Nonlinearity of high-density lipoprotein cholesterol determinations is matrix dependent.

The majority of methods for determining high-density lipoprotein (HDL) cholesterol failed the 1991 College of American Pathologists (CAP) linearity survey (sets LN2-A, B, C). We hypothesized that they failed because of the survey material matrix. We evaluated linearity with dextran sulfate or phosphotungstate as the precipitating reagents for several methods: Ektachem; TDx; Sigma; Dimension; Cobas Fara, with Roche reagents; and the Hitachi 736, with precipitating reagents from Boehringer Mannheim Diagnostics. We tested CAP survey material, Sigma HDL-cholesterol control material, and a fresh serum pool for linearity, using the polynomial method. All of the methods were nonlinear for the CAP material and for the controls precipitated after dilution. Five of eight methods were linear for the control materials precipitated before dilution. All methods but one were linear for the serum pool. These results demonstrate that the source of the nonlinearity is located in the precipitation step and depends on the sample matrix.