Statistical approach for quantifying the evolution of tumor heterogeneity in chronic lymphocytic leukemia (CLL)

Authors

  • Vidhi Vidhi* SIMBA team, Inria Center of University of Lorraine, IECL, University of Lorraine, CNRS, France
  • Ulysse Herbach SIMBA team, Inria Center of University of Lorraine, IECL, University of Lorraine, CNRS, France
  • Coralie Fritsch SIMBA team, Inria Center of University of Lorraine, IECL, University of Lorraine, CNRS, France
  • Nicolas Champagnat SIMBA team, Inria Center of University of Lorraine, IECL, University of Lorraine, CNRS, France

Keywords:

phylogenetic tree, VDJ sequencing, somatic hypermutation, leukemia, intraclonal heterogeneity, variational expectation maximization

Abstract

Intra-tumor heterogeneity, a significant challenge in targeted cancer therapy, is particularly pronounced in CLL. Clonal evolution, driven by VDJ recombination and somatic hypermutation, generates a heterogeneous population of subclones that can confer resistance. Thus, phylogenetic tree reconstruction is key to understanding this process.

Using the weighted uniform distribution [1], we present a probabilistic framework for clonal reconstruction, generating a phylogenetic graph to better account for biological noise and allow sampling of multiple evolutionary trajectories, offering a more flexible and realistic model of tumor evolution than tree-based methods [2, 3]. To further refine this construction, we developed a Variational Expectation-Maximization algorithm, which systematically adds an optimal number of unobserved clones. This improves resolution, leading to a more complete and accurate representation of tumor evolution.

Future work will integrate longitudinal cancer gene mutation data (VCF) into our existing VDJ-based evolutionary graphs, creating a unified graph. This will provide a more comprehensive model of clonal evolution, improving insights into tumor progression and resistance for precision oncology.

Downloads

Published

2025-04-28

Issue

Section

Conference Contributions