Abstract
BACKGROUND: The clinical significance and contribution of the lipid profile in atherosclerosis are well established. However, further investigation is needed in stroke patients, particularly regarding apolipoprotein B100 (ApoB100), a novel non-traditional lipid component in the lipid profile. OBJECTIVES: To explore lipid parameters and their impact on stroke outcomes in patients with and without thrombolysis. METHODS: We prospectively enrolled patients with acute ischemic stroke (AIS) at a single center, including those who did and did not receive thrombolysis. Participants were stratified into improvement (favorable outcome at 2 weeks) and non-improvement groups. Demographic, laboratory, imaging, and clinical scale data were compared between groups. Random forest analyses were used to evaluate the predictive value and importance of individual lipid measures: triglycerides, total cholesterol, high-density lipoprotein (HDL), low-density lipoprotein (LDL), ApoB100, and lipoprotein(a), which better describe the internal characteristics of the profile. RESULTS: Complete data were available for 262 AIS patients, 165 of whom received thrombolysis. Plasma ApoB100 levels were significantly lower in the thrombolysis group (p < 0.001) and decreased ApoB100 levels were independently associated with 2-week stroke improvement (p = 0.009, OR = 0.89, 95% CI: 0.84-0.93). Random-forest feature-importance plots revealed that HDL and ApoB100 (each contributing > 15%) were the strongest lipid predictors of a favorable outcome, outperforming the other lipid variables. CONCLUSIONS: We found that thrombolysis is associated with ApoB100 decrease and a decrease in ApoB100 can predict the 2-week functional improvement in stroke. HDL and ApoB100 emerge as more important determinants of favorable AIS outcomes in this machine-learning analysis. These findings warrant external validation in multi-center trials. TRIAL REGISTRATION: ChiCTR1800018315, 11/09/2018.