A monoclonal antibody-enzyme immunoassay for, serum ...

Proc. NatL Acad. Sci. USA Vol. 80, pp. 3470-3474, June 1983

Medical Sciences

A monoclonal antibody-enzyme immunoassay for, serum carcinoembryonic antigen with increased specificity for carcinomas (nonspecific crossreacting antigens 1 and 2/conventional radioimmunoassay)

ANDERS HEDIN*, LEIF CARLSSONt, ASTA BERGLUNDt, AND STEN HAMMARSTROM* *Department of Immunology, University of Stockholm, S-106 91, Stockholm, Sweden; and tPharmacia Diagnostics AB, Box 17, S-751 03, Uppsala, Sweden Communicated by Werner Henle, February 3, 1983

MATERIALS AND METHODS

ABSTRACT A two-site monoclonal antibody-enzyme immunoassay (MEIA) for carcinoembryonic antigen (CEA) was developed that uses two monoclonal anti-CEA antibodies, which recognize two different epitopes in the peptide moiety of CEA. The assay was sensitive to 0.5 pug/liter and had a measuring range of 0.5-200 pig of CEA per liter. It was highly specific inasmuch as none of three known CEA-related substances, "nonspecific crossreacting antigens 1 and 2" (NCA-1 and NCA-2) and biliary glycoprotein I (BGP I), reacted in the assay. NCA-2 (meconium antigen) is very similar to CEA. None of five commercially available CEA assays were able to differentiate between CEA and NCA-2. With one exception (colon), normal tissue extracts did not react in the MEIA even when tested at very high concentrations. Sera from a total of 180 healthy individuals and patients with malignant and nonmalignant disease were analyzed for CEA levels by using the MEIA and in parallel a conventional radioimmunoassay. A significant increase in specificity for carcinomas was obtained with the MEIA. This was essentially due to a decrease of MEIA CEA values in sera from patients with nonmalignant disease. The CEA values in the group of carcinoma patients (colon, pancreas, lung, and breast) were the same in the two assays.

Monoclonal Antibodies (MAbs). Four anti-CEA MAbs-9, 27, 3851, and 46 (4)-were used in the study. They were all of the IgG1 K subclass and recognized three different epitopes in the peptide moiety of the CEA molecule; epitope A, MAb 9; epitope C, MAb 27; epitope D, MAb 38S1 and MAb 46 (4). Both epitope C and epitope D are present at least twice in the CEA molecule, whereas epitope A appears to be present only once (4). The affinity constants for the interaction between CEA and MAbs 9, 27, and 38S1 were 7.4 X 108, 3.3 X 108, and 1.2 X 108 M-1, respectively (4). The value for MAb 46 has not been determined. None of the antibodies reacted with NCA or biliary glycoprotein I (BGP I) (4). Highly purified MAb was obtained by passage of ascitic fluid over an immunoadsorbent column consisting of CEA conjugated to cyanogen bromide-activated Sepharose 4B (Pharmacia, Uppsala, Sweden). Antibodies were eluted with 6 M guanidine HCl. CEA and Related Materials. CEA, low molecular weight CEA (CEA10,), and NCA (5, 6) were purified from individual liver metastases of colorectal cancer as described (7, 8, 9). The antigens were homogeneous on gel filtration and gave a single relatively broad band on sodium dodecyl sulfate/polyacrylamide gel electrophoresis (7, 8, 9). Highly purified BGP I was prepared from hepatic bile of nonicteric patients as described (10). NCA-2 was purified from pooled meconium (77 g wet weight) of nine infants essentially as described by Matsuoka et al. (3). Briefly, meconium extract was precipitated with 1.0 M perchloric acid and the acid-soluble fraction was applied to a calf antiCEA immunoadsorbent. After washing, semipurified NCA-2 was eluted from the adsorbent with 0.1 M glycine'HCl, pH 2.3. The eluate was then fractionated on a Sepharose 6B column. Most of the CEA-like activity was recovered in two peaks. The elution volume of the material in the first peak was similar to that of CEA. It did not react with specific anti-NCA serum and must therefore be NCA-2 as defined by Matsuoka et al. (3). The material in the second peak reacted with specific NCA serum. Purified NCA-2 was also obtained from P. Burtin (Villejuif, France) (2). Highly purified orosomucoid (al acid glycoprotein) was a gift from H. P. Ekre (Kabi, Stockholm). The perchloric acid-soluble (1.0 M) fraction from the following human tissues or secretions was prepared: liver metastases of colorectal cancer (3 samples), urinary bladder carcinoma (1 sample), prostate carcinoma (1

Although existing carcinoembryonic antigen (CEA) assays can be used successfully for monitoring treatment of cancer and as an independent prognostic tool, they have limited value as an independent test to establish the diagnosis of cancer (1). An important goal in current research on CEA is therefore to improve the cancer specificity of the test. If tumor CEA is chemically different from the molecules responsible for CEA activity in the blood of healthy individuals and patients with various nonmalignant diseases, it should be possible to make monoclonal antibodies against such "discriminating areas" in CEA. Such antibodies may then be used to construct CEA assays with higher disease-related specificity. Three CEA-crossreactive normal tissue components have been identified so far. One of these, nonspecific crossreacting antigen 2 (NCA-2), present in relatively high concentrations in meconium (2, 3), is very closely related to tumor CEA. In fact, most conventional anti-CEA sera do not discriminate between these two macromolecules. We have recently produced a series of monoclonal antibodies against CEA, several of which were of high affinity and directed against different conformation-dependent epitopes of the CEA molecule (4). In this communication we report the-development of a twosite monoclonal antibody-enzyme immunoassay (MEIA) for CEA that shows an increased disease-related specificity for carcinomas compared to a conventional radioimmunoassay (RIA).

Abbreviations: CEA, carcinoembryonic antigen; NCA, nonspecific crossreacting antigen; BGP, biliary glycoprotein; MEIA, monoclonal antibody-enzyme immunoassay; MAb, monoclonal antibody; RIA, radioimmunoassay; ELISA, enzyme-linked immunoadsorbent assay.

The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact. 3470

Proc. Natd Acad. Sci. USA 80 (1983)

Medical Sciences: Hedin et d sample), normal adult colon (1 sample, 5 cm from tumor), normal adult spleen (pool of 20 spleens), normal adult lung (pool of 3 lungs), normal adult liver, muscle, bile, and feces (1 sample each), and meconium (pool from 9 individuals). The material was dialyzed and lyophilized. Saline extracts from normal human spleen, liver, and muscle were also prepared. Patients' Sera. Serum samples from 180 men and nonpregnant women were studied: 40 sera from healthy individuals (20 smokers and 20 nonsmokers), 45 sera from patients with nonmalignant liver disease (cirrhosis associated with alcohol abuse, n = 27; chronic hepatitis, n = 12; Wilson disease, n = 4), 25 sera from ulcerative colitis patients, and 70 sera from patients with malignant disease. The diagnosis and number of cancer patients were as follows: colorectal carcinoma (n = 25, metastases in 13), pancreatic carcinoma (n = 15, metastases in 8), lung carcinoma (n = 15, metastases in 8), breast carcinoma (n = 5, metastases in 1), melanoma (n = 4, metastases in 2), sarcoma (n = 4, metastases in 1), lymphoma (n = 2). In the comparative analyses sera were diluted 1:5 in phosphate-buffered saline, pH 7.4, containing 5% porcine serum prior to assay. MEIA for CEA. Systemic testing of different combinations of the four MAbs-9, 27, 38S1, and 46-showed that the most sensitive assay for CEA was obtained when MAbs 38S1 and 27 were used in combination. These antibodies recognize two different epitopes (Fig. 2). MAb 38S1 in the form of unpurified ascitic fluid was coupled to the solid phase. Four-tenths milliliter of ascitic fluid containing approximately 0.5 mg of MAb was used per g of CNBr-activated paper discs (Pharmacia, Uppsala, Sweden). Coupling was performed in 0.1 M NaHCO3 at 40C for 18 hr in rotating (150 rpm) test tubes. Unreactive groups were blocked with 1 M ethanolamine. The enzyme conjugate was prepared by coupling immunoadsorbent-purified MAb 27 to f3-galactosidase from Escherichia coli (type VI, Sigma) via thiol groups as described by Carlsson et aL (11). Efficient conjugation was obtained with an enzyme-to-MAb ratio of approximately 2. The MEIA was performed as follows: 100 ,ul of serum (or standard CEA) was added to a test tube containing one MAb 38S1-coated paper disc. Then 50 ,ul of MAb 27-enzyme conjugate (2.0 ,ug/ml in phosphate-buffered saline containing 3% bovine serum albumin, 1% normal mouse serum, and 0.5% Tween 20) was added and the reaction mixture was incubated with shaking (1,400 rpm) at 20°C overnight. After three washings with saline, 200 A1 of o-nitrophenyl-,B3D-galactopyranoside (2.1 mg/ml, Fluka AG, Buchs, Switzerland) was added. The tubes were then incubated at 370C for 30-100 min and the reaction was stopped by the addition of 1.0 ml of 0.43 M Na2CO3. The amount of colored product was measured by absorbance at 420 nm. Other CEA Immunoassays. Sera tested in MEIA were also assayed in parallel with a conventional sandwich RIA, Phadebas CEA PRIST (Pharmacia, Uppsala, Sweden). This assay uses specific sheep and rabbit antisera and has a lower limit of sensitivity of =1 jg of CEA per liter. Sera were diluted 1:5 in phosphate-buffered saline plus 5% porcine serum before anal-

SOLID PHASE

-t CEA

3471

MAB 27

FIG. 1. Principle of two-site MEIA. C and D denote groups of independent epitopes on CEA. aGAL, P-galactosidase.

say (ELISA) (12) using antigen-coated microtiter plates (2 Ag/ ml) and a conjugate of rabbit anti-mouse Ig and alkaline phosphatase.

RESULTS MAbs Against CEA. None of the four MAbs crossreacted with NCA or BGP 1 (4). Their reactivity against NCA-2 was now investigated by using ELISA. Microtiter plates were coated with CEA or NCA-2, and the reactivities of different dilutions of the MAbs against the two antigens were compared. MAbs 9 and 27 reacted equally well with CEA and NCA-2, indicating that antigenic sites recognized by these two antibodies-i.e., epitopes A and C (4)-also are present in NCA-2. In contrast, the reactivity of MAbs 38S1 and 46 (both probably recognizing the same epitope, D) against the two antigens was markedly different. The titers against NCA-2 were 1/10 and 1/100 of those against CEA, respectively. Furthermore, MAb 38S1 gave a lower maximal uptake against NCA-2 than against CEA. These findings indicate that the antigenic site in NCA-2 reactive with these antibodies is different from epitope D. Two-Site MEIA. A sensitive assay was obtained when MAb 38S1 was used on the solid phase and MAb 27 was conjugated to the enzyme (Fig. 1). The standard curve is shown in Fig. 2. The assay was sensitive to as little as approximately 0.5 Iug of CEA per liter and the measuring range was 0.5-250 pig/liter. Only at very high CEA concentrations (e.g., >5,000 ,g/liter) was the "hook-effect" (13) observed.

2

-

4

-3 C

c

0 .-

0

C4

0f

1-

-2 N0

1

ysis.

The influence of CEA-related normal tissue components, notably NCA-2, in five commercial CEA immunoassays was also studied. The following assays were investigated: CEA Roche Test Kit (Z-gel assay) and CEA EIA-Test (sandwich test using one MAb), both from Hoffmann-La Roche (Basel, Switzerland); Abbott CEA RIA Diagnostic Kit (sandwich assay) (Abbott Laboratories); Serono Diagnostic CEA Kit (double antibody assay) (Serono, Rome); and Phadebas CEA PRIST. The antibody activity of the MAbs against CEA and NCA-2 was also measured by an enzyme-linked immunoadsorbent as-

L14I 0.5 1.5

5

15

50

51 02

5-1 03

5-1 04

CEA, ,ug/liter

FIG. 2. Dose-response curve for CEA in the MEIA. The values obtained in the normal measuring range 0.5-200 pg of CEA per liter are shown on the left. The curve on the right shows the so-called "hook-effect" obtained at high CEA concentrations. Vertical bars give the standard deviations calculated from four separate experiments run in du-

plicate.

3472

Medical Sciences: Hedin et al

Proc. Nad Acad. Sci. USA 80 (1983)

CEA 52 |

2.0

E

Prostate Ca. HCI04

NCA 1

102

10

104 103 Antigen, ng/ml

Normal spleen I-IC104 lung HC104 liver HC104

BGPI"2 ior

106

FIG. 3. Dose-response curves for CEA and related substances as well as for 1 M perchloric acid (HC104) extracts of normal and tumor tissues in the MEIA. Ca., carcinoma.

To determine the within- and inter-assay coefficients of variation, 21 sera containing different concentrations of CEA were analyzed in quintuplicate on five different occasions. The sera were diluted 1:5 before assay. At low CEA levels (4 ,ug/liter) the interassay variation was 24.2%; at intermediate levels (24 /Lg/liter) the interassay variation was 12.9%; and at high levels (149 .tg/liter) it was 2.8%. The within-assay variation was only slightly lower than the interassay variation. The effect of dilution was investigated by assaying 10 serum

samples containing different concentrations of CEA in a 2-fold dilution series. The calculated CEA values were found to be independent of dilution in all sera tested. The specificity of the MEIA was analyzed by investigating whether purified CEA-crossreactive substances and various tissue extracts gave uptake curves similar to those of CEA. The results are shown in Fig. 3 and Table 1. They can be summarized as follows: (i) All four individual CEA preparations investigated gave identical uptake curves. (ii) Three individual

Table 1. CEA activity of purified CEA-related substances and extracts of normal and tumor tissues as determined by MEIA CEA, CEA, Material Material ,g/mg dry weight Ag/mg dry weight Liver metastases of CEA colorectal carcinoma 1,000 Preparation 47 200 Preparation 45t 1,000 Preparation 48 120 Preparation 48t 1,000* Preparation 50 80 Preparation 50t Preparation 52 1,000

CEA10w Preparation 27 Preparation 41 Preparation 45

850 615 800

NCA Preparation 41 Preparation 50