Treatment-Induced Increase in Eosinophil Count is Associated with Improved Overall Survival in Glioblastoma patients

Abstract

The adaptive and innate immune system play a critical role in tumor control. Methods to improve the anti-tumor immune response through T-cell-specific therapies is the current focus of Immuno-Oncology. Our group and others have demonstrated that treatment-induced lymphopenia correlates with poorer prognosis in newly diagnosed glioblastoma (GBM) patients. We hypothesized that granulocytes, including neutrophils and eosinophils, influence the immunologic response to chemoradiation therapy (CRT) and patient outcomes. We conducted a single-institution, retrospective analysis of 94 patients diagnosed with WHO Grade IV GBM between 2011-2016 treated with surgery followed by CRT (60Gy in 30 fractions, temozolomide 75mg/m2/day) and adjuvant temozolomide. We evaluated the immunologic response to therapy through monthly complete blood count with differential for up to 3 months after CRT. On univariate analysis, the use of immunosuppressive steroids negatively correlated with overall survival (Pearson’s r = -0.382, p<0.01). White blood cell (WBC) count and absolute neutrophil count (ANC) measured at the end, 1 and 2 months after CRT also negatively correlated (r = -0.226 to -0.342, p<0.05) with overall survival. Whereas, eosinophil count measured at the end, 1, 2 and 3 months after CRT (r = 0.227 to 0.366, p<0.05) and absolute lymphocyte count (ALC) measured 3 months after CRT (r = 0.268, p<0.05) positively correlated with overall survival. On partial least squared regression, steroid use and eosinophil count 3 months after CRT were the 1st and 2nd variables of importance in the projection (VIP). In a multivariate analysis, steroid use and eosinophil count 3 months after CRT remained statistically significant (F=5.4, F=6.8, p<0.05). Kaplan-Meier analysis demonstrated that patients with a persistently increased eosinophil count (> 150 cells/uL) 3 months after CRT lived 45% longer than patients with lower (<150 cells/uL) eosinophil count (1165 (CI 663-942) vs. 802 (CI 973-1356) days, p=0.05). Cox regression analysis of patients treated with steroids > or < 59 days confirmed that increased steroid use correlated a 35% decrease in overall survival (639 (CI 513-759) vs. 988 (856-1121) days, p<0.01), independent of initial performance status. We demonstrate that treatment-induced increase in eosinophil count is the strongest hematologic predictor of overall survival in newly diagnosed GBM. Eosinophil count may serve as a proxy for immunologic response to therapy and thus yield more prognostic insight than ALC. We also demonstrate that independent of initial performance status, steroid use was associated with a decrease in overall survival. Though prospective studies are needed, these data suggest that use of immunosuppressive steroids should thus be used judiciously and discontinued when clinically possible.