P01.137 Long term follow-up of neuropsychological functions in patients with high grade gliomas: can cognitive status predict survival or tumor recurrence?

Abstract Background Patients with high grade gliomas (HGG) have a poor prognosis and experience cognitive deficits which depend on location, size, mass effect and edema of the tumor. Neuropsychological functioning (NPF) can be assessed by a battery of tests which objectivise the cognitive status of the patients. These tests are time-consuming and fatiguing for the patients, nevertheless could give relevant information about the course of the disease. Aim of our work was to describe overtime the NPF of patients affected by HGG and to correlate the NPF scores with tumor recurrence and survival. Material and Methods We examined the NPF in 102 patients operated for HGG. Patients were followed longitudinally with the Milano-Bicocca Battery for neuropsychological assessment of low grade gliomas. Patients were evaluated before surgery and few days after surgery. Then, neuropsychological and neuro-oncological follow-up were performed every three months. We first explored the correlational structure of our neuropsychological variables through hierarchical clustering in the R package Hmisc. We then investigated the impact on patients’ outcome of the variables selected through hierarchical clustering via non-linear regression, with fixed/random effects. Time to tumor recurrence/death was selected as the main dependent variable. Results Sixty-seven males and 35 females (median age 59, range 18–81) were included. Starting from a battery of 27 neuropsychological tests, 10 tests could be selected from the statistical analysis (cluster structure, condition number K of 51.16) since they were strictly correlated with the other tests; thus, these 10 tests were used for the further statistical analysis. Pitting patients’ neuropsychological profiles against tumor relapse/death, 5 cognitive tests showed a significant correlation: Token test (p=0.003), Object Naming test (p=0.02), Digit Memory Span (p=0.001), Corsi Block test (p=0.004), Semantic Fluency test (p=0.003). A steep decline for all these tests emerged a few weeks before tumor relapse/death. Analyzing the NPF overtime, all the tests worsened immediately after surgery, recovered steadily within 2–3 months and then remained stable until progression. Conclusion The tests which showed to be more informative about the patients NPF were Token test, Object Naming test, Digit Memory Span, Corsi Block test, Semantic Fluency test. These tests showed a significant decline few weeks before tumor relapse/death.