Abstract
Glioblastomas (GBM) can recur in different ways. While local recurrence is most common, some GBM recur at distant sites or simultaneously at multiple sites. However, the consequences of different regrowth patterns for the clinical course and the factors that might influence regrowth patterns or different modalities of recurrence remain unclear. We wondered (1) whether a more accurate analysis of regrowth patterns helps to detect subgroups of GBM patients, (2) if evaluation of relapse patterns correlates with differences in survival, and (4) whether we can identify predictors for distinct regrowth patterns. Therefore, we retrospectively collected demographic data, as well as tumor- and patient-characteristics, from 251 patients treated at two institutions and characterized their recurrence patterns by analyzing Magnetic Resonance Imaging data. We observed distinct differences in patients' overall and progression-free survival with respect to multicentric and multifocal recurrences, further supporting the hypothesis that these recurrences develop differently. Several tumor relapse patterns were associated with patients' progression-free and overall survival (e.g., unifocal local and multicentric recurrences; p < 0.05), even when other known prognostic factors were taken into account. TTFields were associated with prolonged progression-free survival (mPFS 7.2 months vs. 4.8 months, p = 0.03). They were predictive of a higher frequency of non-local recurrence patterns (OR 0.16, p = 0.02) and longer time to development of a local recurrence after subtotal tumor resection (mPFS 11.1 months vs. 5.2 months; p = 0.01). This can be interpreted as a sign of improved local control.