High-resolution Melting Curve Analysis Technology (HRMA) Testing Stool DNA Mutation To Screen Colorectal Neoplasm For The Evaluation Of Diagnostic Performance

Background and ObjectionIncidence of colorectal cancer （CRC） is listed3in global malignant tumor, and our country is4-6, which is continuously increasing trend. In Guangdong, the incidence of CRC in increase is very fast. The prognosis of CRC is related closely with early diagnosis. The cancer evolution mode of most of CRC is "normal-adenoma-cancer". The evolution from recognizable adenoma to invasive carcinoma need about5-10years, which provided a plenty of time to screen and prevent. The cancerous adenoma mainly resulted from the late adenoma has been confirmed by a lot of studies （refers to:>1cm in diameter or histologically confirmed for villous adenomas or severe atypia hyperplasia of adenomas）. Pre-cancerous lesions such as adenomas are removed in time by clinical screening and can reduce mortality and morbidity of CRC.Colonoscopy is recommended because of the high accuracy and treatment, but colonoscopy requires specialized endoscopic doctor, risky, high cost, difficult to accept for patients, and difficult to used in the screening of asymptomatic people. Fecal occult blood test （FOBT） is to be preferred because it’s simple, noninvasive, economic, and good tolerability, but there are high false positive and false negative rate, and need to repeatedly inspect sample. Although FOBT can reduce cancer mortality, the effect on incidence of cancer is small owing to the poor detection rate of adenoma; stool DNA testing has many advantages, such as only-one submission, easy to accept, high compliance, without effect from medication or dietary control. A large number of studies have shown that the sensitivity and specificity of fecal DNA testing is better than FOBT, especially adenoma detection rate is striking. Stool DNA testing has be listed in The United States guide as candidate program.Current stool DNA testing programs still need to improve the sensitivity and specificity, reduce the costs, and simplify operations. In addition, the further optimization/replacement of current markers, the upgrade of detection methods are also the key. With increasing sensitivity and specificity of stool DNA testing methods. There are some prevalent deficiencies, such as high costs, difficult to operate, too specialized, free to affect on the tested target gene loci and others. Therefore, seeking to a convenient and easy-to-use, cost-effective detection methods is a necessary condition to enhance the practical value of stool DNA testing.The high-resolution melting curve analysis （HRMA） is the closed tube operation was jointly developed by the Wittwer laboratory of the University of Utah and the United States Idaho PCR technology:high-density fluorescence data collection, the melting curve is directly used to detect the small sequence difference in PCR amplified fragment. There are some advantages of HRMA:low cost, high-throughput, non-polluting, efficient, simple to operate, free from the impact of the location variance, good sensitivity and specificity. In2003, HRMA was introduced to detect mutations. The application and development of HRMA have sprung up, and become more and more attention in medical molecular diagnostics. Meta-analysis study showed the sensitivity of HRMA was97.5%（95%confidence interval （CI）,96.8-98.5）, and specificity was95.8%（95%CI:95.3-96.3）. The source of the DNA samples and saturation dye type had no effect on the sensitivity of HRMA. Our previous studies have shown HRMA obtained a higher detection rate of stool DNA mutation detection, and potential clinical value. There are still no relevant studies to evaluate HRMA method as stool DNA testing screening for CRC. Therefore, based on our previous study, DNA sequencing as the gold standard, this research was planned to expand the sample size, to update test design, and to optimize the experimental conditions and operational processes in order to more comprehensively evaluate the accuracy of HRMA. First, for the establishment and optimization of the HRMA operating processes and the experimental parameters, the HRMA was used to detect KRAS and TP53mutation in tissue samples; next, for a comprehensive evaluation of the sensitivity and specificity, the HRMA was used to detect KRAS and TP53mutation in fecal DNA in two independent sample sets （including the stage of "learning set" and "validation set"）; then, the sensitivity of HRMA was assessed in DNA serial dilutions of test; finally, the relationship between fecal KRAS, TP53mutations detected by HRMA and the parameters of clinical pathological features of the colon neoplastic lesions and diagnostics value were investigated.Materials and MethodsExperimental subjectsConsecutive subjects over40years of age, from Huizhou Municipal Central Hospital and Nanfang Hospital （China） were enrolled in this study from January2009to June2010. All subjects were identified with CRCs, AAs or normal control （NC） by colonoscopy. Subjects were matched for age and chosen to represent a mixed distribution of neoplasms from both proximal and distal colorectal sites （Table2）. Individuals were excluded who had any of the following conditions, including contraindication to colonoscopy, personal history of or coexistent with cancer, active therapy with chemotherapy or radiation therapy for a concurrent cancer, personal history of colorectal adenomas or CRC, familial adenomatous polyposis, hereditary non-polyposis CRC, any other high-risk conditions （inflammatory bowel disease, strong family history of CRC （two or more first degree relatives with CRC, or one or more first-degree relatives with CRC younger than age50）, prior colorectal resection for any reason, current pregnancy or lactation） and the weight of stool samples less than5g. At each site, the preparation for, and performance of colonoscopy, was done according to standard operating procedures. CRC and AA initially identified by colonoscopy were further verified by histological diagnosis. Cancers were staged in accordance to Dukes’classification system. This investigation was approved by each clinic institutional review board, and comprised two clinical pilot studies. All patients or their legal representatives signed informed consent.Preparation of fecal samplesWithin48hrs after collection, stool samples were sent to our laboratory and stored in-80℃. Patients were instructed to collect an aliquot of feces （at least5g） using a fecal collection tube, and to store the samples in a hermetically sealed plastic container. More than1week after any colorectal diagnostic procedure or cathartic preparation, and before either endoscopic or surgical neoplasm resection, stools were collected.DNA extraction and dilutionTissue DNA was isolated from formalin-fixed-paraffin-embedded （FFPE） neoplasm tissue using the QIAamp DNA FFPE tissue kit.40neoplasm areas were marked on the haematoxylin and eosin （H&E）-stained slides, and corresponding unstained slides were manually micro-dissected so that at least50%neoplasm cells were contained in the samples. Stool DNA was extracted with QIAamp DNA Stool Mini Kit after homogenization. Processes were operated according to the instructions of the manufacturer. The concentration and purity of extracted DNA were measured using the NanoDrop ND-1000Spectrophotometer and the samples were diluted to a final concentration of50ng/μl. Extracted DNA was excluded if concentration was less than50ng/μl. The absorbance at260/280nm was1.8-2.0, and at260/230nm was2.0-2.5.HRMA assayPrimers for TP53/KRAS were designed from the NCBI RefSeq NG₀17013.1/NG₀07524.1. The primers and the length of all amplicons are listed:KRA Sexon2（92bp） f-TTATAAGGCCTGCTGAAAATGACTGAA, r-TGAATTAGCT G TATCGTCAAGGCACT; exon2（155bp） f-TTATAAGGCCTGCTGAAAATGA C TGAA, r-TGAATTA-GCTGTATCGTCAAGGCACT. Exon3f-GACTGTGTTT CTCCC TTCTC, r-TGTACTGGTCCCTCATTGC; TP53Exon5f-GTGCAGCTGT GGGTTGATT; r-AACCAGCCCTGTCGTCTCT, Exon6f-GATTCCTCACTGAT TGCTCTTAG, r-GG GCACCACCACACTATG,Exon7f-TTGGGCCTGTGTTAT CTCCT, r-TGGCAAGT GGCTCCTGAC,Exon8f-TTGCTTCTCTTTTCCTATCC TGA, r-GCTTCTTGTCCTG CTTGCTT. Two fragments of KRAS exon2with92bp and155bp each, were amplified. PCR was carried out to amplify KR AS exon2（155bp） on a LightCycler480and KRAS exons2（92bp）-3, TP53exons5-8on Rotor-GeneTM6000. Each PCR reaction mixture contained2μl of DNA solution （100ng）,200nmol/L of each primer,2.5mmol/L MgCl2,12.5μl of LightCycler480High Resolution Melting Master/qPCR Master Mix, and ddH2O to a final volume of25μl. PCR paremeters were95℃for5mi n, followed by50cycles of15s at95℃,63℃/60℃for50s to72℃for30s, and10min at72℃. After amplification, the PCR product was denatured at95℃for5min and cooled to40℃for1min to allow heteroduplex form ation. The final HRMA step was performed from65℃to95℃with an increa se in25acquisitions per degree. For sample analysis, after normalization and t emperature-adjustment steps, melting curve shapes were compared between the neoplasm samples and control samples. Wild-type genes were used as a negati ve control. All samples, including wild-type, were plotted according to their m elting profiles. In the different graphs, the melting profiles of each sample wer e compared with those of the wild-type （which were converted to a horizontal line）. Significant deviations from the horizontal line were indicative of sequen ce changes within the amplicon. Samples with aberrant melting curves were re corded as HRMA mutation positive. Data were acquired and analyzed using th e Gene Scanning software1.5（Roche）/the associated RotorGene Series Software （V1.7.25）. All analyses were done by two independent researchers. The final j udgment was concluded by consensus after joint viewing of the melting curves from both researchers. If the first PCR products were not available for the m utational analyses of the melting curves, the second PCR with the same primer s were performed. All samples were amplified and detected in duplicate.DNA sequencing Data from direct DNA sequencing was compared to results from the HRMA assay. Primers used in the HRMA assay were also used for DNA sequencing （Table1）. All PCR products were analyzed by DNA sequencing in order to evaluate the accuracy of HRMA in stool DNA testing. First, the PCR products were purified by digest:1. PCR products of each sample was detected and confirmed the purpose of the strip single using1%agarose gel;2. PCR products with a single strip, the SAP （shrimp alkaline phosphatase） and Exo Ⅰ（exo-nuclease） were used to digest and purify; BioMIGA agarose gel extraction kit was used to recycle by tapping the non-specific bands after agarose gel detected. Then, on the ABI3730DNA sequencing instrument, the one-way Sangers sequencing the method was performed. The3730XL software DNAMAN software were applied to analyze the sequencing data to obtain the final results. DNA sequencing of all samples were sent to the Chaoshi-bio for detection analysis.Statistical analysisThe sensitivity and specificity of the results acquired from HRMA were compared with those results acquired from sequencing. McNemar’s test for matched pairs test was used to compare the sensitivity, specificity, negative predictive value, positive predictive value of the HRMA/mutation with direct sequencing/clinicopathological parameters. Comparison of proportions between various subgroups was based on the Chi-square and Fisher exact test. Comparison between the stool DNA test and tissue DNA test was performed using the kappa coefficient. P≤0.05were considered statistically significant. All data were analyzed using the SPSS version13.0statistical package （SPSS Inc, Chicago, Illinois, USA） and EpiCalc2000statistical software.ResultsClinical and histological characteristics of eligible subjectsInitially,145patients with colorectal neoplasia, and70subjects in the NC group were enrolled. Of all the subjects involved,8were excluded because their stool samples did not meet the requirements （less than5g）;14in the colorectal neoplasia group, and6in the NC group were excluded due to the strict criteria of subject selection. Next, of the187extracted DNA,12were excluded （5CRCs,3AAs,4 NCs）. Eventually,175participants were retained in this study. Among them,63patients had primary CRCs,52patients had AAs, and60patients were age-matched controls. All patients were Chinese. Among the patients, a definitive diagnosis was established by the complete resection specimens of colonoscopy or/and surgical resection specimens, the average age of CRC group was61years （41-87years）, AA group58years （40-80years）, and control group47years （40-73years）; the sex ratio of the case group was69/57, including CRC32/31, AA37/26, and control group24/36; the primary sites of CRC and AA were located in the proximal colorectal88.9%（56/63） and86.5%（45/52）; Duck’s staging of A+B were accounted for60.3%; in case group, KRAS exon2-3mutation rate was26.1%（30/115）, the TP53exon5-8mutations21.7%（25/115）, and3.3%（2/60） in control group.The study in learning set stageIn order to establish and optimize the HRMA experimental conditions and parameters, we evaluated the HRMA detection accuracy of the mutant gene in tissue samples of CRC and AA. First, in40FFPE tissue samples of patients, KRAS exons2-3and TP53exons5-8mutation were confirmed by sequencing; then, the HRMA was applied to detected these gene mutations.22patients with gene mutations were detected, including17cases from29CRCs and5from AAs. The rate of mutation detection was55%（22/40）. Both results from HRMA and sequencing were fully consistent, the sensitivity and specificity of HRMA were100%. Subsequently, based on the above optimized the HRMA methodology, the HRMA detection accuracy of the mutant gene were assessed in referred stool DNA samples, and results showed that18cases （18/40,45%） with mutations were detected by HRMA, including14CRCs （14/29,48.28%） and4（4/11,36.36%） AAs. These detection results are confirmed by sequencing. The sensitivity and specificity of HRMA was100%. The rates of mutation detection in fecal DNA and in tissue DNA samples were18/22（81.82%） with a high degree of consistency, including CRC14/17（82.35%）, and AA4/5（80%）. Pooled Kappa value is0.794, in which CRC and AA were0.794and0.814respectively.The sensitivity analysis of the HRMA In order to evaluate the HRMA reliability of mutations detection in stool DNA from the CRC screening settings, we assess the lowest level of HRMA detection mutation by DNA series dilution test. Fecal samples of known KRAS exon2and TP53exon6-8mutations were selected for DNA serial dilution experiments, and the choice of the wild-type samples used as the background of the dilution experiments. The lowest level of HRMA detection mutation was is1%. but the reliable detection concentration was not less than5%owing to huge variation of samples.The study in validation set stageIn the above assessment study, The sensitivity and specificity of the HRMA were100%and detection limit1%. and encouraged us to further evaluate its accuracy in the validation set.75cases groups, including34cases of CRC,41cases of AA, and60patients in the control group. The frequency of KRAS/TP53mutation detection in the case group（49.3%,37/75） were significantly higher than that of the control group （3.3%,2/60）（P<0.001）; the subgroup analysis in the case group, the rates of KRAS/TP53mutation detection in CRC subgroup （58.8%,20/34） were significantly different from AA subgroup （41.5%,17/41）（P=0.02）.The results of the HRMA are subsequently confirmed by sequencing, its sensitivity and specificity were100%.The relationship between fecal mutations with parameter of clinicopathological features&clinical diagnostic valueCRC/AA originated in the proximal colorectum （88.9%/86.5%） than the remote colorectum （13.5%/11.1%）;Duck’s staging of A+B （60.3%） significantly were more than C+D （39.7%）; there were significant proportions of gender between the case group/CRC subgroup and control group/AA subgroup （P=0.004, P=0.035）, males among the former are more than the latter; the fragments of different lengths （155bp and92bp） in KRAS exon2, mutation detection was is exactly the same.In the CRC group, KRAS mutations were detected in9cases （19/63,30.1%）, and TP53mutations in15cases （15/63,23.8%）; there were no statistically differences between frequencies of mutation and gender, location, degree of differentiation, histological type, size, Duck’s staging, age. In the AA group, KRAS mutations were detected in11cases （11/52,21.2%）, and TP53mutations in10cases （11/52,19.2%）; there were statistically differences in rates of TP53mutation detection among age groups （P=0.036）, the age≥60group32.0%（8/25） was higher than the<60age group7.4%（2/27）; there were no statistically differences between frequencies of mutation and gender, location, degree of dysplasia, histological type, size, Duck’s staging.There were54.0%（34/63） mutation incidences in CRC patients, and40.4%（21/52） in AA patients; the performance of stool KRAS and TP53mutation detections for screening CRC and AA’s was poor, sensitivities were54%[95%CI0.41,0.66] and40%[95%CI,0.27,0.55], the both specificity97%[95%CI0.87,0.99], the accuracy of diagnosis75%and71%.ConclusionThe stage of learning set1. Among tissue samples of CRC and AA patients, the accuracy of the HRMA is consistent with DNA sequencing in detecting KRAS exon2-3/TP53exon5-8mutations, its sensitivity and specificity were100%.2. Among the corresponding stool DNA samples, there is the same accuracy in detecting KRAS exon2-3/TP53exon5-8mutations between HRMA and DNA sequencing, the sensitivity and specificity of HRMA were100%.3. the rates of stool DNA mutation detection is high degree of consistency with tissue DNA, Kappa value is0.794, CRC0.794, and AA0.814.The sensitivity analysisIn the context of the dilution of wild-type gene,1%mutation detection limit can be a rough estimate of the sensitivity of the HRMA, the HRMA detection sensitivity directly is affected by the sequence variation of the samples; the mutation detection limit of HRMA is reliably≥5%.The stage of validation set1. The rates of stool mutation detected by HRMA are49.3%（37/75） in case group, and3.3%（2/60） in control group, KRAS and/or TP53gene mutations were associated with colorectal neoplastic lesions. 2. Among case group, the rates of stool mutation detection in CRC group are significantly higher than in AA group, KRAS and/or TP53gene mutations was closer relations with CRC.3. The results of mutation detection by the HRMA are confirmed by sequencing, the sensitivity and specificity of the HRMA reach100%.The relationship between fecal mutations with parameter of clinicopathological features&clinical diagnostic value1. among all cases of patients, the rates of stool KRAS mutation detection is26.1%, and TP53mutation detection21.7%; the primary sites of CRC, and AA locate in the proximal colorectal more than the distal colorectal; compared with later stage of Duck’s staging（C+D staging）, early lesions （A+B staging） is significantly more; proportions of males in the case group/CRC subgroup is significantly more than the control group/AA subgroup; different lengths of the same gene have no effect on the performance of the HRMA.2. Among the case group, the frequency of KRAS/TP53mutation are correlated with gender, location, degree of differentiation/dysplasia degree, histological type, size, age, and Duke’s staging.3. Among the AA group, TP53mutations were positively related to age group.4. The stool KRAS/TP53mutations are poorly sensitive for screening CRC and AA, the sensitivity is54.0%for CRC;40.0%for AA, and specificity is97%.