NCOG-36. CLINICAL AND DEMOGRAPHIC CHARACTERISTICS PREDICT POSTOPERATIVE BEHAVIORAL DYSFUNCTION FOLLOWING HIGH-GRADE (WHO 2-4) GLIOMA RESECTION

Abstract BACKGROUND Gliomas have the highest incidence (85%) of cancer-related cognitive and behavioral impairments. Few studies have explored preoperative clinical and demographic drivers/predictors of postoperative behavioral impairment. METHODS We analyzed prospectively collected data from patients 1) who underwent glioma (WHO 2-4) resection and 2) who had postoperative Flanker testing, an assessment of inhibitory control and selective attention (NIH Toolbox Cognition Battery). Patients whose age-adjusted Flanker score (last follow-up) was less than the cohort mean were classified as having “poor” outcomes. Univariate analysis, multivariate logistic, and LASSO regressions were performed to identify drivers of poor outcomes. RESULTS In total, 43 patients (mean age at diagnosis 45.9 [range 18-77] years, 35% female) met inclusion criteria. The average age-adjusted Flanker score was 83.3±13.6; twenty (46.5%) patients had below-average scores. On univariate analysis, there were no significant differences in demographic factors (age, sex, education, employment) and tumor characteristics (grade, location). Multivariate logistic regression identified having high school versus higher degree (OR:1.6x10^4, p=0.017), WHO Grade 4 versus Grade 2 (OR:2.36, p=0.026), and parietal lesion (OR:1.6x10^3, p=0.034) as risk factors, while increased age (OR per 1-year increase:0.71, p=0.009) and dominant language hemisphere lesion (OR 0.002, p=0.048) were protective against poor behavioral outcome. LASSO regression (efficiently narrows down predictors) identified higher WHO grade (4 vs. 2), lower education (high school vs. higher), and insular and parietal lesions as predictive of poor outcomes. Increased age at diagnosis, frontal lobe lesions, and lesions on the dominant language hemisphere were protective. CONCLUSION Lower education status, higher WHO grade, and insular and parietal lesions are risk factors for poor behavioral outcomes postoperatively following glioma resection. In contrast, increased age, frontal lobe lesions, and dominant language hemisphere lesions were positive prognostic factors. Neuro-oncologists may use these findings to preoperatively counsel patients and tailor care plans with necessary postoperative support.