Low levels of caveolin-1 predict symptomatic bleeding after thrombolytic therapy in patients with acute ischemic stroke

Abstract

[Abstract] Background and purpose: Experimental models of cerebral ischemia demonstrate that the decrease in the caveolin-1 membrane protein results in an increase in endothelial permeability. Because this phenomenon is responsible for hemorrhagic transformation (HT) after cerebral ischemia, we aimed to determine whether caveolin-1 levels may predict bleeding after recombinant tissue-type plasminogen activator (r-tPA) administration in patients with acute stroke.

Methods: We studied 133 patients with a first hemispheric stroke treated with r-tPA within 4.5 hours of symptom onset. HT was evaluated and classified on cranial computed tomography at 24 hours and was considered as symptomatic HT (sHT) if associated with neurological deterioration. Serum caveolin-1 levels were analyzed before and at 2, 24, and 72 hours post-r-tPA administration in patients and in 40 healthy controls.

Results: Baseline caveolin-1 levels were higher in patients than controls (0.24 [0.17-0.40] versus 0.07 [0.0-0.20] ng/mL; P<0.000). Twenty six (19.5%) patients had HT, which was symptomatic in 7 (5.3%). Patients with parenchymal hemorrhage-2 and sHT had lower baseline caveolin-1 levels than the rest of patients (0.08 [0.04-0.19] versus 0.26 [0.14-0.40]; P=0.019 and 0.08 [0.02-0.17] versus 0.26 [0.13-0.41]; P=0.019, respectively). The levels remained stable in the first 72 hours in patients with parenchymal hemorrhage-2 and sHT, whereas in the rest of patients levels decreased in this time. Caveolin-1 levels ≤0.17 ng/mL had the highest predictive capacity of sHT (86% sensitivity, 65% specificity, 99% negative predictive value, 12% positive predictive value). After adjustment for confounders, caveolin-1 levels ≤0.17 ng/mL independently predicted sHT (odds ratio, 11.6; 95% confidence interval, 11.3-102.8; P=0.027).

Conclusions: Low serum levels of caveolin-1 are an independent predictor of sHT after r-tPA administration. Because of the small sample size, further research is needed to validate these data.