GBM has been recognized as a highly heterogeneous tumor since the first half of the last century (Busch et al., 1947). The significance of glioblastoma intra-tumor and inter-tumor heterogeneity is continually evolving. GBM has been classified into distinct subtypes with specific combinations of genetic alterations, gene expression, methylation profiles, and prognostic survival features. At the single-cell level, GBM was recently reclassified into four states that account for both intra-tumor and inter-tumor heterogeneity. However, the number of cellular and molecular entities present in a single tumor and the interpretation of their boundaries are variable and obviously depend on technologies and thresholds.
Based on our work and that of many groups, one conservative interpretation is that GBM molecular heterogeneity reflects developmental/metabolic cell states/entities, with some level of spatial organization dependent on tissue macroareas. A fundamental question is whether the individual tumor cell identities represent stable and heritable entities or transient cell states. The rigid classification of GBM into subtypes (i.e. close to the entity definition) has evolved into the current plastic cell states classification. Moreover, several clinically relevant covariates, such as differentiation, inflammation, radiotherapy, hypoxia, and infiltration by innate immune cells, have been shown to correlate with glioblastoma phenotypic transitions.
The emerging picture is that GBM subtypes may represent dominant entities within extremely heterogeneous tumors, with mesenchymal glioblastoma being dominant at recurrence. However, causal links between these factors and a specific glioblastoma state have been elusive. In this regard, our work contributed to clarifying the pathophysiological and clinical relevance of one subtype (i.e. the mesenchymal) and its implications for GBM biology and treatment. Mesenchymal glioblastoma is the most aggressive form of these lethal tumors. We showed that this state is adaptive and metastable; it is driven by pro-inflammatory and differentiation cues and DNA damage, but not hypoxia. Importantly, we discovered that innate immune cells promote a proneural-to-mesenchymal transition that confers therapeutic resistance to glioblastoma cells. Thus, we were able to connect cellular and molecular heterogeneity to therapeutic responses. Our views were quickly incorporated into the current conceptualization of the disease, and our claim that innate immune cells drive a mesenchymal glioblastoma transition was later validated in other high-profile work in adult and pediatric brain tumors. These findings were directly informed by the collective work carried out during the iGBMavatar project and the successful achievements of the scientific plan proposed.
