The Study Of Methylation-sensitive High Resolution Melting(MS-HRM) Detecting Stool DNA Methylation For Colorectal Cancer Screening

Background and ObjectiveColorectal cancer (CRC) is among the most common gastrointestinal tumors. In China, CRC becomes the third most common cancer after lung cancer and gastric cancer. The CRC patients are most commonly made a definite diagnosis at middle or advanced stage, because of unawareness of early signs and symptoms. So, the best available strategy to reduce the incidence and mortality of CRC is the detection and removal of precancerous lesions through CRC screening.Currently, colonoscopy, fecal occult blood test (FOBT) and stool DNA (sDNA) testing are the common methods for CRC screening. Colonoscopy is invasive, expensive, risky and difficult to accept for patients. So, it is not a good choice for large-scale CRC screening. To date, FOBT is the most popular in the large-scale CRC screening at home and abroad because it is cost-effective, simple, noninvasive, good tolerability and safe, while it has a low sensitivity and specificity with the ranges of37.1%-79.4%and86.7%-97.7%, respectively. Advanced adenomas (AA, which are defined as size>1cm or having villous histology or high-grade dysplasia), which have a significant potential to transform into invasive adenocarcinomas, is one of the recognised precursors of CRC. It is a key measure to reduce the incidence and mortality rates of CRC by screening and by removal of AA. FOBT has a low detection rate of AA. However, the sDNA test has better performance of detecting AA than FOBT, which potentially improves CRC prevention. A study showed that the sDNA test was significantly more sensitive than FOBT for the detection of AA (18%VS.10%, P=0.001), while specificities between the two tests remained comparable (94%VS.95%, P>0.05). To the above methods, sDNA test with a lot of advantages is to be preferred for CRC screening. The ranges of its sensitivity and specificity are52%～91%and91%～97%, respectively. Another big advantage is patient preference for sDNA test. A survey of4042subjects showed that a higher percentage of patients preferred sDNA testing (45%) to both FOBT (32%) and colonoscopy (15%) for routine screening (P<0.001) and patients willing to undergo colonoscopy, sDNA testing, and FOBT perceive sDNA testing to have a number of advantages over the other two tests.However, the utilization of sDNA test in CRC screening is currently limited by cost considerations mainly resulting from lack of cost-effective detection techniques. So it is time to look for an available technique with good performance.The high-resolution melting curve analysis (HRM) is a relatively new technique for molecular detection. It refers to mutation scanning, methylation analysis, genotyping and sequence matching. HRM with the closed tube operation is faster, simpler, and less expensive than other approaches requiring separations. Based on its low cost, high-throughput, simple to operate, free of contamination, simplicity, flexibility, nondestructive nature, free from the impact of the variant or methylated location, and superb sensitivity and specificity, HRM is quickly becoming the alternative tool to screen patients. In our previous study, we demonstrated that HRM was a simple, reliable, and sensitive method for detecting DNA mutations in the stool samples from patients with colorectal neoplasms. Moreover, it was also validated that methylation-sensitive high resolution melting (MS-HRM) has the excellent potential and reliability for detecting DNA methylation levels in tissue and blood samples. With a very big difference from the tissue and blood samples, human DNA accounts for roughly0.01%of total sDNA and the majority of human DNA is from normal colon mucosa. As a result, there are technical challenges to analyze methylation of human DNA from stool for CRC detection. However, so far, its performance of DNA methylation analysis for screening CRC in stool samples has not yet been elucidated.In the present study, we evaluated the feasibility of using MS-HRM to detect methylation in SFRP2and VIM promoters as were stool-based DNA markers for detection of CRC and precancerous.Methods and resultsExperimental subjectsPatients who were older than40years of age and identified with CRCs, AAs or normal controls (NCs) by colonoscopy at Nanfang Hospital (China) were enrolled in this study from October2011to March2013. Those who were initially identified as CRC and AA by colonoscopy were further verified by histologic diagnosis. The study was approved by the local Ethics Committee. The individuals gave their written informed consent.The primer sequences, the length of amplicons, the annealing temperature and amplified region for MS-HRM, MSP and BSP assays on SFRP2and VIM were listed in Additional table1.Sample collection Stool samples were obtained after commencing bowel preparation the day before either colonoscopy or surgical neoplasm resection in a self-collection manner and sent to the laboratory within24hours after defecation, and stored at-80℃. Stool was collected more than5days after any laxative preparation and before endoscopic or surgical resection. Fresh tissue samples were collected at surgery or endoscopy from Nanfang Hospital. After removal, the tissue samples were immediately snap-frozen in liquid nitrogen and stored at-80℃until analysis.DNA ExtractionDNA was isolated from colonic tissues (15～25mg) and stool samples (180～220mg) using QIAamp DNA Mini Kit and Q1Aamp DNA Stool Mini Kit (Qiagen, Hilden, Germany), respectively, according to the manufacturer’s protocols.The extracted DNA were quantified with the NanoDrop ND-1000Spectrophotometer (Isogen Life Science, IJsselstein, Netherlands) and the samples were diluted to a final concentration of50μg/ml. Extracted DNA was excluded if the concentration was<50μg/ml or the A260/A280ratio was out of1.7～1.9.Sodium bisulfite modification500ng DNA was bisulfite modified using the EZ DNA methylation-Gold kit (Zymo research, USA) to convert all unmethylated cytosines to uracils while leaving methylcytosines unaltered and eluted in lOμL of elution buffer. After sodium bisulfite modification, the DNA is ready for immediate analysis.MS-HRM analysisPCR amplification and HRM were performed on the LightCycler480(Roche Diagnostics, Penzberg, Germany). Each PCR reaction mix consisted of a total volume of25μl containing12.5μl2x EpiTect HRM PCR Master Mix (include HotStarTaq Plus DNA Polymerase, EpiTect HRM PCR Buffer, and dNTPs),1.9μl10μM (each) primer mix (consisting of10μM forward primer and10μM reverse primer),2μl Template DNA (5ng/μl bisulfite converted DNA) and8.6μl RNase-free water. The optimized cycling protocol for HRM analysis on the LightCycler480were as follows:95℃for5min, followed by45cycles at95℃for10sec,54℃(VIM) or60℃(SFRP2) for30sec,72℃for10sec. Run HRM analysis to span a temperature range from65℃～95℃with the ramp rate of0.02℃/s and25acquisitions per second. Finally, cool samples after HRM at40℃for1sec. The standard curves with known methylation ratios were included in each assay and were used to deduce the methylation ratio of each sample. All samples were amplified and detected in duplicate.Using the LightCycler480software (Roche), the melting curves were normalized by calculation of two normalization regions before and after the major fluorescence decrease representing the melting of the PCR product. Output plots were in the form of normalized melting curves. Based on the standard curves, HRM data were classified into different methylation categories by two independent observers.Assessing the Linearity of MS-HRMWe treated the standard serial dilutions of fully methylated DNA (100%): unmethylated (0%) as follows:100:0,75:25,50:50,25:75,5:95,1:99,0:100in triplicates (each containing100ng template DNA) with sodium bisulfite and duplicated MS-HRM assays. The unmethylated DNA (0%) was used as control reference DNA to calculate the differential fluorescence. So there were six data points of differential fluorescence values for each dilution. We evaluated the linearity of MS-HRM assay by linear regression analysis.Measuring the Reproducibility of MS-HRMTo measure reproducibility of bisulfite conversion of DNA and MS-HRM assay, we performed an intra-and inter-assay at three different levels of methylation samples in SFRP2(S1, S2, S3) and VIM (V1, V2, V3) respectively. Four different aliquots from each of the three samples were subjected to bisulfite conversion on4different days. Thus, we ideally obtained four separate bisulfite-converted DNA samples (marked as B1through B4) with the same values for methylation. The MS-HRM assays were performed simultaneously on these four separate bisulfite-DNA samples from each of the three cases. The value for the differential fluorescence peak of each replicate was got using the LightCycler480software. The coefficient of variation (CV) of the differential fluorescence values from B1to B4was measured for each case, which mainly reflects the variations amongst bisulfite modifications. Similarly, we repeated the MS-HRM assays on the three cases, which were simultaneously subjected to bisulfate conversion, four times (marked as M1through M4) on4different days. We calculated the CV of differential fluorescence values from M1to M4, which primarily depend on variations of each MS-HRM assay.Methylation-Specific Polymerase Chain Reaction (MSP)MSP was carried out to amplify the methylated or unmethylated alleles in a blinded manner by using EpiTect MSP Kit (Qiagen). Each PCR reaction mixture consisted of a total volume of50μl containing25μl EpiTect Master Mix (Containing HotStar Taq d-Tect Polymerase, Tris-Cl, KC1,(NH4)2SO4, MgCl2, and dNTPs)(Qiagen),4μl Diluted primer mix (10μM Primer A,10μM Primer B),2μl Template DNA,19μl RNase-free water. The optimized cycling protocol was95℃for10min, followed by40cycles of94℃for15s, specific annealing temperature (in Additional table1) for30s,72℃for30s, and a final extension at72℃for10min. The PCR products were separated by electrophoresis in2.0%agarose gels and visualized by ethidium bromide staining and ultraviolet transillumination. After bisulfite convertion, the universal methylated DNA and unmethylated DNA were used as a positive control and a negative control for methylation, respectively. Bisulfite sequencing PCR (BSP)The primers used for DNA sequencing was shown in Additional table1. Amplification was performed in a final reaction mix of50μl containing5U/μl Taq DNA pol0.8μl,10×PCR Buffer(with Mg2+)5μl,10μM Forward Primer1μl,10μM Reverse Primer1μl,3μl Template and38.2μl ddH2O. The conditions for PCR was as follows:Initial denaturation at98℃for4min, followed by40cycles of denaturation at94℃for45sec, annealing temperature for45sec and extension at72℃for1min, and with a final extension72℃for8min, and finally Hold at4℃for>4min. The products were separated in a3.0%agarose gel to verify the specificity of the product, and the remaining material was purified with a DNA Gel Purification kit and cloned into the pUC18-T vector (Sangon Biotech, Shanghai, China). Ten colonies of each specimen were randomly chosen for plasmid DNA extraction and then sequenced using an ABI3730DNA Analyzer (Applied Biosystems, Foster City, CA).Fecal occult blood immunochemical test (FOBT)An immunochemical FOBT was performed using Fecal Occult Blood Gold Gel Stripe (W.H.P.M.INC, Beijing, China) in accordance with the manufacturer’s instructions.Statistical AnalysisAll data and statistical analysis were performed with SPSS13.0(The Predictive Analytics Company; Chicago, IL). In stage Ⅱ, the sensitivity and specificity of MS-HRM was compared with that of bisulfite sequencing using McNemar’s test for matched pairs test. The κ coefficient between the sDNA tested by MS-HRM and the tissue DNA tested by BSP was calculated using a McNemar’s test to assess the feasibility of MS-HRM for detecting DNA methylation in stool samples. Comparison of the diagnostic sensitivity and specificity between MS-HRM and MSP was also performed using McNemar’s test for matched pairs test. The χ2and Fisher’s exact test were used to compare proportions between various subgroups. The ages and gender between groups were compared by two-sample t test and χ2test, respectively. The linear regression analysis was performed to evaluate the linearity of MS-HRM assay. The coefficient of variation (CV) of the differential fluorescence values was calculated to measure reproducibility of bisulfite conversion of DNA and MS-HRM assay. P<0.05was considered to be statistically significant.ResultClinical and Histopathologic Characteristics of Eligible SubjectsFinally, there are37CRCs,26AAs in stage Ⅱ and40CRCs,36AAs and57age-(P>0.05) and gender-(P>0.05) matched NCs in stage Ⅲ. Among the patients, a definitive diagnosis was established on the complete resection specimens of colonoscopy or/and surgical resection specimens by two independent pathologists. All patients were Chinese. In stage Ⅱ,the average age of CRC and AA groups were65years (43～80years) and67.5years (40～82years), and the sex ratio them were21/16and11/15, respectively. In stage Ⅲ, the average age of CRC, AA and NC groups were8years (41～84years),58years (40～72years) and59years (43～75years), and the sex ratio them were26/14,19/17and34/23, respectively. The difference of age and gender between the case and control group both were insignificant (both P>0.05).The detection limit of MS-HRM and MSP for detecting methylated SFRP2and VIMBoth of SFRP2and VIM MS-HRM assays were able to detect reproducibly1%methylated DNA in a background of unmethylated DNA. But the MSP assays of the two target genes cannot detect<5%methylation.In order to compare directly the detection limit between MS-HRM and MSP, some samples with different methylation levels detected by MS-HRM were chosen to detect by MSP again. Among those samples, C5, CIO and C14were detected methylation levels of<1%by MS-HRM, while they were identified as methylation negative via MSP in the both two target genes. C26, C34and C36were detected in the range of1～5%in SFRP2and C25, C26and A10were also identified in the same range in VIM by MS-HRM, while they had the negative results of MSP reaction. Besides, C3, C8and C31were detected methylation levels of>5%in the two target genes by MS-HRM, and they also had the positive results of MSP.The feasibility of MS-HRM for detecting sDNA methylation in CRC screeningWhen the normalized melting curves of sample was out of the area between0%and1%methylated DNA, the sample was considered methylated.In stage Ⅱ, methylated SFRP2, VIM, or both were also identified by BSP in58of63(92.1%) DNAs from tumor tissue. Specifically,34of37CRC samples (91.9%),24of26AA samples (92.3%), and2of63corresponding normal colonic mucosal samples (3.2%) taken from the same patient were identified as methylation-positive. The methylation levels of the two target genes in tumor tissues were significantly higher than those in adjacent normal tissues (P<0.001).63stool samples and corresponding neoplasm tissues were detected respectively by MS-HRM and BSP. MS-HRM detected55out of the63(87.3%) stool samples at least one target gene methylation including33of the37(89.2%) CRCs and22of the26(84.6%) AAs. The rate of methylation in stool samples detected by MS-HRM was highly consistent with that in corresponding tissue samples detected by BSP (P<0.05, κ=0.843in CRCs; P<0.05,κ=0.629in AAs; P<0.05, κ=0.744in both). All the results of MS-HRM were verified by BSP without false positive or false negative results. The sensitivity and specificity of MS-HRM in stage Ⅱ achieved100%.The performance of MS-HRM for screening colorectal neoplasm by detecting sDNA methylationThe methylation positive frequency of VIM or/and SFRP2detected by MS-HRM were significantly higher in stools from the case group (71/76,93.4%) than the normal control group (5/57,8.8%)(P<0.05). However, the difference in methylation positive frequency between CRCs (37/40,92.5%) and AAs (34/36,94.4%) was insignificant (P>0.05). The MS-HRM diagnostic sensitivity (93.4%,71/76) of combined detection of the two markers was higher than that of MSP in patients (82.9%,63/76; P<0.05). The diagnostic sensitivities for MS-HRM (92.5%,37/40) and MSP (82.5%,33/40) in CRC stool samples were insignificantly different (P>0.05), but were consistent (κ=0.553). No significant difference between the diagnostic sensitivities of MS-HRM (94.4%,34/36) and MSP (83.3%,30/36) in AA stool samples was observed (P>0.05), but the measure of agreement (κ) was statistically insignificant (P>0.05). However, in the NC group the diagnostic specificity of MS-HRM (91.2%,52/57) was significantly higher than that of MSP (73.7%,42/57). In addition, the diagnostic sensitivity (71/76,93.4%) of MS-HRM with combination detection of the two markers was significantly higher than FOBT examination (43/76,56.6%) in patients including CRCs and AAs (P<0.05). The diagnostic sensitivity of MS-HRM was92.5%(37/40) in40CRC cases and94.4%(34/36) in36AA cases, respectively, significantly higher than FOBT examination (67.5%(27/40) in CRC cases (P<0.05) and44.4%(16/36) in AA cases (P<0.05)), while no significant association existed in the specificity between MS-HRM with combination detection rates (91.2%(52/57)) and FOBT (89.5%(51/57))(P>0.05).The linearity and reproducibility of the MS-HRM assayThe values of differential fluorescence peak for the serial dilutions of methylated DNA (from0%to100%) were obtained by the LightCycler480software. And linear regression analysis was carried out. The mean value of six data of each dilution was plotted against the methylation percentage of each dilution to generate a linear fitting curve. The P value and linear regression coefficients were as follows:P<0.001, R2=0.957for SFRP2; P<0.001, R2=0.954for VIM.According to the degree of methylation, the intra-assay CVs of SFRP2and VIM ranged from0.07160to0.1708and0.1428to0.2605and the inter-assay CVs of SFRP2and VIM ranged from0.04454to0.06783and0.1298to0.1970. We found that both bisulfite-to-bisulfite variation and run-to-run variation were very small in the two genes. More importantly, the inter-assay CVs were much smaller than the intra-assay CVs in both target genes. Compared with that of sodium bisulfite conversion, the reproducibility of MS-HRM assay was perfect. ConclusionThe detection limit of MS-HRM and MSP for detecting methylated SFRP2and VIM1. The MS-HRM assays were able to detect reproducibly as low as1%methylated DNA in a background of unmethylated DNA.2. The detection limit of MSP assays were only5%methylated DNA in a background of unmethylated DNA.3. The MS-HRM assays for detecting DNA methylation were more sensitive than MSP.The feasibility of MS-HRM for detecting sDNA methylation in CRC screening1. It is feasible to screen CRC by detecting the methylation of SFRP2and VIM.2. It is also feasible to apply MS-HRM to detect sDNA methylation for CRC screening.The performance of MS-HRM for screening colorectal neoplasm by detecting sDNA methylation1. The sensitivity and specificity of MS-HRM, MSP and FOBT were93.4%and91.2%,82.9%and73.7%, and56.6%and89.5%.2. MS-HRM has better performance in methylated sDNA detection than MSP, a traditional technique, for CRC screening.3. It also has better performance in CRC screening than FOBT, a traditional method, that MS-HRM is applied to detect sDNA methylation.The linearity and reproducibility of the MS-HRM assay1. The MS-HRM assay showed fairly good linearity for sDNA methylation analysis.2. The results detected by MS-HRM are stable and reliable.In conclusion, MS-HRM shows potential for CRC screening.