Evaluating the clinical utility of Tumour Mutational Burden assessment in combination with PD-L1 expression analysis in guiding immunotherapy treatment stratification in a Welsh lung cancer patient group

Roberts, Helen Marie (2024) Evaluating the clinical utility of Tumour Mutational Burden assessment in combination with PD-L1 expression analysis in guiding immunotherapy treatment stratification in a Welsh lung cancer patient group. Doctoral thesis (DClinSci), Manchester Metropolitan University.

Preview

Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (4MB) | Preview

Abstract

The immunotherapy, pembrolizumab, is not effective in the treatment of all lung cancer patients. Stratification of the use of this drug in non-small cell lung cancer (NSCLC) patient care is currently performed within the NHS using immunohistochemistry (IHC)-based PD-L1 expression analysis, but patient response rates following stratification remain low at 45%. A more accurate predictor of immunotherapy response is desirable to minimise the use of ineffective and costly therapy. Tumour Mutational Burden (TMB), defined as the number of somatic mutations found within a tumour, has been identified in numerous research studies as a potential new biomarker for immunotherapy stratification in lung cancer patients either alone or in combination with PD-L1 expression analysis. Publications show that high TMB, quantified using either Whole Exome Sequencing (WES), or targeted Next Generation Sequencing (NGS), is associated with immunotherapy response. Following the recent United States Food and Drug Administration (FDA) approval of the TMB-stratified use of pembrolizumab for solid tumours including lung cancers, UK approval for TMB-based immunotherapy stratification via National Institute for Health and Care Excellence (NICE) may be granted in the coming years, replacing or supplementing the suboptimal PD-L1 expression analysis. With such approval would come the requirement for NHS Genomics laboratories to deliver TMB services. Despite the considerable international interest in TMB as a biomarker, there remains a lack of consensus in how TMB is calculated. TMB-focused studies to date show differences in the NGS panels used to determine variant number, the variants included within the TMB assessment, and the definition of ‘high TMB’ via the use of different TMB thresholds to separate likely responders from non-responders. The clinical impact of these variables has to be understood and controlled prior to service implementation within the NHS to ensure high quality services are provided to patients. This research study aimed to produce novel data regarding the impact of these variables on TMB score and TMB status. This study has provided increased understanding in this area by demonstrating the impact on TMB estimation and TMB high status when three NGS panels (Illumina TruSightTM Oncology 500 panel, Agilent SureSelect Community Design Glasgow Cancer Core panel, and Nonacus Cell3TM Target: Pan Cancer panel) targeting varying proportions (1.58-1.94Mb) of the genome, were used to determine TMB using different TMB quantification methods on the same cohort of Welsh NSCLC patients with high PD-L1 expression status and known pembrolizumab response status. TMB quantification for all three NGS panels was performed using the Institut Curie TMB tool. TMB values were generated following the application of different variant filtering parameters based on the inclusion/exclusion of sequencing artefacts, which is an area not well-researched currently in terms of impact on TMB, and the inclusion/exclusion of synonymous variants, which is an area of difference within TMB publications. The utility of ROC curve generated TMB high thresholds for immunotherapy response prediction were evaluated alongside a 10 variants/Mb threshold, which is a threshold used in a number of TMB publications. This evaluation enabled the primary research question to be answered by demonstrating the potential clinical utility of a combined TMB and PD-L1 biomarker for immunotherapy response. Sequencing data from the Illumina and Nonacus panels highlighted an increase in sensitivity for the separation of responders and non-responders when a combined TMB and PD-L1 biomarker was used compared to the use of PD-L1 expression analysis alone. The Agilent NGS panel failed to produce any sequencing data above the minimum coverage level. The study identified elements of analysis providing optimal TMB quantification, and generated suggestions for minimising the clinical impact of panel- and analysis-dependent TMB variation to improve the clinical utility of TMB as a biomarker. Given the small size of the cohort (n=17), limited by the cost of NGS and the financial constraints of this research, this thesis represents a pilot study. The findings could be used to shape the design of future larger scale research studies evaluating the utility of different panel/analysis combinations, or to drive further research into the clinical utility of TMB in a larger Welsh cohort, which would be more representative of the Welsh population as a whole and would provide more weighting to the findings of this small pilot study. The study makes recommendations that could guide NHS Genomic laboratories in how to progress TMB service validations, and which could contribute to future best practice guidelines for TMB service delivery. These recommendations support the use of: NGS panels >1.6Mb in size, the Institut Curie TMB tool, and ROC curves in TMB evaluation, whilst the need for artefact removal prior to TMB calculation is not favoured. The feasibility of TMB service implementation within the NHS environment was highlighted by the potential cost neutral status of a TMB service and the recent launch of an External Quality Assurance (EQA) pilot scheme for TMB quantification (Abate 2020); recommendations for future EQA schemes are provided.