Abstract 1071P
Background
Tumor mutation burden (TMB) has emerged as a major biomarker for predicting sensitivity to immune checkpoint inhibitors. Ideally, TMB should be measured by whole-genome or whole-exome sequencing (WES), but TMB is routinely estimated from targeted gene panel sequencing data. However, it remains unclear whether a certain subset of patients is more vulnerable to the bias of panelbased TMB estimation than others.
Methods
Using the WES somatic mutation data of ∼10,000 TCGA patients, we performed a comprehensive analysis to assess the performance of TMB estimation using 11 popular targeted gene panels. We compared the panel-based TMB to WES-based TMB using both conventional correlation analysis and rank-based group classification. Focusing on serious mistakes in rank-based group classifications, we further examined the TMB estimation bias of different patient groups of sex or age.
Results
The performance of different panels across cancer types was generally similar. Importantly, even for some cancer types (e.g., lung and breast cancers) with an overall high panel-WES TMB correlation, patients of specific sex or age tend to be misclassified when using certain gene panels. While the sex bias appears to be weak, the age bias in certain cancer types is significant and consistent across different panels.
Conclusions
Our results suggest significant bias of sex and age-based patient groups in estimating TMB from targeted panel sequencing data, which is underappreciated and should be given serious attention. This study provides valuable information on how to choose a suitable gene panel to obtain less biased TMB estimation for a specific patient.
Clinical trial identification
Editorial acknowledgement
Legal entity responsible for the study
Precision Scientific Biomedicine Co., Ltd.
Funding
Department of Finance, Jiangsu Province, China; Government of New District, Suzhou, Jiangsu Province, China.
Disclosure
S-J. Zhang: Shareholder/Stockholder/Stock options, Full/Part-time employment: Precison Scientific Biomedicine Co., Ltd. All other authors have declared no conflicts of interest.