LAUSR

Recurrent glioblastoma (rGBM) prognosis is dismal. In the absence of effective adjuvant treatments for rGBM, re-resections remain prominent in our arsenal. This study evaluates the impact of reoperation on post-progression survival (PPS) considering rGBM genetic makeup. To assess the genetic heterogeneity and treatment-related changes (TRC) roles in re-operated or medically managed rGBMs, we compiled demographic, clinical, histopathological, and next-generation genetic sequencing (NGS) characteristics of these tumors from 01/2005 to 10/2019. Survival data and reoperation were analyzed using conventional and random survival forest analysis (RSF). Patients harboring CDKN2A/B loss (p = 0.017) and KDR mutations (p = 0.031) had notably shorter survival. Reoperation or bevacizumab were associated with longer PPS (11.2 vs. 7.4-months, p = 0.006; 13.1 vs 6.2, p < 0.001). Reoperated patients were younger, had better performance status and greater initial resection. In 136/273 (49%) rGBMs undergoing re-operation, CDKN2A/B loss (p = 0.03) and KDR mutations (p = 0.02) were associated with shorter survival. In IDH-WT rGBMs with NGS data (n = 166), reoperation resulted in 7.0-month longer survival (p = 0.004) than those managed medically. This reoperation benefit was independently identified by RSF analysis. Stratification analysis revealed that EGFR-mutant, CDKN2A/B-mutant, NF1-WT, and TP53-WT rGBM IDH-WT subgroups benefit most from reoperation (p = 0.03). Lastly, whether or not TRC was prominent at re-operation does not have any significant impact on PPS (10.5 vs. 11.5-months, p = 0.77). Maximal safe re-resection significantly lengthens PPS regardless of genetic makeup, but reoperations are especially beneficial for IDH-WT rGBMs with EGFR and CDKN2A/B mutations with TP53-WT, and NF1-WT. Histopathology at recurrence may be an imperfect gauge of disease severity at progression and the imaging progression may be more reflective of the prognosis.