Proximity to Brainstem at Recurrence is Associated with Decreased Survival in IDH Wild-Type Glioblastoma.

<bold>Purpose/objective(s): </bold>Recent autopsy data from patients with glioblastoma (GBM) has demonstrated that pronounced brainstem infiltration is now a common pattern of disease progression near the end of life. The present study evaluated the association of disease proximity to the brainstem on imaging, both at diagnosis and at recurrence, with overall survival (OS).<bold>Materials/methods: </bold>140 patients with IDH wild-type GBM treated definitively with resection and adjuvant chemoradiation at a single institution from 2013-2019 were retrospectively analyzed. Disease proximity to brainstem was calculated from measurements made on T1 post-contrast MRI brain obtained at diagnosis postoperatively and at recurrence. Tumor volume was approximated by the resection cavity plus adjacent extrinsic enhancement, measured during target delineation at time of treatment delivery. The Kaplan-Meier method was used to estimate OS, and multivariable analysis was performed using the Cox proportional hazards model. Logistic regression was used to assess the effect of tumor volume on brainstem involvement, and the relationship between tumor volume and proximity to brainstem was assessed by the Spearman correlation.<bold>Results: </bold>Median distance of disease to brainstem was 22.4 mm (range 0-54) at diagnosis, with 2.1% (n = 3) having brainstem involvement. At 56-month median follow-up, 87% of patients had recurrence; 2-year OS was 38.6%. Median distance of disease to brainstem was 20.8 mm (range 0-63) at recurrence, with 5.7% (n = 8) having brainstem involvement at recurrence. Controlling for resection status, MGMT methylation, and Karnofsky performance status, there was no noted association of brainstem invasion (P = .60) or proximity at diagnosis (P = .14) with OS. However, at recurrence, distance from the brainstem was significantly associated with improved OS when measured continuously (HR: .97, 95% CI: .95-.99, P = 0.001) and when compared by group (20-40 mm vs. ≤20 mm: HR .59, 95% CI: .35-.99, P = 0.05; > 40 mm vs. ≤20 mm: HR .16, 95% CI: 0.05-.55, P = 0.003). While volume of tumor at diagnosis was not found to be independently associated with brainstem involvement (P = .99) or OS (P = .90) at recurrence, tumor volume at diagnosis was found to be significantly associated with disease proximity to brainstem at the time of recurrence (P = 0.037), with larger volume disease more likely to recur close to the brainstem.<bold>Conclusion: </bold>GBM proximity to the brainstem at diagnosis was not found to be prognostic, whereas disease proximity to the brainstem at recurrence was associated with worse OS. Tumor volume at diagnosis was found to be associated with disease proximity at recurrence, with higher volume disease at diagnosis associated with greater proximity to the brainstem at recurrence. These findings may have ramifications for upfront and salvage radiation treatment design.<bold>Author Disclosure: </bold>A. Bajaj: None. E.T. Benishay: None. I.B. Helenowski: None. R. Savoor: None. M.R. Drumm: None. N. Rammohan: None. S. Sachdev: None. C.M. Horbinski: None. T. Kruser: Employee; Northwestern Memorial Hospital. Honoraria; Elsevier, OncLive. Speaker's Bureau; AstraZeneca. Advisory Board; AstraZeneca. [ABSTRACT FROM AUTHOR]