Comparison of plasma total homocysteine measurements in 14 laboratories: an international study.
With the introduction of tris(2-carboxyethyl)phosphine (TCEP) as a reducing agent (6, 7), the comparability of HPLC results obtained with different reducing/ derivatizing agents has become important. Furthermore, the first fully or partially automated kits for the determination of tHcy were introduced recently: a fluorescence polarization immunoassay (FPIA) on the IM[R] analyzer [Abbott Laboratories (8)], a microtiter plate enzyme immunoassay [EIA; Bio-Rad Laboratories (9)], an HPLC kit with fluorometric detection (HPLC-FD) after trialkylphosphine reduction and 4-(aminosulfonyl)-7-fluoro-2,1,3-benzoxadiazole (ABD-F) derivatization [Bio-Rad Laboratories (10)], and an HPLC kit with electrochemical detection [HPLCED; Bioanalytical Systems (11)].
To assess the performance and the comparability of these methods, the CDC invited national and international clinicians and laboratorians actively involved in homocysteine research, as well as manufacturers of commercially available assay kits, to participate in a roundrobin interlaboratory comparison study. Each participant was asked to analyze 53 samples each on 2 separate days.
Materials and Methods
We invited laboratories to analyze two identical, blinded sets of 50 human plasma samples with normal and moderately increased homocysteine ([less than or equal to] 30 [micro]mol/L), as well as 3 CDC in-house plasma quality-control (QC) pools on 2 days. We attempted to include two or more laboratories performing each of the following methodologies: HPLC-ED, HPLC-FD [subdivided into tributylphosphine (TBP)/ammonium 7-fluoro-2,1,3-benzoxadiazole-4-sulfonate (SBD-F), TCEP/SBD-F, trialkylphosphine/ABD-F, and sodium borohydride (NaBH4)/monobromobimane (MBrB)], GC/MS, EIA, and FPIA. Of the 14 participating laboratories, 10 were in the US. The participants included three manufacturers, two government laboratories, eight academic laboratories, and one clinical research facility. Each of three laboratories participated with two different methods.
Under a CDC agreement with the Emory University Hospital Blood Collection Service (including an omnibus informed consent and Human Subject Review protocol), blood (~20 mL each) was collected from 50 apparently healthy subjects into EDTA-anticoagulated tubes (Becton-Dickinson) and cooled in ice water immediately after being drawn. The plasma was separated by centrifugation within 30 min of venipuncture and stored for a maximum of 1 month at -70[degrees]C before shipment. An aliquot of each specimen was analyzed for tHcy in the CDC NHANES laboratory by an HPLC-FD assay (7) with plasma QC pools at three concentrations (6.9, 13.4, and 29.2 [micro]mol/L). Five specimens were selected for addition of homocystine and were thawed, pooled, and redivided into five aliquots; L-homocystine was added to four of these aliquots to concentrations of 5.0, 9.9, 14.7, and 19.5 [micro]mol/L free thiol. Two random numbers were assigned for each sample. All 50 specimens were dispensed in 0.25-mL aliquots into 2.0-mL Nalge cryovials and placed promptly at -70[degrees]C until shipment. Each laboratory was sent a shipment on dry ice containing two boxes with 53 samples each. Extra aliquots were retained by CDC.
We tested for outliers by calculating the all-laboratory consensus mean [+ or -] 3 SD for each sample and comparing each individual result with this range. One sample from laboratory 9 was outside of this range and was excluded from all further calculations.
All evaluations of imprecision, method differences, and recovery, except for those for among-run variability, were based on the mean between the day 1 and day 2 results. The following measures of imprecision were evaluated: among-laboratory, among-method, withinmethod, and among-run. For each sample, we calculated the among-laboratory CV of the participating laboratories. We expressed the among-laboratory variation as the mean CV (SD) for each sample type (plasma, plasma with added homocystine, and QC pool samples). For calculation of the among-method variation, all laboratories were nested within a method group based on the method they performed. The among-method variation was expressed as the mean CV (SD) of the eight method groups over the three sample types. The within-method variation was calculated for only five of eight method groups because three groups were represented by only one laboratory each. The among-run within-laboratory and among-run within-method variability was expressed as the mean CV (SD) for each laboratory and each method, respectively. P <0.05 was considered statistically significant.
In the absence of target values for the samples analyzed, we considered the GC/MS method arbitrarily as a reference method. Difference plots were used to assess the agreement between tHcy results obtained by GC/MS and all other methods (12,13). Possible systematic biases were assessed by computing the 95% confidence intervals (mean difference [+ or -] 2 SE) for the mean differences between GC/MS and each of the methods included. Limits of agreement were assessed by calculating the central 0.95 intervals (mean difference [+ or -] 2 SD). The mean differences and the mean between GC/MS results and the results for each test method were correlated to test for a relationship between these two variables. There was no relationship for the 46 plasma samples. To assess the mean proportional bias between GC/MS and each test method, we calculated the relative ratios of the GC/MS results and test method results.
Recoveries were calculated individually for each sample with added homocystine: recovery (%) = (plasma with added homocystine--plasma without added homocystine)/added concentration of tHcy. Recovery results were reported as the mean (SD) of the four samples with added homocystine analyzed on both day 1 and day 2.
To test for methodological differences, we grouped laboratories by method and performed a two-way ANOVA with laboratory and method as variables, using the SAS GLM procedure and the Bonferroni test.
PARTICIPANTS AND METHODS
The methods used by the 14 participating laboratories are listed in Table 1. Almost all of the laboratories performing chromatographic assays use homocystine as the calibrator and calibrate at least once a day (laboratories 1, 3-5, and 7-11, once a day; laboratory 2, twice a day; laboratory 6, every 20-25 samples); most calibrate in plasma or serum. Two laboratories (laboratories 8 and 11) use defined plasma samples as the calibrator. Laboratories using immunoassays calibrate with S-adenosyl-homocysteine in buffer and calibrate daily (laboratory 3B) or biweekly (laboratories 4B, 8B, 13, and 14). Laboratory 12, which uses GC/MS, calibrates weekly and uses deuterated homocystine as the internal standard. The required sample volume varies from 30 to 200 /,L. Of the 11 laboratories performing HPLC assays, 4 (laboratories 1, 3, 10, and 11) use no internal standard, 4 (laboratories 2, 4, 5, and 6) use a disulfide (i.e., cystamine), and 3 use a thiol [i.e., cysteamine (laboratory 7), N-acetyl-cysteine (laboratory 8), and 2-mercaptopropionylglycine (laboratory 9)]. Eleven of the 14 laboratories routinely measure samples in single determinations; the other 3 laboratories routinely measure samples in duplicate.
Shown in Table 2 are the mean among-laboratory, amongmethod, and within-method variations for plasma samples (n = 46), samples with added homocystine (n = 4), and QC pools (n = 3). The mean among-laboratory variation was 7.6% for the QC pools. There was no difference in the CV as a function of the concentrations of tHcy: 7.9% for QC low (6.9 /,mol/L), 7.2% for QC medium (13.4 [micro]mol/L), and 7.6% for QC high (28.6 /,mol/L). The mean among-method variation was ~1% lower than the among-laboratory variation. Here, the CV was a function of the concentration of tHcy: 7.7% for QC low (6.9 /,mol/L), 6.5% for QC medium (13.3 /,mol/L), and 4.9% for QC high (28.3 [micro]mol/L).
The within-method variation was lowest for the FPIA assay (laboratories 4B, 8B, 13, and 14) and highest for the HPLC-FD assay using NaB[H.sub.4]/MBrB (laboratories 10 and 11). However, laboratory 10 reported results that were in general higher than the results of all other laboratories. Because of the small number of laboratories included in a method, the within-method variation data must be interpreted cautiously.
The among-run variation (Table 3) was highest for laboratory 9 (>10%) and for some laboratories (i.e., 6-8, 10, and 3B) came close to 10%. The mean among-run within-laboratory variation was 5.6% for plasma samples, 4.9% for samples with added homocystine, and 4.2% for the QC pools, respectively. Interestingly, for a ~70% of the laboratories, among-run CVs were lower for the QC pools and for the samples with added homocystine than for the plasma samples. This indicates that it is difficult to obtain the same kind of information when different procedures are used to control the analysis.
Recoveries were 85-115% for laboratories other than 10 (121.1% [+ or -] 20.9%), 9 (66.0% [+ or -] 18.8%), and 6 (75.9% [+ or -] 12.0%; Table 3). The high recovery of laboratory 10 and the low recovery of laboratory 6 indicate that these laboratories might have a calibration problem. The low and inconsistent recovery of laboratory 9 paired with the highest among-run variability among all laboratories indicates that this laboratory might have a performance problem.
DIFFERENCES AMONG METHODS AND LABORATORIES
Among-method and among-laboratory differences were assessed with plasma samples (without added homocystine) to avoid unbalanced results by a few samples with high concentrations. To visualize among-method inaccuracy, Fig. 1 shows scatter plots of observed measurement differences between the averaged results of each method type and GC /MS against the mean of GC /MS and the test method (used by one or more laboratories). The mean (SD) differences between GC/MS results (laboratory 12) and each laboratory's results (among-laboratory differences) as well as between GC/MS and the averaged results of each method type (among-method differences) are reported in Table 4. Using the standard errors of the mean differences, we computed the 95% confidence intervals; these showed an apparent positive bias for HPLC-ED, HPLC-FD using NaB[H.sub.4]/MBrB (however, only for laboratory 10), and EIA, and an apparent negative bias for HPLC-FD using trialkylphosphine/ABD-F. The two HPLC-FD methods using TCEP/SBD-F or TBP/SBD-F and the FPIA method showed no apparent bias with respect to the GC/MS method; however, TBP/SBD-F produced a relatively wider scatter of difference data points.
[FIGURE 1 OMITTED]
The central 0.95 interval (mean difference [+ or -] 2 SD) gives an indication of the agreement between GC /MS and the other methods to measure plasma tHcy. Ninety-five percent of tHcy determinations by HPLC-FD using trialkylphosphine/ABD-F were 2.05 to 0.33 [micro]mol/L lower than concentrations determined by GC/MS. This corresponds to a proportional bias of -16.1%. Ninety-five percent of tHcy determinations by HPLC-ED and by EIA were 0.20 and 0.76 /,mol/L lower to 1.26 and 1.79 [micro]mol/L higher than concentrations determined by GC/MS. This corresponds to a proportional bias of 7.5% and 7.4%, respectively.
When we grouped laboratories by method type and performed a two-way ANOVA with laboratory and method as variables to compare methods individually, we found no significant differences among the methods except that HPLC-FD trialkylphosphine/ABD-F (laboratory 6) gave significantly lower results than HPLC-FD NaBH4/ MBrB (laboratories 10 and 11) and HPLC-ED (laboratories 1-3). However, the results of laboratories 1 and 11 were not significantly different from the results of laboratory 6 or any other laboratory. Only laboratories 2, 3, and 10 reported significantly higher results than laboratory 6. Because HPLC-FD trialkylphosphine/ABD-F was used by only one laboratory, we could not conclude whether their significantly lower results were method-specific or part of the among-laboratory variation, as was seen with the method HPLC-FD NaB[H.sub.4]/MBrB.
This interlaboratory comparison study for plasma tHcy is unique in that it brought together 14 highly esteemed national and international laboratories performing the most frequently used methods to assess plasma tHcy and also included the new commercially available assays. The study design, asking for the analysis of 53 plasma samples on 2 days, allowed us to make direct method comparisons in addition to among-laboratory and among-run comparisons. Furthermore, by including plasma samples with added synthetic L-homocystine, we were able to assess recoveries. However, despite this elaborate design, methodological differences can only be assessed with statistical significance if more than one laboratory participates with each method. In addition, within- and among-method variability and method-specific bias comments and conclusions in this study are to a certain extent speculative because of the small number of laboratories using each method.
The mean among-laboratory CV of 9.2% for the 46 plasma samples (without added homocystine) is in good agreement with the 9% among-laboratory CV obtained by Moller et al. (5) for one EDTA plasma sample analyzed by nine laboratories with GC/MS and HPLC methods. In the present study, in which each laboratory and method used its own calibrators, the results of the two laboratories performing HPLC-FD with NaBH4/MBrB demonstrated that laboratory-to-laboratory variations within one method can exceed among-method variations. Although one laboratory showed virtually no apparent bias relative to GC/MS, the other laboratory reported results that were on average 18% higher than the GC/MS results.
When compared with GC/MS, HPLC-FD using TCEP/ SBD-F and FPIA showed virtually no apparent bias and narrow limits of agreement. HPLC-FD using trialkylphosphine/ABD-F and EIA showed the highest disagreement with GC/MS.
Objective analysis of whether the imprecision and bias of a method are satisfactory is difficult to perform. Some have proposed using the biological variation as the basis for analytical quality specifications (14). For tHcy, the [CV.sub.within-subject] and the [CV.sub.between-subject] were 7% and 33.5%, respectively (15). A widely held view is that analytical imprecision (CVA) should be <0.25 [CV.sub.within-subject] for optimum performance, <0.50 [CV.sub.within-subject] for desirable performance, and <0.75 [CV.sub.within-subject] for minimum performance (16). In our analysis, this would require an analytical imprecision of <1.75%, <3.5%, and <5.3% for optimum, desirable, and minimum performance, respectively. As shown in Table 3, none of the laboratories obtained among-run variations <1.75% for all three types of samples, and only two laboratories (11 and 13) had overall among-run variations <3.5%. The among-run variations of six laboratories (6-10 and 3B) exceeded the required CV for minimum performance of 5.3% for at least two types of samples. Three methods performed best regarding analytical imprecision: GC/MS, FPIA, and HPLC-FD using TCEP/SBD-F. They reached the required CV for minimum performance for at least two types of samples.
For optimum performance, the bias of a method (BA) should be <0.125([[CV.sub.within-subject.sup.2] + [[CV.sub.between-subject.sup.2]).sup.1/2] for desirable performance, it should be <0.25([CV.sub.within-subject.sup.2] + [[CV.sub.between-subject.sup.2]).sup.1/2] and for minimum performance, it should be <0.375([CV.sub.within-subject.sup.2] + [[CV.sub.between-subject.sup.2]).sup.1/2] (16). For our analysis, this would mean a bias of <4.3%, <8.6%, and <12.8% for optimum, desirable, and minimum performance, respectively. As shown in Table 4, several laboratories met the requirement for optimum performance (4,5,8, 9,11,13,14,4B, and 8B) compared with GC/MS. However, two laboratories (6 and 10) did not meet the requirement for minimum performance. The following three methods performed best regarding apparent analytical bias (vs GC/MS): FPIA and HPLC-FD using TCEP/SBD-F or TBP/SBD-F. It must be cautioned, however, that the GC/MS method may itself be biased and that until high-level reference methods for tHcy are available, little can be said about a laboratory's or a method's bias.
Our analysis for methodological differences in mean concentrations showed significant differences for two method comparisons: HPLC-FD trialkylphosphine/ ABD-F compared with HPLC-FD NaB[H.sub.4]/MBrB and compared with HPLC-ED. Although we could not confirm that these differences were method specific, Dias et al. (17) reported a significant negative bias of 5.2 [micro]mol/L when they compared the new Bio-Rad HPLC method with a SBD-F assay.
Gilfix et al. (6) introduced TCEP as a novel and more suitable reductant for the routine determination of plasma tHcy. In a small method comparison study, they found that the TCEP method yielded values ~21% higher than the TBP method. We previously directly compared results obtained with these two reducing agents and found that although the use of TCEP produced higher relative fluorescence intensities, the calculated concentrations of plasma tHcy were not significantly different if calibration was performed in plasma and cystamine was used as the internal standard (7). The 20% difference seen by Gilfix et al. (6) could have been a result of the different calibration matrices that were used in that comparison (saline for TCEP vs plasma for TBP). We showed that if calibration was performed in saline and no internal standard (which could correct for the matrix effect) was added, tHcy concentrations were ~20% higher than when calibration was performed in plasma (7). The present round-robin showed very good agreement between results obtained with TCEP as reductant and those obtained by most other methods.
The performance of the Abbott homocysteine assay is of great interest to many clinical laboratories because this FPIA is fully automated and requires no sample pretreatment step. In a large method comparison study (n = 811 plasma and serum samples), we previously found a very good correlation between the FPIA assay and our HPLC assay with fluorometric detection (y = -0.110 + 0.992x; [r.sup.2] = 0.986) (18). In the present interlaboratory comparison study, all four participating laboratories (13, 14, 4B, and 8B) that used the Abbott homocysteine assay produced results that were in good agreement with each other and with results obtained by most other methods.
Although we found no correlation between the performance of the different methods and the use of an internal standard or between the use of plasma calibration vs aqueous calibration, we did find that laboratories whose results did not agree well with the GC/MS results also usually showed higher among-run imprecision and lower and more variable recoveries.
This international round-robin for plasma tHcy showed good agreement between some of the most experienced laboratories performing different methods. It answered important questions concerning the performance of TCEP as a novel reductant and the Abbott homocysteine assay. Although the results indicate that some of the methods tested could be used interchangeably, the analysis for analytical quality specifications have shown that overall there is an urgent need to improve analytical imprecision and, for some methods, an apparent need to decrease analytical bias to guarantee that laboratories throughout a homogeneous population area can use the same reference intervals. We believe that improvement can be aided by the introduction of standard reference materials and more external quality assessment programs.
This work was supported by the National Center for Environmental Health, CDC. We thank the following laboratorians for their participation and support: Richard Edwards, Bio-Rad Laboratories, Benicia, CA; Chester Duda and Bruce Solomon, Bioanalytical Systems, West Lafayette, IN; Dan Huff and Patricia Yeager, CDC NHANES Laboratory, Atlanta, GA; Donald Jacobsen and Patricia Di Bello, Cleveland Clinic Foundation, Cleveland, OH; Rene Malinow, Barbara Upson, and Erik Graf, Oregon Regional Primate Research Center, Beaverton, OR; Joseph McPartlin and Paudy O'Gorman, St. James Hospital, Dublin, Ireland; Joshua Miller and Ralph Green, University of California Davis, Sacramento, CA; Jan Moller and Karsten Rasmussen, Aarhus University, Aarhus, Denmark; Patricia Mueller and Shuenae Smith, CDC/NCEH/DLS/Molecular Biology, Atlanta, GA; Keith Mullen and Don Carter, Vascular Disease Intervention & Research Laboratory, Edmond, OK; Sean O'Broin, St. James Hospital, Dublin, Ireland; Helga Refsum, University of Bergen, Bergen, Norway; Jessie Shih and Shelley Holets-McCormack, Abbott Laboratories, Abbott Park, IL; and Michael Tsai, Fairview-University Medical Center, Minneapolis, MN. We also thank the staff of the NHANES Laboratory, CDC, for their contributions in preparing and shipping these materials to the participating laboratories.
Received January 13, 1999; accepted May 17, 1999.
(1.) Kang SS, Wong PWK, Malinow MR. Hyperhomocyst(e)inemia as a risk factor for occlusive vascular disease. Annu Rev Nutr 1992; 12:279-98.
(2.) Ueland PM, Refsum H, Brattstrom L. Plasma homocysteine and cardiovascular disease. In: Francis RB Jr, ed. Atherosclerotic cardiovascular disease, hemostasis, and endothelial function. New York: Marcel Dekker, 1992:183-236.
(3.) Malinow MR. Homocyst(e)ine and arterial occlusive diseases. J Intern Med 1994;236:603-17.
(4.) Boushey CJ, Beresford SAA, Omenn GS, Motulsky AG. A quantitative assessment of plasma homocysteine as a risk factor for vascular disease. JAMA 1995;274:1049-57.
(5.) MollerJ, Christensen L, Rasmussen K. An external quality assessment study on the analysis of methylmalonic acid and total homocysteine in plasma. Scand J Clin Lab Investig 1997;57: 613-9.
(6.) Gilfix BM, Blank DW, Rosenblatt DS. Novel reductant for determination of total plasma homocysteine. Clin Chem 1997;43:687-8.
(7.) Pfeiffer CM, Huff DL, Gunter EW. Rapid and accurate HPLC assay for total plasma homocysteine and cysteine in a clinical laboratory setting. Clin Chem 1999;45:290-2.
(8.) Shipchandler MT, Moore EG. Rapid, fully automated measurement of plasma homocyst(e)ine with the Abbott IMx[R] analyzer. Clin Chem 1995;41:991-4.
(9.) Frantzen F, Faaren AL, Alfheim I, Nordhei AK. Enzyme conversion immunoassay for determining total homocysteine in plasma or serum. Clin Chem 1998;44:311-6.
(10.) Ueland PM, Refsum H, Stabler SP, Malinow MR, Andersson A, Allen RH. Total homocysteine in plasma or serum. Methods and clinical applications. Clin Chem 1993;39:1764-79.
(11.) Solomon BP, Duda CT. Homocysteine determination in plasma. Curr Sep 1998;17:3-7.
(12.) Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307-10.
(13.) Bland JM, Altman DG. Comparing methods of measurement: why plotting difference against standard method is misleading. Lancet 1995;346:1085-7.
(14.) Fraser CG, Petersen PH. Desirable standards for laboratory tests if they are to fulfill medical needs. Clin Chem 1993;39:1447-55.
(15.) Garg UC, Zheng ZJ, Folsom AR, Moyer YS, Tsai MY, McGovern P, Eckfeldt JH. Short-term and long-term variability of plasma homocysteine measurement. Clin Chem 1997;43:141-5.
(16.) Fraser CG, Petersen PH. Analytical performance characteristics should be judged against objective quality specifications [Editorial]. Clin Chem 1999;45:321-3.
(17.) Dias VC, Bamforth FJ, Forward L, Tesanovic M, Hyndman ME, Parsons HG, Cembrowski GS. Evaluation of the new Bio-Rad high performance liquid chromatographic method for plasma total homocysteine [Abstract]. Clin Chem 1998;44(Suppl):A172-3.
(18.) Pfeiffer CM, Twite D, Shih J, Holets-McCormack S, Gunter EW. Method comparison for total plasma homocysteine between the Abbott IMx analyzer and an HPLC assay with internal standardization. Clin Chem 1999;45:152-3.
(19.) Malinow MR, Sexton G, Averbuch M, Grossman M, Wilson D, Upson B. Homocyst(e)inemia in daily practice: levels in coronary artery disease. Coron Artery Dis 1990;1:215-20.
(20.) Ubbink JB, Vermaak WJH, Bissbort S. Rapid high-performance liquid chromatographic assay for total homocysteine levels in human serum. J Chromatogr 1991;565:441-6.
(21.) Vester B, Rasmussen K. High performance liquid chromatography method for rapid and accurate determination of homocysteine in plasma and serum. Eur J Clin Chem Clin Biochem 1991;29: 549-54.
(22.) Jacobsen DW, Gatautis VJ, Green R, Robinson K, Savon SR, Secic M, et al. Rapid HPLC determination of total homocysteine and other thiols in serum and plasma: sex differences and correlation with cobalamin and folate concentrations in healthy subjects. Clin Chem 1994;40:873-81.
(23.) Fiskerstrand T, Refsum H, Kvalheim G, Ueland PM. Homocysteine and other thiols in plasma and urine: automated determination and sample stability. Clin Chem 1993;39:263-71.
(24.) Stabler SP, Lindenbaum J, Savage DG, Allen RH. Elevation of serum cystathionine levels in patients with cobalamin and folate deficiency. Blood 1993;81:3404-13.
 Nonstandard abbreviations: tHcy, total homocysteine; GC/MS, gas chromatography/mass spectrometry; TCEP, tris(2-carbo)cyethyl)phosphine; FPIA, fluorescence polarization immunoassay; EIA, enzyme immunoassay; HPLCFD, HPLC with fluorometric detection; ABD-F, 4-(aminosulfonyl)-7-fluoro2,1,3-benzoxadiazole; HPLC-ED, HPLC with electrochemical detection; QC, quality control; TBP, tributylphosphine; SBD-F, ammonium 7-fluoro-2,1,3-benzoxadiazole-4-sulfonate; NaB[H.sub.4], sodium borohydride; and MBrB, monobromobimane.
CHRISTINE M. PFEIFFER, * DAN L. HUFF, S. JAY SMITH, DAYTON T. MILLER, and ELAINE W. GUNTER
National Center for Environmental Health, Centers for Disease Control and Prevention, Atlanta, GA 30341.
* Address correspondence to this author at: Centers for Disease Control and Prevention, 4770 Buford Hwy., NE, MS F-18, Atlanta, GA 30341-3724. Fax 770-488-4609; e-mail firstname.lastname@example.org.
Table 1. Participating laboratories and methods. Lab Method Red uction/precipitation Calibrator /derivatization agents 1 HPLC-ED NaBH (4)/TCA (a)/-(b) L-Homocystine 2 HPLC-ED Proprietary/Proprietary/- Homocystine 3 HPLC-ED Proprietary/Proprietary/- Homocystine 4 HPLC-FD TCEP/TCA/SBD-F L-Homocystine 5 HPLC-FD TCEP/PCA/SBD-F Homocystine 6 HPLC-FD TAP/TCA/ABD-F Homocystine 7 HPLC-FD TBP/TCA/SBD-F L-Homocystine 8 HPLC-FD TBP/PCA/SBD-F Plasma 9 HPLC-FD TBP/PCA/SBD-F L-Homocystine 10 HPLC-FD NaBH(4)/PCA/MBrB Homocystine 11 HPLC-FD NaBH(4)/-/MBrB Plasma 12 GC/MS DTT/AX/TFAA DL-Homocystine 3B (d) EIA DTT/-/- SAH 13 FPIA DTT/-/- SAH 4B (d) FPIA DTT/-/- SAH 8B (d) FPIA DTT/-/- SAH 14 FPIA DTT/-/- SAH Sample Lab Calibration in volume, mL Ref. 1 Plasma 0.2 (c) 19 2 Plasma or water 0.20 11 3 Water 0.20 11 4 Plasma 0.05 7 5 Water 0.10 6 6 Serum 0.25 10 7 Water 0.15 20 8 Plasma 0.05 (c) 20 9 Plasma 0.1 (c) 21 10 Plasma 0.10 22 11 Plasma or serum 0.03 23 12 Water 0.20 24 3B (d) Buffer 0.05 9 13 Buffer 0.05 8 4B (d) Buffer 0.05 8 8B (d) Buffer 0.05 8 14 Buffer 0.05 8 (a) TCA, trichloroacetic acid; PCA, perchloric acid; TAP, trial kylphosphine; DTT, dithiothreitol; AX, Bio-Rad AG-MP-1 AX resin (sample clean up); TFAA, N-methyl-N-(t butyldimethylsilyl)-trifluoroacetamide. (b) No agent used. (c) Samples are prepared routinely in duplicate. (d) B indicates second method used by this laboratory. Table 2. Imprecision: among-laboratory, among-method, and within-method variation. n Plasma Plasma + Hcy QC Pools Among-laboratory 17 9.2 (1.5) 8.8 (2.7) 7.6 (0.4) CV, mean (SD), % Among-method 8 8.7 (1.7) 6.9 (2.3) 6.4 (1.4) CV, mean (SD), % Within-method CV, mean (SD), HPLC-ED 3 4.8 (3.8) 5.8 (2.2) 7.7 (3.7) HPLC-FD TCEP/SBD-F 2 4.3 (2.7) 3.0 (2.3) 5.6 (2.1) HPLC-FD TBP/SBD-F 3 9.8 (5.0) 15 (7.6) 6.7 (3.9) HPLC-FD NaBH4/MBrB 2 13 (4.3) 9.5 (6.0) 13 (2.7) FPIA 4 4.9 (2.4) 3.2 (1.2) 3.2 (1.5) Table 3. Among-run variation for each laboratory and method, and recoveries. Among-run CV, mean (SD), % Method/Lab Plasma Plasma + Hcy QC pools HPLC-ED (a) 6.2 (3.7) 6.7 (2.3) 7.4 (2.7) Lab 1 (b) 5.3 (4.3) 6.5 (2.3) 2.4 (1.7) Lab 2 (b) 3.4 (3.3) 4.4 (3.9) 2.5 (3.0) Lab 3 (b) 4.8 (5.5) 3.7 (3.1) 1.3 (1.2) HPLC-FD TCEP/SBD-F (a) 5.9 (2.7) 4.3 (3.1) 5.2 (1.3) Lab 4 (b) 5.2 (3.0) 3.9 (1.8) 3.2 (1.0) Lab 5 (b) 4.8 (4.0) 3.7 (5.5) 2.5 (0.4) HPLC-FD using trialkylphosphine/ABD-F Lab 6 (b) 6.7 (5.1) 5.7 (6.1) 9.3 (6.0) HPLC-FD TBP/SBD-F (a) 12 (4.8) 16 (6.2) 8.5 (2.0) Lab 7 (b) 7.0 (4.5) 9.9 (6.3) 6.6 (1.0) Lab 8 (b) 5.6 (4.4) 8.2 (5.4) 5.1 (4.3) Lab 9 (b) 14 (12) 13 (11) 6.9 (5.5) HPLC-FD NaBH4/MBrB (a) 12 (4.1) 8.3 (4.8) 13 (0.6) Lab 10 (b) 6.0 (5.3) 3.3 (1.2) 8.7 (4.8) Lab 11 (b) 2.5 (2.9) 2.6 (2.4) 2.5 (2.2) GC/MS Lab 12 (b) 3.8 (3.5) 1.6 (1.2) 3.7 (0.9) EIA Lab 3B (b,c) 9.3 (7.4) 9.3 (7.6) 2.0 (1.6) FPIA 6.0 (2.7) 3.5 (1.3) 4.7 (1.5) Lab 13 (b) 2.7 (2.3) 0.6 (0.2) 0.8 (0.5) Lab 4B (b,c) 4.8 (5.7) 2.9 (3.0) 3.5 (1.4) Lab 8B (b,c) 3.5 (3.2) 1.7 (0.9) 3.7 (3.4) Lab 14 (b) 4.9 (4.8) 2.1 (2.0) 7.3 (3.0) Recovery, Method/Lab mean (SD), % HPLC-ED (a) 104.0 (14.1) Lab 1 (b) 99.0 (9.6) Lab 2 (b) 113.2 (16.6) Lab 3 (b) 99.7 (11.8) HPLC-FD TCEP/SBD-F (a) 95.5 (8.0) Lab 4 (b) 94.6 (3.0) Lab 5 (b) 96.5 (11.3) HPLC-FD using trialkylphosphine/ABD-F Lab 6 (b) 75.9 (12.0) HPLC-FD TBP/SBD-F (a) 86.4 (21.7) Lab 7 (b) 106.1 (13.0) Lab 8 (b) 87.2 (10.0) Lab 9 (b) 66.0 (18.8) HPLC-FD NaBH4/MBrB (a) 108.7 (19.7) Lab 10 (b) 121.1 (20.9) Lab 11 (b) 96.3 (6.4) GC/MS Lab 12 (b) 94.0 (3.6) EIA Lab 3B (b,c) 99.8 (17.5) FPIA 91.6 (6.6) Lab 13 (b) 94.3 (3.9) Lab 4B (b,c) 86.5 (4.4) Lab 8B (b,c) 93.4 (6.3) Lab 14 (b) 92.1 (8.6) (a) Among-run among-laboratory (within one method group). (b) Among-run within-laboratory variation. (c) B indicates second method used by this laboratory. Table 4. Differences of methods relative to GC/MS. Mean (SD) (a) difference Method Lab/mean from GC/MS HPLC-ED Lab 1 0.33 (0.65) Lab 2 0.65 (0.80) Lab 3 0.59 (0.32) Mean 0.53 (0.36) HPLC-FD TCEP/SBD-F Lab 4 -0.13 (0.27) Lab 5 0.28 (0.37) Mean 0.08 (0.26) HPLC-FD Lab 6 -1.19 (0.43) trialkylphosphine/ABD-F HPLC-FD TBP/SBD-F Lab 7 0.50 (0.34) Lab 8 -0.11 (0.39) Lab 9 -0.25 (1.08) Mean 0.05 (0.39) HPLC-FD NaBH4/MBrB Lab 10 1.36 (0.55) Lab 11 -0.13 (0.22) Mean 0.62 (0.31) EIA Lab 3B (c) 0.52 (0.64) FPIA Lab 13 0.01 (0.32) Lab 4B (c) -0.27 (0.37) Lab 8B (c) -0.10 (0.29) Lab 14 0.21 (0.44) Mean -0.04 (0.23) 95% Limit of 95% Cl (b) of Method agreements the means (a) Lab/mean HPLC-ED -0.97 to 1.64 0.15 to 0.53 Lab 1 -0.94 to 2.25 0.42 to 0.89 Lab 2 -0.05 to 1.24 0.50 to 0.69 Lab 3 -0.20 to 1.26 0.42 to 0.64 HPLC-FD TCEP/SBD-F Mean -0.67 to 0.42 -0.21 to -0.02 Lab 4 -0.46 to 1.01 0.17 to 0.39 Lab 5 -0.44 to 0.59 -0.01 to 0.15 HPLC-FD Mean -2.05 to -0.33 -1.32 to -1.06 trialkylphosphine/ABD-F Lab 6 HPLC-FD TBP/SBD-F -0.18 to 1.17 0.40 to 0.60 Lab 7 -0.90 to 0.67 -0.23 to 0.01 Lab 8 -2.41 to 1.92 -0.57 to 0.08 Lab 9 -0.72 to 0.82 -0.07 to 0.16 HPLC-FD NaBH4/MBrB Mean 0.27 to 2.45 1.20 to 1.52 Lab 10 -0.57 to 0.31 -0.20 to -0.07 Lab 11 -0.01 to 1.24 0.52 to 0.71 EIA Mean -0.76 to 1.79 0.33 to 0.71 FPIA Lab 3B (c) -0.63 to 0.65 -0.09 to 0.10 Lab 13 -1.01 to 0.46 -0.38 to -0.16 Lab 4B (c) -0.69 to 0.48 -0.19 to -0.02 Lab 8B (c) -0.67 to 1.09 0.08 to 0.34 Lab 14 -0.49 to 0.41 -0.11 to 0.03 Proportional Method Lab/mean difference, % HPLC-ED Lab 1 5.2 Lab 2 9.0 Lab 3 8.3 Mean 7.5 HPLC-FD TCEP/SBD-F Lab 4 -1.6 Lab 5 3.6 Mean 1.0 HPLC-FD Lab 6 -16.1 trialkylphosphine/ABD-F HPLC-FD TBP/SBD-F Lab 7 6.5 Lab 8 -1.6 Lab 9 -2.6 Mean 0.8 HPLC-FD NaBH4/MBrB Lab 10 18.3 Lab 11 -2.0 Mean 8.1 EIA Lab 3B (c) 7.4 FPIA Lab 13 0.1 Lab 4B (c) -3.6 Lab 8B (c) -1.7 Lab 14 3.2 Mean -0.5 (a) tHcy (I mol/L). (b) CI, confidence interval. (c) B indicates second method used by this laboratory.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||General Clinical Chemistry|
|Author:||Pfeiffer, Christine M.; Huff, Dan L.; Smith, S. Jay; Miller, Dayton T.; Gunter, Elaine W.|
|Date:||Aug 1, 1999|
|Previous Article:||Determination of the sum of bilirubin sugar conjugates in plasma by bilirubin oxidase.|
|Next Article:||Validation of accuracy-based amino acid reference materials in dried-blood spots by tandem mass spectrometry for newborn screening assays.|