Clin Biochem, Vol. 21, pp. 311-314, 1988 0009-9120:88 $5.~,,5 Printed in Canada. All rights reserved. Copyright © 1988 The Canadian Society of Ciinicat Che~, :~

One-Step Sandwich Enzyme Immunoassay for Insulin Using Monoclonal Antibodies

W. B O R G I J M. B R I N E R , ~ N. F R A N K E N , ~ a n d A . - C H . K E S S L E R 2

1 D e p a r t m e n t o f L a b o r a t o r y M e d i c i n e , K a n t o n s s p i t a l A a r a u , S w i t z e r l a n d , a n d 2 B o e h r i n g e r M a n n h e i m G m b H , B i o c h e m i c a l R e s e a r c h C e n t e r , D - 8 1 3 2 T u t z i n g , F R G

An enzyme-linked immunosorbent assay for the measurement of insulin in human serum has been developed. The test is based on the sandwich technique with two monoclonal antibodies directed against two different epitopes of insulin using coated plastic tubes as the solid phase and horse radish peroxidase as the label. The immunoreactions are completed in one step within 2 h. The horse radish peroxidase activity bound to the tube wall is measured pho- tometrically after an additional 1 -h incubation with the substrate. The standards used cover the range from 0 to 260 mU insulin/L. Em- ploying the Enzymun-Test ® System ES 22 modular batch analyzer, the detection limit was found to be 3.7 mU insulin/L. Coefficients of variation (CV's) between 1.4-7.8% for intraassay precision and 5.6- 10% for interassay precision were obtained over the concentration range of 17-107 mU Insulin/L. The correlation between the proce- dure described here (y) and a commercially available double anti- body radioimmunoassay (x) is expressed by the following equation: y = 1.07x + 1.14 mU insulin/L.

Those procedures, however, are tedious depending oil the number of incubation steps and the bound-free per- formance. Furthermore, they suffer from additional drawbacks if they use polyclonal antibodies (6).

In contrast, the application of monoclonal antibodies offers a number of advantages associated with their homogeneity, specificity and availability.

The present communication describes a rapid and convenient one-step sandwich enzyme immunoassay for insulin quantification in human serum using two monoclonal antibodies. A preliminary report of this test has recently been published (7).

Materials and methods

KEY WORDS: insulin; enzyme immunoassay; monoclonal antibodies; evaluation.

In~oducf ion

T he clinical picture of diabetes mellitus can be char- acterized in terms of absolute or relative insulin

deficiency. Determination of insulin secretion charac- teristics provides an insight into the type of defect in glucose homeostasis and permits clinical classification into insulin-dependent (type I) and non-insulin-depen- dent (type II) diabetes mellitus (1).

Since the first description of radioimmunoassay (RIA) for the determination of insulin in the late 1950s by Berson and Yalow (2), significant improvements have been made in reliability and practicability of the measurement of this analyte. Although RIAs continue to be the most widely used assay technique, alternative labels have been introduced in insulin immunoassays in order to avoid the disadvantages inherent with the application of radio isotopes. Use of enzyme labels with photometric (3), fluorometric (4), and luminometric (5) measurement of activity in both competitive and sand- wich assay techniques (4,5) have been reported.

Correspondence: Prof. Dr. med. W. Bttrgi, Department of Laboratory Medicine, Kantonsspital Aarau, CH-5001 Aarau/ Switzerland.

Manuscript received August 17, 1987; revised March 9, 1988, accepted April 6, 1988.

REAGENTS

Enzymun-Test ® Insulin monoclonal test kit was ob.- tained from Boehringer Mannheim GmbH, D-6800 Mannheim, FRG. The package contains incubation buffer (40 mmol/1 phosphate, pH 7.0); anti-insulin-an- tibody horse radish peroxidase (HRP) conjugate (>30 U/l); substrate buffer (100 mmol/1 phosphate-citrate, pH 4.4, 3.2 mmol/1 sodium perborate); di-ammonium 2, 2-azino-bis (3-ethylbenzothiazoline-6-sulphonate) (ABTS ®) (1.9 mmol/1); 5 insulin standards in bovine serum, their concentrations ranging from 0 to 260 mUi 1; 100 plastic tubes coated with anti-insulin-antibody (binding capacity > 100 ng insulin per tube). The work~ ing solutions are prepared according to the manufac. turer's instructions.

Human monocomponent insulin (cat. No. 400), hu. man C-peptide (cat. No. 820), porcine insulin and proin-. sulin (cat. Nos. 480 and 710) and bovine insulin (cat~ No. 500) were obtained from Novo Industry, Denmark. Human proinsulin was a kind gift of Ely Lilly, USA. To determine the assay specificity bovine serum wa:~ spiked with these antigens in concentrations of 100 .... 2000 pmol/1.

The standards have been calibrated against the In. ternational Reference Preparation of Insulin, humar: for Immunoassay 66/304.

APPARATUS

The assay was performed on Enzymun-Test ® Syster:~ ES 22, a modular batch analyzer for semi-automated~

CLINICAL BIOCHEMISTRY, VOLUME 21, OCTOBER 1988 31L

BORGI, BRINER, FRANKEN, AND KESSLER

analysis (Boehringer Mannheim GmbH) (8). The sys- tem includes a pipett ing and rinsing uni t with a trans- port chain, a filter photometer, a flow-through cuvette, and a desk-top computer (Epson HX20).

1.4

S P E C I M E N S 1.2

Venous blood was drawn in Vacutainer ® tubes and allowed to clot. The serum obtained after centrifugation was stored at minus 80°C unti l used for this study. There was no special selection of the patients.

For the assessment of intraassay precisions and daily quali ty control, the control sera Precinorm ® IM, Pre- cipath ® IM (Boehringer Mannheim GmbH, cat. Nos. 649945 and 649937), CON 6 immunoassay Tri-Level control (Diagnostic Product Corporation) and Omega ® Ligand control sera (Cooper Biochemical) were used.

1.0

E f" .8

O

¢i .6

Assay procedure

Samples and standards (100 ~l) in duplicate were incubated with 1 ml anti-insulin-conjugate solution in the coated tubes for 2 h at room temperature.

The tube contents were aspirated and the tubes rinsed 3 times with tap water. 1000 ~1 of substrate buffer containing ABTS ® was added to the tubes and incubated for 1 h at room temperature.

Absorbance was read at 405 nm and insulin concen- trat ions were calculated from the calibration curve. A new calibration curve for each assay was constructed.

Evaluation protocol and statistics

Intraassay and interassay precision in human sera, control sera, and standards were calculated and ex- pressed in mean values (~), s tandard deviations (sd), and coefficients of variat ion (CV).

The lower detection limit of the assay was determined by performing 20 single measurements of the zero-stan- dard and calculating the average of the absorbance plus the two-fold and three-fold s tandard deviations (9).

The comparison with the commercially available ra- dioimmunoassay was carried out using the biometric regression procedure established by Passing and Bab- lok (10).

Results

Figure i represents the s tandard curve, which shows a l inear relationship between OD and insulin concen- t rat ion of at least 266 mU/L. The precision of all stan- dards was assayed in one run (n = 10). The absorbances of the zero s tandard showed a CV of 8%, the standards with the concentration of 17.5, 78.5,164.5, and 266 mU/ L were measured with CV's of 3.9, 1.6, 3.4, and 3.8% respectively.

The results of the intraassay precision are presented in Figure 2. Human serum pools and commercial con- trol sera spanning the range from 8.5 to 218 mU in- sulin/L have been determined. With the exception of two sera (CV's of 7%) all specimen have been measured with CV's less than 4.5%.

312

.4

.2

o ~ 0 50 100 150 200 250

Insulin concentration [mU/I]

Figure 1--Standard curve of Enzymun-Test ® Insulin. Each point represents the mean of 10 single determinations on each standard solutions containing 0, 17.5, 78.5, 164.5, and 266 mU/L of human insulin. The vertical lines indicate +-2 SD.

8 - -

_

_

_

0

0

. • • 0

0 0

e

I I / / ' I 50 100 220

INSUL IN C O N C E N T R A T I O N ( m U / I )

• H U M A N S E R A o C O N T R O L S E R A

Figure 2--Intraassay precision of Enzymun-Test ® Insulin. Each point represents the mean of 10 single determinations. The insulin concentration is shown on the x-axe, the coeffi- cient of variation are demonstrated on the y-axe.

CLINICAL BIOCHEMISTRY, VOLUME 21, OCTOBER 1988

INSULIN EIA

TABLE 1 Interassay Precision of Enzymun-Test ® Insulin

N series Control sera duplicates mU/1 sd CV%

BM 1 16 17.6 1.83 10.4 DPC 13 22.2 1.82 8.2 BM 2 16 107.1 5.96 5.6

Note: The serial determinations were performed on consecutive working days.

The interassay precision was determined from du- plicate determinations of three commercial control sera, over a period of 13 to 16 days (Table 1). Mean values of 17.6, 22.2, and 107.1 mU/L were obtained and the corresponding CV's were found to be 10.4, 8.2, and 5.6%.

Dilution experiments using human sera of different insulin concentrations, control serum, and four stan- dard solutions (Figure 1) revealed the results shown in Table 2. The lowest recoveries were found in the ex- periments with human serum diluted with control serum, the other experiments resulted in recovery val- ues between 95.2 and 104.5% which confirms the lin- earity of the assay procedure.

The lower detection limit was assayed in five iden- tical experiments at different times. These determi- nations revealed mean variations of the zero standards corresponding to values between 2.4 and 3.7 mU/L (Ta- ble 3). The highest ~ + 3 sd value (3.7 mU/L) resulted from experiment No. 5. Thus, it appears reasonable to assume a lower detection limit of 3.7 mU/L.

The studies on the specificity of the insulin assay demonstrated a reactivity with porcine insulin approx- imately three fold higher than that of human insulin (Figure 3). Human and porcine proinsulin as well as bovine insulin were found to be almost three fold less reactive than human insulin. It should be added that

TABLE 3 Sensitivity of Enzymun-Test ® Insulin

Absorbances Concentration Experiment zero standard Insulin [mU/1]

No. X(n = 20) sd X + 2sd X + 3sd

1 0.025 0.027 1.6 2.4 2 0.034 0.037 1.8 2.8 3 0.033 0.037 1.9 2.9 4 0.032 0.036 1.9 3.0 5 0.033 0.037 2.35 3.7

Note: Each of 5 identical experiments carried out at different times comprised 20 intra-assay insulin determinations of the zero standard.

no immunoreact ivi ty was obtained with human C-pep- tide.

A correlation study between Enzymun-Test ® Insu- lin and a commercially available radioimmunoassay was performed assaying 60 human sera. The correla- tion was shown to be highly significant (r = 0.96; y = 1.07x - 1.14 (Figure 4).

D i s c u s s i o n

The present paper describes a new test procedure for the quanti tat ive determination of human serum insulin with a total incubation time of 3 h at room temperature. The assay features a one-step sandwich enzyme im- munoassay using two different monoclonal antibodies (7). One of these antibodies is bound to the inner tube wall, while the other is conjugated with horse radish peroxidase and solubilized in buffer. The monoclonal antibodies have been shown to be directed against two different epitopes of insulin (11).

The mechanization offered by the automatic system Enzymun-Test ® System ES 22 with the computer con- trolled incubation, pipetting and rinsing steps as well

TABLE 2 Recovery in Dilution Experiments

Sample dilution steps Expected values [mU/1] Measured values [mU/1] Recovery [%]

Human serum/human serum 14.2 mUll/64.4 mU/1

1 + 0.5 30.9 29.6 95.8 1 + 2 47.7 46.3 97.1 1 + 5 56.0 54.9 98.0

Human serum/human serum 37.3 mU/1/186 mU/1

1 + 0.5 86.9 90.8 104.5 1 + 2 136.4 141.7 103.9 1 + 5 161.2 165.0 102.3

Human serum/control serum 20.7 mU/1/103.6 mU/1

1 + 1 62.2 57.8 92.9 1 + 3 82.9 75.8 91.4 1 + 5 89.8 89.5 99.7

Standards c/d 1 + 1 121.5 126.0 103.7 b/e 1 + 1 141.8 144.0 101.6 d/e 1 + 1 215.3 205.0 95.2

Note: Recovery Studies of Enzymun-Test ® Insulin. The standard solutions used were the same as in Figure 1.

CLINICAL BIOCHEMISTRY, VOLUME 21, OCTOBER 1988 313

BIJRGI, BRINER, FRANKEN, AND KESSLER

E e-

t O O q¢

d 6

2.25 Porcine Insulin

2.0

1.75 Human Insulin

1.5

1.25

1.0 Porcine Proinsulin

j Human 0.75 Proinsulin

Bovine Insulin 0.5

as evaluation of measured absorbances allows assay of 150 samples in one run performed by one person. In separate experiments three different sera were ana- lyzed in duplicate by the manual procedure. The pre- cision was found to range between 4.1 and 6.9% CV. When the same sera were analyzed by the ES 22, the corresponding CV's were from 2.7 to 4.9%.

With the present sandwich enzyme immunoassay, lower detection limit of 3.7 mU/L was found. This value represents a mean of five separate experiments. Values determined by the enzyme immunoassay (y) and by double antibody radioimmunoassay (x) correlated highly (r = 0.96, y -- 1.07x - 1.14; p < 0.05, N = 60) which demonstrates the reproducibility in different techniques.

R e f e r e n c e s

0.25 ~ C-peptide

120 500 1000 1500 2000 pMol/I

Figure 3--Crossreactivity of different insulins with Enzy- mun-Test ® Insulin.

1 0 0 -

90 w

~E 80 -

70-

50- I,u 4 0 -

~ 2 0 -

1 0 -

/ . / "

I I I I I I I I i 0 10 20 30 40 50 60 70 80 90 100

RIA INSULIN [mUll]

Figure 4--Comparison of Enzymun-Test ® Insulin (y) with RIA insulin (x) in 60 human sera.

1. Rausch-Stroomann JG. Die Bestimmung der Insulin- sekretion und ihre diagnostische Bedeutung. Diagnose und Labor 1986; 36: 133-9.

2. Berson SA, Yalow RS. Assay of plasma insulin in human subjects by immunological methods. Nature (London) 1959; 184: 1648-9.

3. Yoshioka M, Taniguchi H, Kawaguchi A, et al. Evalua- tion of a commercial enzyme immunoassay for insulin in human serum and its application. Clin Chem 1979; 2511: 35-8.

4. Imanaga M, Hashida S, Ishikawa E, et al. A highly sen- sitive sandwich enzyme immunoassay for insulin in hu- man serum developed using capybara anti-insulin FA'- horseradish peroxidase conjugate. Analytical Letters 1983; 16: 1509-23.

5. Strasburger CJ, Fricke H, Gadow A, Klingler W, Wood WG. Luminescence immunoassays--Alternativen zum Radioimmunoassay. Artzl Lab 1983; 29: 75-82.

6. Hubbuch A. Enzyme-immunoassays: a review. In: Prog- ress in clinical biochemistry and medicine. Vol. 4. Pp. 109-143. Berlin, Heidelberg: Springer Verlag, 1986.

7. Franken N, Haug H, Deeg R, Wahlefeld AW. A new one- step enzyme immunoassay for insulin. Clin Chem 1986; 32: 1066.

8. Meyer HD, Braun SL. Mechanisierung von heterogenen Enzym-immunoassays mit einem modular aufgebauten Batch-Analyzer. Arztl Lab 1985; 31: 308-12.

9. Kaiser H. Zum Problem der Nachweisgrenze. Z Analyt Chem 1962; 209: 1-18.

10. Passing H, Bablok W. A new biometric procedure for test- ing the equality of measurements from two different an- alytical methods. J Clin Chem Clin Biochem 1983; 21: 709-20.

11. Schroer JA, Bender T, Feldmann RJ, Kim KJ. Mapping epitopes on the insulin molecule using monoclonal anti- bodies. Eur J Immunol 1983; 13: 693-700.

314 CLINICAL BIOCHEMISTRY, VOLUME 21, OCTOBER 1988