Quantification of spatial subclonal interactions enhancing the invasive phenotype of paediatric glioma

Abstract

Intra-tumour heterogeneity is an intrinsic property of all cancers. In some cases, such variation can be maintained by interactions between tumour subclones with distinct molecular and phenotypic characteristics. In paediatric gliomas, interactions can take the form of enhanced invasive phenotype, a hallmark of these malignancies. However, subclonal interactions are hard to quantify and difficult to distinguish from spatial confounding factors and experimental bias. Here we combine spatial computational modelling of cellular interactions and invasion, with co-evolution experiments of clonally disassembled primary glioma lines derived at autopsy. We design a Bayesian inference framework to quantify spatial subclonal interactions between molecular and phenotypically distinct lineages with different patterns of invasion. We show how this approach could discriminate genuine subclonal interactions where one clone enhanced the invasive phenotype of another, from apparent interactions that were only due to the complex dynamics of subclones growing in space. This study provides a new approach for the identification and quantification of spatial subclonal interactions in cancer.