Multicenter study of homocysteine measurement-performance characteristics of different methods, influence of standards on interlaboratory agreement of results

After the demonstration that moderate hyperhomocysteinemia is associated wi th thrombosis, many hematological labs are becoming interested in total hom ocysteine (tHcy) measurement. This prompted us to organize a collaborative study to investigate the performance of methods used in this setting and to assess the between-lab comparability of results. Two pairs of pooled plasm a (A(1)-A(2) and B-1-B-2) were prepared at the coordinating Center. tHcy le vels were normal in A(1)-A(2) and moderately high in B-1-B-2. Within each p air tHcy levels were similar but not identical. Aliquots were taken from ea ch pool to prepare sets of 100 samples (coded from 1 to 100). Each set cons isted of 25 replicates for each pool. Samples were frozen and shipped in dr y ice to 16 labs with a common frozen aqueous standard. Labs were asked to measure (in blind) tHcy with their methods and standards. Results were sent to the coordinating Center both as raw readings and as tHcy levels. The fo llowing methods were used: High Pressure Liquid Chromatography (HPLC) in 12 labs (home-made in 10 and commercial in 2); Enzyme Immune Assays (EIA) in 2; Fluorescence Polarization Immunoassay (FPIA) in 2 and Capillary Electrop horesis ICE) in one. Results for paired pools (A(1)-A(2) and B-1-B-2) were analyzed by the Student t test to assess for the ability to discriminate be tween similar but not identical tHcy levels. Results for each pool were use d to assess within-lab reproducibility and between-lab comparability. Withi n-lab reproducibility expressed as median CV ranged from 12.6 to 13.9% (hom e-made HPLC); from 9.2 to 11.4% (commercial HPLC); from 21.8 to 24.2% (EIA) , from 2.7 to 3.3% (FPIA) and from 11.2 to 22.0% (CE). AII labs, except one using CE and 2 using home-made HPLC, were able to discriminate between sim ilar tHcy levels in the normal range (pools A(1)-A(2)). Ten labs (4 using h ome-made HPLC, 2 commercial HPLC, 2 FPIA, one EIA and one CEI were able to discriminate between similar moderately high tHcy levels (pools B-1-B-2). B etween-lab comparability expressed as CV was 14.0% 13.9%, 15.6% acid 14.5% for pools A,. A(2), B-1, and B-2. These values were considerably lower (CV values < 5.2%) when a common plasma standard was used for calculation of tH cy levels, while the use of a common aqueous standard failed to achieve the necessary harmonization. In conclusion, performance characteristics of the FPIA method compare favorably with the well-established HPLC methods. It i s simpler and more suitable to be used by general hematological labs. Betwe en-lab comparability of results is still a problem. The establishment of an international plasma standard would be of help to harmonize tHcy measureme nt across laboratories.