Back

Beyond tumor volume: connectomic measures of tumor burden provide superior prognostic information across tumor compartments in glioblastoma

Falco-Roget, J.; Ielo, A.; Janus, A.; Lillo, S.; Fasano, A.; Matteoli, M.; Pessina, F.; Politi, L. S.; Coletta, L.; Vavassori, L.; Sarubbo, S.; Avesani, P.; Argasinski, J. K.; Cacciola, A.

2026-09-15 oncology
10.64898/2026.09.14.26362015 medRxiv
Show abstract

Glioblastoma (GBM) is increasingly recognized as a disease that perturbs and interacts with distributed brain networks, yet its imaging-based prognostic assessment remains largely focused on local tumor extent. Here, we ask whether brain-wide tumor involvement provides a more robust representation of prognostic burden than conventional tumor volume. We analyze preoperative imaging from 999 patients with newly diagnosed GBM across multiple institutions, comparing volumetric measures with the lesion-tract density index (L-TDI), a tractography-derived measure of tumor involvement of white-matter pathways. Systematic survival analyses across tumor compartments and stratification thresholds reveal that the prognostic value of tumor volume is strongly dependent on analytical design and is largely confined to the contrast-enhancing compartment. In contrast, L-TDI shows robust associations with overall survival across thresholds and tumor compartments, outperforming volumetric measures and retaining prognostic value after adjustment for established clinical variables. These findings identify brain-wide tumor involvement as a robust dimension of GBM prognosis and delineate the conditions under which conventional volumetric measurements provide reliable survival information. These results highlight the prognostic relevance of considering GBM as a brain-wide disease, beyond its local anatomical extent.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.