Effects of intravenous human albumin, enteral cilostazol, and combination therapy on the reduction of delayed cerebral ischemia in patients with aneurysmal subarachnoid hemorrhage

Background: Delayed cerebral ischemia (DCI) and cerebral infarction remain major causes of disability and death after aneurysmal subarachnoid hemorrhage (aSAH). Enteral cilostazol and intravenous (IV) 25% human albumin have each been investigated as candidate therapies targeting potentially overlapping DCI mechanisms. Objective: This article aims to estimate the plausible effects of albumin, cilostazol, and combination therapy on reducing DCI after aSAH using Bayesian prior-predictive simulation and factorial component-effect decomposition. Methods: We defined four treatment classes: control, albumin-only, cilostazol-only, and combination therapy. The primary endpoint was the National Institute of Neurological Disorders and Stroke Common Data Elements-defined DCI (CDE-d-DCI) on computed tomography on days 10 to 14. A prior-predictive Monte Carlo model used a control DCI-risk prior centered on the Cilostazol Albumin Treatment in Subarachnoid Hemorrhage (CATS) protocol-planning value of 31%, a cilostazol prior anchored to randomized-trial meta-analysis estimates, and an albumin prior informed by the Albumin in Subarachnoid Hemorrhage (ALISAH) dose-escalation data with wider uncertainty because ALISAH was uncontrolled. No observed CATS outcomes were analyzed. External consistency, control-prior sensitivity, weaker albumin priors, correlated-prior structures, and factorial component effects were evaluated. Results: Under base-case priors, median absolute risk reductions (ARRs) versus control were 9.1 percentage points for albumin (90% simulation interval, -3.3 to 18.1; prior-predictive probability of ARR > 0, 89.8%), 18.1 percentage points for cilostazol (11.0 to 25.4; 99.9%), and 21.7 percentage points for combination therapy (12.4 to 29.9; 99.8%). Although combination therapy had the largest projected reduction versus control, the incremental benefit of adding albumin to cilostazol was small and uncertain (3.5 percentage points; -3.0 to 8.9), and the additive interaction was also uncertain (-5.5 percentage points; -11.9 to 2.8). Illustrative two-group sample sizes were 343, 76, and 49 participants per group for albumin, cilostazol, and combination therapy, respectively, versus control. Conclusion: These simulations provide planning-level projections and do not establish efficacy. The findings support continued randomized evaluation of cilostazol and albumin, with concurrent single-agent, combination, and control arms to estimate component effects and interaction effects.