Webmail.nvkc.nl

Uit de laboratoriumpraktijk
Analytical performance comparison of five new generation immunoassay analyzers
H.A. HENDRIKS1*, W. KORTLANDT2 and W.M. VERWEIJ1 We evaluated the analytical performance of the new
nation with relatively short reaction times. This is generation immunoassay analyzers (ACS:Centaur®,
especially relevant in view of the goal of shortening Architect TMi2000, Elecsys®2010, Immulite®2000 and
the patients stay in hospitals and the increasing scale Vitros ECi) for the following analytes: TSH, FT4, vit-
amin B12, ferritin, folate, CEA, HCG, PSA, estradiol,
The introduction of a number of large scale analyzers LH, FSH, prolactin, and progesterone. The charac-
prompted us to evaluate them for analytical per- teristics evaluated were: within-run precision, carry-
formance and efficiency and capacity. The evaluation over and comparison of methods with instruments
of the efficiency and productivity of the five new currently in use; ACS:180, AxSYM and Immulite1.
generation analyzers tested (ACS:Centaur, Architect, The within-run precision of the test results for most
Elecsys 2010, Immulite 2000 and Vitros ECi) has assays was good (CV’s of between 2 and 4%), and for
been described elsewhere (1). Here we focus on the some assays on the Vitros ECi (HCG, PSA, FSH and
analytical performance of the five new generation prolactin) and Elecsys 2010 (FT4, FSH and pro-
immunoassay analyzers, which comprised a carry- lactin) even very good precision (< 2%) was mea-
over study, evaluation of the within-run precision and sured. Relatively higher CV’s for the anemia assays
comparison of the new generation analyzers assays (ferritin, vitamin B12 and folate), as observed in the
with the in-house ACS:180, Immulite 1 and AxSYM low concentration range, and for some fertility assays
assays in order to study the problems involved in (progesterone on the ACS:Centaur and the Immulite
arriving at the use of one universal set of reference 2000 and estradiol on the Architect), were deter-
mined. No sample carry-over was found for any of
the analyzers. Most methods showed good correlation
MATERIALS and METHODS
(r>0.97). Statistical significant slope differences were
measured for the prolactin and FSH assays on the
Analyzers
Elecsys 2010 and significant intercept differences
The ACS:Centaur (Software version 1.2, Chiron were measured for the prolactin assay on the Vitros
Diagnostics) is a fully automated random access ECi and the vitamin B12 assay on the Elecsys 2010.
immunoassay analyzer that uses paramagnetic solid- The new generation analyzers do not diminish the
phase particles and an acridinium ester based direct efforts that have to be made to harmonize the values
chemiluminescent tracer, coupled to antibodies in a of different laboratories using different equipment.
second reagent. Luminescence is initiated by additionof acid and base reagent. The ACS: Centaur has a Keywords: immunoassay analyzers; ACS:Centaur®; reagent capacity of 30 reagent packs (2).
ArchitectTMi2000; Elecsys®2010; Immulite®2000 and The fully automated random access analyzer Archi- Vitros ECi; within-run precision; carry-over; com- tect i2000 (Software version 1.00, Abbott) uses chemiluminescent immunoassay technology incorpo-rating an acridinium derivative tracer. This analyzer Immunoassay testing has progressed over the years.
uses paramagnetic microparticles as solid phase.
Since the introduction of the first radioimmunoassay, After exposure to pre-trigger and trigger reagent, the several alternative and less hazardous detecting labels acridinium undergoes a decomposition reaction and have been developed and the methods of antibody the emitted light is amplified and processed. The production have been improved. Besides these chemi- Architect has 25 reagent pack positions.
cal improvements, automation has enabled a further The fully automated analyzer Elecsys 2010 (Software increase in reproducibility and sensitivity, in combi- version 3.08, Roche- Boehringer Mannheim) incor-porates an electrochemiluminescence (ECL) detec-tion cell. The streptavidine coated paramagnetic Stichting Artsen Laboratorium1 and Diakonessenhuis2, beads, are coupled to the ruthenium-labeled antigen- antibody complex. Following the addition of tri-propylamine, a voltage is applied and the resulting Correspondence to: H.A. Hendriks, Stichting Artsen Laborato- luminescence is measured (4,5). The Elecsys 2010 rium, Noorderstraat 8, 3512 VX Utrecht, The Netherlands.
Received: 13.01.00 encompasses 15 reagent positions and 12 different assays which are available simultaneously. Ned Tijdschr Klin Chem 2000, vol. 25, no. 3 The assays of the fully automated continuous random RESULTS and DISCUSSION
access analyzer Immulite 2000 (Software version 1.2,Diagnostic Products Corporation) are based on an Precision
alkaline phosphatase label and a chemiluminescent Table 1 shows the mean concentrations and within- substrate and employ a centrifugal wash method.
run precision data expressed as coefficient of varia- The Immulite 2000 encompasses 24 reagent positions tion (CV,%). The within-run CV’s of the thyroid hormones were less than 4%, with exception of the The Vitros ECi (Software version 2.0, Ortho Clinical low concentration TSH assays on the ACS:Centaur Diagnostics) is a fully automated random access (10% at a concentration of 0.06 mIU/L), Immulite immunoassay system that utilizes an enhanced 2000 (10% at a concentration of 0.07 mIU/L) and the chemiluminescence technology. This analyzer uses Vitros ECi (9% at a concentration of 0.17 mIU/L).
streptavidin coated plastic wells as solid phase. Horse Very good precision (CV’s <2%) were measured for radish peroxidase is used as label and a luminogenic medium and high concentration TSH assays on the substrate (a luminol derivative and a peracid salt) as Vitros ECi and Elecsys 2010. The Elecsys 2010 signal detection. The Vitros ECi has a reagent showed very good precision for all levels with the FT4 assay. The FT4 assay on the Immulite 2000 Descriptions of the in-house analyzers ACS:180, showed CV’s of above 4% for all levels. These higher AxSYM and Immulite 1 are given in the respective CV’s are in accordance with the within-run data as given by the supplier for this assay.
The tumor markers showed CV’s of 4% or less on all Serum samples
analyzers except for the low CEA levels on the Anonymous patient sera were collected, aliquoted ACS:Centaur and Elecsys 2010. Very good precision, and stored at -20°C. To exclude effects of re-thawing with CV’s under the 2%, was measured for all levels a set of 50 aliquots per assay was made available for each analyzer to perform correlation studies. The CV’s of between 2 and 4% (up to 8% at low levels) samples are distributed over the range that is encoun- were observed with most of the fertility hormones, tered in routine application. The samples were stored with exception of the methods as mentioned below.
in the same manner for all of the analyzers to avoid Very good precision, with CV’s under or equal to 2%, storage conditions becoming a variable. The samples were measured for a large number of Vitros ECi were thawed and homogenized immediately before assays, in particular for all levels of the HCG, FSH and prolactin assays. The Elecsys 2010 also showedvery good precision for all levels with the LH, FSH Precision
and prolactin assays. With exception of the lowest Within-run precision was determined using commer- level, the FSH assay on the ACS:Centaur was very cial control sera (Lypocheck Immunoassay Plus Con- good. Only the progesterone assay on the Immulite trol Levels 1, 2 and 3 and Lypocheck Anemia 2000 exhibited significant higher within-run CV’s Control, Bio-Rad). For each assay, replicate measure- from 7 to 11%. For the progesterone assay on the ments (n=20) were performed in one run for each ACS:Centaur and the estradiol assay on the Architect level. This procedure was processed three times CV’s of between 4.5 and 6.7% were measured. within a period of 4-7 days, using a single reagent lot The anemia assay vitamin B12 was only available on and a single calibration. The within-run precision the ACS:Centaur, the Architect and the Elecsys 2010 data are expressed as coefficients of variation (CV%).
and folate was only available on the ACS:Centaur andthe Elecsys 2010. On the Architect moderate CV’s of Carry-over
between 5 and 7% were measured for all levels of the Assay carry-over was tested on the analyzers with vitamin B12 assay, and for the low levels of the HCG assays. Ten aliquots of low concentrate sample vitamin B12 and folate assays on the ACS:Centaur.
(~ 5 mIU/l) were measured to generate a base line con- For the Elecsys 2010 a CV higher than 10% was centration. Then one aliquot of high concentration measured for the medium level of the folate assay.
sample (~ 100000 mIU/l) was followed by five aliquots The relatively higher CV’s for these anemia assays of the low concentration sample. This sequence was reflect the present state of art for these methods in repeated twice (17). Carry-over was calculated as general (20). The within-run CV’s of the ferritin follows: (concentration of a low sample after pipetting assays on all of the analyzers can be generally rated the high samples minus base line concentration) as being good, with CV’s between 2 and 4%. With divided by (concentration of high sample) (13). exception of the low levels on the ACS:Centaur andImmulite 2000 and the high level on the Vitros ECi Comparison of methods
were CV’s of respectively 7.4, 4.2 and 5.7 were mea- Comparisons of ACS:Centaur, Architect i2000, Elec- sys 2010, Immulite 2000 and Vitros ECi assays withthe in-house ACS:180, Immulite 1 and AxSYM Carry-over
assays were performed. For the complete matrix of No carry-over (< 10-6) could be demonstrated, on method comparisons the Passing and Bablok regres- each of the five analyzers. Though sample or reagent- sion lines and the coefficients of correlation were cal- dependent carry-over possibly caused by pipetting steps is eliminated by the use of disposable tips in the Ned Tijdschr Klin Chem 2000, vol. 25, no. 3 ACS:Centaur, Elecsys 2010 and Vitros ECi, a poten- Comparison of methods
tial source of carry-over is present in the Architect In the comparison of thyroid methods (FT4 and TSH) and Immulite 2000 , as well as the measuring cell in between all eight assays, each showed correlation the Elecsys 2010 system. No carry-over was detected, coefficients of more than 0.97 (Table 2). The mean although on the Elecsys 2010 the results of the low slope for the TSH assay on the Elecsys 2010 was concentration samples measured immediately after 1.32, resulting in higher results. Comparable higher the high concentration samples were systematically slopes were found in the studies of Forest et al (4) flagged with potential carry-over by the instrument and Ebert et al (21) for the Elecsys 2010 TSH assay software. When a signal is lower than a calculated with respect to the AxSYM TSH assay.
limit in respect to prior measurement the sample will Table 3 shows the regression results of the fertility be flagged with potential carry-over by the instru- assays. Method comparison of the Vitros ECi LH ment software. In our experiment this flag appeared assay with methods on the other analyzers showed large scattering compared with the regression data of Table 1. Within-run precision of thyroid and fertility hormones, tumormarkers and anemia assays measured on ACS:Centaur, Archi-
tect, Elecsys 2010, Immulite 2000 and Vitros Eci
a: anemia control instead of Immunoassay level 1; b: investigational use only; -: test not available at time of study Ned Tijdschr Klin Chem 2000, vol. 25, no. 3 Table 2. Method comparisons of the thyroid assays TSH and FT4
Table 3. Method comparisons of the hormone assays LH, FSH, progesterone, prolactin, oestradiol and HCG
Ned Tijdschr Klin Chem 2000, vol. 25, no. 3 Table 3. Continue
Ned Tijdschr Klin Chem 2000, vol. 25, no. 3 Table 4. Method comparisons of the tumor markers PSA and CEA
the other instruments. The large scattering observed for the Vitros Eci resulted in significant higher pro- by the comparison of the Vitros ECi LH assay with lactin values, although all prolactin assays were cali- the other assays could be a result of the difference in brated to the same standard. With progesterone, good method of measurement. The correlation coefficients correlation coefficients (>0.97) were obtained in all of the Vitros ECi LH assays and the other assays cases tested. No correlation coefficients were calcu- were between 0.90 and 0.96. Between the other ana- lated for the Elecsys, because on this instrument 35 lyzers the range of correlation coefficients for the LH of the 50 samples were under detection level (<0.48 nmol/l). Good concordance between all of the ana- The FSH methods had correlation coefficients of lyzers for the HCG assays observed, was found. 0.99, except for the Elecsys assays where correlation In table 4, PSA and CEA method comparison studies coefficients of 0.95 were determined. The mean slope of the regression lines, observed by comparing the Results of the method comparison of the anemia FSH assay of the Elecsys 2010 to the other FSH assays are represented in table 5. The ferritin methods methods, was 1.58. Resulting in higher results for this showed very good correlation with coefficients of > assay on the Elecsys 2010 analyzer. Comparable dif- 0.99. Correlation of the Vitamin B12 methods exhib- ferences were found for the values of the control ited coefficients between 0.97 and 0.99, except for materials measured in the precision experiment.
the Elecsys 2010 Vitamin B12 assay where corre- The relatively poor precision results of the estradiol lation coefficients of between 0.94 and 0.96 were methods clearly has demonstrated the difficulty with observed. Additionally, the Elecsys 2010 Vitamin this analyte (20) and therefore can also explain the B12 assay showed intercepts of more than 100 pmol/l large number of correlation coefficients of below 0.97. against other methods. Considerably lower folate Whilst comparing the prolactin assays on the results were measured on the Elecsys 2010 compared AxSYM, correlation coefficients of between 0.91 and to all of the analyzers and 32 out of the 50 samples 0.97 where observed, whilst generally, correlation were under detection level (<1.1 nmol/l). The re- coefficients of more than 0.97 were obtained between maining sample size was too small for regression all other analyzers. Against all methods, a significant analysis. The other available assays showed correla- average slope value of 1.36 was observed for the tion coefficients of between 0.95 and 0.98, with about Elecsys prolactin assay resulting in systematic higher 30% higher values for the Immulite 1 in comparison values. The average intercept value of 65.15 mIU/l to assays of the two other suppliers.
Ned Tijdschr Klin Chem 2000, vol. 25, no. 3 Table 5. Method comparisons of the anemia assays ferritin, vitamin B12 and folate.
Slope and intercept differences in the method com- In conclusion: on each of the five analyzers no carry- parison data demonstrated that not all methods are over was measured. Most of the test results of the interchangeable unless prior reference studies are per- new immunoassay analyzers showed good within-run formed. Deviations in some methods referenced to reproducibility (CV <4%) with a few exceptions in the same standard, such as the TSH methods that are the low concentration range of some assays and with all referenced to the 2nd IRP WHO 80/558 standard, all concentration ranges of some fertility assays. The showed that calibration against the same standard Vitros ECi and Elecsys 2010 even exhibited very does not automatically guarantee inter-method agree- good CV’s (CV<2%) for a large number of assays.
ment. Recent publications in the Netherlands (22) do For most of the method comparisons, good correla- at least suggest that there is one universal set of refer- tion coefficients were determined (>0.97).
ence values for laboratory tests. Our results indicate Significant slope and intercept differences were mea- the problems that have to be solved to achieve this sured, for some fertility hormones and anemia assays, Ned Tijdschr Klin Chem 2000, vol. 25, no. 3 The new generation analyzers do not diminish the 15. Babson AL. The Immulite automated immunoassay amount of effort that has to be applied to harmonize system. J Clin Immunoassay 1991; 14: 83-88.
the values of different laboratories using different 16. Costongs GMPJ, Oers RJM van, Leerkes B, Janson PCW.
Evaluation of the DPC IMMULITE random access immunoassay analyzer. Eur J Clin Chem Clin Biochem1995; 33: 887-892.
Acknowledgement
17. Patterson W, Werness P, Payne WJ, Matsson P, Leflar C, We acknowledge Abbott, Chiron Diagnostics, Diagnostic Melander T, Quast S, et al. Random and continuous - Products Corporation, Ortho Clinical Diagnostics and Roche access immunoassays with chemiluminescent detection by Boehringer Mannheim for placing their systems at our dis- Access® automated analyzer. Clin Chem 1994; 40: 2042- posal, giving technical support and providing the necessary reagents. We also thank Jolanda Huizing for work in carrying 18. Passing H, Bablok W. A new biometrical procedure for out the analysis, Malcolm Stimmler and René van Oers for testing the equality of measurement from two different contributions in improving this manuscript and everyone of analytical methods. Application of linear regression proce- Stichting Artsen Laboratorium Utrecht who helped with dures for method comparison studies in Clinical Chem- aliquoting the large numbers of samples. We also like to thank istry, Part I. J Clin Chem Clin Biochem 1983; 21: the laboratories of Diakonessenhuis Utrecht, Lucas Andreas Ziekenhuis Amsterdam and Reinier de Graaf Gasthuis Diag- 19. Passing H, Bablok W. Comparison of several regression nostisch Centrum SSDZ Delft for collecting sera and the Stg.
procedures for method comparison studies and determina- Deventer Ziekenhuizen for their cooperation. tion of sample sizes. Application of linear regression pro-cedures for method comparison studies in Clinical Literature
Chemistry, Part II. J Clin Chem Clin Biochem 1984; 22:431-445.
Hendriks HA, Kortlandt W, Verweij WM. Standardized 20. SKZL. LWBA rondes 1999, Dutch collaborative external comparison of processing capacity and efficiency of five new-generation immunoassay analyzers. Clin Chem 2000; 21. Ebert Ch, Bieglmayer Ch, Igari J, Chan D, Torralba A, Müller Ch, Veys A, et al. Elecsys® TSH, FT4, T4, T- Girgensohn S, Liedtke R, Balzer G, Castor S, Hauser M.
uptake, FT3 and T3. Clinical results of a multicentre Performance of the ACS:CentaurTM High-capacity, random- study. Wien Klin Wochenschr 1998 [suppl] 3; 110: 27-40.
access immunoassay system. Clin Lab 1997; 43: 975-983.
22. Leusden van HAIM, Boer A, Bijlsma JA, Bosman PJ, Hartke Th. Future aspects of Elecsys® automation - The Denteneer AE, Grijseels EWM, Loenhout van JWA, et al.
modular approach Wien Klin Wochenschr 1998 [Suppl] 3; Diagnostisch Kompas 1999/2000. 2e ed. Utrecht: Roto Forest JC, Massé J, Lane A. Evaluation of the analyticalperformance of the Boehringer Mannheim ElecsysÒ2010 immunoanalyzer. Clin Biochem 1998; 31: 81-88.
Sapin R, Gasser F, d’Herbomez M, Wemeau JL, Ebert C, Analytical performance comparison of five new generation Schlienger JL. Elecsys thyrotropin (TSH) assay evaluated.
immunoassay analyzers. Henriks HA, Kortlandt W and Verweij WM. Ned Tijdschr Klin Chem 2000; 25: 170-177. Vankrieken L, De Hertogh R. Rapid, automated quanti- We evaluated the analytical performance of the new generation fication of total human chorionic gonadotropin in serum immunoassay analyzers (ACS:Centaur®, Architect TMi2000, by a chemiluminescent enzyme immunometric assay. Clin Elecsys®2010, Immulite®2000 and Vitros ECi) for the fol- lowing analytes: TSH, FT4, vitamin B12, ferritin, folate, CEA, Babson AL, Olson DR, Palmieri T, Ross AF, Becker DM, HCG, PSA, estradiol, LH, FSH, prolactin, and progesterone.
Mulqueen PJ. The Immulite assay tube: a new approach to The characteristics evaluated were: within-run precision, heterogeneous ligand assay. Clin Chem 1991; 37: 1521- carry-over and comparison of methods with instruments cur- rently in use; ACS:180, AxSYM and Immulite1. The within- Horst van der FA, Bos W. Modified sample-cup holder run precision of the test results for most assays was good improves work flow in immulite immunoassay system.
(CV’s of between 2 and 4%), and for some assays on the Vitros ECi (HCG, PSA, FSH and prolactin) and Elecsys 2010 Saw S, Sethi S, Aw TC. Technical evaluation of thyroid (FT4, FSH and prolactin) even very good precision (< 2%) assays on the Vitros ECi. Clin Chem 1999; 45: 578-580.
was measured. Relatively higher CV’s for the anemia assays 10. Boland J, Carey G, Krodel E, Kwiatkowski M. The Ciba (ferritin, vitamin B12 and folate), as observed in the low con- Corning ACS:180TM benchtop immunoassay analyzer.
centration range, and for some fertility assays (progesterone on the ACS:Centaur and the Immulite 2000 and estradiol on 11. Mora-Brugués J, Gascón-Roche N, Rodríguez-Espinosa J, the Architect), were determined. No sample carry-over was Cortés-Rius M, González-Sastre G. Evaluation of Ciba found for any of the analyzers. Most methods showed good Corning ACS:180TM automated immunoassay system.
correlation (r>0.97). Statistical significant slope differences were measured for the prolactin and FSH assays on the 12. Römer M, Haeckel R, Capelli M, Rocipon J. The analyti- Elecsys 2010 and significant intercept differences were mea- cal performance of the Ciba Corning ACS:180 automated sured for the prolactin assay on the Vitros ECi and the vitamin immunoassay system. A multicentre evaluation. Eur J Clin Chem Clin Biochem 1994; 32: 395-407.
The new generation analyzers do not diminish the efforts that 13. Smith J, Osikowicz G. Abbott AxSYM random and con- have to be made to harmonize the values of different laborato- tinuous access immunoassay system for improved work- flow in the clinical laboratory. Clin Chem 1993; 39: Keywords: immunoassay analyzers; ACS:Centaur®; ArchitectTM i2000; Elecsys®2010; Immulite®2000; Vitros ECi; within-run 14. Costongs GMPJ, Oers RJM van, Leerkes B, Hermans W, precision; carry-over; comparison of methods. Janson PCW. Evaluation of the Abbott automated random,immediate and continuous access immunoassay analyser,the AXSYMTM. Eur J Clin Chem Clin Biochem 1995; 33:105-111.
Ned Tijdschr Klin Chem 2000, vol. 25, no. 3

Source: http://webmail.nvkc.nl/tijdschrift/content/2000/nr%203/p170/2000-3-170.pdf