Major blood lipids and risk of ischaemic stroke and intracerebral haemorrhage in Chinese adults

Background Stroke is the second most common cause of premature death worldwide, and the leading cause of death and disability in China. Stroke is a clinical syndrome that includes three distinct pathological types (i.e., ischaemic stroke [IS], intracerebral haemorrhage [ICH] and subarachnoid haemorrhage [SAH]) that have a different underlying pathophysiology. Previous observational studies and randomised trials have reported conflicting results on the associations of major blood lipids, including low-density lipoprotein cholesterol (LDL-C), high density lipoprotein cholesterol (HDL-C), and triglycerides with stroke pathological types. Moreover, the causal relevance of many of these associations is still not fully understood. Methods The China Kadoorie Biobank (CKB) prospective study collected questionnaire, physical measurement data, and blood samples on 512 891 adults from 10 diverse regions in China in 2004-2008. By 1st January 2016, over 32 000 incident IS, and 8000 ICH cases were recorded. A nested case-control study of stroke included 16 541 adults (5475 IS, 4776 ICH, and 6290 controls), with measurements of baseline plasma concentrations of LDL-C, HDL-C, and triglycerides, and genotyping of known lipid-related genetic variants. Repeat measurements of blood lipids were obtained in random samples of participants at resurveys (2008 and 2013-2014) to correct for regression dilution bias. The associations of major blood lipids with stroke pathological types were examined after adjustment for confounders and correction for regression dilution bias. The causal relevance of these associations was assessed by Mendelian randomisation (MR) using genetic risk scores (GRSs) involving lipid-related genetic variants identified in Western populations. Multivariable logistic and Cox regressions were used to yield odds ratios (ORs) and hazard ratios (HRs), respectively, relating major blood lipids or genetic variants with risks of stroke pathological types. Results Overall, the incidence rates of first IS and ICH were 761 and 187 cases per 100 000 person-years, respectively, with ∼20% of the incident strokes attributable to ICH. The mean (SD) plasma concentrations of LDL-C, HDL-C, and triglycerides were 2.3 (0.8) mmol/L, 1.3 (0.3) mmol/L, and 1.6 (1.1) mmol/L, respectively. The regression dilution ratios at the mid-point of follow-up (around 3.7 years after baseline) were 0.68 for LDL-C, 0.72 for HDL-C, and 0.62 for triglycerides, respectively. Throughout the range of LDL-C examined (1.7-3.2 mmol/L), there was a positive log-linear association of usual LDL-C with risk of IS (HR 1.17; 95% CI 1.10-1.25 per 1 mmol/L higher LDL-C), and an inverse association with ICH (0.86; 0.80-0.92). Likewise, after adjusting for non-lipid vascular risk factors, usual plasma triglycerides were positively associated with risk of IS (1.02; 1.00-1.04 per 30% higher triglycerides), albeit weaker than for LDL-C, and were inversely associated with risk of ICH (0.94; 0.92-0.96), independent of LDL-C. HDL-C was inversely associated with risk of IS (0.93; 0.89-0.97 per 0.3 mmol/L higher HDL-C), but not with ICH. The proportional relevance of LDL-C, triglycerides, and HDL-C with IS differed significantly from those for ICH (p heterogeneity: <0.00001, <0.00001, 0.0006, respectively). In the MR analyses, an externally weighted GRS of 8 known LDL-C related genetic variants was associated with plasma LDL-C (p=1.15x10-272), but not associated with other lipid fractions or non-lipid major vascular risk factors. Each 1 mmol/L higher genetically-determined LDL-C was associated with an OR of 1.28 (1.05-1.56) for IS, similar to the risk estimates in both observational studies and randomised trials. There was no significant association of genetically-determined LDL-C with ICH (OR 1.04; 0.87-1.25), but the estimates did not differ significantly from those in the observational studies (p heterogeneity=0.06), and randomised trials (p heterogeneity=0.8). Neither genetically-determined HDL-C nor triglycerides were significantly associated with either IS or ICH. Conclusion The associations of major blood lipids with stroke differed qualitatively between pathological types. Lower LDL-C is causally associated with lower risk of IS, and probably, higher risk of ICH. Better genetic instruments are needed to assess the causal relevance, if any, of the observed associations of HDL-C, and of triglycerides for stroke pathological types presented in this thesis.