MSK-IMPACT (Integrated Mutation Profiling of Actionable Cancer Targets) is an NGS large panel product of Memorial Sloan Catlin Cancer Center (MSK) in the United States. It has been used by MSK to formulate products since January 2014. Treatment options for patients with advanced cancer. This Panel has been iteratively upgraded, and the number of genes covered has been growing, from 341 genes in 2014 to 468 genes in 2017. Later, new tumor driver genes, genes related to immunotherapy and clonality were added. As of 2020, 505 genes have been included. MSK-IMPACT? was approved by the FDA on November 16, 2017, becoming the first large panel product approved to detect multiple cancer types and genes in tissues.
MSK-IMPACT can extensively detect tumor gene mutations in the genome by performing panel capture, NGS sequencing and bioinformatics analysis on tumor tissue FFPE samples and Normal samples. The sequencing platform it uses is Illumina HiSeq? 2500 Sequencer (qualified by MSK). The mutation types it detects include SNVs, INDEL (<30bp), and MSI. It cannot detect CNVs, rearrangements, and TMB. This article mainly summarizes and introduces the technical details of 468 gene versions of MSK-IMPACT that have been approved by the FDA.
Appendix 1a lists the detection hotspots of MSK-IMPACT, Appendix 1b lists 468 genes and corresponding transcript information, and Appendix 1c lists 73 Panel coverage depths. Low, no gene and exon region information is reported.
Quality control requires that the tumor cell content is greater than 10%, the section content is greater than 20%, and the MSI detection tumor cell content is greater than 25%
Omitted
Using homozygous sites Point consistency is tested for the pairing of Normal samples and Tumor samples from the same sample. It can be expected that the inconsistency of homozygous sites in paired samples is low (<5%), and that the inconsistency of homozygous sites in unpaired samples is high (in around 25%). If the matched sample has a high level of homogeneous site inconsistency (>5%), it is marked as a "possibly unpaired" type.
Detect the frequency of different alleles at the heterozygous site. If the frequency is abnormal, mark the sample as a "possible contamination" type.
Normal samples should not contain SNVs and INDELs. The analysis process will detect mutation hotspots in Normal samples. If the mutation frequency of a hotspot is >1%, the sample will be marked
SNVs (MuTect software) and INDELs (SomaticIndelDetector software). The CALL mutation process requires Normal Pair samples with Tumor. If Normal samples are missing or the average sequencing depth of Normal samples is low (<50X), select a Normal sample from the batch sample as a control.
Analysis of NC samples is used to confirm the absence of contamination and analysis of PC samples is used to ensure the sensitivity of the analysis
Tumor sample quality control requires that 98% of the exon coverage depth is greater than 200X, The paired Normal sample depth is greater than 50X
The original SNV and INDEL mutations need to go through the following series of filters to ensure the accuracy of reported mutations.
(1) VFtumor/VFnormal >=5, AD>=5,VF>=1%
(2) Whether the mutation is in the hotspot mutation library, Appendix 1a is the hotspot list
(3) Hot spots: DP>=20,AD>=8,VF>=2%; non-hot spots: DP>=20,AD>=10,VF>=5%
(4) Filter the annotation results to retain non-synonymous mutations and frameshift mutations located in the exon region
Use MSIsensor software to calculate the status of all MSI sites covered by MSK-IMPACT. MSIsensor will calculate a score. If the threshold is greater than 10, it will be marked as MSI-H.
Patient tissues or peripheral blood are used as Normal samples. If unpaired Normal samples are used in batches, there may be some rare germline mutations that are treated as somatic mutations
Positive control It is a mixture of 3 positive samples (each of the 3 samples contains different types and frequencies of mutation information). The normal sample of the positive control is a mixture of Normal samples from the same batch. After process analysis, the analysis results are checked. (1) Whether known mutations are detected (2) Whether the detected mutation frequency is consistent with the mutation frequency of the original sample. Table 2 is a sample with known mutation frequency.
The negative control sample is a mixture of FFPE normal samples confirmed by previous repeated experiments to be free of tumor contamination and free of germline copy number mutations. Previous analyzes have identified normal sample-specific polymorphisms and frequencies. The mutation frequencies of the 862 SNPs generated by the pipeline were compared with the expected mutation frequencies, and Pearson concordance was calculated. The expected mutation frequency should be greater than 0.9
No Template Control(NTC)
Sample quality control metric:
Power analysis is used to calculate the required coverage depth here. The concept of (power analysis)
1a Determination of exon coverage depth:
When setting a potential minimum detection line such as 2%, a certain significant value can be calculated through power analysis The depth of reads below the level. For example, when the mutation frequency is 10%, under the 500X depth 95% confidence interval, the mutation frequency ranges between 7.5% and 13%.
Therefore, an equimolar mixture of 10 unrelated Normal samples of FFPE creates a series of preset mutation frequencies, with the lowest mutation frequency being 5%. The relationship between the preset mutation frequency and the real mutation frequency was then verified using 862 SNPs. The final experimental data supported the use of 5% as a lower limit for reporting detected mutations, with the actual potential frequency being 10%.
Repeat within and between runs, 5 replicates for each sample
1a Panel precision verification (verification of standard)
10 samples ( 9 FFPE samples, one commercial cell line) representing different tumor types, different mutation types (mutation type is known), different mutation frequencies (mutation frequency is known).
2b Panel precision and repeatability consistency verification (validation of standards)
Table 6 When evaluating the precision of known mutations, the results detected in duplicate samples are also Mutations other than known mutations were verified for repeat consistency.
Consistency of the number of all mutations detected in 5 replicate samples
Consistency of mutation frequencies detected in 5 samples
Table 7 lists 10 Results of precision verification of test samples (5 replicates for each sample)
LoD is the mutant allele of a certain mutation that can be stably detected in 95% of all replicates. frequency. This part is to verify the LoD of the putative mutation. The test is divided into two parts. The first part is gradient dilution to determine the reliable lowest mutation frequency; the second part is to verify the LoD in many replicates.
1a Gradient dilution
Select 10 FFPE samples, each selection The 5 most covered exons and the 5 least covered exons. Prepare 5-8 gradient dilutions
Detection results Table10
Comparison of 1a SNV/indel mutation methods
MSK-IMPACT method and MSK-IMPACT method are correct The original results after cross-validation were compared. ***Detected 48 exons of 20 genes in 433 FFPE samples and detected 267 unique mutation sites. Known mutations in 433 samples*** were detected in 432 samples, and the EGFR gene in one sample was missed by INSERT.
Consistency of detection Table15A-C.
The final mutations reported by MSK are divided into two types: one is a cancer mutation with clinical significance, and the other is a mutation with potential significance of cancer mutations.
1a Validation of clinical samples
MSK has tested more than 10,000 patients with advanced tumors and stored medical record information and mutation information on the following website.
https://www.cbioportal.org/study/summary?id=msk_impact_2017
https://www.accessdata.fda.gov/cdrh_docs/reviews/DEN170058.pdf