Establishment And Application Of High-throughput Sequencing Platform For Patients With AML

Background and ObjectivesAcute myeloid leukemia （AML） is a hematological malignant tumor with high heterogeneity. Abnormality in cytogenetics is one of the most important independent indicators in diagnosis, prognostic evaluation and therapeutic selection in patients with AML. At present chromosome analysis technology is applied universally in clinic, but 50 percent of adult patients remain failing to detect any cytogenetic abnormality in microscope. They are considered as AML patients with normal karyotype （CN-AML）, and divided into intermediate risk group. However, the clinical outcomes are different in these patients, it indicates that great heterogeneity is present among patients with CN-AML. Therefore, abnormalities detected by molecular biology contribute to morbidity and prognosis evaluation in CN-AML.The development of molecular biology technology provides scientists with a wider field in leukemic mechanism from microcomic. Many characteristic molecular abnormalities have been discovered in AML, including gene expression, mutation, modification of gene transcription, and so on, many new detected gene abnormalites were identified constantly. Scientists has demonstrated that the relationship between gene mutations and clinical prognosis can not been explained by a sigle gene mutation after many gene abnormalities appear. Different combinations of mutations result in different clinical significance. It is better to indicate the real prognosis after the varieties of mutations are considered in AML. So clinicians depend on more detail in gene mutation spectrum to judge prognosis and guide individual therapy.Now clinical research is focused on how to obtain the valuable genetic abnormalities quickly to decide reasonable therapy. In the clinical molecular biological laboratory, traditional Sanger sequencing was used to screen gene mutations. It can detect all kinds of mutations, but the detection susceptibility is only 20%. Although the susceptibility of site-specific PCR is 1‰～1%, it can just detect the specific sites. It will take a longer time if several gene fragments are detected at the same time. Comprehensive routine screening using conventional or first-generation sequencing platforms in a clinical molecular diagnostics laboratory is very challenging since it requires large amounts of sample, more time and expense. So traditional sequencing platform is impossible to satisfy the requirement in clinic, and become a bottleneck in clinical application in study on AML. Clinicians hoped that they can obtain more information using a little amount of sample in a short time, and try to relieve patient’s pain suffering from extracting bone marrow repeatedly. Therefore, a new technology is established with high efficacy, shortcut, sensitivity and little expanse to demonstrate the value in individual therapy. It is very urgent and necessary.The next generation sequencing （NGS） technologies has the quality of high throughput, short time, high accuracy, low expense and amount of information, researcher can spend little time in analyzing what he is interested in, it has a broad perspective to apply in clinical. A new sequencing technology-Ion Torrent was developed by Life Technologies, Carlsbad, CA, USA) recently. The Ion Torrent method was a new generation sequencing technology based on semiconductor array, which has the characteristics:high accuracy, standard DNA polymerase sequencing with unmodified dNTPs and low background interference. It is based on detection of hydrogen ions released during every cycle of DNA polymerization. Each nucleotide incorporated into the growing complementary DNA strand causes the release of a hydrogen ion that is sensed by a hypersensitive ion sensor. Ion Torrent Personal Genome Machine （PGM） can currently generate 10-100 Mb pairs （Mbp） of sequence data on various chips within several hours of instrument’s run time. It has broad applications covering Sanger sequencing and other next generation sequencing in genomic DNA sequencing. RNA-sequencing, epigenetic assays, targeted DNA sequencing and so on.Based on Multiplex PCR, thousands of target genes can be amplificated within one single pipe. At present, the new sequencing technology-Ion Torrent has not used well in AML research. Commercialization Cancer Hotspot Panel involved gene mutations of solid tumors mostly and less about AML. Researchers can select a subset of the target genes and design a customized Ion AmpliSeqTM Panels by Ion AmpliSeqTM online. We add some AML-specific gene mutations to Ion AmpliSeq Cancer Hotspot Panel V2 and design a customized Ion AmpliSeqTM Panels for AML. The genes are WT1、DNMT3A、CEBPA、RUNX1、PHF6、ASXL1、MLL、JAK1、 PAX5、SF3B1、SH2B3、SRSF2、TET2、EBF1、DNMT1、DNMT3B、DNMT3L. We hope that the new panel could include the genes with clinical significance as much as possible and we can identify novel molecular abnormalities. The aim of part one in this study was to evaluate the methodology reliability of the Ion Torrent PGMTM sequencing platform and the customized Ion AmpliSeqTM Panels in clinical diagnosis of AML.Acute myeloid leukemia with t （8; 21） （q22; q22） is one of the most common AML, and which occupied about 10-15% of adult AML.90% of AML with （8; 21） is found in M2, others are in M0, M1, M4, MDS and therapy-related AML. Although this AML subtype is generally associated with a favorable prognosis, about 30%-50%of patients will relapse and appear resistance to chemotherapy. The prognosis of these patients is heterogeneous. These patients may have unfavourable molecular abnormality, and researchers have to look for the new marks and change therapy based on it. However, routine usage of Sanger technologies leaves us with several limitations such as the cost of entry, long processing time, and sample scalability. Few research compared the pattern of molecular mutations between initial diagnosed and relapsed AML with t （8;21）. In the second part of this study, hotsports of cancer-related genes were detected by Ion Torrent PGMTM and gene mutation patterns were compared between initial diagnosed and relapsed AML patients with t （8; 21）. We hope that this research can provide the molecule basis for relapsed AML with t （8; 21）.Materials and MethodsPart1Tumor specimenA total of 27 patients diagnosed with CN-AML were involved in present study from 2012.10 to 2013.12 in NanFang Hospital. DNA was extracted from 14 fresh bone marrow samples and 13 bone marrow smears.Methods1. Target gene screening and primers designing for mulitiplex PCR Primers for mulitiplex PCR of 67 target genes were designed by Ion AmpliSeq Designer （www.ampliseq.com）. Information of 50 genes came from Ion AmpliSeqTM Cancer Panel V2 and the other 17 genes came from literature.2. DNA preparationDNA was isolated from both fresh bone marrow and bone marrow smears using the QIAamp DNA Mini Kit （Qiagen） and manufacturer’s instructions were followed.3. Next-generation sequencing1) Library Preparation For multiplex PCR amplification,10 ng of DNA （quantified by QubitTM Fluorometer） was used and the customized Ion AmpliSeq panel was processed with Ion AmpliSeq Library kit 2.0 according to the manufacturer’s instructions. The resulting amplicons were treated with FuPa Reagent to partially digest the primers and phosphorylate the amplicons. The resulting library was purified and the concentration and size of the library were determined by Ion Library Quantitation Kit.2) Emulsion PCR and template-positive enrichment Sample emulsion PCR, emulsion breaking, and enrichment were performed using the Ion PGMTM Template OT2 200 Kit and Dynabeads（?） My OneTM Streptavidin C1 Beads according to the manufacturer’s instructions. The two steps were carried out on Ion OneTouchTM ES.3) Sequencing was performed on Ion 316 chips using the Ion PGMTM 200 Sequencing Kit according to the manufacturer’s instructions. The raw signal data were analyzed using Torrent Suite v.4.0.2 （Life Technologies）. The pipeline included signaling processing, base calling, quality score assignment, adapter trimming, read alignment to human genome GRCh37 references, mapping quality control, and coverage analysis. Several filtering steps were used to generate final variant calling.4. Sanger sequencingFLT3-ITD, NPM1 and DNMT3A specific PCR fragments were analyzed by Sanger sequencing.Part 2Tumor specimenA total of 11 relapsed AML patients with t （8; 21） were included in the study. Paired bone marrow samples from initial diagnosed and relapsed patients were available to study.Methods1. Next-generation sequencing （see Part 1）2. Sanger sequencingC-KIT （exons 8 and 17） specific PCR fragments were analyzed by Sanger sequencing.3. Pathogenic predict of mutationpolyphen-2 （http://genetics.bwh.harvard.edu/pph2/）, mutationtaster （http://www.mutationtaster.org/cgi-bin/MutationTaster/MutationTaster69.cgi） and MAPP-MMR （http://mappmmr.blueankh.com/） were used to predict the pathogenic of gene mutation.Results1. Primers designing for mulitiplex PCRPrimers for mulitiplex PCR of target genes were designed by Ion AmpliSeq Designer.652 amplicons covered 98.03% target regions. Two primer pools （332 amplicons and 320 amplicons） were designed. Amplicon ranges were between 125bp and 275bp.2. Quality control of NGSTaking one chip as an example, we found total bases were 409Mb, IPS Loading was 80% and enrichment of effective library was 100%. Usable Reads was 57% （2,884,511 total reads） with the exception of polyclonal （34%）, test fragments （1%）, adapter （1%）and low quality （12%）. The mean read length was 139 bp. Strength of H+ flows were uniform. Accuracy at position in read was more than 98%. Total number of bases with AQ20 was 352Mb. Result of one sample showed coverage overview was uniform. Other parameters were as follows:number of mapped reads were 493,804, percent reads on target were 96.15%, average base coverage depth was 517, uniformity of base coverage was 85.15%, percent base reads on target were 94.51%, target base coverage at 20X were 97.47%, target bases with no strand bias were 81.92%, target GC% was 50% and amplicons reading end-to-end were 56.90%.4. Summary of identified variants by NGSWe used the COSMIC （version64） （http://cancer.sanger.ac.uk/cancergenome/ projects/cosmic/）. MyCancerGenome（http://www.mycancergenome.org/）. SNP:（http: //www.ncbi.nlm.nih.gov/snp/） and some publications to assess mutations reappearing in CN-AML.18 variants were identified by Ion Torrent PGMTM in 67 mutational hotsports. Multiple mutations （single base substitution, insert, deletion） were observed in mutation genes. NPM1, DNMT3A, FLT3 and TET2 mutations were detected most frequently （22.2%,22.2%,18.5% and 18.5% respectively）. NPM1 exon 12 mutations were detected frequently. Two sequence variants （c.859₈60insTCTG and c.861₈62insTGCA） were observed. The most frequent mutation （which we have called mutation A） was a duplication of a TCTG tetranucleotide insert, which resulted in a frame shift of C-terminal portion of the NPM protein. Internal tandem duplication of the FLT3 gene was detected in 3 cases, and a mutation at aspartic acid residue 835 （D835） in FLT3 was detected in 2 cases. One patient carried both internal tandem duplication and D835 mutation. Single base substitution mutations were detected frequently in DNMT3A and TET2. The most common of DNMT3A mutations was predicted to affect amino acid R882 （4 cases of R882H and 1 R882C）. The type of gene mutations, the gene locations on the chromosome, as well as the proportion of mutation cells can be obtained from the NGS sequencing, which can be used to monitor minimal residual disease.5. Validation using Sanger sequencing4 cases with FLT3-ITD,6 cases with NPM1 and 5 cases with DNMT3A in 27 patients were detected by Sanger sequencing. Within the same target region, the consistency of two methods in NPM1 and DNMT3A detection was 100%, and the one variant in FLT3-ITD was detected by Sanger, but not by Ion Torrent PGMTM.Part 21. Quality control of NGSAll results of chips were qualified. The sequencing coverage for each of the regions was between 100X and 399X. Percent reads on target were more than 90%. Target base coverage was uniform.2. Summary of identified variants by NGSPaired bone marrow samples from initial diagnosed and relapsed AML patients with t （8; 21） were available and 67 mutational hotspots of cancer-related genes were detected by Ion Torrent PGMTM. Heterozygous mutations were identified frequently. The proportions of mutation cells between the initial and the relapse were similar. 1-4 variants were identified in initial diagnosed patients. Most common variants were C-KIT mutations （6/11）, followed by ASXL1, MLH1, TET2 （each n=2）, FBXW7, DNMT3A, NRAS and DNMT3B mutations （each n=1）. C-KIT mutations were distributed in detail as the following:D816 mutations （exon 17） in 3 cases; N822 point mutations （exon 17）, M541 （exon10） and R815_D816ins （exonl7） in 1 case respectively. Among KIT D816, the following mutations were found:D816Y, D816H and D816V in 1 case respectively. In 4 cases, C-KIT mutations remained stable at the time of relapse. Loss of the C-KIT mutation at relapse was observed in 2 cases. In 6 patients, the initial molecular mutation pattern changed at relapse. Mutations commonly gained at relapse were KMT2A, C-KIT, TET2 （each n=2） and SH2B3 （n=1）. Loss of a mutation at relapse was observed in C-KIT （n=2） and NRAS （n=1）.3. Validation using Sanger sequencing5 cases with C-KIT mutations （D816 point mutations in 3 cases, N822 point mutations in 1 case and R815_D816ins in 1 case） were detected in initial diagnosed patients by Sanger sequencing.3 cases with C-KIT mutations （D816, N822 and R815_D816ins in 1 case, respectively） were detected in relapsed patients. No mutations were detected in KIT exon 8. Within the same target region, the consistency of two methods detection was 100%.4. Prediction of recurrent mutation MLH1MLH1 c.1151T>A were detected in two t （8; 21） AML patients （paired initial diagnosed and relapsed sample） and 3 patients with CN-AML. Sequencing depth of the heterozygous mutation were>100X. The variant located in the middle of its amplicon. Three softwares were used to predict the pathogenicity of the variant, scores were as the following:polyphen-2-Probably Damaging （0.998）, mutationtaster-Disease Causing, MAPP-MMR-5.120 （>4.55）.Conclusion1. In this study, a new PCR array kit based on detecting gene mutation of solid tumor was originally structured by Ion Torrent PGMTM.17 gene mutations with predictive value to prognosis in AML were added in this new kit. Compared with conventional sequencing technique, Ion Torrent PGMTM is a good method to identify gene mutation detection with high sensitivity, specificity, reproducibility and high throughput.2. The detection of long segment of the insertion/deletion and homopolymer by Ion Torrent PGMTM were still needed to be further improved.3. Different mutation types from the initial were detected in most relapsed AML patients with t （8; 21）. Furthermore, C-KIT mutation is the most frequent genetic variation by using novel technique.