Clin Biochem, Vol. 25, pp. 251-254, 1992

0009-9120/92 $5.00 + .00 Copyright ¢ 1992 The Canadian Society of Clinical Chemists.

Printed in the USA. All rights reserved.

Evaluation of the Boehringer Mannheim ES 300 Immunoassay Analyzer and Comparison With Enzyme Immunoassay, Fluorescence Polarization Immunoassay, and Radioimmunoassay Methods PAUL D. CAMARA, KIM VELLETRI, MAUREEN KRUPSKI, MARK ROSNER, and WILLIAMS C. GRIFFITHS Roger Williams Hospital, Department of Pathology and Laboratory Medicine, Clinical Chemistry Section, 825 Chalkstone Avenue, Providence, RI 02908, USA The ES 300 (Boehringer Mannheim Diagnostics, Indianapolis, IN) is a new automated immunoassay analyzer intended for the quantitative determination of a wide range of analytes. We compared its performance to enzyme immunoassay (EIA), fluorescence polarization immunoassay (FPIA), and radioimmunoassay (RIA) methods for cortisol, digoxin, ferritin, prolactin, T4-uptake, totaI-T3, and TSH. The ES 300 methods showed excellent precision and the manufacturers' linearity claims were met in all cases. Cortisol, prolactin, total-T3, and TSH showed no bias and acceptable correlation with other methods. Digoxin, ferritin and total T4 showed positive bias but acceptable correlation. The ES 300 T4-uptake correlated poorly with the TDx method and showed positive bias; however, these assays appear comparable (although deficient) in diagnostic sensitivity when compared to TSH and T4 data for the same patient population. In all, we found the ES 300 to be an acceptable instrumental alternative for the high volume immunoassay laboratory.

KEY WORDS: immunoassay; ligands; comparative study; digoxin; thyroid; cortisol; prolactin; ferritin; statistical bias.

Introduction

ne of the most active areas of instrument and methodologic development in clinical chemistry today is the automation of immunoassays. The evidence for this is particularly striking in cancer diagnosis, endocrine testing, infectious disease testing, and therapeutic drug monitoring, in which immunoassays are the methods of choice. Recently, Boehringer Mannheim Diagnostics introduced a new automated immunoassay analyzer to the American market. The ES 300 is intended for

O

Correspondence: W i l l i a m s C. Griffiths, Roger W i l l i a m s Hospital, D e p a r t m e n t of Pathology and L a b o r a t o r y Medicine, Clinical C h e m i s t r y Section, 825 C h a l k s t o n e Avenue, Providence, RI 02908, USA. M a n u s c r i p t received A p r i l 15, 1991; revised A u g u s t 22, 1991 and J a n u a r y 7, 1992; accepted F e b r u a r y 14, 1992. CLINICAL BIOCHEMISTRY, VOLUME 25, AUGUST 1992

the in vitro quantitative determination of a wide range of ligands using a nonisotopic, ELISA (enzyme-linked immunosorbent assay) coated-tube technology; it employs the same reagent formulation and analytical technology as its predecessor, the ES 600. Analytes are measured using competitive binding, modified competition, and sandwich type immunologic reactions. Incubation times range from 45 min (digoxin) to 3.5 h (TSH) as shown in Table 1. The modified competition technique is used exclusively for the T4-uptake assay. The reagent formulation in this assay is configured to measure thyroxine binding (rather than triiodothyronine binding), the measurement of which includes the thyroid-binding capacity of prealbumin, in addition to the other thyroid-binding proteins. TABLE 1 Comparison of Incubation Times and Analytical Range Incubation Times (min)" Analyte Cortisol (nmol/L) Prolactin (ng/mL) T 3 (nmol/L) TSH (~LU/mL) Ferritin (ng/mL) Digoxin (nmol/L) T4 T4-uptake (Units)

Analytical Ranges

ES 300

Method Comparedb

ES 300

Method Compared b

60

45 (RIA)

" 28-1300

28-1380

60 55 210 165

65 60 150 (EIA) 90

2-400 0.4-9.2 0.07-30 5-800

2-250 0.8-12.3 0.05-60 10-800

45 55

24 (FPIA) 20

0.4-6.4 6.5-310

0.6-6.4 39-310

55

25

0.2-1.6

0.4-2.0

"Times for ES 300, RIA, and EIA methods include only actual immunologic and detection incubations for any given number of samples. Times indicated for FPIA assays represent the time needed to complete an entire run of a full carousel (20 samples). b Actual method descriptions are given in the text. 251

CAMARA, VELLETRI, KRUPSKI, ROSNER, AND GRIFFITHS TABLE 2

ES 300 Run-to-Run Imprecision Studies Level I

Level II

Analyte

Mean

SD

CV (%)

No.

Mean

SD

CV (%)

No.

Cortisol (nmol/L) Digoxin (nmol/L) Ferritin (ng/mL) Prolactin (ng/mL) T3, total (nmol/L) T4, total (nmol/L) T4-uptake (Units) TSH (~U/mL)

458 1.77 39.1 7.5 0.29 98.7 1.22 4.16

16.7 0.12 5.1 0.49 0.02 6.7 0.06 0.28

3.6 6.5 13.0 6.5 5.9 6.8 4.9 6.6

25 7 48 13 5 12 11 13

737 4.45 190.6 29.9 0.56 268 0.7 7.68

30.4 0.19 9.4 1.8 0.02 14.5 0.03 0.4

4.1 4.3 4.9 6.0 3.1 5.4 4.9 5.2

25 7 46 14 5 12 11 13

We evaluated the ES 300 and compared its performance with methods for cortisol (RIA, Diagnostic Products Corp., Los Angeles, CA), digoxin, total-T4, and T4-uptake (FPIA, Abbott TDx©, Abbott Laboratories, N. Chicago, IL), thyroid-stimulating hormone (TSH), and ferritin (EIA, Abbott Quantum, Abbott Laboratories, N. Chicago, IL), prolactin (RIA, MAIAclone, Serono Diagnostics, Norwell, MA), and total-T 3 (RIA, Clinical Assays, Dade Div., Baxter/Travenol, Cambridge, MA).

Materials and Methods ES 300 SYSTEMDESCRIPTION The major system components of the ES 300 include a sample rotor (150 sample cup capacity, with space for 10 cups of multiconstituent control of 5.0 mL each), reagent rotor (12 reagent capacity, plus cleaning solution and universal substrate), incubator (162 tube capacity), a robotic arm for the various dispensing and washing steps, a tip washer, automatic pipettor, and a photometer for the optical measurements. The various instrumental functions are programmed and controlled through an external desktop computer (Hewlett Packard Vectra, with a 386-processor, 4 Mb of RAM and a 40 Mb hard drive). The software allows the user to access available results from a run in progress, or to program the next run while the current run is still underway.

The ES 300 employs a transmission filter (422 nm) photometer, with a halogen source (350-800 nm emission spectrum). The cuvette holder and cuvette are temperature controlled by a Peltier device which maintains the same temperature as the incubator. The cuvette is a flow-through cell with a 3 mm lightpath to minimize carryover. The ES 300 may perform up to 12 different assays in a single run. The operator is required to prepare the reagents (some reconstitution is necessary), program the run, load the reagents, standards, controls, unknowns, and the coated tubes for each assay. Once programmed, the computer optimizes the run for fastest execution by calculating the appropriate pipetting sequences allowing the assays to be performed in the shortest possible time. Assays based on different principles (competition, sandwich, etc.) can be performed in the same sample run. Analyte concentrations in patient and control samples are determined by comparing their absorbances with standard absorbances after statistical manipulation using the Rodbard 4-parameter calculation model (1). COMPARISON STUDY AND INSTRUMENT VALIDATION

The comparison study was conducted using serum specimens collected during morning rounds at the Roger Williams Hospital, Providence, RI. Blood was collected into Vacutainer tubes without anticoagulant and allowed to clot at room temperature. Serum

TABLE 3

Bias and Regression Data for Paired Serum Samples Analyte

z-Value

Bias

Slope

y-Intercept

r

No.

Cortisol (nmol/L) Digoxin (nmol/L) Ferritin (ng/mL) Prolactin (ng/mL) T3, total (nmol/L) T4, total (nmol/L) T4-uptake (Units) TSH (p~U/mL)

- 0.75 5.21 2.07 0.45 -0.40 9.54 8.03 1.68

Absent Present Present Absent Absent Present Present Absent

0.97 1.00 1.24 0.94 0.81 1.03 0.59 1.09

- 14.8 0.39 11.9 2.31 0.36 20.7 0.45 0.45

0.990 0.931 0.979 0.982 0.891 0.962 0.693 0.987

48 174 62 41 63 204 187 253

252

CLINICAL BIOCHEMISTRY, VOLUME 25, AUGUST 1992

COMPARISON OF BMD ES 300 WITH EIP, FPIA, AND RIA

was separated from cells by centrifugation at 1000 x g for 12 min at 4 °C. Serum specimens were stored for no longer t h a n 14 days at 0 °C before analysis. All analyses on the ES 300, and all comparative methods were performed exactly as described by the manufacturers. Isotopic assays were counted on an LKB Compugamma 1302 (Pharmacia/LKB, Paramus, NJ). ES 300 precision was assessed using Precinorm© IM and Precipath© IM quality control materials (Boehringer Mannheim Diagnostics, Indianapolis, IN). Assay linearity was determined by running manufacturers' standards as unknowns and plotting the actual v s the observed concentration.

run precision near the low limit of sensitivity (mean = 0.16, SD = 0.03, CV = 18%, n = 10). The comparisons of the data from the ES 300 and the reference methods are shown in Table 3 and displayed graphically in Figures 1 and 2. Cortisol, pro300

a

/

z . 9,54. Pm Bias; N = 204

250

/"

y - ~.7 + I .Q3x r = 0'~62

/" / / / /

200

[]

150

i

a

,,

'o"

[-

STATISTICAL ANALYSIS

511

Results from the ES 300 and from the comparative methods were compared using a procedure for quantifying bias between analytical methods (2). For our purposes, data from the ES 300 were considered "test" data, and the results from the comparative methods were considered the "reference" data. The reference and test method data were used to calculate a z-value;

o

ToCabT~q n m o l / L FP[AI IA a 14 I:

,,~

y - I~o

Z--S--~-~

a

g I

i

I

,

I

,

I

*

T i - U p t a k e IL'nil& F P I A ] /

z = l 68. No Bbts; N ~ L53 70

0/ ,'"

y = 04~* I 09~ r = O987

o

//

60

i. "~ 4o ss °

~

.30

o

10 0

211

Ill

~l

411

50

64)

70

8tl

TSH (~lU/mL. EIA) 12 z ° -0 40, No BLas; N = 63

," "'"

y = 036 + O ~ h r=0891

,"

//

Results and discussion

CLINICAL BIOCHEMISTRY, VOLUME 25, AUGUST 1992

g

o.

80

lO

ES 300 assay linearity, as assessed using manufacturers' standards, was excellent and met the manufacturers' claims (given in Table 1) in all cases. Data validating this statement are not presented but are available from the authors or from the editor. The ES 300 methods also showed excellent precision with the greatest imprecision seen in the ferritin assay at a low concentration, the coefficient of variation (CV) being 13.0% at 39.1 ng/mL. CVs for all other assays, at two levels, ranged from 3.0 to 6.7%. Interassay imprecision data for all analytes are shown in Table 2. TSH, marketed as a second generation assay, also showed acceptable within-

,'"

./

Lq~

Ii ~.

where y = the mean of the raw test method results, i~o = the mean of the raw reference methods results, s = the standard deviation of the raw test method results, and n = the number of pairs of results. This is a simplification of the Two Sample Student t-test in which only the test method data are used to estimate the standard deviation. The z-value gives a quantitative estimation of the significance of the observed bias. When z > 1.96, the test method exhibits positive bias; when z < - 1.96, the test method exhibits negative bias. Bias between methods was also assessed graphically by comparing the points to a line of identity (the line passing through the origin with a slope of 1.0). Linear regression statistics were also calculated.

,"

o a o o o o

,,"

i

o

.

II

2

4

~

S

10

12

Total-T a (nmoUL, RIA)

Figure 1 -- Graphical representation of analytical data from the comparison of the BMD ES 300 with FPIA, EIA and RIA methods for Total-T4, T4-uptake, Thyrotropin (TSH), and Total Ta. The line of identity is illustrated as (- -), and linear regression as ( ). 253

CAMARA, VELLETRI, KRUPSKI, ROSNER, AND GRIFFITHS

:

~7~ \oiha, ] 4 m . o~Tx

\

,'"

4s /

o

i

7"~1

," o

"d

o

2N) {i

2=41

5(81

7;(I

Corlisol

1~)

Inmol/L

12N)

IN:~I

RIA)

120

100

z - 0 4 5 , No Btas: N = 41 y • 231 * 0 94x r • 0982

,'" ," 0

,,""

o w .S

4o 20

0

,

I

.

i

211

,

i

4it

,

t,41

i

,

,~tl

i

,

lilt)

12o

I ' r o l a , ti/I l l l g / l ~ l l , I~I,~O

BOO

i ~ / / / ,

,,

z = 2 0 7 . P ~ Bta~.N : o ,,= 1 1 9 + l ~ 4 x

I

r = Oe'~ _

bOO

1 400

oO~ "'"

"""

200

~

~'~

](~)

I).'"

21)11

~4)11

Ivrrltin

41~)

Ingtml.

glXI

~R]

7(RI

z = 5~A, Pos Btas; N = 174 y . 0 3 9 • 100~

.'"

,, o

r - 0.931

o co o

"~

gllll

I IAI

o

""

o/

o ,'" ° ~ ° "e

o

h,,.~

e~

I

2 I)igoxin

a

4

Inml)l/L. IPIA}

Figure 2 - - Graphical representation of analytical data from the comparison of the BMD ES 300 with EIA, RIA, and FPIA methods for cortisol, digoxin, ferritin, and prolactin. The line of identity is illustrated as (- -), and linear regression as ( ).

254

lactin, total-T3, and TSH showed no methodological bias and acceptable correlation with the comparative methods. Among these analytes, the poorest correlation was seen between the ES 300 and the radioimmunoassay method for total-T 3 (r = 0.891, n = 63). Digoxin, ferritin, and total-T4 showed some positive bias (z = 5.21, 2.07, and 9.54 respectively) but acceptable correlation with the Abbott EIA and TDx methods. ES 300 T4-uptake correlated poorly with the TDx-FPIA method (r = 0.693) and showed some positive bias (z = 8.03). Although the ES 300 T4-uptake data correlated poorly with the TDx T4-uptake, the diagnostic sensitivities of both assays were comparable, but weak, when considered alongside the TSH and T4 results for the same patients. A review of both the ES 300 and Abbott TDx and EIA data from 43 patients in this study who had TSH, T4, and T4-uptake ordered concurrently, revealed 30 patients with TSH and T 4 data within the reference range, 7 distinctly hyper-, and 6 distinctly hypothyroid patients (TSH and T4 were well correlated). In all these cases, the T4uptake results from the TDx and the ES 300 were either normal or equivocal (borderline high or low). There were 13 cases for which extensive clinical data were available in a chart record. In all these cases, the laboratory results supported the clinical data. In addition to the analytical findings, we found the analyzer to be sophisticated yet simple enough so that training was accomplished quickly (5 days). Although some reagent reconstitution is necessary, the effort is minimal and the time spent is offset by the genuine hands-off fashion in which this analyzer operates: a distinct advantage over labor-intensive, manual i m m u n o a s s a y methods. Routine maintenance is also minimal: essentially, all that is required is an automated 25 min monthly cleaning performed entirely by the analyzer once instructed to do so by the operator. There is no required daily or weekly maintenance. Additionally, we experienced no downtime during the 2-month evaluation period. In all, we found the ES 300 to be a reliable, efficient, and technically excellent instrumental alternative for the high volume immunoassay laboratory. References

1. Duncan T, Engelberth L, LaBrash B. The Boehringer Mannheim ES 300 immunoassay system. J Clin Immunoassay 1991; 14: 105-10. 2. Griffiths WC, Camara PD, Diamond I, Pezullo JC. A procedure for quantifying bias between quantitative analytical methods. J Autom Chem 1986; 8: 147-50.

CLINICAL BIOCHEMISTRY, VOLUME 25, AUGUST 1992

Evaluation of the Boehringer Mannheim ES 300 immunoassay analyzer and comparison with enzyme immunoassay, fluorescence polarization immunoassay, and radioimmunoassay methods.

The ES 300 (Boehringer Mannheim Diagnostics, Indianapolis, IN) is a new automated immunoassay analyzer intended for the quantitative determination of ...
334KB Sizes 0 Downloads 0 Views