The association between lipid levels and leukocyte count: A cross-sectional and longitudinal analysis of three large cohorts

Abstract

Background: Relationships between dyslipidaemia and leukocyte counts have been investigated in several studies, demonstrating limited evidence of associations in humans. As such, studying a diverse range of cohorts will ensure evidence is robust. This study focused on investigating cross-sectional and longitudinal relationships in three large-scale cohorts.

Methods: The cross-sectional analysis included a total of 27,566 participants with valid data on lipid measures and leukocyte counts from three study cohorts: National Health and Nutrition Survey (NHANES), Korean National Health and Nutrition Survey (KNHANES) and Treating to New Targets (TNT) trial. The longitudinal analysis included 9323 participants with valid data on lipid measures and leukocyte counts at baseline and one year with statin treatment. Associations between lipid levels and leukocyte counts were analysed by multivariable linear regression and adjusted for basic demographic and cardiovascular risk factors.

Results: Cross-sectional data from NHANES demonstrated the association of lower high-density lipoprotein (HDL) cholesterol and higher triglycerides with higher leukocyte count (0.9% lower and 0.3% higher count per 10 mg/dL increase in HDL cholesterol and triglycerides respectively, both p < 0.001). Similar trends were found in TNT trial (both p < 0.001), but not in KNHANES. In the TNT trial, 10 mg/dL increase in triglycerides over one year was also associated with a 0.09 × 10³/μL increase in leukocyte count over the same period.

Conclusions: The findings of this study are consistent with those of previous human studies, supporting weak yet noteworthy associations between dyslipidaemia and leukocytosis.

Keywords: Atherosclerosis; Cholesterol; Inflammation; Leukocytes; Lipids.

Fig. 1

Participant flow in the three datasets. Three data registries were used in this study. In the NHANES (1999–2010) and KNHANES (2013) a total of 82,091 and 8018 participants were included, respectively. After exclusion of participants with missing data on lipid levels and blood cell counts 12,457 participants aged ≥20 years were included in the NHANES final analysis; 5115 participants aged >20 years were included in the KNHANES final analysis. In the TNT trial, a total of 10,001 participants aged 35–75 with known coronary artery disease (CAD) were recruited. Of these, 9994 participants had valid data on leukocyte count and were included in the final analysis.

Fig. 2

Association of lipid levels with total leukocyte count in the NHANES, KNHANES and TNT trial. For (A) NHANES and (B) KNHANES, regression coefficient B (95% CI) is expressed as ln-transformed cell counts (×10³/μL) per 10 mg/dL increase in lipid measures. In the (C) TNT study, a pilot analysis using diagnostic plot of residues does not suggest any improvement in model fitting after taking ln-transformation of leukocyte count. Therefore, the regression coefficient B (95% CI) is expressed as leukocyte count (×10³/μL) per 10 mg/dL increase in lipid measures. Data from all three registries were adjusted for age, gender, BMI, smoking status, diabetes, LDL cholesterol (except for analysis of total cholesterol and LDL cholesterol), HDL cholesterol (except for analysis of HDL cholesterol) and triglycerides (except for analysis of triglycerides). Data in the NHANES and KNHANES were further adjusted for education, alcohol consumption, systolic blood pressure, use of anti-hypertensive medication, use of lipid-lowering medication, history of CVD, RBC count and platelet count. Beyond this the NHANES was further adjusted for race/ethnicity, ratio of family income to poverty, physical activity and ln-transformed CRP, and the KNHANES was further adjusted for HbA1c. Data in the TNT study were further adjusted for race/ethnicity, hypertension and blood urea nitrogen. Abbreviations: B, regression coefficient; HDL, high-density lipoprotein; LDL, low-density lipoprotein; TC, total cholesterol; TG, triglycerides. *Remained significant after adjusting for multiple comparisons using a false discovery rate of <0.05.

Fig. 3

Association of lipid levels with differential cell counts in the NHANES study. The regression coefficient B (95% CI) are expressed as ln-transformed cell counts (×10³/μL) per 10 mg/dL increase in lipid measures. Data were adjusted for ln-transformed total leukocyte count (except for analysis of total leukocyte count), age, sex, race/ethnicity, education, ratio of family income to poverty, BMI, smoking, alcohol consumption, exercise, systolic blood pressure, use of hypertensive medication, use of cholesterol lowering medication, diabetes, history of CVD, ln-transformed CRP, LDL cholesterol (except for analysis of total and LDL cholesterol), HDL cholesterol (except for analysis of HDL cholesterol), triglycerides (except for analysis of triglycerides), RBC count and platelet count. Abbreviations: B, regression coefficient; HDL, high-density lipoprotein; LDL, low-density lipoprotein; TC, total cholesterol; TG, triglycerides. *Remained significant after adjusting for multiple comparisons using a false discovery rate of <0.05.

Fig. 4

Association of change in lipid levels with change in leukocyte count in TNT trial cohort over 12 months. This data reflects all participants (n = 9318), including those who took 10 mg atorvastatin (n = 4683) and 80 mg atorvastatin (n = 4635). The regression coefficient (B) and 95% confidence intervals (CI) were expressed as increase in cell counts (×10³/μL) over one year per one mg/dL increase in lipid levels over the same period. Data were adjusted for age, sex, Caucasian race, BMI, smoking status, hypertension, diabetes, LDL cholesterol (except for analysis of total cholesterol and LDL cholesterol), HDL cholesterol (except for analysis of HDL cholesterol), triglycerides (except for analysis of triglycerides), blood urea nitrogen, leukocyte count at baseline and treatment allocation.