PT  - JOURNAL ARTICLE
AU  - Jiuhong Pang
AU  - Hongai Xia
AU  - Shijun Mi
AU  - Wen Zhang
AU  - Danielle Pendrick
AU  - Christopher Freeman
AU  - Helen Fernandes
AU  - Mahesh Mansukhani
AU  - Susan J Hsiao
TI  - Benchmarking bioinformatics approaches for tumour mutational burden evaluation from a large cancer panel against whole-exome sequencing
AID  - 10.1136/jcp-2022-208385
DP  - 2022 Jul 29
TA  - Journal of Clinical Pathology
PG  - jclinpath-2022-208385
4099  - http://jcp.bmj.com/content/early/2022/07/29/jcp-2022-208385.short
4100  - http://jcp.bmj.com/content/early/2022/07/29/jcp-2022-208385.full
AB  - Tumour mutational burden (TMB) is used to predict response to immunotherapies. Although several groups have proposed calculation methods for TMB, a clear consensus has not yet emerged. In this study, we explored TMB calculation approaches with a 586-gene cancer panel (1.75 Mb) benchmarked to TMB measured by whole-exome sequencing (WES), using 30 samples across a range of tumour types. We explored variant allelic fraction (VAF) cut-offs of 5% and 10%, population database filtering at 0.001, 0.0001 and 0.000025, as well as different combinations of synonymous, insertion/deletion and intronic (splice site) variants, as well as exclusion of hotspot mutations, and examined the effect on TMB correlation. Good correlation (Spearman, range 0.66–0.78) between WES and panel TMB was seen across all methods evaluated. Each method of TMB calculation evaluated showed good positive per cent agreement and negative per cent agreement using 10 mutations/Mb as a cut-off, suggesting that multiple TMB calculation approaches may yield comparable results.