Abstract C001: PIONEER: Computational Probing of dIfferences in symptOms and fuNction of divErsE brain and spine tumoR populations

Abstract

Abstract Background: Cancer outcomes studies primarily enroll white patients, and symptom care standards are based overwhelmingly on white patients’ experiences, even though many analyses have shown ethno-racial disparities in symptom burden. This study seeks to better understand ethno-racial disparities in the symptom burden of brain and spine tumor patients using novel computational network analysis (NA) approaches. NA identifies complex symptom co-severity patterns across large patient cohorts. This is the first study to use NA to analyze ethno-racial disparities in cancer symptom burden. Methods: Symptom severity data reported using the MD Anderson Symptom Inventory from a two-institutional cohort of 1651 brain and spine tumor patients was analyzed. Symptom data from white (n = 1,382) and non-white patients (n = 269) were analyzed separately. All non-white patients were analyzed together due to sample size limitations. Gaussian Graphical Model networks were constructed for each group. Network characteristics were analyzed and compared using permutation-based statistical tests. Results: Networks for white and non-white patients were constructed with high and moderate accuracy, respectively. Assessment of strength centrality, a measure of how core a symptom is to overall symptom burden, and betweenness centrality, a measure of how much a symptom contributes to the severity of other symptoms, revealed several key symptoms for each group. For white patients, fatigue, nausea, and distress/feeling upset had the highest strength, whereas fatigue, nausea, and pain had the highest betweenness. For non-white patients, drowsiness, disturbed sleep, and sadness had the highest strength, whereas change in appetite, disturbed sleep, sadness, and pain had the highest betweenness. The two network architectures were statistically different from each other in the Network Comparison Test (p = 0.033). White patients had significantly higher strength for fatigue (p = 0.032) and betweenness for nausea (p = 0.005), whereas non-white patients had significantly higher strength for sleep disturbance (p = 0.005). Independent t-test analysis only identified drowsiness as being significantly different between the two groups (p = 0.032), emphasizing how NA can reveal differences in symptom burden undiscoverable by traditional analyses. Discussion: Our results demonstrate that white and non-white patients experience different symptom co-severity patterns, with symptoms such as sleep disturbance being more influential in non-white patients and symptoms such as fatigue and nausea more significant in white patients. These results underscore the importance of considering the differences (and similarities) in symptomatology of patients from different racial and ethnic backgrounds. This will help address disparities and provide more personalized and effective care for diverse brain and spine tumor patient populations. Citation Format: Brandon H. Bergsneider, Elizabeth Vera, Mark R. Gilbert, Terri S. Armstrong, Orieta Celiku. PIONEER: Computational Probing of dIfferences in symptOms and fuNction of divErsE brain and spine tumoR populations [abstract]. In: Proceedings of the 15th AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; 2022 Sep 16-19; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2022;31(1 Suppl):Abstract nr C001.