Targeting clonal complexity in glioblastoma

Targeted therapies have revolutionized cancer treatment. While they are highly efficient in some cancer types, others have not yet benefited from this new treatment paradigm. The major reason for the observed differences in outcome is tumor heterogeneity, where therapeutic targets are often present in only a subset of tumor cells or clones. While tumor heterogeneity grossly limits therapeutic response and promotes early tumor recurrence, it is rarely assessed in the routine diagnostic setting as a basis for treatment decisions. Glioblastoma is the prototypic example of a molecularly heterogenous tumor. In this project, we aim to dissect its clonal composition through quantifying the number of sub-clones and assessing their spatial and temporal distribution. Our core hypothesis is, the more complex the clonal composition, the higher the treatment resistance. To test this hypothesis, we will combine deep sequencing technology with advanced computational modeling and an ex-vivo model that fully retains its clonal complexity. Successful completion of the research will help understand the clonal evolution of this tumor, evaluate clonal complexity as a prognostic marker, and thereby advance a new therapeutic concept for this lethal cancer.

The present project explored the spatial genetic architecture of glioblastoma, a prototypic example of a molecularly heterogenous cancer with high endogenous treatment resistance. In our approach, we combined deep whole exome profiling of 75 multi-regional samples obtained from 16 patients with disease phenotyping and functional validation in perturbed tumor slices, which were profiled using spatial transcriptomics. Phylogenetic evolution grossly followed two patterns, linear versus branched, with an average of 6 clonal clusters per tumor. The emergence of new clones defined by genetic driver alterations seemed to be further shaped by tumor location and patient age. Similarly, tumors varied in their molecular diversity as measured by Nei's genetic distance, which correlated with physical distance only in tumors with clustered spatial patterns (fewer cellular dispersion likely due to less motile clonal populations), and which was reversed by extent of necrosis. In this preliminary setting, we found early evidence for a clinical relevance for genetic diversity with shorter time to progression in patients with tumors showing high genetic distance (p=0.013) and a non-significant trend towards worse survival in those with random spatial heterogeneity (p=0.35). Functional experiments further allowed us to link the genetic diversity with diverging signaling pathways between treatment responders and non-responders prominently including EMT and angiogenic signaling. Taken together, glioblastoma tumors differ in molecular diversity and its spatial patterning with early evidence for a clinical relevance, which needs to be validated in larger prospective cohorts. The findings support a translation of multi-regional tumor profiling to routine diagnostics and highlight tumor cell motility as one potential Achilles' heel to revert random into clustered spatial heterogeneity to increase treatment efficacy. Complementary findings from other cancers including non-brain cancers suggest a shared principle relevant to patients with several or many types of cancer.