Original Article Ann Clin Biochem 2000; 37: 194±198 Evaluation of the Abbott IMx TM uorescence polarization immunoassay and the Bio-Rad enzyme immunoassay for homocysteine: comparison with high-performance liquid chromatography James G Donnelly 1,2 and Claire Pronovost 1 From the 1 Department of Laboratory Medicine, Ottawa Hospital ± Civic Campus, 1053 Carling Avenue, Ottawa, Ontario, Canada K1Y 4E9, and the 2 Faculties of Biochemistry, Microbiology and Immunology, Pathology and Laboratory Medicine, University of Ottawa School of Medicine, Ottawa, Ontario, Canada SUMMARY. We evaluated the precision, linearity and accuracy of the Abbott IMx TM and Bio-Rad (Axis) homocysteine assays. Both assays make use of S-adenosylhomocysteine hydrolase and excess adenosine, to convert homocysteine to S-adenosylhomocysteine (SAH). A monoclonal anti-sah antibody is then used to quantify SAH. The IMx assay measures the uorescence polarization of a conjugated SAH analogue for the nal analytical step, whereas the Bio-Rad method uses a microplate enzyme immunoassay (EIA) employing an anti-mouse antibody± peroxidase conjugate. The Abbott procedure is completely automated whereas the Bio-Rad EIA is performed manually. Between-run coef cient of variation using commercial controls was 2 6% at 7 mol/l, 2 5% at 13 mol/l and 1 7% at 24 mol/l for the Abbott method, and 19 7% at 6 4 mol/l, 15 9% at 11 0 mol/l and 14 5% at 23 4 mol/l for the Bio-Rad method. Both assays correlated well with a highperformance liquid chromatography (HPLC) procedure for homocysteine: Bio-Rad EIA=1 03HPLC+1 0 mol/l, r=0 98, s y/x =0 51; Abbott IMx= 1 02 HPLC+0 7 mol/l, r=0 99, s y/x =0 33. Both methods were linear up to 50 mol/l homocysteine. The IMx assay had superior precision as well as the technological advantage of being completely automated. Both immunoassays exhibited greatly improved throughput compared with our existing HPLC method. Homocysteine was rst associated with thrombosis in the early 1960s, when it was isolated from the urine of paediatric patients. 1,2 More recently, moderately elevated plasma homocysteine concentrations have been shown to be a signi cant risk factor for coronary artery disease and venous thrombosis. 3 Subsequent to these reports, there has been increased demand for plasma homocysteine measurements, but few laboratories have been able to offer this assay routinely due to the need for high-performance liquid chromatography (HPLC) equipment. Correspondence: Dr James G Donnelly. E-mail: jdonnelly@civich.ottawa.on.ca Many HPLC assays have been developed for homocysteine using electrochemical detection or uorescence derivatization. 4±6 Bio-Rad (Hercules CA, USA) have developed a commercial HPLC assay for homocysteine with uorescence derivatization. Axis (Oslo, Norway) recently introduced a novel method for the measurement of homocysteine. Homocysteine is enzymatically converted to S-adenosylhomocysteine (SAH) by S-adenosylhomocysteine hydrolase in the presence of excess adenosine. A mouse monoclonal anti-sah antibody then recognizes SAH. Abbott Laboratories (Abbott Park IL, USA) have adapted this system to the IMx analyser, which uses uorescence polarization immunoassay (FPIA). 7 The assay has also been adapted to 194
Immunoassays for homocysteine 195 a microplate enzyme immunoassay (EIA) format by Axis and is marketed and distributed through Bio-Rad Laboratories. The initial steps ± reduction of homocystine to homocysteine, enzymatic conversion of the reduced species to SAH and then binding to the anti- SAH antibody ± are similar in both systems, but the assays have different detection procedures. The Bio-Rad assay is currently available as a manual EIA with multiple incubation and pipetting steps and peroxidase-based detection, whereas the IMx assay is a completely automated FPIA. We evaluated both methods for precision, linearity and correlation with our existing HPLC method. Throughput and ease of use were considered, and the detection limit of the FPIA was determined. METHODS All procedures were performed in accordance with the guidelines of the Institutional Review Board of the Ottawa Hospital. Abbott IMx TM homocysteine assay kits, calibrators and quality control materials were a generous gift from Abbott Diagnostics Limited (Missisauga, Canada). Bio-Rad EIA homocysteine assay kits, calibrators and quality control materials were a generous gift from Bio- Rad Laboratories (Missisauga, Canada). For the HPLC assay, chlorobimane and all other chemicals were obtained from Sigma Chemical Co. (St. Louis MO, USA). The solvents used for HPLC were from Fisher Scienti c (Ottawa, Canada) and the C-18 HPLC cartridges were from Brownlee-Perkin Elmer (Norwalk CT, USA). The Abbott IMx and Bio-Rad EIA assays were performed according to the manufacturers instructions. An EIA plate reader (Diagnostics Pasteur, Anthos Labtec Instruments, Salzburg, Austria) was used to measure absorbance and a Kinecalc v.2.03 computer program (Bio-Tech Instruments, Mississauga, Canada) was used to generate a four-parameter logistic calibration curve. The HPLC procedure was modi ed from Jacobson et al. 6 and has been described elsewhere. 8 Precision study IMx homocysteine The Abbott Diagnostics quality control materials (three levels) were run on 12 different days in duplicate, both morning and afternoon, using a single IMx analyser. Bio-Rad homocysteine Bio-Rad quality control materials at three levels were analysed with 11 different calibration curves on separate days. The absorbance data from the calibration materials were also used to generate precision data for the immunoassay step, without the pretreatment procedure. Analytical sensitivity (IMx only) The zero calibrator was analysed 20 times for homocysteine on the IMx analyser on each of three separate days. The mean value +2 standard deviations (SD) was considered the lowest reportable value. Dilution recovery Specimens with homocysteine concentrations both greater than and less than 50 mol/l were measured neat and diluted 1:2, 1:4 and 1:8 with zero calibrator. Correlation study Whole-blood specimens were obtained in tubes containing EDTA from patients in our thrombosis, atherosclerosis and haemodialysis clinics. Each specimen was chilled after collection and the plasma was separated from the cells within 1 h. The plasma specimens were stored at 20 C until ready for analysis using the Abbott, Bio- Rad and HPLC assays. The specimens of 64 patients were measured for the Abbott IMx correlation study and 57 for the Bio-Rad study. RESULTS The precision and correlation data of the three assays are presented in Table 1 and Fig. 1. The mean homocysteine concentration of the Abbott zero calibrator was 0 15 mol/l, with an SD of 0 108 mol/l. Using 2 SD above the mean as the de nition of the detection limit gives a detection limit of 0 37 mol/l. DISCUSSION The commercial availability of non-hplc assays for homocysteine will allow many more laboratories to offer this test. Inter-laboratory concordance will be important in developing reference intervals and monitoring guidelines for this atherogenic and thrombotic risk factor. The use of common methods will help in the validation of monitoring guidelines that are
196 Donnelly and Pronovost TABLE 1. Analytical performance characteristics of the Abbott IMx TM FPIA, Bio-Rad EIA and HPLC methods Assay Within-run CV* (%) Day-to-day CV (%) [concentration, mol/l] Upper limit of linearity Throughput (reportable results/h) Estimated relative cost per reportable result** Abbott IMx FPIA 4 1 2 6 [7 0] 50 mol/l 36 0 43 2 5 [13 0] (r=0 99) 1 7 [24 0] Bio-Rad EIA 2 3 19 4 [6 4] 50 mol/l 21 3 0 61 15 9 [11 0] (r=0 92) 14 5 [23 0] HPLC 4 0 8 4 [8 7] 100 mol/l 1 3 1 0 6 7 [14 1] (r=0 99) 3 8 [19 4] *Coef cient of variation (CV) of duplicates. **Based on cost of HPLC assay. Estimate includes labour and is based on our quoted manufacturer s kit pricing at a level of 1000 tests per year. Based on two levels of quality control material per carousel. Assuming two EIA plates run simultaneously in one shift (7 5 h). If only one plate is used then 14 2 reportable results/h can be obtained. Based on three levels of quality control material per plate. EIA=enzyme immunoassay; FPIA= uorescence polarization immunoassay; HPLC=high-performance liquid chromatography. applicable to a wide range of laboratories and clinics. In this study, we compared two commercial assays with our current HPLC assay. All three assays provided similar results within the reference range and in the concentration range of patients with moderate hyperhomocysteinaemia. At homocysteine values (by HPLC) greater than 20 mol/l, however, the Bio-Rad EIA assay showed some discordance [Fig. 1(d)]. This probably relects the imprecision of the EIA assay, since there was no overall positive or negative bias. The Abbott assay and the HPLC assay both demonstrated superior precision compared with the Bio-Rad EIA. The Bio-Rad EIA method suffered from imprecision in our laboratory, even under optimized conditions and ensuring that the manufacturer s protocol was followed exactly. For the duration of this study, one experienced technologist performed the assay repeatedly, but we were unable to improve the precision of the quality control materials. We attempted to isolate the problem by running a series of calibrators as specimens over several days in duplicate. An improvement in precision under these circumstances would indicate that the imprecision originated in either the reduction or enzymatic conversion step, prior to the immunoassay steps. However, the coef cient of variation (CV) remained greater than 12% for all calibrators, suggesting that the imprecision encountered was inherent to the immunoassay steps. Increasing the number of wash steps might improve the precision, but we did not deviate from the recommended protocol for this evaluation. Sensitivity of the plate to atmospheric humidity may be an issue. The manufacturer recommends that the plates be left at room temperature for several hours before use, and we kept the plates on the bench overnight before use. It is possible that the performance characteristics of this EIA may be improved with an automated pipetting station and plate washer and reader. One limitation of this study is that we have fewer than the 20 runs suggested in the NCCLS guidelines for precision analysis. 9 This may prevent us from observing sporadic outliers. However, the Abbott IMx assay CV over 6 months of routine use has remained consistently below 5% at all levels. The three assays showed no signi cant differences in results over the range studied. Both the commercial assays employ SAH calibrators, whereas the HPLC assay uses a homocysteine standard. Each HPLC specimen was measured with and without a known amount of homocysteine, and the differences in
Immunoassays for homocysteine 197 FIGURE 1. Regression lines (a, b) and difference plots (c, d) for the Abbott IMx TM and Bio-Rad homocysteine assays versus homocysteine measurement by highperformance liquid chromatography (HPLC). IMx=1 02HPLC+0 7 mol/l, r=0 99, s y/x =0 33, n=64. Bio-Rad=1 03 HPLC+1 0 mol/l, r=0 98, s y/x =0 51, n=57. EIA=enzyme immunoassay.
198 Donnelly and Pronovost peak areas between the spiked and untreated specimens were used to determine the concentration of the endogenous homocysteine. This provided a recovery control. Although it is possible to speculate that the use of SAH as a calibrator may result in the underestimation of patient homocysteine if the reduction and conversion steps do not go to completion, we saw no evidence of this in either assay. The throughput for both assays was signi cantly higher than that of our HPLC assay, which has a throughput limited to 40 injections over approximately 12 h. Because each specimen is run with a recovery control this further limits throughput to 20 patients samples per day. In contrast, the IMx is capable of processing 20 specimens and controls in 1 h, including the loading of the instrument; the throughput for the Bio-Rad method is comparable, with a 96-well plate being processed in approximately 4±5 h. We did not study interference from other thiol-containing metabolites. Shipchandler and Moore 7 and Pfeiffer et al. 10 have shown that the IMx procedure does not cross-react with a variety of thiols. Haemolysed plasma specimens should not be analysed for homocysteine, as we have demonstrated increases in plasma homocysteine when the cells are not removed promptly. This is probably due to intracellular conversion of methionine to homocysteine via the methylation pathway, and it reasonable to assume that this reaction can also take place in haemolysates. Lipaemic specimens are also not analysed for homocysteine in our laboratory. CONCLUSION The Abbott and Bio-Rad methods both exhibited signi cantly increased throughput in our laboratory. This saving in labour is partially offset by the increased reagent costs compared with HPLC analysis. The Bio-Rad EIA assay requires a plate reader, has many manual steps and exhibited poorer precision. The Abbott assay is fully automated but requires an IMx FPIA analyser. Both commercial immunoassays will help to meet the increasing demand for homocysteine testing. It would be helpful to establish consensus guidelines for preparing patients, specimen collection and processing, and reporting of test results. Acknowledgements We thank R Musani, M O Conner and P Stoddard for the HPLC measurement of homocysteine. REFERENCES 1 Gerritsen T, Vaughn JG, Waisman HA. The identi cation of homocysteine in the urine. Biochem Biophys Res Commun 1962; 9: 493±6 2 Carson NA, Cusworth DC, Dent CE, Field CM, Neill DW, Westall RG. Homocysteinuria: a new inborn error of metabolism associated with mental de ciency. Arch Dis Child 1963; 38: 425±36 3 Graham IM, Daly LE, Refsum HM, Robinson K, Brattstrom LE, Ueland PM, et al. Plasma homocysteine as a risk factor for vascular disease. The European Concerted Action Project. JAMA 1997; 277: 1775±81 4 Daskalakis I, Lucock MD, Anderson A, Wild J, Schorah CJ, Levene MI. Determination of plasma total homocysteine and cysteine using HPLC with uorescence detection and an ammonium 7- uoro- 2,1,3-benzoxadiazole-4-sulphonate (SBD-F) derivatization protocol optimized for antioxidant concentration, derivatization reagent concentration, temperature and matrix ph. Biomed Chromatogr 1996; 10: 205±12 5 Evrovski J, Callaghan M, Cole DE. Determination of homocysteine by HPLC with pulsed integrated amperometry. Clin Chem 1995; 41: 757±8 6 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 7 Shipchandler MT, Moore EG. Rapid, fully automated measurement of plasma homocyst(e)ine with the Abbott IMx analyzer. Clin Chem 1995; 41: 991±4 8 Donnelly JG, Livesey JF, Ooi DS. An HPLC method for quantitation of serum (plasma) homocysteine. Clin Chem 1997; 43: S168 9 National Committee for Clinical Laboratory Standards. Evaluation of Precision Performance of Clinical Laboratory Devices: Tentative Guideline, 2nd ed. Document EP5-T2. Wayne, PA: NCCLS, 1992 10 PfeifferCM, Twite D, Shih J, Holets-McCormackSR, 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 Accepted for publication 7 July 1999