Matrix effects on proficiency testing materials. Impact on accuracy of cholesterol measurement in laboratories in the nation's largest hospital system.

The objective of this collaborative study with the Department of Veterans Affairs (VA), College of American Pathologists (CAP), and the Centers for Disease Control and Prevention (CDC) was to quantitate the matrix-induced biases of cholesterol measurements on the CAP Comprehensive Chemistry Surveys materials used in proficiency testing (PT). A total of 174 VA Medical Centers outpatient clinics and clinical laboratories participate in the VA-CDC National Cholesterol Standardization and Certification Program. This study was conducted in 112 VA laboratories that have been standardized for measuring cholesterol accurately (within +/- 3.0% of the CDC reference-method values) using fresh, unfrozen, unadulterated human serum samples. Fresh serum samples and 1990 CAP Surveys materials were sent by overnight mail, and the laboratories were asked to analyze them simultaneously in triplicate in a single analytic batch run. The results showed significant matrix-effect biases with the CAP Surveys materials with six of the eight major peer groups, despite the fact that accuracy of cholesterol measurements was maintained with fresh serum samples. The magnitude and direction (positive or negative) of the matrix-effect biases were instrument, reagent, and method specific using the following peer groups: du Pont Dimension (-8.9%); Beckman CX4, CX5, and CX7 (-5.5%); Kodak Ektachem 400, 500, and 700 (+4.4%); Instrumentation Laboratory Monarch (-3.1%); Baxter Paramax (-2.4%); Technicon SMAC and RA (+1.3%); Hitachi/BMD 704 through 747 (+0.4%); and Abbott Spectrum (-0.3%). The CAP PT materials used currently do not behave in a manner identical to fresh human serum when measuring cholesterol on many, but not all, analytic systems. The observed biases due to "matrix effects" with PT materials will cause incorrect conclusions about the accuracy of many laboratory procedures performed on fresh patient specimens. This matrix-effect phenomenon will severely hamper interlaboratory accuracy transfer, standardization efforts, and monitoring performance of a laboratory's testing accuracy with the use of the current survey materials used in PT programs. Collaborative efforts are needed to (1) improve PT fluids to analytically behave more like fresh, human serum; (2) improve instrument design and reagent formulation; and (3) select methods and methodologic parameters that are more "robust" and less sensitive to the exact character of processed calibrators, quality control, and PT materials.