Prognostic associations of steroid use in patients with glioblastoma: A population-based study

Abstract Background Steroids are broadly used in patients with brain tumors. A possible detrimental association of corticosteroid treatment with outcome in patients with glioblastoma has been reported but not explored on a population level. Methods Clinical data, including corticosteroid use and related complications, of 223 patients diagnosed with isocitrate dehydrogenase wild-type glioblastoma in the years from 2010 to 2014 in the Canton of Zurich, Switzerland, were retrospectively analyzed. Survival data were estimated with the Kaplan-Meier method. Cox proportional hazards regression models were used to test the association of clinical variables with survival. Results Most patients (N = 201, 95%) received steroids at least once in the first year after diagnosis, the majority (N = 151, 69%) in the perioperative setting (hospitalization for surgery ± 7 days), and 92 patients (61%) in the first 3 months. Steroid use at 3 months was associated with steroid use at 6 months (OR = 7.66, P < .001), and the latter was associated with inferior survival even after correction for known prognostic and predictive factors (P = .003). Conclusion Steroid use at the 6-month landmark after diagnosis is independently associated with inferior survival in patients with glioblastoma. Limitations of the study include its retrospective design and the confounding factor that 1 frequent indication of steroid treatment is a symptomatic tumor. The biological mechanisms mediating this effect remain incompletely understood. Our findings support a cautious use of steroids and highlight the importance of minimizing dose and duration whenever clinically feasible.