Skip to Main content Skip to Navigation

Statistical methods for deciphering intra-tumor hereterogeneity : challenges and opportunities for cancer clinical management

Abstract : Accessing the repertoire of cancer somatic alterations has been instrumental in our current understanding of carcinogenesis. However, efforts in genomic characterization of cancers are not sufficient to predict a patient's outcome or response to therapy, which is key to inform their clinical management. This failure is partly attributed to the evolutionary aspect of cancers. Indeed, as any biological population able to acquire heritable transformations, tumor cells are shaped by natural selection and genetic drift, resulting in a mosaic structure, where several subclones with distinct genomes and properties coexist. This has important implications for cancer treatment as those subpopulations can be sensitive or resistant to different therapies, and new resistant phenotypes can keep emerging as the diseases progresses further. An important number of mathematical or statistical methods have been developed to detect and quantify the intra-tumor heterogeneity (ITH), but no systematic evaluation of their performances and potential for clinical application has been performed. Our first contribution consists in a survey of existing approaches to decipher ITH, that allows to navigate the different underlying ideas easily. We have also proposed a framework to assess the robustness of those approaches, and their potential for use in patient stratification. This survey has allowed us to identify an unexploited type of data in the process of ITH reconstruction, and our second contribution fills remedies to this shortfall. Indeed, besides observed prevalences of somatic mutations within a tumor sample that allow us to distinguish several clones, the nucleotidic context of those mutations reveals the unknown causative mutational processes. We illustrate on both simulated and real data the opportunity to jointly model those two aspects of tumor evolution. In conclusion, we highlight the need to reinforce data integration from several sources or samples to harness the potential of tumor evolution for cancer clinical management.
Document type :
Complete list of metadata
Contributor : ABES STAR :  Contact
Submitted on : Tuesday, May 18, 2021 - 12:20:08 PM
Last modification on : Wednesday, November 17, 2021 - 12:31:07 PM


Version validated by the jury (STAR)


  • HAL Id : tel-03228677, version 1


Judith Abecassis. Statistical methods for deciphering intra-tumor hereterogeneity : challenges and opportunities for cancer clinical management. Bioinformatics [q-bio.QM]. Université Paris sciences et lettres, 2020. English. ⟨NNT : 2020UPSLM065⟩. ⟨tel-03228677⟩



Record views


Files downloads