Patterns of Failure after Stereotactic Radiosurgery vs. Whole Brain Radiotherapy for Resected Brain Metastases: Central Imaging Review of the N107C/CEC.3 (Alliance) Phase III Clinical Trial

Purpose/Objective(s) Whole brain radiotherapy (WBRT) has historically been the main instrument to decrease intracranial progression after resection of brain metastases. The previously reported NCCTG N107C/CEC.3 (Alliance) randomized phase III clinical trial of WBRT vs. stereotactic radiosurgery (SRS) demonstrated that overall survival did not significantly differ between arms, while SRS was associated with less cognitive decline than WBRT. However, the initial analysis also showed that local tumor bed control (LC) was significantly worse with SRS than WBRT, with an unexpectedly low 12-month LC rate of 60.5% in patients receiving post-operative SRS. To further investigate LC, a central imaging review was performed. Materials/Methods In N107C/CEC.3, 194 patients were randomized to SRS (n=98) or WBRT (n=96). SRS dose was dependent on resection cavity volume and ranged from 12 Gy to 20 Gy all in a single fraction, and WBRT dose was 30 Gy in 10 fractions or 37.5 Gy in 15 fractions. For this analysis, patients with local failure (LF) or leptomeningeal failure that had imaging available were centrally reviewed by two radiation oncologists to verify pattern of failure. Patient and treatment characteristics were assessed for association with updated outcomes after central review using Cox proportional hazards, Pearson's chi-square, or Kruskal-Wallis tests. Results Seventy-three patients with LF or leptomeningeal failure determined by their local site were centrally reviewed. After adding patients with no LF or leptomeningeal failure, 185 (95% of total enrolled) patients were included in this analysis as 9 patients with LF did not have imaging available for central review. Of 59 patients initially determined to have LF, 20 were determined on central review to be free of LF. Accordingly, after central review post-operative SRS was no longer associated with significantly lower rates of LC than WBRT (79.2% vs. 86.5% 12-month LC, respectively, p=0.099). After central review, the interrelated variables of tumor diameter, target volume, and prescription dose were associated with risk of LF. Patients with resection cavities > 3 cm had higher rates of LF than those with smaller cavities (23.5% vs. 12.6% 12-month LF rate, respectively; p=0.010). Larger target volume was associated with increased risk of LF (HR: 1.05, 95% CI: 1.01 – 1.09, p=0.008), while higher prescription dose was associated with lower risk of LF (HR: 0.81, 95% CI: 0.67 – 0.96, p=0.018). Conclusion In contrast to the initial report, after central review SRS was not associated with significantly higher rates of LF than WBRT. Patients with larger surgical cavities had higher rates of LF, though it is unclear whether this is related to cavity size, biology, or radiation dose. Randomized data comparing single to multifraction SRS in larger surgical cavities is awaited.