Establishment And Application Of Reference Material Preparation Platform For Next-generation Sequencing-based Tumor Gene Mutation Detection

Malignant tumors are one of the main causes of death in China,which seriously endanger the health of our population.With the continuous development of precision medicine,suitable targeted drugs for patients based on the gene mutation information of cancer patients have become a vital part of clinical cancer diagnosis and treatment.Due to the in-depth research of tumor-associated gene mutations,clinical molecular detection technology is constantly updated.The advent of next-generation sequencing（NGS）has made it possible to detect multiple gene mutations in large quantities at the same time.However,compared with traditional molecular detection methods,NGS is more complicated.When errors emerge in "wet bench" with operating steps or a "dry bench" for biochemical analysis and variant interpretation,it may lead to the "unreliable" detection results.In order to ensure the accuracy of genetic mutation detection results,clinical laboratories must select appropriate reference material to establish a complete quality management system,including validation for laboratory-developed tests（LDTs）,verification for commercial kits,performing internal quality control（IQC）,and regularly participating in external quality assessment（EQA）or proficiency testing（PT）.Till now,there is no unified standard for the preparation of reference materials for NGS-based tumor gene mutation detection.However,the most ideal reference materials are from patient tumor tissue samples,but they are difficult to obtain in large batches and have tumor heterogeneity.So they cannot be used to perform a comprehensive performance evaluation on NGS-based detection methods,and fail to meet the requirements for laboratory quality assurance（including IQC and EQA）.In order to solve this problem,our study established a preparation platform for NGS-based tumor gene mutation detection reference materials（see Part I）.By using this platform,the prepared reference material has the following advantages:First,the reference material can be applied either on the "tumor-only"-based small tumor gene detection panel or on the"tumor-normal"-based large tumor gene detection panel.Specifically,we used human genomic DNA（Genome DNA,gDNA）extracted from cell lines as template DNA,and generated multiple DNA fragments with the same genomic background but containing different mutations by overlap extension polymerase chain reaction（OE-PCR）method.The DNA fragment was mixed uniformly with the genomic DNA to generate as the tumor gene mutation sample.This sample can be used independently as a reference material for the "tumor-only"-based detection panel.The non-mutagenized gDN A was used as a normal paired sample,and it can be used together with the tumor gene mutation sample for the"tumor-normal"-based detection panel.Second,the background cell line selected in this study was the GM12878 cell line with known genomic sequence information,which can also be used alone to verify the accuracy of the gene mutation results in clinical laboratories.Third,the reference material prepared by this platform can cover multiple variant types,and the variant allele frequency（VAF%）can be preset for each mutation.In this study,we totally constructed 15 kinds of DNA fragments containing different mutations,and the variant types contain single nucleotide variant（SNV）,deletion,insertion and copy number variation（CNV）.These DNA fragments were verified by bacterial liquid PCR electrophoresis and Sanger sequencing,and the DNA fragments containing mutation sites were obtained by double digestion of the plasmids.According to the different VAF%preset for each mutation,the volume of each DNA fragment was calculated based on the copy number concentration of each DNA fragment and the genomic DNA.Fourth,the reference materials showed good homogeneity and stability.In our study,the homogeneity and stability for the reference materials were evaluated and the results showed good uniformity and stability.Fifth,the reference material was not only suitable for the evaluation of NGS-based gene mutation detection process,but also can be used as the internal quality controls（IQC）for other conventional gene mutation detection methods,such as amplification refractory mutation system（ARMS-PCR）,digital PCR（dPCR）and real-time fluorescence PCR（qPCR）.EQA/PT is an important aspect of clinical laboratory quality management.Through a comprehensive comparison of the EQA/PT feedback results from laboratories,the problems either in "wet bench" or "dry bench" in the NGS-based tumor gene mutation detection can be easily found,which can help laboratories to improve and comprehensively optimize their detection pipeline,thus providing accurate and reliable variant detection reports for the clinicians.In order to evaluate the performance of NGS-based variant detection across laboratories in China,the National Center for Clinical Laboratories launched an EQA in 2017（see part II）.The two most common tumor types in targeted therapy,Non-small cell lung cancer（NSCLC）and breast cancer were selected and two rounds of EQA were organized.Based on the platform established in the Part Ⅰ,the NGS-based NSCLC and breast cancer gene mutation detection reference material sample panel were prepared.Each sample panel contained one matched normal sample and six tumor samples.The variant types in NSCLC reference material sample panel included SNV,small insertion and deletion（InDel）and gene fusion,while the SNV,InDel,and CNV were included in the breast cancer sample panel.The concentration of each sample was greater than 30ng/μL,and the total amount of nucleic acid was at least 750ng,which can meet the requirements of various NGS platforms,various targeted capture methods,and various library preparation methods.VAF%for most mutations in the sample panel was set between 10%and 45%,which met the detection capability level for most laboratories.In addition,the sample panel covered variants with different levels of clinical significance,including variants with significant clinical significance,variants with potential clinical significance,and variants with unknown clinical significance.We then distributed the EQA sample panels to the participating laboratories.The feedback results were analyzed to explore the problems in the variant detection and variant interpretation process across laboratories.A total of 102 clinical laboratories nationwide participated in this EQA,of which 89 laboratories returned the results.In the NSCLC NGS-based gene mutation detection EQA,62.9%（56/89）of the laboratories had eligible EQA scores,of which only 48（53.9%）laboratories had a completely correct detection result,while five laboratories（5.6%,5/89）got zero points.The results of the breast cancer NGS-based gene mutation detection EQA gradually improved,and 87.6%（78/89）of the laboratories had an eligible score and 76（85.4%）laboratories had completely correct detection results,but there is still one laboratory（1.1%,1/89）with a score of zero.The variant detection results reported by laboratories were analyzed and found that false positives and false negatives were the main sources of errors.In the NSCLC NGS-based gene mutation detection EQA,three laboratories reported more than 20 false positives.Mostly false-negative results were reported by laboratories using hybrid capture methods based on the Illumina platform.We also found that the poor detection ability to detect the InDel variant,such as ERBB2（NM₀04448.3）:c.2326delGinsTTAT（p.Gly776delinsLeuCys）,the correct detection rate of this variant in the two EQA samples was only 67.9%（57/84）and 69.0%（58/84）.Besides,laboratories should improve the ability to correctly identify the base positions of InDel variants.In addition,the variant detection results in some clinical laboratories also reflect their neglect of the standardized HGVS gene mutation nomenclature way of naming mutations.After two rounds of EQA,each laboratory found the problems in the variant detection process through the first round of EQA and improve them,so that the results for the second round of EQA gradually improved.This also showed that EQA can help clinical laboratories improve their quality management system.We also analyzed the variant interpretation process across laboratories.In the NSCLC NGS-based gene mutation detection EQA,we mainly analyzed variant interpretation information fed back from 34 correct detection results and consistent detection range laboratories,which included database information,consistency of variant clinical significance classification,and the consistency of clinical evidence and targeted drug information.The results showed that except for the variants with strong clinical significance and clearly targeted drugs,such as EML4 exon 13-ALK exon 20 and EGFR c.2235₂249delGGAATTAAGAGAAGC（p.Glu746_Ala750del）,the clinical significance classification results and targeted drug information for other mutations were inconsistency.We analyzed the reasons mainly in the following two aspects:First,some laboratories did not refer to the US Food and Drug Administration（FDA）or authoritative guidelines when interpreting variants,and led to the insufficient clinical evidence.So the variant interpretation results may be problematic in the variant clinical significance and the submitted targeted drug information.Second,the laboratories used different combinations of public databases or in-house databases to obtain variant clinical evidence.Because each type of database provided only limited and different variant-related information according to its database characteristics,the clinical evidence and clinical significance classification for each mutation in different public databases may differ.Therefore,laboratories using different types of public databases may produce different levels of variant clinical evidence,resulting in inconsistent variant interpretation results.In the breast cancer NGS-based gene mutation detection EQ A,we mainly summarized and evaluated the accuracy and adequacy for the targeted drug information fed back from the laboratories.The results showed that the targeted drug information provided by the laboratory was inaccurate,mainly because some laboratories provided drugs that showed low efficacy for the mutations.Secondly,some laboratories also submitted information on targeted drugs that were not compatible with their detected mutations.In addition,laboratories provided insufficient targeted drug information.Compared with the clinically relevant targeted drug information provided by the Ion Torrent Oncomine Knowledgebase Reporter v.3.1.0 system,no laboratory could completely cover all the targeted drug information displayed in the system.61.8%（21/34）of the laboratories only provided less than ten targeted drug information.The main reason was that the laboratory mainly provided the drugs approved by the guidelines or the FDA,and ignored the drug information that emerged in the different phases of clinical trials.Clinical laboratories should evaluate the variant clinical significance and provide targeted drug information from clinical validity and clincial actionality.Therefore,our study provided a variant interpration pipeline.Different levels of clinical evidence were obtained through different types of databases and literatures,which carried out accurate variant classification and provided adequate and comprehensive targeted drug information.To sum up,our study established a NGS-based tumor gene mutation detection reference material prepartation platform,and successfully prepared the reference materials in a simple,accurate,and economical way.It has the advantages of known and identical genomic background,good homogeneity stability and can be large-scale prepared according to pre-designed mutation and VAF%.The prepared reference material is applicable not only to the "tumor-only"-based small tumor gene detection panel,but also to the "tumor-normal"-based large tumor gene detection panel,which can perform evaluation both on the "wet-bench" operation steps and the "dry-bench" bioinformatics analysis.After two rounds of EQA,each clinical laboratory had found the problems existing in the first-round EQA and improved them,resulting in an improvement score in the second-round EQA.This showed that EQA can help laboratories to improve quality and improve their quality management system.In addition,NGS-based tumor gene mutation detection should not only ensure accurate variant detection results,but also the accurate and comprehensive variant interpretation results.The laboratory should combine different levels of clinical evidence from different types of databases to classify the variant clinical significance accurately and provide accurate and sufficient targeted drug information,so as to provide a reliable clinical genetic mutation detection report for the clinicians,which could make a great promotion to achieve the "precision medicine".The innovation of this study is as follows.（1）The NGS-based tumor gene mutation detection reference material prepartation platform is established by using a method of site-directed mutagenesis of human cell line genome DNA.The prepared reference materials contain various types of variants with different VAF%and are originated from the same genomic background,and can be not only suitable for "tumor-only"-based small tumor gene detection panel,but also suitable for "tumor-normal"-based large tumor gene detection panel;（2）The NGS-based tumor gene mutation detection EQA was formally launched in China for the first time,which used the reference material sample panels prepared by our established platform.Through a comprehensive comparison of the EQA/PT feedback results from laboratories,the problems in the NGS-based tumor gene mutation detection can be easily found,which can help laboratories to improve and comprehensively optimize their variant detection and interpretation pipeline,thus providing accurate and reliable variant detection reports for the clinicians,which contributed to the normalization and standardization.（3）In addition to the above-mentioned EQA,the prepared reference material can also be used to the verification of LDTs,validation of commercial kits,and IQC.