Analysis of 100,000 human cancer genomes reveals the landscape of tumor mutational burden

Analysis of 100,000 human cancer genomes reveals the landscape of tumor mutational burden Zachary R. Chalmers Caitlin Connelly David Fabrizio Laurie M. Gay Siraj M. Ali Riley Ennis Alexa B. Schrock Brittany Campbell Adam Shlien Juliann Chmielecki Franklin W. Huang Yuting He James Sun Uri Tabori Mark Kennedy Daniel S. Lieber Steven Roels Jared White Geoffrey A. Otto Jeffrey S. Ross Levi A. Garraway Vincent A. Miller P. J. Stephens Garrett M. Frampton

Tag layer of the DHSS Meta-Tagging Project. The article text is reproduced from the staged copy named in sourceDesc and is not altered; the tags are held separately in standOff so that overlapping and nested spans are preserved.

Analysis of 100,000 human cancer genomes reveals the landscape of tumor mutational burden 2017 Genome Medicine 10.1186/s13073-017-0424-2 chalmers2017-analysis-of-100-000-human-cancer-g chalmers2017-analysis-of-100-000-human-cancer-g.txt, the project's staged copy bd44f95137a283b9a3d141804b2cf78c866bca3b cc-by medicine

the paper's own words

the tag layer; a sentence written by the tagger about the paper, always anchored to a quotation but not itself a quotation

0 spans located by searching the stored text for the quote's exact words, allowing any whitespace between them tumour mutational burden as a biomarker - how it is counted, how little genome is enough to count it, and how it is distributed across cancer types Quotation withheld because this record is not cleared for redistribution. whether a 1.1 Mb targeted panel can stand in for whole exome sequencing when measuring mutational burden, and what the burden looks like across a hundred thousand clinical specimens Quotation withheld because this record is not cleared for redistribution. a validated targeted measurement of TMB, a downsampling analysis showing where it stops working, and the distribution of TMB across more than 100 tumour types Quotation withheld because this record is not cleared for redistribution. existence Quotation withheld because this record is not cleared for redistribution. a clinical category - high mutational burden - whose membership depends on an assay's size, a list of exclusions, and a threshold stated once in a figure caption Quotation withheld because this record is not cleared for redistribution. omics-profiling omics-profiling Quotation withheld because this record is not cleared for redistribution. TMB is a rate estimated from a sample of the genome, so the fewer megabases read the noisier the estimate; the effect is worst where the underlying rate is low, because the expected count in the sampled region approaches zero sampling variance rising as the panel shrinks Quotation withheld because this record is not cleared for redistribution. a synonymous change cannot itself create a neoantigen, but it is produced by the same mutational process that produced the nonsynonymous ones elsewhere, so counting it raises the sample size of the rate estimate without changing what the rate is about synonymous mutations as a process signal Quotation withheld because this record is not cleared for redistribution. with no normal tissue sequenced, inherited variants must be removed by matching against reference databases - so the filter's accuracy is the databases' coverage of the patient's ancestry germline filtering without a matched normal Quotation withheld because this record is not cleared for redistribution. the germline filter depends on reference databases, so it will be least accurate for patients from ethnic backgrounds not well represented in them - stated by the authors Quotation withheld because this record is not cleared for redistribution. the MSI cut points were set by manual unsupervised clustering against a reference that was itself partly APPROXIMATED by the same pipeline's homopolymer indel counts Quotation withheld because this record is not cleared for redistribution. the accuracy figure rests on 29 specimens with matched whole exome sequencing; the two 0.98 correlations are the assay against itself and an in-silico subsetting Quotation withheld because this record is not cleared for redistribution. known mismatch-repair and replication genes account for less than 10% of high-TMB cases, so the molecular basis of most of the phenomenon is unexplained Quotation withheld because this record is not cleared for redistribution. the PMS2 promoter mutations are not shown to be causal, and the TOP2A association rests on eight cases Quotation withheld because this record is not cleared for redistribution. targeted profiling measures the burden but cannot identify the neoantigens the mechanism depends on, since those may occur in any gene Quotation withheld because this record is not cleared for redistribution. principal components analysis, carried from multivariate statistics to collapse 228 repeat-length measurements into a single MSI score that can be cut into categories Quotation withheld because this record is not cleared for redistribution. resampling simulation, carried from survey methodology to ask how small a genomic panel may be before the rate it estimates stops being the rate Quotation withheld because this record is not cleared for redistribution. the same biopsy specimens measured by both the targeted panel and whole exome sequencing, so the panel is checked against a different measurement rather than against itself eliminates: the targeted panel's own gene selection as the explanation of the burden it reports - the same biopsy measured by a method that selects no genes gives a correlated answer · outcome: partial Quotation withheld because this record is not cleared for redistribution. 60 replicate pairs run through the same assay, separating precision from accuracy eliminates: assay noise as the explanation of the remaining disagreement with whole exome sequencing - the assay repeats itself at R2 0.98, so what is left is between methods and not within one · outcome: held Quotation withheld because this record is not cleared for redistribution. simulating the measurement at smaller panel sizes to find where sampling variance makes the estimate unusable, rather than asserting that 1.1 Mb is enough eliminates: sampling variance as a hidden contributor at the panel size actually used - the simulation measures it rather than assuming 1.1 Mb is enough, and names where it stops being enough · outcome: partial Quotation withheld because this record is not cleared for redistribution. an independent check on 8,917 TCGA exomes, comparing total mutations against those in the 315 assay genes eliminates: this laboratory's own cohort and pipeline as the source of the agreement - the same comparison on 8,917 independently generated exomes reproduces it · outcome: held Quotation withheld because this record is not cleared for redistribution. cohort filtering declared in advance - duplicates removed, coverage floor, and a minimum of 50 specimens before a cancer type is analysed at all eliminates: duplicate specimens and under-covered samples as contributors to the reported distribution · outcome: held Quotation withheld because this record is not cleared for redistribution. checking whether more than the first principal component of the MSI score was needed eliminates: further principal components as carriers of microsatellite signal the one-dimensional score discards - none of the others explains more than 4% of the variance · outcome: held Quotation withheld because this record is not cleared for redistribution. TMB TMB Quotation withheld because this record is not cleared for redistribution. mutation load mutation load Quotation withheld because this record is not cleared for redistribution. biomarker biomarker Quotation withheld because this record is not cleared for redistribution. Synonymous mutations Synonymous mutations Quotation withheld because this record is not cleared for redistribution. MSI-High MSI-High Quotation withheld because this record is not cleared for redistribution. MSI score MSI score Quotation withheld because this record is not cleared for redistribution. CGP CGP Quotation withheld because this record is not cleared for redistribution. WES WES Quotation withheld because this record is not cleared for redistribution. neoantigens neoantigens Quotation withheld because this record is not cleared for redistribution. percentage deviation percentage deviation Quotation withheld because this record is not cleared for redistribution. germline polymorphisms germline polymorphisms Quotation withheld because this record is not cleared for redistribution. effect size effect size Quotation withheld because this record is not cleared for redistribution. "high TMB" - AS USED throughout the abstract, results, discussion and conclusions: an unquantified clinical category of patients who might benefit from immunotherapy Quotation withheld because this record is not cleared for redistribution. "high TMB" - AS DEFINED: more than 20 mutations per megabase. The threshold appears once in the paper, in the caption of Fig. 3, and nowhere in the prose Quotation withheld because this record is not cleared for redistribution. agreement of targeted CGP with whole exome - AS ACCURACY: R2 = 0.74, from 29 specimens where both measurements were made on the same biopsy Quotation withheld because this record is not cleared for redistribution. agreement of targeted CGP with whole exome - AS PRECISION: R2 = 0.98, the assay against itself on 60 replicate pairs. The paper labels this one precision, and it is the number that looks like agreement without being it Quotation withheld because this record is not cleared for redistribution. qualifies → PD-1/PD-L1 expression by immunohistochemistry as the companion diagnostic for checkpoint inhibitors Quotation withheld because this record is not cleared for redistribution. extends → whole exome measurement of tumour mutational burden, by showing a 1.1 Mb targeted panel reproduces it Quotation withheld because this record is not cleared for redistribution. qualifies → neoantigen load as the predictor of immunotherapy response - correlated, but with no recurrent neoantigens found that predict response Quotation withheld because this record is not cleared for redistribution. MSI-High —part-of→ TMB MSI-High TMB the relation itself is separately evidenced · strength of the relation: 0.85 · the relation stated: MSI-High samples are largely a subset of the high-TMB population, and not the reverse Quotation withheld because this record is not cleared for redistribution. mutation load —same-as→ TMB mutation load TMB the relation itself is separately evidenced · strength of the relation: 0.9 · the relation stated: one quantity under two names, given in apposition at first use Quotation withheld because this record is not cleared for redistribution. Targeted sequencing identifies high-TMB specimens well, but deviation rises at intermediate TMB, especially below 0.5 Mb sequenced. Targeted sequencing identifies high-TMB specimens well, but deviation rises at intermediate TMB, especially below 0.5 Mb sequenced. Quotation withheld because this record is not cleared for redistribution. 83% of MSI-High samples have high TMB but only 16% of high-TMB samples are MSI-High - two markers of genomic instability that do not substitute for each other 83% of MSI-High samples have high TMB but only 16% of high-TMB samples are MSI-High - two markers of genomic instability that do not substitute for each other Quotation withheld because this record is not cleared for redistribution. the overlap between MSI-High and high TMB depends on the cancer type - near-total in gastrointestinal cancers, rare in melanoma the overlap between MSI-High and high TMB depends on the cancer type - near-total in gastrointestinal cancers, rare in melanoma Quotation withheld because this record is not cleared for redistribution. known mismatch-repair and replication mutations account for under 10% of high-TMB cases known mismatch-repair and replication mutations account for under 10% of high-TMB cases Quotation withheld because this record is not cleared for redistribution. the number of somatic, coding, base substitution and indel mutations per megabase of genome examined, with synonymous alterations counted and seven classes of alteration excluded TMB Quotation withheld because this record is not cleared for redistribution. more than 20 mutations per megabase - stated once, in the caption of Fig. 3, for a category used throughout the article high TMB Quotation withheld because this record is not cleared for redistribution. the first principal component of the means and variances of repeat lengths at 114 intronic homopolymer loci, projected onto one dimension MSI score Quotation withheld because this record is not cleared for redistribution. the panel size below which measurement variance increases significantly 0.5 Mb Quotation withheld because this record is not cleared for redistribution. MSI-High samples that also had high TMB 83% Quotation withheld because this record is not cleared for redistribution. high-TMB samples classified as MSI-High - the converse, and much smaller 16% Quotation withheld because this record is not cleared for redistribution. the initial clinical cohort, filtered to 92,439 102,292 Quotation withheld because this record is not cleared for redistribution. R2 = 0.74 R2 = 0.74 Quotation withheld because this record is not cleared for redistribution. R2 = 0.98 R2 = 0.98 Quotation withheld because this record is not cleared for redistribution. principal components analysis principal components analysis Quotation withheld because this record is not cleared for redistribution. false discovery rate false discovery rate Quotation withheld because this record is not cleared for redistribution. percentage deviation percentage deviation Quotation withheld because this record is not cleared for redistribution.

The full text of this article is not redistributable, so it is not included here. The evidence sentences are withheld with it; the tag layer and its grounding status remain. The article itself is at https://doi.org/10.1186/s13073-017-0424-2.