Tumor mutational burden assessment and standardized bioinformatics approach using custom NGS panels in clinical routine

Célia Dupain,Tom Gutman,Elodie Girard,Choumouss Kamoun,Grégoire Marret,Zahra Castel-Ajgal,Marie-Paule Sablin,Cindy Neuzillet,Edith Borcoman,Ségolène Hescot,Céline Callens,Olfa Trabelsi-Grati,Samia Melaabi,Roseline Vibert,Samantha Antonio,Coralie Franck,Michèle Galut,Isabelle Guillou,Maral Halladjian,Yves Allory,Joanna Cyrta,Julien Romejon,Eleonore Frouin,Dominique Stoppa-Lyonnet,Jennifer Wong,Christophe Le Tourneau,Ivan Bièche,Nicolas Servant,Maud Kamal,Julien Masliah-Planchon
DOI: https://doi.org/10.1186/s12915-024-01839-8
IF: 7.364
2024-02-22
BMC Biology
Abstract:High tumor mutational burden (TMB) was reported to predict the efficacy of immune checkpoint inhibitors (ICIs). Pembrolizumab, an anti-PD-1, received FDA-approval for the treatment of unresectable/metastatic tumors with high TMB as determined by the FoundationOne®CDx test. It remains to be determined how TMB can also be calculated using other tests.
biology
What problem does this paper attempt to address?
### Problems the paper attempts to solve This paper aims to solve the standardization problem of tumor mutational burden (TMB) assessment methods in clinical routine. Specifically, the authors attempt to use a custom - made next - generation sequencing (NGS) panel and bioinformatics methods to calculate TMB and compare it with the existing FoundationOne® (FO) algorithm. The main objectives include: 1. **Develop a TMB calculation method applicable to different sample types**: - The authors developed a custom - made TMB calculation algorithm (IC algorithm) and applied it to formalin - fixed paraffin - embedded (FFPE) and frozen samples. - Determine the variant allele frequency (VAF) thresholds for different sample types (FFPE and frozen) to reduce false - positive results. 2. **Compare the TMB calculation results of different algorithms**: - Compare the TMB calculation results of the custom - made IC algorithm and the FoundationOne® (FO) algorithm on the same NGS data, and evaluate the differences and consistencies between the two. 3. **Verify the biological authenticity of high - TMB cases**: - Manually perform biological verification on high - TMB cases, exclude false - positive results caused by technical or biological factors, and re - classify some cases as microsatellite instability high (MSI - H) or POLE - mutated tumors. 4. **Propose a standardized bioinformatics tool**: - Develop a highly customizable bioinformatics tool (pyTMB) to adapt to different NGS panels and sample types and achieve standardized TMB calculation. ### Main findings 1. **Influence of TMB calculation methods**: - Using the IC algorithm, the optimal VAF threshold for FFPE samples is 10%, while for frozen samples it is 5%. - In microsatellite - stable/POLE wild - type (MSS/POLE WT) tumors, the median TMB score is 8.8 mut/Mb, while in MSI/POLE - mutated tumors it is 45 mut/Mb. 2. **Comparison of different algorithms**: - The TMB value of FFPE samples calculated using the FO algorithm is significantly higher than that of the IC algorithm (40 mut/Mb vs 8.2 mut/Mb, p < 0.001). - All TMB values calculated using the FO algorithm exceed 10 mut/Mb, which is inconsistent with the results reported in the literature, indicating that the details of the FO algorithm are difficult to fully reproduce. 3. **Biological verification of high - TMB cases**: - Among the top 10% non - MSI - H cases with the highest TMB, 21% of the cases were re - classified as MSI - H or POLE - mutated tumors. - Among the remaining high - TMB cases, 49% could not find a biological explanation, but 30% of these cases had pathogenic variants in the TP53, PTEN and ARID1A genes. ### Conclusions By developing and applying a custom - made TMB calculation algorithm, the authors demonstrated the influence of different calculation methods on TMB values. They proposed a highly customizable bioinformatics tool (pyTMB) to adapt to different NGS panels and sample types, thereby achieving standardized TMB calculation. These research results are helpful for optimizing the application of TMB as a predictor of the efficacy of immune checkpoint inhibitors (ICIs). Future research needs to further verify the applicability and effectiveness of these methods in larger patient cohorts.