Diagn. 7(2), 117- 5 131.

Rao et al., Chromosomal amplifications, 3q gain and deletions of 2q33-q37 are the frequent genetic changes in cervical carcinoma. BMC Cancer 4(1), 5 (2004).

Rebhan, M. et al. (1997) GeneCards: Integrating information about genes, proteins and diseases. Trends in Genetics, 13, 163. 0 Shiraisi, M et al. Biol Chem. 1999, 380(9): 1127-1131.

Straub, J. et al., A64- AACRMD (2007): Base5, a versatile, highly integrated high- throughput methylation profiling platform for Methylation-Specific PCR based marker identification applied to CRC.

Suzuki Y, Yamashita R, Sugano S, Nakai K. DBTSS, DataBase of Transcriptional5 Start Sites: progress report 2004. Nucleic Acids Res 2004; 32(Database issue):D78-81.

Thomassin H. et al. MethylQuant: a sensitive method for quantifying methylation of specific cytosines within the genome. 2004 . Nuc Acid Res 32, 21.

Thompson JD, Higgins DG, Gibson TJ. CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position- 0 specific gap penalties and weight matrix choice. Nucleic Acids Res 1994; 22(22):4673-80.

Tost, J. et al. Analysis and accurate quantification of CpG methylation by MALDI mass spectrometry. Nuc. Acid Res, 2003, 31(9): e50.

Trinh B. et al. DNA methylation analysis by MethyLight technology. Methods. 2001 5 Dec; 25(4).

Trooskens G, De Beule D, Decouttere F, Van Criekinge W. Phylogenetic trees: visualizing, customizing and detecting incongruence. Bioinformatics 2005; 21(19):3801-2.

van DongenJ.J. et al. (2003) Design and standardization of PCR primers and protocols for detection of clonal immunoglobulin and T-cell receptor gene recombinations in suspect lymphoproliferations: report of the BIOMED-2 Concerted Action BMH4- CT98-3936. Leukemia, 17, 2257-2317.

Virmani et al. (2001) Aberrant methylation during cervical carcinogenesis. Clin Cancer Research, Mar; 7(3):584-9.

Xiong, Z.G. and Laird, P.W. (1997) COBRA: A sensitive and quantitative DNA methylation assay. N ^' ucleic Acids Research, 25, 2532-2534.

Zeschnigk M. et al. A novel real-time PCR assay for quantitative analysis of methylated alleles (QAMA): analysis of the retinoblastoma locus. 2004. Nuc Acid Res

32, 16.