Interlaboratory variation of plasma total homocysteine measurements: results of three successive homocysteine proficiency testing surveys.

BACKGROUND Numerous studies have demonstrated that increased plasma total homocysteine (tHcy), whether measured after fasting or after a methionine load, is associated with increased risk for cardiovascular and thromboembolic diseases. However, little information is available regarding interlaboratory variation of tHcy measurements, especially at the increased tHcy concentrations observed after loading. METHODS We conducted three Homocysteine Proficiency Testing Surveys at 6-month intervals. Sets of five plasma pools with endogenous tHcy concentrations ranging from 5 to 48 micromol/L were sent to participants. We received 11, 23, and 17 responses in the first, second, and third surveys, respectively. The following methods were used by participating laboratories: fluorescence polarization immunoassay (FPIA); HPLC with fluorescent detection (HPLC-FD), further subdivided by type of reduction/derivatization; HPLC with electrochemical detection (HPLC-ED); amino acid analyzer with ninhydrin detection; and liquid chromatography-electrospray tandem mass spectrometry (LC-MS/MS). RESULTS In surveys 1 and 2, no notable differences among the mean tHcy values obtained by the different methods performed were observed. In survey 3, tHcy values obtained by the FPIA method were significantly lower (P <0.05) at increased tHcy concentrations (34 micromol/L) compared with values obtained by HPLC-FD regardless of reduction/derivatization agents used. Our laboratory confirmed the observation that tHcy values obtained by the FPIA method differed from those obtained by HPLC-FD at increased tHcy concentrations by reanalyzing each pool 10 times by FPIA and HPLC-FD using tributylphosphine-ammonium 7-fluoro-2,1,3-benzoxadiazole-4-sulfonate (P <0.001 for tHcy >19 micro mol/L). The mean among-method variations in surveys 1, 2, and 3 were 19%, 12%, and 9.6%, respectively. When results of the three surveys were combined, the mean among-method variation on 170 samples was 13%. Within-method variation was lowest for the FPIA method (4.4%), and ranged from 11-20% for HPLC methods. CONCLUSIONS Various degrees of imprecision and lack of correlation among tHcy methods indicate that there is a need to improve analytical precision, decrease analytical difference, and standardize tHcy measurements among laboratories.