The Nonlinearity Seen for LDL-Cholesterol with Lyophilized Material Is a Matrix Effect

¹ Department of Pathology, Division of Clinical Chemistry, The Johns Hopkins University School of Medicine, 600 North Wolfe St, Meyer B-125, Baltimore, MD 21287-7065,
² Department of Clinical Pathology, Clinical Chemistry Service, Clinical Center, National Institutes of Health, Building 10, 2C-407, Bethesda, MD 20892
^a Author for correspondence. Fax 410-955–0767; e-mail mkroll{at}pathlan.path.jhu.edu.

To the Editor:

Recently, Genzyme Corporation (Cambridge, MA) developed and patented a direct method for LDL-cholesterol (LDL-C) determination. The method uses an antibody to separate LDL from VLDL and HDL. Because the method is analogous to that for HDL-cholesterol (HDL-C) in that it requires a separation step and that matrix effects have been documented to interfere with the assessment for linearity for that analyte, we investigated the effect matrix has on the evaluation of linearity for the direct measurement of LDL-C (1).

We determined LDL-C on the Cobas FARA (Roche Diagnostic Systems), using cholesterol reagents: cholesterol esterase and oxidase (Boehringer Mannheim Diagnostics) and an LDL-C kit (Sigma Diagnostics). We tested linearity, using the following materials: CAP Linearity Survey material LN2–21 (College of American Pathologists, Northfield, IL); reconstituted according to instructions, mixed high- and low-concentration vials to make five, equally-spaced in concentration solutions; HDL-C and LDL-C control material (Sigma Diagnostics); and Enzyme diluent (DuPont Clinical Diagnostic Systems). In addition, we used Centriprep 100 concentration tubes (Amicon) to concentrate serum and a Sorvall RT 6000 Centrifuge with H1000B swinging bucket rotors to separate the phases in the LDL-C method.

Serum was added to the LDL-C separation tubes, each tube containing latex beads coated with affinity-purified goat polyclonal antisera against specific apolipoproteins. HDL and VLDL adhere to the latex beads, allowing one to centrifuge the samples in the Sorvall centrifuge and pour off the filtrate solution containing the remaining LDL. We then determined cholesterol with the COBAS FARA.

We followed a previously published method of preparing samples for linearity studies, producing nine concentrations (1). In addition, we concentrated serum, centrifuging it in the Centriprep 100 for 20 min at 3000 rpm at a 12-cm radius (1000g) in the Sorvall centrifuge. We reconstituted HDL-C and LDL-C control materials following the manufacturers' directions, and diluted them as above. In addition, we reconstituted LDL-C controls, separating the LDL-C. We made nine dilution concentrations (1) and determined cholesterol to measure the LDL-C fraction.

We evaluated all data using the polynomial evaluation method for linearity, as described by Kroll and co-workers (1)(2). In the polynomial method for linearity, the nonlinear beta coefficients are statistically analyzed by use of the t-test. If the nonlinear beta coefficients are statistically significant, the method is nonlinear; if not, the method is linear.

The lyophilized materials, CAP Linearity Survey material, and the LDL and HDL controls all had statistically significant nonlinear coefficients (P <0.02) and were nonlinear. The routine serum and concentrated, postprecipitant pools did not have statistically significant beta coefficients and therefore were linear. The nonlinearity of the HDL-C and LDL-C controls and the CAP samples was sigmoid in shape, whereas the serum, concentrated serum, and LDL-C control material precipitated before dilution, all appeared linear (Fig. 1 ).

View larger version (14K): [in this window] [in a new window]   Figure 1. Concentration of cholesterol, representing LDL-cholesterol, plotted against tube number (relative dilution). LDL control (), HDL control (), CAP linearity materials (), serum (), concentrated serum (), and LDL-C control separated before dilution (). The materials represented by solid symbols show sigmoid curves, whereas those represented by open symbols are linear.

Previous studies showed that methods for the determination of HDL-C were nonlinear with CAP linearity survey material as well as with the quality-control materials supplied with each method (1). These HDL methods were linear, however, when fresh sera was used for the linearity studies, indicating that the observed nonlinearity was a result of the material matrix and not the method itself. In this study, we observed similar results for LDL-C, as determined with Sigma reagents. The Sigma Diagnostics method for the direct determination of LDL-C is linear for serum samples. The reportable range can be extended by concentrating a serum pool. Lyophilized, processed materials, as seen with controls or CAP survey material, are nonlinear with this method, but the nonlinearity is a matrix effect of the material. The nonlinearity of the matrix effect occurs at the separation step.

One of the common factors for processed materials is that many of these materials are lyophilized. Many studies have shown that lyophilized materials exert a matrix effect on the determination of cholesterol (3)(4)(5)(6). These studies suggest that the act of lyophilization alters the native state of lipoproteins. The nonlinearity seen with the processed materials for the HDL and LDL methods strongly suggests that the observed matrix effects are dependent on the concentrations of the analyte as well as the other lipoproteins. We encourage those who use the LDL-C method to establish their reportable range with fresh serum pools. One may concentrate the sera, as performed in this study, to further extend the reportable range.

References

1. Kroll MH, Chesler R. Nonlinearity of high-density lipoprotein cholesterol determination is matrix dependent. Clin Chem 1994;40:389-394. [Abstract/Free Full Text]
2. Kroll MH, Emancipator K. A theoretical evaluation of linearity. Clin Chem 1993;39:405-413. [Abstract/Free Full Text]
3. Kroll MH, Lindsey H, Greene J, Silva C, Hainline A, Jr, Elin RJ. Bias between enzymatic methods and the reference method for cholesterol. Clin Chem 1988;34:131-135. [Abstract/Free Full Text]
4. Miller WG. Matrix effects in the measurement and standardization of lipids and apolipoproteins. Curr Opin Lipidol 1992;3:361-364.
5. Myers GL, Waymack PP. Matrix effects in biological reference materials used in the standardization of cholesterol measurements. Fresenius' J Anal Chem 1990;338:538-542.
6. Ross JW, Myers GL, Gilmore BF, Cooper GR, Naito HR, Eckfeldt J. Matrix effects and the accuracy of cholesterol analysis. Arch Pathol Lab Med 1993;117:393-400. [Web of Science][Medline] [Order article via Infotrieve]