Comparison of lipoprotein fractionation by sequential density gradient ultracentrifugation with precipitation or fast phase liquid chromatography

Lipoprotein fractions, i.e. very low density (VLDL, ρ=0.93−1.019), low density (LDL, ρ=1.019−1.063), and high density lipoprotein (HDL, ρ=1.063−1.21) were separated by sequential gradient density ultracentrifugation (SDGU) and their cholesterol values were determined and compared with values determined by fast phase liquid chromatography (FPLC) or HDL cholesterol by phosphotungstic acid-MgCl2 (PTMg) precipitation in hamsters fed diets containing various levels of cholesterol. The correlation coefficient (r) for LDL between the SDGU and FPLC methods in plasma from hamsters fed 0–3% cholesterol diets was 0.21–0.51, n=45. The FPLC method over estimated (+45%) LDL and the HDL values were under estimated (−18%). The agreement between FPLC and SDGU methods was also evaluated by plotting mean values against the differences between the values obtained by the two methods. FPLC method overestimated LDL 22–49% (mean 36%) and HDL was underestimated 14–27% (mean 20%). This was significant systematic bias with the FPLC method in VLDL, LDL and HDL values with the level of cholesterol in the diet. As FPLC is a fast method, it could be used in intervention type experiments to monitor the progress, however final results may need validation for research studies with the SDGU method. HDL determined by the SDGU method and phosphotungstic acid MgCl2 precipitation in hamsters (n=26) fed 0.25% cholesterol diets was represented by two significantly different (P⩽0.05) divergent lines when a regression fitted model and a one-to-one relationship model by the two methods were plotted. The data suggest that in hamsters fed either no added cholesterol or cholesterol-containing diets, lipoprotein fractions determined by the precipitation method or by FPLC need to be validated against a SDGU for critical samples.