Js03.7.a interplay between self-reported and performance-based symptoms in patients with glioma and meningioma - a network analysis and comparison between diagnoses

Abstract BACKGROUND Patients with brain tumors show complex symptomatology often involving self-reported complaints, e.g., concerning self-perceived cognitive function (SCF), fatigue or mood, as well as deficits in cognitive performance. Understanding the interplay between symptoms across diagnoses is crucial for intervention, but remains challenging. This study used network analysis to 1) uncover unique relationships between various multidimensional self-reported symptoms and cognitive performances in patients with meningioma and glioma, and 2) investigate differences in symptom networks between diagnoses. MATERIAL AND METHODS We used pre-surgical data from patients with meningioma (n=240) and glioma (n=208). Two Gaussian graphical models were constructed to explore conditional relationships among symptoms in the two samples. Included as symptoms in the networks were ten patient-reported measures: SCF (3 nodes), Anxiety, Depression, and Fatigue (5 nodes), and 5 cognitive performance measures: Inhibition, Cognitive flexibility, Verbal memory, Processing speed and Psychomotor speed. We evaluated the importance of the symptoms with three centrality indices; node strength, betweenness and closeness. We compared features of the networks between the diagnoses with a Network Comparison Test. RESULTS In both samples, particularly strong connections existed between Anxiety and Depression, Forgetfulness and False Triggering (SCF nodes), and General Fatigue and Physical Fatigue. These positive relationships were significantly stronger than most of the other relationships. The most central symptoms, based on the strength of their connections with other symptoms, were Forgetfulness, General Fatigue and Inhibition in glioma patients, and Forgetfulness, Physical Fatigue and Depression in meningioma patients. Activity-related fatigue appeared as the least central symptom in the glioma network, but it was in the top 5 in the meningioma network. The difference in the number of relationships found (meningioma: 49 vs. glioma: 41), was not significant. Visual inspection of the networks suggested additional connections between multiple fatigue and cognition nodes in meningioma that were absent in glioma. Significant differences were found between the networks regarding strength of the relationships General Fatigue-Physical Fatigue (stronger in glioma), and Physical Fatigue-Activity-related Fatigue (stronger in meningioma), p’s < .01. CONCLUSION While symptom networks of patients with glioma and meningioma show similarities, they appear to differ with regard to which symptoms play a more central role and how strongly some symptoms are related. Particularly fatigue subtypes may carry different importance. Information that multidimensional symptom networks provide about how symptoms interact can guide treatment decisions for patients with different diagnoses.