To determine if soluble ST2 (sST2) is associated with neurotoxicity after chimeric antigen receptor T-cell (CAR-T) therapy.

sST2 is a member of the interleukin-1 receptor family known to predict outcome in ischemic and hemorrhagic stroke. Its role in neurotoxicity following CAR-T therapy has not yet been reported.

sST2 concentration was measured on plasma samples from a cohort of 43 patients who underwent CAR-T infusion for lymphoma at Dana-Farber Cancer Institute. Blood was drawn at baseline and 7 days after infusion. sST2 was measured using a commercially-available ELISA (Critical Diagnostics, San Diego, CA). Neurotoxicity was defined as grade 1 or greater on the Common Terminology Criteria for Adverse Events rating scale. The association between sST2 and neurotoxicity was determined using logistic regression and receiver operating curve (ROC) analysis. Spearman rank correlation was used to determine relationships between sST2 level and transcranial doppler velocities (TCDs).

Patients with neurotoxicity had similar levels of sST2 at baseline, but higher levels on day 7 (100.9 ng/mL [IQR 49.7-200] vs. 47.1 [30.4-58.8], p = 0.02). ROC analysis demonstrated that sST2 > 80.0 ng/mL predicted neurotoxicity with an area under the curve of 0.657. After adjusting for clinical covariates, sST2 remained an independent predictor of neurotoxicity in a final model with age and C-reactive protein. Serum interleukin 6 was collinear with sST2 in the model, and the two were significantly correlated (rho = 0.71, p < 0.0005). Lastly, sST2 level was found to be correlated with peak MCA, PCA, and ACA TCD velocities (MCA rho = 0.73, p < 0.005; PCA rho = 0.83, p < 0.0001; ACA rho = 0.74, p < 0.001).

Plasma sST2 is associated with neurotoxicity and TCD velocities after CAR T-cell therapy. These results support the role of sST2 as a biomarker for neurovascular injury.