Evaluation of AutoDELFIA and Access Automated ...

Hebles-Duvison et al.: Evaluation of systems for third generation thyrotropin assays

881

Eur J Clin Chem CJin Biochem 1995; 33:881-885 © 1995 Walter de Gruyter & Co. Berlin · New York

Evaluation of AutoDELFIA and Access Automated Immunoassay Systems for Third Generation Assays for Thyrotropin By Maria Hebles-Duvison, Marina Cruz-Ruiz, Reyes Vazquez-Rubio, Isabel Macarro-Sancho, Cristina Silva-Mejias and Fernando Recio-Quijano Servicio Analisis Clinicos, Laboratorio de Bioquimica, H. U. de Valme, Sevilla, Spain (Received April 25/July 17, 1995)

Summary: We evaluated the technical performance of two new commercial automated immunoassay systems of third-generation assays for thyrotropin. The interassay CV was 2.8% for AutoDELFIA and 3.25% for Access at thyrotropin concentrations of approximately 1.3 mIU/1. The lower detection limits of the assays were 0.023 mIU/1 for AutoDELFIA and 0.0096 mIU/1 for Access, and the functional sensitivity for a CV of 20% was 0.027 mIU/1 and 0.028 mIU/1, respectively. Sample to sample carry over was negligible (0.0016% for AutoDELFIA and 0.005% for Access). The range of linearity was acceptable for Access (102—115%) but not for low thyrotropin concentrations in AutoDELFIA (143% for a thyrotropin value of 0.48 mIU/1). Correlation between AutoDELFIA and Access was adequate (r = 0.999). We conclude that both automated immunoassays offer good reliability, practicability and performance characteristics.

Introduction Immunoassays are a valuable analytical technique, due to their low detection limit and good precision. Totally automated immunoassays which use non-isotopic labels are being incorporated into the routine work of clinical laboratories. These automated systems combine heterogeneous test designs with a variety of solid phases and separation methods, leading ultimately an improvement in the low detection limit. This aspect is particularly important when determining hormone concentration such as thyrotropin. The introduction of second generation immunometric assays in the mid-1980s (functional sensitivity limit 0.1-0.2 mIU/1), followed more recently by the third and fourth generations (functional sensitivity of 0.01-0.12 and 0.001-0.002 mIU/1, respectively), has made a valuable contribution to the diagnosis of hyperthyrqid patients (thyrotropin values < 0.1 mIU/1), enabling their differentiation from euthyroid patients with low thyrotropin values caused by non thyroid-related diseases, or by the effect of certain drugs (1—5). In this study we evaluated a third generation thyrotropin assay carried out on two automated systems which use Eur J Clin Chem Clin Biochem 1995; 33 (No 11)

non-isotopic labels. One of the assays is based on a solid phase of magnetic particles with chemiluminometric detection (Access, Sanofi Diagnostics Pasteur, Inc. Chaska, MN 55318 USA). The other automated system is based on a sandwich-type heterogeneous time-resolved immunofluorimetric assay which uses europium ions as signal (AutoDELFIA, Pharmacia Wallac Oy, Turku, Finland). Materials and Methods AutoDELFIA This is a multichanel autoanalyzer (manufacturer see section Introduction) with a load capacity of 432 samples (36 racks with 12 samples/rack), which uses primary tubes with bar code identification. Up to eight analytes can be determined in parallel from each sample and the analysis may also be carried out on one to four replicates of each sample. The reactions take place in microtitre strips, which enables optimization of the sample (25 μΐ) and reagent (200 μΐ) volumes. The technology used is an automated adaptation of the DELFIA manual equipment. The improved lower detection limit obtained with the thyrotropin test is due to the use of three monoclonal antibodies directed against three separate antigenic determinants of the thyrotropin molecule.

Hebles-Duvison et al.: Evaluation of systems for third generation thyrotropin assays.

882 Access

Access (manufacturer see section Introduction) is an automated system with continuous accessibility of samples; results are obtained in 30 minutes, with a maximum of 100 tests/hour. Up to 24 assays may be carried out simultaneously on one serum sample. It uses primary as well as secondary tubes and 100 μΐ of sample for each determination. It is based on a two-site immunoenzymometric (sandwich) assay, in which an immune complex is formed between a monoclonal antibody fixed to paramagnetic particles, the analyte molecule and a second antibody conjugated with alkaline phosphatase. The chemiluminescent substrate, spiroadamantane dioxetane phosphate (dioxetane), generates light which is proportional to the analyte concentration. Specimens

tropin sample with standard zero up to concentrations close to the detection limit. The procedure was repeated the following day by diluting the same patient sample again with standard zero. The results of the day with the poorer recovery of thyrotropin were used for calculation. Nevertheless, both replicates were very similar. Carryover: The sample-to^sample carryover was studied by performing two consecutive determinations'of a high-concentration human serum pool (H) followed by five consecutive determinations of standard zero (L). The relative carryover (%) was calculated according to the formula (Li - L5) X 100/(H - L5). Patient comparison: Fifty two patient sera were assayed for thyrotropin by both AutoDELFIA and Access systems. Tlie concentrations were distributed within the expected clinical thyrotropin range, and they included low and high values. Results were evaluated by the non-parametric regression procedure of Passing & Bablok (8).

The following materials were used: Patient sera for which thyroid function tests had been requested. Control sera (Lyphochek® Bio Rad), at three concentrations, L (low, 1.14 mIU/1), M (medium, 6.96 mIU/1) and H (high, 24.3 mIU/1). These values represent the assigned concentrations from the manufacturer, for the Delfia system.

Tab. 1 Imprecision study with three control sera (20 fold determinations). AutoDELFIA

Access

χ (mIU/1)

CV (%)

χ (mIU/1)

CV (%)

Imprecision within-run L 1.23 Μ 6.96 Η 24.34

1.79 1.24 1.48

1.09 7.89 27.36

3.25* 3.21* 4.08*

The evaluation of both methods was carried out following the recommendations of IFCC and ECCLS (6,7).

Interserial imprecision L 1.31 Μ 7.10 Η 24.62

2.8 4.6 3.7

1.09 7.73 27.60

3.25 3.98 1.81*

Within-run imprecision was assessed by analysing three control sera 20 times each within 1 day.

* Significant differences (p < 0.05) when compared with AutoDELFIA.

Standard solutions included in the commercial kits, with a concentration range (excluding the zero) from 0.03 to 100 mIU/1 for AutoDELFIA and from 0.1 to 100 mIU/1 for Access. Both standards were calibrated by the manufacturers against the 2nd International Reference Preparation 80/558. Evaluation procedures

Between-run imprecision was calculated by daily determination of each one of the control materials over a period of 20 days. The Snedecor F was used for the study of the significance of the imprecision differences between the two methods. Functional sensitivity is considered as the minimum quantified concentration with less than 20% interassay imprecision. It was evaluated with 6 aliquots of a standard thyrotropin preparation (MRC 80/558) diluted with human serum containing an undetectable thyrotropin concentration. The concentrations obtained were 0.005, 0.01, 0.02, 0.03, 0.04, 0.06, 0.07, 0.08, 0.1, 0.2, 0.3, 0.4 and 0.5 mIU/1. Measurements were made with both devices over a period of 6 days. The variances for each method at each concentration were evaluated by the F-test. CV values were calculated, and the thyrotropin concentration with a CV value below 20% was taken as the functional sensitivity limit for each device. To determine whether the actual values measured (at functional sensitivity level) were different from the quantities really present in these samples, Student's t-test was used. Limit of detection: The response of the blank calibrator was analysed by measuring it ten consecutive times in 2 analytical series, after a complete calibration. The limit of detection for each method was calculated according to the formula limit of detection = χ + 3 SD, where χ is the mean concentration and SD the standard deviation. Linearity: Special attention was paid to the behaviour of assays with low thyrotropin concentrations, where diagnosis is more problematic. This was done by processing five dilutions of a high thyro-

Control

Tab. 2

Functional sensitivity.

Expected value (rnIU/1) 0.005 0.01 0.02 0.03* 0.04 0.05 0.06 0.07 0.08

0.1 0.2 0.3 0.4 0.5

Access

AutoDELFIA

χ (n = 6) (mIU/1)

CV (%)

χ (n = 6) (mIU/1)

CV (%)

0.009

45.4

0.016 0.028* 0.049 0.05 0.06 0.075 0.083 0.101 0.149 0.350 0.380 0.419

21.8 11.3*

0.003 0.0085 0.02 0.027*

23 27 25

-^

—

4.9 7.9 3 3.7 9 6.7 4.6 6.1 9.8 7.3

—

0.043 . 0.059 0.075 0.083 0.118

—

0.299 0.41 0.446

8.5*

— 5

11.7

13 5.7 5 — 9.7 6.5 3.4

The functional sensitivity is considered as the value with a CV < 20% (in bold type). * Student's t-test: not significant. The 0.01 mIU/1 level for Access, 0.04 and 0.2 mIU/1 for AutoDELFIA were not considered because of technical problems. Eur J Clin Chem Clin Biochem 1995; 33 (No 11)

883

Hebles-Duvison et al.: Evaluation of systems for third generation thyrotropin assays

0.005

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.1

0.2

0.3

0.4

0.5

Thyrotropin [mil/I]

Interassay precision profiles for human serum sample measured by both methods. +—+ AutoDelfia; a—a Access

Fig. 1 Functional sensitivity.

Results

Limit of detection

Imprecision study

The lowest detectable concentration not equal to zero was 0.023 mIU/1 for AutoDELFIA and 0.0096 mIU/1 for Access.

The results of the imprecision studies are listed in table 1. The within-run CV for the lowest concentration of thyrotropin (mean = 1.23 mIU/1 for AutoDELFIA and 1.09 mIU/1 for Access) were 1.79% and 3.25%, respectively; the between-run CVs were 2.8% and 3.25%, respectively. In intra-serial imprecision studies, CVs were significantly different in both systems at the three control levels analysed. However, only the CV of the high level control was different in the inter-serial imprecision studies.

Linearity Table 3 shows the high and low thyrotropin concentrations measured by both methods; the linearity was studied as well as the percentage recovery. Recovery varied from 109% to 143% for AutoDELFIA and from 102% to 115% for Access equipment.

Functional sensitivity The functional sensitivity of both automated systems is shown in table 2 and figure L.For a variation coefficient lower than 20%, the limit is 0.027 mIU/1 for AutoDELFIA and 0.028 mIU/1 for Access (tab. 2). Method variances were significantly different only for 0.06 and 0.07 mHJ/1, the better values being shown by Access. Thyrotropin concentrations found by both methods at the functional sensitivity level were not statistically different from the theoretical values of the standard dilutions (tab. 2). Eur J Clin Chem Clin Biochem 1995; 33 (No 11)

Tab. 3 Assay linearity. Dilution

1/5 1/20 1/40 1/160 1/640

AutoDELFIA

Access

exmeare- ' pected sured covery (mIU/1) (mIU/1) (%)

exmearepected sured covery (mIU/1) (mIU/1) (%)

1134 283 1417 354 88

1097 274 137 343 86

129 335 155 472 126

114 118 109 133 143

126 301 151 36 88

115 110 110 105 102

Hcbles-Duvison et al.: Evaluation of systems for third generation thyrotropin assays

884 Carryover studies

Discussion

Sample-to-sample carryover was 0.0016% for AutoDELFIA and 0.005% for Access (tab. 4).

Although it is known that significant differences exist between the thyrotropin concentrations of hyperthyroid and euthyroid patients, there is no defined cut-off point for separating these two groups (1-5). In fact, about 4% of patients with a non-thyroid-related disease show thyrotropin concentrations lower than 0.1 mIU/1 (4). Therefore, when faced with an unselected patient population, we cannot draw absolute conclusions about whether a patient is or is riot hyperthyroid, based solely on his/her thyrotropin concentration; on the contrary, we must rely on the free thyroxine quantification and on clinical examination (1, 4). However, a method with a lower detection limit will contribute to an improved differentiation in patients under thyroxin treatment and in hyperthyroid patients with very low serum thyrotropin values (5).

Patient comparison Figure 2 summarizes the statistical quantities of the correlation between the different thyrotropin concentrations measured by AutoDELFIA and by Access. The correlation coefficient was r = 0.999 and the fitted line equation was y = -0.031 + l.OSOx.

Tab. 4

Carryover assay. Measured thyrotropin (mIU/1)

Sample

Serum human

STO

(H) L, L2 L3 L4 L5

'

Carryover*

AutoDELFIA

Access

190

60.13 0.003 0.001 0.001 0.000 0.000

0.006 0.006 0.005 0.002 0.003 0.0016%

0.005%

* Carryover (%) = (L, - L5) X 100/(H - L5). STO = Standard Zero

In this respect, the two methods tested are equally valid for the diagnosis of thyroid disease. The imprecision in both methods is acceptable (below 5%) in spite of the fact that significant differences do exist, AutoDELFIA showing the better intra-assay variation (below 2%). The functional sensitivity (inter-assay variation coefficient less than or equal to 20%) was 0.028 mIU/1 for Access and 0.027 mIU/1 for AutoDELFIA.

100 -

10 -

...s*'

1 -

I« 1 0.01 -

0.001 0.001

0,01

i

1

0,1 1 log Thyrotropin (AutoDelfia) [mlU/I]

10

100

Perpendicular lines drawn to X and Υ axes represent detection limit for each procedure Fig. 2 Patient comparison, r = 0.999, y = - 0.031 + 1.030 χ Eur J Clin Chem Clin Biochern 1995; 33 (No 11)

885

Hebles-Duvison et al.: Evaluation of systems for third generation thyrotropin assays

The limits of detection were 0.0096 mIU/1 and 0.023 mIU/1 for Access and AutoDELFIA respectively, somewhat higher than those specified by the manufacturers (0.006 and 0.005 mIU/1, respectively). The recovery obtained in the linearity study was better with Access. A possible explanation for this difference could be a matrix effect produced by the type of diluents used in this procedure. To test this hypothesis, working buffer provided by the AutoDELFIA manufacturer was used to dilute a new serum sample. The percentage recovery improved slightly (110-128%), but was still poorer than that obtained with Access (102-115%). The results obtained for functional sensitivity, limits of detection and linearity studies show that for low concentrations of thyrotropin the Access assay is slightly better than AutoDELFIA.

In relation to the interchangeability study, significant differences did not exist between the concentrations measured by either method. Neither proportional nor constant errors were found. Results from both systems would therefore be completely interchangeable. With regard to practicability, Access produces results faster, allows programming of emergency samples, has random access, and its calibration is very stable (28 days according to the manufacturers). Access is therefore suitable even for laboratories with a small work load. AutoDELFIA requires programming prior to execution and cannot, therefore, be used for stat samples. The calibration is stable at least for one month. The assays take a longer time and at least two duplicate calibration points are needed. However, its high load capacity (432 samples simultaneously) makes it appropriate for laboratories with a high work load.

Sample-to-sample carryover was acceptable in both methods, since results showed less than twice the intraassay variation coefficient.

References 1. Surks MI, Chopra IJ, Mariash CN, Nicoloff JT, Solomon DH. American Thyroid Association guidelines for use of laboratory test in thyroid disorders. J Am Med Ass 1990; 263:1529-32. 2. Spencer CA, Schwarzbein D, Guttler RB, Lo Presti J, Nicoloff JT. Thyrotropin (TSH)-releasing hormone stimulation test response employing third and fourth generation TSH assays. J Clin Endocrinol Metab 1993; 76:494-8. 3. Spencer CA, Lo Presti J, Patel A, Guttler RB, Eigen A, Shen D, et al. Applications of a new chemiluminometric thyrotropin assay to subnormal measurement. J Clin Endocrinol Metab 1990; 70:453-60. 4. Taimela E, Tahtela R, Koskinen P, Nuutila P, Forsstrom J, Taimela S, et al. Ability of two thyrotropin (TSH) assays to separate hyperthyroid patients from euthyroid patients with low TSH. Clin Chem Endocrinol 1994; 40:101-5. 5. Wilkinson E, Rae PWH, Thomson KIT, Toft AD, Spencer CA, Beckett GJ. Chemiluminiscent third generation assay (amerlite TSH-30) of thyroid-stimulating hormone in serum or plasma assessed. Clin Chem 1993; 30:2166-73.

Bur J Clin Chem Clin Biochem 1995; 33 (No 11)

6. European Committee for Clinical Laboratory Standards. Guidelines for the evaluation of analysers in chemistry. ECCLS document 1986; Vol. 3, No. 2. Berlin: Beuth Verlag. 7. International Federation of Clinical Chemistry. Approved recommendation (1978) on quality control in clinical chemistry. Part 2. Assessment of analytical methods for routine use. J Clin Chem Clin Biochem 1980; 18:78-88. 8. Passing H, Bablok W. A new biometrical procedure for testing the equality of measurements from the different analytical methods. Application of linear regression procedures for method comparison studies in clinical chemistry. Parti. J Clin Chem Clin Biochem 1983; 21:709-20. Maria Hebles-Duvison Servicio Analisis Clinicos Laboratorio de Bioquimica H. U. de Valme Carretera de Cadiz s/n E-41014 Sevilla Spain

Costongs et al.: Evaluation of the DPC IMMULITE

887

Eur J Clin Chem Clin Biochem 1995; 33:887-892 © 1995 Walter de Gruyter & Co. Berlin · New York

Evaluation of the DPC IMMULITE Random Access Immunoassay Analyser By Guido M. P. J. Costongs1, Rene J. M. van Oers2, Ben Leerkes2 and Piet C. W. Janson1 1 2

Maaslandziekenhuis, Sittard, The Netherlands Ziekenhuis Centrum Apeldoorn, Apeldoorn, The Netherlands

(Received April 10/July21, 1995)

Summary: The automated immunoassay analyser, IMMULITE, developed by DPC, was evaluated. IMMULITE is an automated system that allows random access in combination with immediate and continuous access. In this study we evaluated the IMMULITE on four panels of analytes: thyroid, fertility, tumour and "non-routine" markers. We observed good within-run reproducibility (ranging from 2.3-15.9%CV, for low controls, to 2.7-8.7%CV for high controls) as well as between-day imprecision (ranging from 3.7—24.6%CV for low controls, to 2.5- 11.8%CV for high controls). The analytical sensitivity of the assays ranged generally from very sensitive to acceptable. 1 I

\ i,

The dilution curves for all assays were nearly linear i. e. the maximum deviation of the observed from the expected recovery was 8%. . Correlation between IMMULITE and other assays (AxSYM, IMx, TDx, AIA-1200, DPC-c. a. c., Medgenix) varied from r = 0.931 to r = 0.994, except for lutropin and parathyrin with coefficients of correlation of r = 0.594 and r = 0.591. The slopes of the regression lines ranged from 0.745 to 1.327, except for parathyrin where a slope of 2.389 was found. Inter-laboratory correlation was very good between the two locations (Sittard and Apeldoorn) and varied from r = 0.984 to r = 0.999; slopes of the regression lines varied from 0.924 to 1.086. We conclude that the DPC JMMULITE system is suitable for testing routine analytes in random access mode as well as for testing "non-routine" analytes on an automated immunoassay analyser.

Introduction Furthermore it is DPC strategy to introduce not only ., ^v·-,. i ^ · r routine analytes on the IMMULITE but also analytes T In our laboratory for binding-analysis, concentrations of . ess A en ^ ,, - . . _. r ΤΧ^Λ^ΤΤΤ , ι j · · * °" requested by physicians. Therefore IMMULJ many hormones, tumour markers and vitamins are mea- T __ /! . ., j . , 1 " * j n · j «r u u ITE is suitable for running with random access, series sured in serum and other body fluids. We have been ° . ' . . , j · of patient samples either for routine and/or for "nonr using continuous- and random-access immunoassay ana*. „ , ^ _. . ~ TX _ _ ττ T_ , 6 ·,_' , „ j · ^u .· r i routine" analytes. These options of IMMULITE gave us r lysers with a bar-coded primary tube option for several . ^ ' . , . , . - , ". ^ ,. / . j . , the opportunity to improve the service to the rpatient by years for those analytes which are requested m large ,r. ' r r . , , , „ J , .^ ,?, ι ΆΓ producing results for routinely tested analytes as well as TT numbers. However, many analytes which are less fre, ^ . „ , ., . . . . ·, ul results of Mnon-routine analytes within one-working xl" A , quently requested are at present not available onΛthese c,™^ j , .,. ^ , ^" day. For STAT requests we can rproduce results within analysers. ' ^ 0 J 2 hours. IMMULITE (DPC-Cirrus, New Jersey) is an immunoassay analyser (1—2), with continuous- and random ac- We evaluated the following routine and "non-routine" cess capabilities and positive patient identification with analytes: the thyroid panel of thyrotropin and free thybar-codes, without the facility for loading primary tubes, roxine; the fertility panel (lutropin, follitropin, prolactin, Eur J Clin Chem Clin Biochem 1995; 33 (No 11)

Costongs et al.: Evaluation of the DPC IMMULITE

888 progesterone, oestradiol, total hyman chorionic gonadotropin and dehydroepiandrosterone-sulphate), as well as tumour markers (carcinoembryonic antigen, prostate specific antigen, 2-microglobulin, CA-125 (ΟΜΜΑ) and other analytes (parathyrin, cortisol and human growth hormone). In this study, we evaluated IMMULITE at both locations (Sittard and Apeldoorn) for its flexibility, ease of use, as well as assay performance (precision, analytical sensitivity, correlation with the current routine methods and correlation between both locations).

Materials and Methods System description The IMMULITE system features an automated, continuous random access, benchtop immunoassay analyser built around a patented assay tube that encloses the assay-specific coated bead. A centrifugal wash accomplishes efficient separation of free from bound material. The IMMULITE uses enzyme-amplified chemiluminescent chemistry for the detection. The system allows the use of a variety of assay methods: immunometric, competitive and sequential assays. Incubation time and number of reagent additions can be different for each test. During a 30 or 60 minute incubation at 37 °C with intermittent agitation, analyte in the sample and an alkaline phosphatase-labelled second antibody or alkaline phosphatase-labelled competitive reagent is bound to the antibody coated bead to form a complex. Unbound conjugate is then removed during a centrifugal wash, in which the test unit is spun around its vertical axis, forcing excess liquid into a coaxial waste sump integral with the tube. The chemiluminescent substrate, a phosphate ester of adamantyl dioxetane, is added, the temperature controlled, and the test unit incubated for a further 10 minutes. As the substrate undergoes hydrolysis in the presence of alkaline phosphatase, the continuous production of an unstable intermediate results in the sustained emission of light. The photon output, as measured by the luminometer, is proportional to the concentration of analyte in the sample. The IMMULITE automatically handles sample and reagent additions, incubations, separation steps and measurements of the photon output via the temperature-controlled luminometer. Samples and reagents are bar-coded. STAT access is possible at any time regardless of other analytes on the system. On-board reagent stability is approximately 90 days. Up to 12 assay reagent wedges can be stored on the IMMULITE, and substrate stability is 30 days on-board. The IMMULITE PC software graphics display the real time status of each assay in progress. The software calculates analyte concentrations for controls and patient samples from the measured photon signal, using a stored master curve generated by DPC and provided with each reagent lot. Supplied adjustors eliminate the need for calibration: by assaying the adjustors periodically, in the same manner as patient samples, the counts obtained from the adjustors on the user's instrument are correlated with those from the instrument used by DPC to generate the master curve. The software also stores quality control and patient data, which can be archived, searched and displayed in a number of ways. L.I.S. interfacing, in compliance with A.S.T.M. specifications, is available for unidirectional-, bidirectional- or host query communications.

Reagents used for method comparison For the correlation study, samples from outpatients and hospitalised patients between 18 and 80 years old were analysed with the routine method and the DPC IMMULITE. For lutropin, follitropin, prolactin, total human chorionic gonadogropin, human somatotropin, 2-microglobulin (sandwich format), we routinely used the TOSOH ΑΙΑ-1200 system (TOSOH, Tokyo, Japan). For oestradiol, progesterone, dehydroepiandrosterone-sulphate we routinely used RJA-DPC-CAC assays, parathyrin was determined with IRMA Medgenix, cortisol with the TDx assay (FPIA), CA-125 with the IMx assay (MELA), and free thyroxine and thyrotropin with the AxSYM™ assays (MEIA) from Abbott. Precision Within run precision was determined by running 20 replicates of the low assay control, followed by 20 replicates of the medium and 20 replicates of the high assay control during two or three different days. Between day precision was determined by running a 3 level control serum for 10-47 days. Intra- and inter-run precision are expressed as standard deviation and coefficient of variation (3). Sensitivity To calculate analytical sensitivity, 10 replicates of zero calibrator or sample diluent and two replicates of a low yet clearly measurable sample were run on 3 different days (4). Mean rate values obtained from the zero calibrator plus 2 standard deviations were converted into concentrations by interpolation between the zero (diluent) and low sample (concentration plotted versus rate values). Correlation Regression coefficients were calculated according to the Passing & Bablok orthogonal regression method (5-7). Dilution study Dilution curves were prepared for those analytes where, in our opinion, a linear dilution of samples is of clinical relevance. For each analyte tested, we made 2-fold dilutions of three different patient samples. These dilutions were assayed in duplicate. The IMMULITE diluent was used to prepare the dilutions.

Results and Discussion Imprecision Within-run imprecision For the thyroid panel, the intra-run variation for free thyroxine was between 6.6% and 9.5% and for thyrotropin between 2.7% and 3.6% and for the low concentration (0.044 mlU/1) 10.0% (tab. la). For the fertility indicators we obtained an intra-run variation between 3.4% and 9.2%, except for total human chorionic gonadotropin whose low control (7.5 mlU/1, i. e. close to the lower limit of the working range for this assay) showed a CV of 12.6%, and for dehydroepiandrosterone sulphate which gave a CV of 14.2% (mean concentration: 4.0 μιηοΐ/ΐ) (tab. Ib). Eur J Clin Chem Clin Biochem 1995; 33 (No 11)

889

Costongs et al.: Evaluation of the DPC IMMULITE

For the tumour markers we found a within-run precision between 2.3% and 9.8% (tab. Ic).

AI A-1200 and AxSYM™. Free thyroxine of the AxSYM™ has a better intra-run CV.

Other hormones showed an intra-run variation between 2.7 and 9.8% (tab. Id). Compared with our previously published results obtained with the TOSOH AI A-1200 (8) and the ABBOTT AxSYM™ (9), the intra-run CVs in the lower concentration range obtained with the IMMULITE for thyrotropin were better than those from the

Lutropin, follitropin, prolactin and human chorionic gonadotropin show an equivalent intra-run CV, with the exception of the low control of human chorionic gonadptropin from IMMULITE which shows a high intra-run CV (12.6%). Dehydroepiandrosterone-sulphate showed a rather high %CV especially in the lower con-

Tab. 1 Within run and between days variation for the DPC-IMMULITE assays of thyroid, fertility, tumour and "non-routine" markers measured in three different control samples. The control

samples were locally produced in Sittard or bought from Biorad (Lyphochek), Data shown here were obtained from either Sittard or Apeldoom and shown as mean values with their corresponding CV.

Analyte

η

Within-run imprecision Control Low

Medium

Mean (pmol/1) CV (%) Mean (mIU/1) CV (%)

0.044 10

14.7 9.5 2.4 2.7

26.4 6.6 7.9 3.6

Mean (IU/1) CV (%) Mean (IU/1) CV (%) Mean (IU/1) CV (%) Mean (IU/1) (CV %) Mean (nmol/1) CV (%) Mean (nmol/1) CV (%) Mean (μιηοΐ/ΐ) CV (%)

2.7 7.2 2.3 5.5 0.13 3.8 8 12.6 10.3 7.2 0.14 15.9 1.4 *

10.4 5.8 15.5 4.8 0.70 3.4 65 4.6 56.0 7.2 0.56 7.9 4.0 14.2

39.9 5.4 69.8 4.9 1.97 4.3 4019 5.4 109.8 9.2 3.55 7.0 17.8 8.7

Mean fag/l) CV (%) Mean fag/l) CV (%) Mean (kU/1) CV (%) Mean (mg/1) CV (%)

2.1 7.5 1.4 6.1 21 9.8 1.26 2.3

13.2 5.9 3.1 5.9 77 6.1 5.07 3.8

57.1 4.0 30.6 8.2 377 4.5 13.56 5.8

Mean (μπιοΐ/ΐ) CV (%) Mean (ng/1) CV (%) Mean (mIU/1) CV (%)

0.12 9.8 28 5.0 1.2 6.2

0.47 6.3 159 4.0 7.0 5.9

0.89 8.2 435 4.0 20.4 2.7

Between-day imprecision Control Low

High

Medium

High

a. Tliyroid hormones Free thyroxine Thyrotropin (2nd generation)

10 10

4.7 14.4 0.16 3.7

15.8 5.1 8.02 2.9

42.3 8.0 35.83 6.2

1.2 7.4 0.7 8.6 0.13 6.1 4.8 17.7 1.6 24.6 0.51 12.6 1.6 *

2.6 5.5 17.6 6.1 0.56 8.9 129.0 9.8 4.4 13.9 0.24 12.0 4.3 13.7

27.8 6.3 23.9 5.9 1.63 11.7 180.6 9.0 25.1 5.4 1.46 8.5 19.6 11.8

2.2 5.8 1.2 7.4 13.8 5.2 0.90 10.5

13.5 4.8 2.6 5.5 48.0 4.3 2.59 7.6

56.1 2.9 27.8 6.3 62.9 2.5 2.59 6.5

0.23 9.1 31.9 18.0 2.0 6.3

0.44 7.9 43.0 6.0 5.2 3.7

0.75 9.0 106 4.3 16.0 4.0

b. Fertility hormones Lutropin Follitropin Prolactin Human chorionic gonadotropin Progesterone Oestradiol Dejiydroepiandrosterone sulphate

44 44 46 46 13 47 28

c. Tumour markers Carcinoembryonic antigen Prostate-specific antigen CA-125 (ΟΜΜΑ) 2-Microglobulin

10 10 12 14

d. Others Cortisol Parathyrin Human somatotropin

= below detection limit, η = number of patient samples Bur J Clin Chem Clia Biochem 1995; 33 (No 11)

22 13 13

890 centration range. The intra-run CV for the tested tumour markers was comparable to the observations made with AI A-1200 and/or AxSYM. Cortisol showed CVs varying between 6.3 and 9.8%, which are rather high for intra-run CVs. Parathyrin and human somatotropin showed acceptable intra-run CVs.

Costongs et al.: Evaluation of the DPC IMMULITE

tigen, we checked the claims of the manufacturer. Our calculated results (thyrotropin 0.0035 mU/1, human chorionic gonadotropin 1.5 U/l, carcinoembryonic antigen 0.06 μg/l and prostate-specific antigen 0.03 μg/l) were without exception similar to or better than those claimed by the manufacturer in the package inserts. In some cases, our calculated sensitivity was even far better. For lutropin the specification was 0.7 mlU/1, whereas we found a sensitivity of 0.04 mlU/1.

Between-day imprecision For our thyroid panel, we observed over a period of at least 10 days inter-run precision (CV) varying from 2.9 to 14.4%. Between-run precision was generally below 8.0%, except for the low control of free thyroxine (tab. la). Compared with our previously published data on a similar panel assayed on the ΑΙΑ-1200 (8) and the AxSYM™ (9), we found slightly higher to similar values for free thyroxine and lower values for thyrotropin, especially at lower concentrations. In our hands, the inter-run CV for the fertility panel varied from 5.5% to 24.6%. Between-run precision was generally below 10% with the exception of the low controls of human chorionic gonadotropin (17.7%), oestradiol (12.8% and 12%), dehydroepiandrosterone-sulphate (13.7%), and progesterone (24.6%), and the high control of prolactin (11.7%) (Tab. Ib). Overall, the control samples for the fertility panel gave inter-run CV values higher than the AxSYM™ and the ΑΙΑ-1200. The oestradiol between run CVs were similar to the values found with our current routine-methods (DPC-RIA-C. A. C). The between-run CVs for progesterone and dehydroepiandrosterone-sulphate were higher than the CVs obtained with the DPC-C. A. C. assays. Between-run CVs for tumour markers were very good, ranging from 2.5% to 7.6%, except for the 2-microglobulin at low concentrations (tab. Ic). The between run precision for the remaining analytes (cortisol, parathyrin, human somatotropin) varied from 4.3%, up to 18.9% for the low control of parathyrin. Inter-run precision was generally below 9.1% (tab. Id). These results are equivalent to those found with our routine methods, except for cortisol (TDx) where higher CVs were observed on IMMULITE. Sensitivity Analytical sensitivity as claimed by the manufacturer is acceptable. For thyrotropin, human chorionic gonadotropin, carcinoembryonic antigen, and prostate-specific an-

Dilution study Dilutions were made as described previously, for CA125 (ΟΜΜΑ), human chorionic gonadogropin, carcinoembryonic antigen and prostate specific antigen. The dilution curves were without exception nearly linear. The maximum deviation of the observed from the expected recovery was 4.6% for CA-125 (ΟΜΜΑ), 7.7% for human chorionic gonadotropin, 7.2% for carcinoembryonic antigen and 4.9% for prostate-specific antigen. Correlation Comparison of the DPC-IMMULITE assays with our current routine methods showed a correlation coefficient > 0.89 between the respective assays (tab. 2). However, IMMULITE lutropin and parathyrin, compared with the ΑΙΑ-1200 and Medgenix IRMA, showed correlation coefficients of 0.594 and 0.591 respectively. Others, however, have reported a good correlation between the IMMULITE and the Nichols parathyrin IRMA (10) and between the IMMULITE and the Medgenix lutropin IRMA (11); we could not explain these anomalies. However, it is well known that discrepant results between assays from different manufacturers can be caused by differences in assay format, reagent composition and calibration (12). Generally these results suggest that it is possible to change from our routine methods to IMMULITE assays (lutropin, follitropin, prolactin, progesterone, oestradiol, human chorionic gonadotropin, CA-125 (ΟΜΜΑ), β2microglobulin, cortisol and parathyrin) without generating any problems regarding correlation and slope. For parathyrin a change of the reference range interval was needed. Interlaboratory correlation Comparison of the IMMULITE assays at Apeldoorn and Sittard showed closely comparable regression statistics, a correlation coefficient > 0.98 and a slope around 1 (tab. 3), except for free thyroxine which gave a slope of 0.924. This indicates that equivalent results are obtained with the same assays at different locations. Eur J din Chem Clin Biochem 1995; 33 (No 11)

891

Costongs et l.: Evaluation of the DPC IMMULITE

Practicability and ease of use The IMMULITE system consists of several instruments: the actual IMMULITE analyser, a personal computer with dot-matrix printer and bar-code pen, and an uninterruptable power supply. The system requires about 2.5 metres of bench space. In the near future DPC will be adding "tutorial" software, explaining the function of each part of the analyser and reagents. After installation, DPC made some technical changes to the IMMULITE analyser: modification of tubing and of Hamilton syringes, and installation of a substrate preheater. The IMMULITE requires about 5 minutes startup and maintenance time each day and about 5 minutes extra for weekly maintenance. The monthly maintenance takes about 15 minutes. When working with the IMMULITE analyser we found the positive aspects of the system to be:

— the system is easy to operate; routine use requires only a short training of the technologist. — the system allows true random access operation. — the status screen gives the operator useful information on the status of sample, test and reagents in process. Furthermore, the 17 most recently produced results are shown on the status screen. — the reagent level status is kept up to date by level sensing of the pipettor. When the system runs out of reagent a visible and audible alarm is given and the system pauses to allow the user to insert new reagent. — STAT requests can be added at any time, only limited by the maximum of 12 different tests that can be loaded in the reagent carousel. Negative aspects of the IMMULITE system, in our opinion, were: — the handling of test-units is time consuming.

Tab. 2 Correlation between the DPC-IMMULITE assays of thyroid, fertility, tumour and "non-routine" markers and the respective current routine methods. Analyte Free thyroxine Thyrotropin Lutropin Follitropin Prolactin Human chorionic gonadotropin Progesterone Oestradiol Dehydroepiandrosterone sulphate Carcinoembryonic antigen Prostate-specific antigen CA.125 (ΟΜΜΑ) 2-Microglobulin Cortisol Parathyrin Human somatotropin

Routine method AxSYM AxSYM ΑΙΑ- 1200 AIA-1200 AIA-1200 AIA-1200 DPC-c. a. c. DPC-c. a. c. DPC-c. a. c. .AxSYM

IMx IMx

AIA-1200

TDx

Medgenix AIA-1200

η

Slope

Intercept

Γ

Regression interval

51 43 78 101 51 46 75 106 40 66 56 51 81 94 81 49

0.964 1.01 0.789 0.845 0.745 1.03 0.890 1.02 0.970 1.38 0.916 0.985 1.21 1.01 2.39 1.33

1.25 0.072 -0.084 -0.931 0.014 -0.028 0.321 0.007 0.124 -0.627 -0.044 -5.08 -0.932 0.016 8.50 0.196

0.958 0.986 0.594 0.990 0.987 0.994 0.984 0.941 0.978 0.942 0.988 0.979 0.977 0.931 0.591 0.991

6.05- 48.4 pmol/1 0.00- 60.85 mIU/1 0.7 - 59 IU/1 0.1 - 106 IU/1 0.013.00 IU/1 1.0 -7835 IU/1 0.8 - 132 nmol/1 0.051.7 nmol/1 0.05- 23.05 μηιοΐ/ΐ 0.2 - 748 μg/l 0.00- 160.0 μg/l 1.0 - 727 kU/1 1.4 - 13.6 mg/1 0.031.30μιηο1/1 4.0 - 907 ng/1 0.8 - 45.1 mIU/1

χ =? Routine method, y = Immulite, η = number of patient samples, r = regression coefficient

Tab. 3 Inter-laboratory correlation for several analytes between the two DPC-IMMULITE analysers in Sittard and Apeldoorn. Analyte

n

Slope

Intercept

r

Regression interval

Free thyroxine Thyrotropin Lutropin Follitropin Prolactin Human chorionic gonadotropin Progesterone Oestradiol Carcinoembryomc antigen Prostate-specific antigen CA-125(OMMA)

59 46 54 55 56 56 59 54 66 38 54

0.924 1.00 0.972 0.926 1.00 1.04 1.01 1.09 0.985 0.982 1.04

-0.183 -0.100 0.232 0.001 0.000 0.985 0.164 -0.011 -0.188 0.008 -0.371

0.983 0.995 0.998 0.996 0.998 0.985 0.992 0.996 0.994 0.999 0.995

2.0 - 67.6 pmol/1 0.00- 59.30 mIU/1 1.0 - 94.7 IU/1 0.6 - 26.3 IU/1 0.061.80 IU/1 173 -6840 IU/1 1.0 - 125 nmol/1 0.044.35 nmol/1 0.2 - 853 μg/l 0.25- 39.1 μg/l 1 -1577 kU/1

χ = Apeldoorn, y — Sittard, η = number of patient samples, r = regression coefficient Eur J Clin Chem Clhi Biochem 1995; 33 (No 11)

892

Costongs et al.: Evaluation of the DPC IMMULITE

- the software is not always juser friendly because, when working off-line, a lot of typing is required for instance when requesting an adjustment, control or patient sample. Also the access to, and presentation of, patient results, and data from quality control sera is not very user friendly. However when working on-line some of these disadvantages will be eliminated.

tine method; however, these results were still acceptable. For progesterone and dehydroepiandrosterone-sulphate we used the first kit lot released by DPC, and found the analytical variation for these kits to be to high (CV generally about 10%). The analytical sensitivity for the analytes we tested is similar to or better than the sensitivity claimed by DPC.

In our experience the DPC IMMULITE system has positive and negative aspects. We think that this system is, regarding user friendliness, very suitable for running smaller series of samples in random access mode, and for running "non-routine" tests on an automated immunoassay analyser.

We found good results for the analytes tested for linear dilution. The correlation with the routine methods was acceptable, except for lutropin and parathyrin; we only had to change our reference-interval for parathyrin. Inter-laboratory concordance was very good between Sittard and Apeldoorn except for free thyroxine whose regression slope was rather low (0.924).

Conclusions The analytical variation for most of the analytes tested was good, except for prolactin, human chorion gonadotropin and cotfisol where CVs were higher than the rou-

The DPC IMMULITE menu offers not only the routine analytes but also the more esoteric tests not yet available on other random access immunoassay analysers. These features make DPC-MMULITE suitable for running not only routine but also "non-routine" analytes On an automated system.

References 1. Babson AL. The Immulite automated immunoassay system. J Clin Immunoassay 1991; 14:83-8. 2. Hand C. DPC Immulite. In: Wild D, editor. The immunoassay handbook. New York: Stockon Press, 1992:170-4. 3. Swinsow TDV. Statistics at square one. London, British Medical Assoc 1982:1-60. 4. Smith J, Osikowicz G, Tayi R, Walker D, Martin R, Vaught J., et al. Abbott AxSYM™ random and continuous access immunoassay system for improved workflow in the clinical laboratory. Clin Chem 1993; 39:2063-9. 5. Passing H, Bablok W. A new biometrical procedure for testing the equality of measurements for two different analytical methods. J Clin Chem Clin Biochem 1983; 21:709-20. 6. Passing H, Bablok W. Comparison of several regression procedures for methods: comparative studies and determination of sample sizes. J Clin Chem Clin Biochem 1984; 22:431-45. 7. Bablok W, Passing H, Bender R, Schneider B. A general regression procedure for method transformation. J Clin Chem Clin Biochem 1988; 26:783-90. 8. Costongs GMPJ, Janson PCW. Comparison of the automated random access immunoassay analyser ACS-180 (Ciba-Corning) and ΑΙΑ-1200 (Tosoh). Eur J Clin Chem Clin Biochem 1993; 31:701-6. 9. Costongs GMPJ, van Oers RJM, Leerkes B, Hermans W, Janson PCW. Evaluation of the Abbott automated random, immediate and continous access immunoassay analyser, the AxSYMTM. Eur J Clin Chem Clin Biochem 1995; 33:105-11.

10. Durham AP, Sustarsic D, Lei JD, El Shami AS., et al. DPC selected scientific posters and abstracts. An automated chemiluminescent immunometry assay for intact parathyrin (PTH) on the IMMULITE system. 46th National Meeting and Clinical Laboratory Exposition. New Orleans, Louisiana, American Association for Clinical Chemistry 1994:29-32. 11. Sustarsic D, Lei JD, Durham AP, El Shami AS. DPC selected scientific posters and abstracts. Fertility panel on the automated DCP-IMMULITE system. In: American Association for Clinical Chemistry. 46th National Meeting and Clinical Labo^ ratory Exposition, New Orleans, Louisiana 1994:36-7. 12. Kuroki M, Haruno M, Arakawa F, Wakisaka M, Matsuoka Y. Reaction profiles of seven enzyme immunoassay kits for carcinoembryonic antigen analyzed with purified preparations of CEA and related normal antigens. Clin Biochem 1992; 25:29-35. 13. Vankrieken L, De Hertogh R. Rapid, automated quantification of total human chorionic gonadotropin in serum by a chemiluminescent enzyme immunometric assay. Clin Chem 1995; 41:36-40. Dr. G. M. P. J. Costongs Dept. Clinical Chemistry Maaslandziekenhuis Walramstraat 23 NL-6131 BKSittard The Netherlands

Eur J Clin Chem Clin Biochem 1995; 33 (No 11)