Comparison of plasma total homocysteine measurements in 14 laboratories: an international study.

BACKGROUND Information on interlaboratory variation and especially on methodological differences for plasma total homocysteine is lacking. METHODS We studied 14 laboratories that used eight different method types: HPLC with electrochemical detection (HPLC-ED); HPLC with fluorescence detection (HPLC-FD) further subdivided by type of reducing/derivatizing agent; gas chromatography/mass spectrometry (GC/MS); enzyme immunoassay (EIA); and fluorescence polarization immunoassay (FPIA). Three of these laboratories used two methods. The laboratories participated in a 2-day analysis of 46 plasma samples, 4 additional plasma samples with added homocystine, and 3 plasma quality-control (QC) pools. Results were analyzed for imprecision, recovery, and methodological differences. RESULTS The mean among-laboratory and among-run within-laboratory imprecision (CV) was 9.3% and 5.6% for plasma samples, 8.8% and 4.9% for samples with added homocystine, and 7.6% and 4.2% for the QC pools, respectively. Difference plots showed values systematically higher than GC/MS for HPLC-ED, HPLC-FD using sodium borohydride/monobromobimane (however, for only one laboratory), and EIA, and lower values for HPLC-FD using trialkylphosphine/4-(aminosulfonyl)-7-fluoro-2,1,3-benzoxadiazole. The two HPLC-FD methods using tris(2-carboxyethyl) phosphine/ammonium 7-fluoro-2,1,3-benzoxadiazole-4-sulfonate (SBD-F) or tributyl phosphine/SBD-F, and the FPIA method showed no detectable systematic difference from GC/MS. CONCLUSIONS Among-laboratory variations within one method can exceed among-method variations. Some of the methods tested could be used interchangeably, but there is an urgent need to improve analytical imprecision and to decrease differences among methods.