Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
(CANCER RESEARCH 50, 6229-62.14. October 1. 1990| An Enzyme-linked Immunosorbent Assay for Cancer Procoagulant and Its Potential as a New Tumor Marker1 Stuart G. Gordon2 and Barbara A. Cross Departments of Medicine fS. G. G., B. A. C.] and Pathology fS. G. G.J and University of Colorado Cancer Center fS. G. G./, University of Colorado Health Sciences Center. Denver, Colorado 80262 ABSTRACT Cancer procoagulant (CP) is a M, 68,000 cysteine proteinase that initiates blood coagulation and is expressed by a variety of malignant cells but not by normally differentiated cells. Polyclonal immunoglobulin G and monoclonal immunoglobulin M antibodies were developed to purified CP and used to develop an enzyme-linked immunosorbent assay to analyze the antigen in human serum samples. The purpose of this preliminary study was to determine whether or not the analysis of CP in the serum might be a useful tumor marker. Pure CP was added to normal serum to establish a quantitative standard curve; the correlation coefficient of seven standard curves was 0.99. The upper limit of the normal range was established with 46 normal sera (mean ±2 SD = 0.57 /ig/ml). A total of 128 blinded serum samples were analyzed: 54 were from cancer patients (29 with gastrointestinal cancer, 22 with lung cancer, and three with urogenital cancer); 20 were from benign disease patients; and 54 were from normal individuals. All of the 13 early stage cancers were >0.57 Mg/ml(positive), 31 of 41 (76%) of the late stage cancers were positive; overall, 44 of the 54 cancers (81%) were positive. Forty-nine of 54 (91%) of the normal sera and 16 of 20 (SI)'/ ) of the benign disease sera were negative. Overall, the assay had a sensitivity of 81% and a specificity of 88%. INTRODUCTION Despite many therapeutic advances, early detection of malig nancy has great potential as a means of affecting outcome and survival of cancer patients. All too often we find the disease too far advanced and therapeutic options limited. Earlier diagnosis and treatment can affect outcome as evidenced by screening Papanicolou smears for squamous cell carcinoma of the cervix (1). The development of an assay with similar potential for early detection of other malignancies would be of considerable aid to clinical oncologists. Recently, investigators have identified sev eral substances that are produced by tumor cells for use as diagnostic markers of cancer. These include the germ cell markers human choriogonadotropin and a-fetoprotein, which are very specific markers, and less sensitive and specific markers such as CEA,3 CA-12.5, CA-19.9, CA-15.3, PAP, and others. Recent reviews describe the use of these and other tumor markers in diagnostic and therapeutic applications (2-6). During our studies to identify a substance that initiated the Received 1/23/90; accepted 7/5/90. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1This work was supported by a grant from the National Cancer Institute, NIH, contracts from Western Biomedicai Development Corp. and E. I. DuPont deNemours & Company, and a gift from L. G. Hickey. 2To whom requests for reprints should be addressed. 3The abbreviations used are: CEA, carcinoembryonic antigen; PAP, prostatic acid phosphatase; CP, cancer procoagulant; ELISA, enzyme-linked immunosor bent assay; IgG, immunoglobulin G; IgM, immunoglobulin M; SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis; PAGIF, polyacrylamide gel isoelectric focusing; HPLC, high-pressure liquid chromatography: CV, coefficient of variation; COPD, chronic obstructive pulmonary disease; GI, gas trointestinal; GU. gastrourogenital; PBT, 0. l Mphosphate buffer (pH7.4) contain ing 0.05 % Tween -20. abnormal blood coagulation associated with malignant disease, we discovered (7), purified, and partially characterized CP (8, 9). To determine the distribution of CP activity in various malignant and normal tissues, a variety of extracts of serumfree culture media and cell suspensions of malignant or trans formed cells as well as their normal counterparts were examined (10-13). It was found that CP activity existed in the extracts of malignant cells, but not in extracts from normal tissue or serumfree media from normal cells in tissue culture. Careful enzy matic and immunological characterization of extracts of malig nant and benign melanocytic tissue revealed CP in the mela noma extracts and not in the benign nevi extracts (14). In acute nonlymphocytic leukemia, CP was identified in most of the cytological subtypes but not in control mononuclear cells from either peripheral blood or bone marrow aspirates (15). In ad dition, a virtually identical procoagulant protein has been pu rified from human amnion-chorion tissue (fetal origin), imply ing that CP may be an oncofetal protein (16, 17). These data were sufficiently suggestive of the unique expression of CP by malignant cells to warrant the preliminary study to evaluate the potential of CP as a tumor marker. A double antibody "sand wich" ELISA was developed and used to assay for CP in the serum of cancer patients and of noncancerous control individ uals. These preliminary results suggest that CP has potential as a tumor marker, and they are sufficiently encouraging to warrant additional work on the assay and to expand the study. MATERIALS AND METHODS Purification of Cancer Procoagulant Antigen. Cancer procoagulant was purified from extracts of rabbit V2 carcinoma and human amnionchorion by immuno- and p-chloromercurial benzoate affinity chroma tography as described previously (9). Briefly, CP was purified by immunoaffinity chromatography in which the immunoaffinity resin was prepared by coupling purified goat polyclonal IgG to cyanogen bromideactivated Sepharose. Extracts were equilibrated overnight at 5°C,the column was washed thoroughly with 20 mivi barbila! buffer (pH 7.4), and bound antigen was eluted with 3 M NaSCN (Sigma, St. Louis, MO). Trace contaminants were removed by adsorbing CP to an organomercurial resin, washing contaminants off, and eluting CP with 2 HIM and 5 mM HgCl2. The CP sample was a single band on SDS-PAGE and PAGIF. It had the enzymatic properties of CP, including procoagulant activity in normal and Factor Vll-deficient human plasma (7, 8); it directly activated human Factor X in a 2-stage assay (8); and its activity was inhibited by mercury, iodoacetamide, and peptidyl diazomethyl ketone and was reactivated by KCN (8, 9, 18). The purified CP preparation was used to immunize mice and to screen hybrid cells for antibody. Polyclonal Antibodies to Cancer Procoagulant. The polyclonal goat antibody was raised by standard methodology (9). Briefly, the goat was immunized and boosted with s.c. injections of between 50 and 100 ^g of purified enzyme from rabbit V2 carcinoma suspended in Freund's adjuvant. The goat IgG fraction was partially purified from the serum by standard purification techniques, including ammonium sulfate fractionation and DEAE-cellulose ion exchange chromatography. A trace contaminant of an antibody to normal rabbit serum protein was re moved by coupling normal rabbit serum proteins to cyanogen bromide- 6229 Downloaded from cancerres.aacrjournals.org on August 3, 2017. © 1990 American Association for Cancer Research. ENZYME-LINKED IMMINOSORBENT activated Sepharose and passing the antibody preparation over this "normal rabbit serum column." Potential cross-reactivity to normal human serum proteins was eliminated in the same way, by adsorbing the antibody preparation to human serum protein-resin. The purified antibody was tested for cross-reactivity with normal rabbit serum and normal human serum by crossed immunodiffusion; it was not innminoreactive with either. Hybridoma Antibodies to Cancer Procoagulant. BALB/c mice (Charles River) were immunized with cancer procoagulant to induce a B-cell immune response, according to standard procedures (19, 20). Briefly, spleen lymphocytes from immunized mice were hybridized with the P3/X63-AG8.653 variant of mouse myeloma cells with 50% poly ethylene glycol (21). The hybrids were plated with 2 x 10' normal mouse spleen cells/ml as a feeder layer, and the unhybridized myeloma cells were eliminated by hypoxanthine aminopterin thymidine suicide. Cells from antibody-positive wells were expanded and cloned at 0.1 to 0.5 cells per well with a feeder layer of normal spleen cells. Antibodies from positive wells were classified using specific subclass anti-mouse immunoglobulins according to the technical information provided in the Zymed Lab Mono Ab-ID EIA kit (22). Antibodies were screened by a standard single antibody microtiter plate immunoassay using purified CP absorbed to microtiter wells. Alkaline phosphatase-labeled anti-mouse IglVlantibody was used as the signal antibody, and the color was read at 405 nm on a Dynatech microtiter plate reader. Antibody Production and Purification. The anti-CP IgM was produced in the ascites of Pristane-treated BALB/c mice (23). During the analysis of immunoglobulin in the ascites fluid, procoagulant activity was dis covered in the cell-free ascites. This procoagulant had the characteristics ascribed to cancer procoagulant, including inhibition by mercury, ini tiation of coagulation of Factor VII-depleted human plasma, and immunoprecipitation with the polyclonal goat antibody (9). To effectively utilize the antibody in immunoassays or other immunological experi ments, it was necessary to separate the antigen from the antibody. This was accomplished by making the ascites 6 M urea and running the sample over a 1.5-m agarose (Biorad, Richmond, CA) column (1.5 x 60 cm) that was preequilibrated in 6 M urea. The fractions containing the IgM protein peak were pooled, dialyzed against 3 changes of 20 imi barbila! buffered saline (pH 7.4), and concentrated over an Amicon PM-50 membrane. An alternative procedure used to separate antigen from the IgM was to make the ascites 6% Polyethylene Glycol 6000 to precipitate the IgM. The precipitate was recovered by centrifugation. decanting the supernatant, and redissolving it in a minimal volume (about I ml/10 ml of ascites) of 0.1 M phosphate-buffered saline (pH 7.4). The solubilized IgM was made 50% with ethylene glycol and stored at 5°Cuntil it was further purified by HPLC (Waters, Inc.). A semipreparative GF450-XL (DuPont) was preequilibrated in 0.1 M sodium phosphate buffer in 50% ethylene glycol, up to 2 ml of the antibody preparation were applied and eluted at 1 ml/min, and fractions were collected to separate IgM from the antigen. The purified IgM was concentrated on an Amicon PM-50 ultrafiltration membrane, made 50% with glycerol, and stored at —20°C until it ASSAY FOR CANCER PROCOAGULANT Each step of the assay was checked to determine the optimum conditions for the assay. The dilution of monoclonal antibody was checked from 1:1 dilution to 1:50,000 dilution. Human serum albumin (0.1%) was as good a blocking agent as if not better than, the same concentration of bovine serum albumin, ovalbumin, normal goat serum, normal human serum, and normal rabbit serum. Higher (to 5%) and lower (to 0.01%) concentrations of each of the protein solutions were tested without improving the assay results. The time (1, 2, 4, and 16 h) and temperature (5°,25°,and 37°C)for coating and blocking the wells were determined as well as each incubation. A variety of washing buffers and concentrations of nonionic detergents were tested to determine the most effective way to remove substances that increased nonspecific binding. Thus, each of the conditions was carefully checked to arrive at the sequence of steps that provided the best assay system. Standard Curve and Calibration for the Assay. A standard curve was prepared by adding known amounts of pure CP to normal serum. Concentrations from 0 to 30 Mg/ml were used. Each microtiter plate contained a standard curve, 6 to 8 normal control sera and a high and a medium control pooled sera, and the unknown serum samples. Analysis of Human Serum Samples. The initial serum samples in cluded 46 normal volunteers from the Health Sciences Center that were processed to establish the normal range. The remaining samples were blinded and analyzed without knowing the patient's identity or diag nosis. After the assay was complete, the results were analyzed, and the samples were identified and compared with the results of the analysis. In the 128 serum samples that were analyzed in this blinded fashion, there were 54 sera from cancer patients, 54 from normal individuals, and 20 from individuals with benign disease (Table 1). The samples were usually analyzed in triplicate in at least 3 different assays. A small group of samples were collected in duplicate or triplicate to determine the effect of anticoagulants and hemolysis on the assay. Heparin, citrate, and NaF plasma samples had essentially the same CP values as did their serum counterparts, but the results had greater variability. EDTA-plasma samples had assay results that were about 50% lower than their serum counterparts and greater variability. Sam ples with slight hemolysis had comparable results to unhemolyzed serum, but extensive hemolysis resulted in high analysis results. Lipemic serum had high analysis results, but the addition of clearing agents, such as Liposolve, provided samples that appeared to have reliable assay values. RESULTS The identification of procoagulant activity in the ascites indicated the need to purify the antigen from the antibody. Ascites was made 6 M urea to dissociate antigen-antibody 1100 you 2 600 60 was used in an assay. Testing of the ascites IgM antibody and the purified IgM antibody clearly demonstrated that the purified IgM had greater immunoreactivity, as expected. Development of a Double Antibody Immunoassay. The wells of an luminimi II 96-well microtiter plate were coated with 0.19 ^g/well of monoclonal antibody (1:80 dilution of stock IgM) for 2 h at 25°C,and «700 'S £ „500 emptied and washed 3 times with PBT, and 50 ¿il (0.26 nM) of alkaline phosphatase-conjugated goat anti-CP antibody (24) in phosphate-buff ered saline were added and incubated for l h at 37°C.Finally the pnitrophenyl phosphate substrate was added, incubated at 37°Cfor 400 40 « 200 20° U the plate was washed 3 times with PBT. Open sites in the well were blocked with 0.1% normal human serum albumin in O.I M phosphate buffer (pH 7.6) and stored in a vacuum dessicator at 5°Cuntil it was used in an assay. Serum (25 ^1) was diluted in an equal volume of PBT In the microtiter wells and incubated for 2 h at 5°C.The wells were < 300 9 17 25 33 41 49 57 65 Fraction Number Fig. 1. An elution profile from a I.S-m agarose chromatography column showing the separation of hybridoma IgM antibody from CP antigen. Four ml of ascites were made 6 mol with crystalline urea, applied to the column that was preequilibrated in 6 M urea, and eluted with 6 M urea in 20 HIMbarbital buffer. Protein was monitored by measuring absorbance at 280 nm ( ). Fractions were dialy/cd against 20 HIMbarbital buffer, and IgM was measured in a microtiter plate assay using alkaline phosphatase-labcled goat anti-mouse IgM antibody ( ). The clotting activity was measured with the recalcification clotting time of citrated normal human plasma ( ). between 30 and 60 min, and read on a microtiter plate reader at 405 nm. 6230 Downloaded from cancerres.aacrjournals.org on August 3, 2017. © 1990 American Association for Cancer Research. ENZYME-LINKKD IMMl NOSORBKN I ASSAY FOR CANTER PRCX'OAGULANT complex. The two proteins were then physically separated on a 1.5-m agarose gel filtration column (Fig. 1) with an exclusion limit of 1.5 x IO7 daltons which can separate M, 850,000 IgM from M, 68,000 cancer procoagulant. The HPLC Chromato graphie profile of the GF-450-XL purification of IgM was essentially the same as that in Fig. 1; using ethylene glycol as the dissociating agent provided IgM with better stability and immunoreactivity, so this procedure was adopted as the method of choice. The purified IgM fractions were assayed for CP by (a) crossed immunodiffusion against polyclonal anti-CP goat antibody, (b) assaying the sample for iodoacetamide sensitivity and Factor VII-independent procoagulant activity, and (c) as saying for IgM-CP complex in a modified double antibody immunoassay by determining the reactivity of the purified IgM (containing IgM-CP) with the goat antibody in the microtiter wells and SDS-PAGE. Assay Characteristics. Each sample was assayed in triplicate on a microtiter plate, and the CV for a representative 56 sets of triplicate wells was 7.9 ±6.0%. The CV for 44 duplicate samples analyzed in the same assay was 8.6 ±5.9%. The interassay CV for 35 samples that were assayed in from 2 to 5 different assays was 13.2 ±9.4%. There were 7 pairs of samples that were blinded with different code numbers and integrated into the pool of samples; the mean CV was 18.1 ±5.8%. Hand pipetting of all steps in the assay probably contributed to the elevated CVs in the assay. A standard curve was prepared from the samples of normal serum that contained known amounts of CP. This standard curve was used to establish linearity of the assay, to quantitate the level of CP in the unknown serum samples, and to determine the reproducibility of the assay. Fig. 2 shows the mean ± standard error of the mean of 7 standard curves. The regression analysis of the data shows a correlation coefficient of 0.99. The results of each plate were analyzed by checking the linearity of the standards. The mean and one standard deviation of the 46 normal serum samples was 0.36 i¿g/m\±0.11 Mg/ml. Thus, the mean + 2 standard deviations was used to establish the normal range of 0 to 0.57 ¿ig/ml;results above 0.57 Mg/ml were considered positive. Analysis of Blinded Samples. A total of 54 normal samples from volunteers were "blinded" with code numbers and inte grated into the pool of 54 blinded cancer patient samples and 20 benign disease patient samples. Of the 54 normal sera, 49 (91%) were negative, while 16 of 20 benign disease sera were negative. In the overall noncancer population, 65 of the 74 sera (88%) were negative; i.e., the results were <0.57 ¿ig/ml. Table 1 shows some of the clinical details of the 54 cancer patients and 20 benign disease patients from whom serum samples were obtained. Fig. 3 shows the distribution of data for each of the groups of samples. In the 29 samples from patients with gastrointestinal cancers, all of which were coloTable 1 List of clinical profiles of patients that provided blood for the blinded serum samples analy:ed in the study, including information about the patient 's type and stage of cancer or the patient's type of benign disease and the number of samples in each group There were 54 cancer patient sera, 20 benign disease sera, and 54 normal sera analyzed in the study: there were 46 nonblinded normal control scrum samples used to establish the "normal range": they are not included in the tabulation of the results. Stage III Malignancies Gastrointestinal Colorectal adenocarcinoma Colorectal carcinoma Lung Squamous cell Small cell carcinoma Urogenital Prostate Ovarian 2 IV All 25 28 7 15 1 2 Benign disease Lung Asthma Pneumonia coro 4 (1 patient also had BPH)° 3 6 5 (1 patient also had asbestosis) Emphysema Other Inflammatory bowel I Deep vein thrombosis ' BPH. benign prostatic hypertrophy. 3422 •Gl 85cJO3 480 r2 =O.99 91%•• 4a'54 = 80%• & GU 78%••... 25-32 = 19/22 -86%X• 28(u O>«-~ ~^§- 2 380 .£ ° °>1 71 i1'71i_ 280 •"• ••X• ^-^ 0)o Ift O *,,c <o 0n 180 .• x"• • i-14 0x •* •rtfS&fBENIGN1670 • ." •• ••* ^ ;•• ' •' 80 • •••• 0 •X <**x• * •.*. •.•LUNG• .NORMAL -20 eo 180 270 360 450 ng 0.1 ml Cancer Procoagulant Fig. 2. The mean of seven standard curves in which pure CP antigen was added to normal human serum and analyzed by the ELISA procedure described in the text. The blank signal ut the normal serum was subtracted from each sample to calibrate the signal for each level of antigen. Each standard was assayed in triplicate on each microtiter plate, as described in the text. These standards were used to quantitate the level of antigen in the normal and blinded serum samples. The correlation coefficient was 0.99, and the equation of the leastsquares fit line was y = -9.43 + 7.69.V.Points, mean; bars. SEM. Fig. 3. Analysis of 128 cancer and noncancer sera, the results of the blinded samples analyzed with the ¡mmunoassay.The normal + 2 SDs (0.57 fig/ml) was used as the upper limit of normal. The samples with CP levels less than 0.57 >ig/ ml were considered normal, and those above that level were considered positive. The left column shows the 54 normal serum samples that were assayed, and 49 of them were normal. The second column from the left shows the results of 20 benign disease patient sera: 16 of them were normal. An ,Vin the benign disease group is a patient with asbestosis and COPD. who may have cancer. The third column from the left shows 29 gastrointestinal cancer patient sera and 3 urogenital cancer patient sera; 22 of IheGI samples and all of the GU samples were positive. Three gray tone dots with white centers represent the GU cancer samples. The last column on the right shows 22 lung cancer patient serum samples; 19 of them were positive. A' represents the early stage (Stage I and II) cancers. 6231 Downloaded from cancerres.aacrjournals.org on August 3, 2017. © 1990 American Association for Cancer Research. ENZYME-LINKED IMMUNOSORBENT ASSAY FOR CANCER PROCOAGULANT rectal cancers, there were 22 (76%) positive (>0.57 Mg/ml) sera. There were 22 serum samples from patients with lung tumors, 7 with squamous carcinoma and IS with small cell carcinoma. Nineteen of these sample (86%) were positive. All 3 sera from patients with urogenital cancer were positive. Table 2 shows the data analyzed by stage of disease. All of the Stage A and B gastrointestinal cancer patients. Stage I and II lung cancer patients, and the patient with Stage II ovarian carcinoma were positive. Thus, 13 of 13 sera (100%) from early stage cancers were positive. In contrast, only 19 of 27 (73%) of the later stage gastrointestinal cancer patient sera and 10 of 13 (77%) of the Stage III and IV lung cancer patient sera were positive. Both of the late stage urogenital cancer patient sera were positive. There were 20 serum samples from patients with benign disease; 18 of them had various types of pulmonary disease. Two patients had more than one type of benign disease; one patient with asthma also had benign prostatic hypertrophy. A second patient with COPD had a diagnosis of asbestosis and was a 2-pack/day smoker; this was one of the 4 benign disease patients that tested positive in the assay. A second positive benign disease patient had severe COPD and was also a heavy cigarette smoker. Statistical analysis of the results (25) indicates that the ability of the ELISA to correctly identify patients with cancer (i.e., the sensitivity of the test) was 76% for colorectal tumors, 86% for lung tumors, and 100% for the urogenital malignancies, with an overall sensitivity of 81%. The sensitivity for early stage cancers was 100%. The specificity of the test (the percentage of noncancer patient sera that were negative) was 88%. antigens that have been identified to date are also present in sera from some normal individuals and from a spectrum of benign disease conditions, making them inadequate for accept able detection or screening procedures. However, they have been clinically useful for monitoring cancer therapy; most no table among these are CEA and «-fetoprotein. Recently the list of tumor markers has grown substantially with the addition of PAP, tissue polypeptide antigen, pancreatic oncofetal antigen, CA-19.9, CA-12.5, CA 15.3, and others. There are recent reviews of tumor markers and tumor-associated antigens (2-6). Many of these markers have not been fully tested or widely accepted for clinical use. The list of "current tumor markers" changes frequently, and few tumor antigens hold the place of prominence and security that CEA and a-fetoprotein have held for the past decade. Carcinoembryonic antigen, first described by Gold and Freedman (26) and by Fuks and Gold (27), is a high-molecular-weight (M, 180,000 to 200,000) glycoprotein that is present in various types of malignant disease, including tumors of the GI tract (30%), stomach (72%), pancreas (88%), breast (24%), and res piratory tract (30%) (28). Elevated plasma CEA is associated with a variety of nonmalignant disorders, such as chronic obstructive pulmonary disease (57%) and most inflammatory diseases, including pneumonia (46%) and hepatitis (40%). CEA is elevated in about 10% and 15% of normal nonsmokers and smokers, respectively. a-Fetoprotein is a M, 70,000 protein (29, 30) that is elevated in the serum of about 80% of patients with hepatic tumors, almost all those with teratocarcinomas (31, 32). CP was originally found during studies to identify a substance that initiated the abnormal blood coagulation associated with malignant disease. Abnormalities of the coagulation system DISCUSSION (33-35), including disseminated intravascular coagulation (36, It is widely believed that malignant cells produce substances 37), have been repeatedly identified as pathological features of that are not produced by normal cells, but finding and identi malignant disease. Furthermore, fibrin is deposited in and fying these substances have been very difficult. With the enor around solid tumors (38-42), is associated with blood-borne mous medical promise and human effort devoted to the quest malignant cells (43, 44), and is thought to participate in various for reliable tumor markers, it is not surprising that many tumor aspects of the malignant process (45-50). antigens have been identified and partially characterized during These documented abnormalities in the coagulation system the past 20 yr in the hope of providing methods for early in malignant disease led investigators to look for procoagulants detection and therapeutic monitoring of cancer. However, many in neoplastic tissue (51-54). We identified, purified to homo substances that are found in cancer cells are also found in some geneity, and partially characterized CP, a unique cysteine pronormal and noncancerous (benign) diseases. Thus, many of the teinase (7-9, 55) that initiates coagulation by directly activating Factor X in the coagulation cascade. CP is a single polypeptide Table 2 Summary of the results of the study protein with a molecular weight of 68,000 and an isoelectric "Early" stage cancers refer to Stage I and II cancers, and "late" stage cancers point of 4.8. It is rich in serine and glycine and contains no refer to Stage III and IV cancers. The number correctly identified is in the numerator, and the total number of samples analyzed is in the denominator. For measurable carbohydrate (9). Its physical, chemical, and enzy "all" cancers samples, correct identification means a positive result (>0.57 ¿ig/ matic properties are different than other coagulation enzymes ml), and for all "noncancer" samples, correct identification means a negative (9). result (<0.57 ¿ig/ml).Calculations of the sensitivity (the ability to correctly identify patients with cancer) and specificity (the ability to correctly identify During the course of these studies, it became important to individuals that do not have cancer) are derived from these results. These values establish whether or not CP was found in association with test the efficiency of this test in accurately identifying people with cancer from normal tissues (11) and cells (10, 12); it was found in neither those that are free of disease. intact normal cells, extracts, nor culture media. Furthermore, a procoagulant protein was purified from human amnion-chorion tissue (of fetal origin), and its properties are virtually Cancer patients identical, with the exception of minor differences in amino acid (76) Gl tumors 10/13(77) 19/22(86) Lung tumors 9/9(100) composition, to CP from rabbit V2 carcinoma (16, 17), B16 1/1 (100) 2/2(100) 3/3(100) GU tumors melanoma, and all the human tumor extracts that have been TotalNoncancer 13/13(100)49/54(91)31/41 (76)All22/2944/54(81) analyzed. This suggests that the procoagulant protein is highly patients conserved across species lines and tumor types and that CP Normal may be an oncofetal protein. 16/20(80) Benign TotalEarly3/3(100)° 65/74 (88)StageLate19/26(73) During the development of the hybridoma antibody, we had °Numbers in parentheses, percentage of samples correctly identified. trouble obtaining consistent results with the screening assay 6232 Downloaded from cancerres.aacrjournals.org on August 3, 2017. © 1990 American Association for Cancer Research. ENZYME-LINKED IMMIINOSORBENT ASSAY FOR CANCER PROCOAGULANT and getting an IgM with adequate immunoreactivity from ascites samples that contained from 5 to 7 ¿/gof IgM/ml of ascites. The ascites contained a procoagulant that had all the enzymatic properties of CP (8, 9) and cross-reacted with antiCP polyclonal goat antibody on an Ouchterlony immunodiffusion plate (9). Subsequently, it found that the tissue culture medium from both the parent P3/X63-AG8.653 cell line and other hybrid cell clones from colleagues in the Health Sciences Center contained CP with immunoreactive and enzymatic iden tity. Although unexpected, it was logical since both the parent cell line and the hybrid cell lines have the malignant phenotype and should produce the CP antigen if CP is an oncofetal protein as suggested above. It was necessary to separate the antigen and IgM so that the IgM could be used in an immunoassay. We chose to take advantage of the difference in the physical size of the two molecules for the separation. Two methods were used to dis sociate CP from IgM: 6 M urea and 50% ethylene glycol; both methods facilitated dissociation so the IgM could be resolved from the CP on gel filtration chromatography (1.5-m agarose and GF-450 HPLC). However, IgM purification in 50% ethyl ene glycol resulted in more stable IgM that was free of CP. In this preliminary study, antibodies to CP were developed and used in an immunoassay to measure CP in serum from cancer and noncancer patients. The assay correctly identified 44 or 81% of the 54 serum samples from cancer patients. Ninety-one % of the normal sera and 80% of the benign disease sera were negative; 88% of the 74 noncancer sera were correctly identified. The mean + 2 standard deviations seems to be a reasonable estimator of the 90th percentile of the sample dis tribution (the limit excludes the upper 10% of the observations). Thus, the percentage of false-positive results for the normal sera is within the limits of the expected value. In a previous pilot study (56), 140 blinded serum samples were assayed from patients with colorectal (26), lung (27), breast (35), prostate (6), lymphoma (6), osteosarcoma (4), and gall bladder cancer (3) and from 70 normal individuals and 10 patients with benign disease. Approximately 90% of the sam ples were correctly identified. Thus the results of both studies are similar and support the hypothesis that CP may be a tumor marker. Comparison of these data with data from an extensive 1974 clinical trial of CEA (57) shows that CEA had a sensitivity of 72% for colorectal cancer, 91 % for pancreatic cancer, 74% for lung cancer, and 47% for breast cancer (58). CEA had low specificity (ranging from 30% to 82%) due to the large number of benign conditions that led to positive serum CEA levels. In contrast, CP had a sensitivity of 76% for colorectal cancer and 86% for lung cancers. The specificity of the CP assay was 88%. In addition, the assay correctly identified 100% of the early stage cancers. These statistics compare favorably with CEA. Thus, the data support the hypothesis that CP may be an oncofetal protein with the distribution necessary to be an effec tive tumor marker. There was a substantial percentage of false-negative cancer patient sera in both the gastrointestinal and lung cancer groups of patients with late stage cancer. It was noted that several of these "false negative" patients were terminal at the time the blood samples were obtained and that they died from their malignancy shortly thereafter. We examined the blood samples for endogenous anti-CP antibodies because such antibodies in the serum of these cancer patients would probably interfere with the immunoassay. These serum samples were reacted with pure CP on Ouchterlony-crossed immunodiffusion, and 3 of the 4 showed immunoprecipitation bands, suggesting the pres ence of endogenous antibodies. Alternatively, there is evidence for a decreased production of CP activity in the serum of rats with late stage epithelioma (59), suggesting that CP production in late stage cancers may decrease. In the false negative samples, CP may be associated with a serum protein or other substance produced by the tumor that binds CP, blocks the epitope, and inhibits the immunoreactivity of CP in the assay. The false negative results in late stage cancer will be the subject of continued studies. The objective of this study was to develop a basic immunoas say system and to obtain adequate preliminary evidence from blinded cancer and noncancer sera to test the concept that CP may be a useful tumor marker. Although we concluded that the immunoassay for CP had reasonable clinical potential as a useful tumor marker test, there were several difficulties that need to be resolved by refinements of the assay. These include the following: (a) developing a more stable monoclonal IgG hybridoma antibody to replace the IgM antibody used in the assay; (b) establishing a method to screen serum samples for endogenous anti-CP antibody; (c) doing follow-up studies on the normal and benign disease populations if they have high CP levels in their blood to determine whether or not they have malignant disease; and (<•/) studying a large number of benign diseases to develop a list of noncancerous conditions associated with elevated serum CP levels. These are questions that cannot be answered in the present study but will be examined very closely in the next phase of evaluation of CP as a tumor marker. In spite of these problems, we feel that the immunoassay for CP in serum may have reasonable clinical potential as a cancer diagnostic or monitor ing test for cancer. REFERENCES 1. Lunt, R. Worldwide early detection of cervical cancer. Obstet. Gynecol.. 63: 708-713. 1984. 2. Pohl. A. L., Francesconi, M., Ganzinger, U. C., Graninger, W., Lenzhofer, R. S.. and Moser, K. V. Present value of tumor markers in the clinic. Cancer Detect. Prevent., 6: 7-20, 1983. 3. Klavins, J. V. Advances in biological markers for cancer. Ann. Clin. Lab. Sci.. 13: 275-280. 1983. 4. Sulitzeanu, D. Human cancer-associated antigens: present status and impli cations for immunodiagnosis. Adv. Cancer Res., 44: 1-42, 1985. 5. Virji, M. A., Mercer, D. W., and Herberman, R. B. Tumor markers in cancer diagnosis and prognosis. CA. 38: 105-126, 1988. 6. Sikorska. II.. Schuster. J., and Gold. P. Clinical application of carcinoembryonic antigen. Cancer Detect. Prevent.. 12: 321-355. 1988. 7. Gordon. S. G., Franks, J. J., and Lewis, B. Cancer procoagulant A: a factor X-activating procoagulant from malignant tissue. Thrombos. Res., 6: 127137, 1975. 8. Gordon, S. G., and Cross, B. A. A factor X-activating cysteine protease from malignant tissue. J. Clin. Invest., 67: 1665-1671. 1981. 9. Falanga, A., and Gordon, S. G. Isolation and characterization of cancer procoagulant: a cysteine proteinase from malignant tissue. Biochemistry, 24: 5558-5567. 1985. 10. Gordon. S. G.. and Lewis, B. Comparison of procoagulanl from normal and transformed fibroblasts. Cancer Res., 38: 2467-2472. 1978. 11. Gordon. S. G., Franks. J. J.. and Lewis. B. J. Comparison of procoagulants in extracts of normal and malignant human tissue. J. Nati. Cancer Inst., 62: 773-776. 1979. 12. Gordon, S. G., Gilbert, L. C., and Lewis, B. J. Analysis of procoagulant activity of intact cells from tissue culture. Thrombos. Res., 26: 379-387, 1982. 13. Gordon. S. G., and Benson, B. A. Analysis of serum cancer procoagulant activity and its possible use as a tumor marker. Thrombos. Res., 56: 431440. 1989. 14. Donati. M. B., Passerini, C. G., Casali, B., Falanga. A.. Vannotti, P., Fossati, G.. Semeraro, V. and Gordon. S. G. Cancer procoagulant in human tumor cells: evidence from melanoma patients. Cancer Res., 46: 6471-6474, 1986. 15. Falanga. A.. Alessio. A. G., Donati. M. B., and Barbui, T. A new procoagulant in acute leukemia. Blood. 71: 870-875, 1988. 6233 Downloaded from cancerres.aacrjournals.org on August 3, 2017. © 1990 American Association for Cancer Research. ENZYME-LINKED IMMUNOSORBENT ASSAY FOR CANCER PROCOAGULANT 16. Gordon. S. G., Hashiba, U.. Cross, B. A.. Poole, M. A., and Falanga. A. A cysteine proteinase procoagulant from amnion-chorion. Blood. 66: 12611265, 1985. 17. Falanga. A., and Gordon, S. G. Comparison of properties of cancer procoagulant and human amnion-chorion procoagulant. Biochim. Biophys. Acta, 831: 161-165, 1985. 18. Falanga, A.. Shaw, I .. Donati, M. B., Consonni, R., Barbui. T.. and Gordon. S. G. Inhibition of cancer procoagulant by peptidyl diazomethyl ketones and pcptidyl sulfonium salts. Thrombos. Res., 54: 389-398, 1989. 19. Köhler.G., and Milstein. C. Derivation of specific antibody-producing tissue culture and tumor lines by cell fusion. Eur. J. Immunol.. 6: 511-519, 1976. 20. Yelton. D. E., Margulies. D. H.. Diamond. B.. and Scharff. M. D. Plasmacytomas and hybridomas. Development and applications. In: R. H. Kennet!, T. J. McKearn, and K. B. Bechtol (eds.). Monoclonal Antibodies, pp. 3-17. New York: Plenum Press. 1980. 21. Kohler. G., and Milstein, C. Continuous cultures of fused cells secreting antibody of predefined specificity. Nature (Lond.). 256:495-497. 1975. 22. Technical Bulletin, pp. 12 to 13. San Francisco, CA: Zymed Laboratory. Inc.. 1984. 23. Johnslone. A., and Thorpe. R. Immunochemistry in Practice, pp. 27-40. Boston: Blackwell Scientific Publications. 1982. 24. Engvali, E. Enzyme immunoassay ELISA and EMIT. Methods Enzymol., 70:419-439, 1980. 25. Archer, P. G. The predictive value of clinical laboratory test results. Am. J. Clin. Pathol., 69: 32-35, 1978. 26. Gold. P.. and Freedman, S. O. Demonstration of tumor-specific antigens in human colonie carcinomata by immunological tolerance and absorption techniques. J. Exp. Med., 121: 439-462. 1965. 27. Fuks. A., and Gold, P. Carcinoembryonic antigen. J. Tumor Oncol., I: 111,1986. 28. Beatty. J. D., and Terz. J. J. Value of carcinoembryonic antigen in clinical medicine. Prog. Clin. Cancer, 8: 9-29, 1982. 29. Alpen, E., Drysdale. J. W., Isselbacher, K. J., and Schur. P. H. Human alpha-fetoprotein. J. Biol. Chem., 247: 3792-3798, 1972. 30. Watabe. H. Purification and chemical characterization of alpha-fetoprotein from rat and mouse. Int. J. Cancer. 13: 377-388. 1974. 31. Ruddon. R. W. Tumor markers in the recognition and management of poorly differentiated neoplasms and cancers of unknown primary. Semin. Oncol., 9: 416-426, 1982. 32. Mclntire. K. R.. Waldmann. T. A.. Moertel, C. G.. and Go, V. L. W. Serum a-fetoprotein in patients with neoplasms of the gastrointestinal tract. Cancer Res., 35: 991-996, 1975. 33. Miller, S. P., Sanchez-Avalos, J.. Stefanski, T., and Zuckerman, L. Coagu lation disorders in cancer. I. Clinical and laboratory studies. Cancer (Phila.), 20:1452-1465, 1967. 34. Pochedly, C., Miller. S. P.. and Mehta. A. Hypercoagulable state in children in acute leukemia or disseminated solid tumors. Oncology. 28: 517-522, 1973. 35. Rickles. F. R.. and Edwards. R. L. Activation of blood coagulation in cancer: Trousseau's syndrome revisited. Blood. 62: 14-31, 1983. 36. Gore, J. M.. Appelbaum, J. S.. Greene. H. L.. Dexter. L.. and Dalen. J. E. Occult cancer in patients with acute pulmonary embolism. Ann. Intern. Med., 96: 556-560, 1982. 37. Rasche. H., and Dietrich. M. Hemostatic abnormalities associated with malignant disease. Eur. J. Cancer, 13: 1053-1064, 1977. 38. Franks. J. J.. Gordon, S. G., Kao. B., Sullivan. T., and Kirch. D. Increased fibrin formation with tumors and its genesis. In: R. Bianchi. G. Mariani. and A. S. McFarlane (eds.). Plasma Protein Turnover, pp. 423-440. London: Macmillan Press, 1976. 39. Ogura, T., Tsubura, E., and Yamamura, Y. Localization of radioiodinated 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. fibrinogen in invaded and metastasized tumor tissue of Walker carcinosarcoma. Gann. 61: 443-449. 1970. Dvorak, H. F., Dvorak, A. M.. Manseau. E. J.. Wilberg, L., and Churchill. W. H. Fibrin gel investment associated with line 1 and line 10 solid tumor growth, angiogenesis. and fibroplasia in guinea pigs. Role of cellular immu nity, myofibroblasts. microvascular damage, and infarction in line 1 tumor regression. J. Nati. Cancer Inst., 62: 1459-1477. 1979. O'Meara. R. A. Q., and Jackson, R. D. Cytological observations on carci noma. Irish J. Med. Sci., 391: 327-328, 1958. Day, E. D., Planinsek, J. A., and Pressman, D. Localization in viro of radioiodinated anti-rat-fibrin antibodies and radioiodinated rat fibrinogen in the Murphy rat lymphosarcoma and in other transplantable rat tumors. J. Nati. Cancer Inst., 22: 413-426, 1959. Donati. M. B.. Poggi. A., and Semeraro, N. Coagulation and malignancy. In: L. Poller (eds.). Recent Advances in Blood Coagulation, pp. 227-259. Edin burgh: Churchill-Livingstone, 1981. Wood, S., Jr. Experimental studies of the intravascular dissemination of ascitic V2 carcinoma cells in the rabbit with special reference to fibrinogen and fibrinolytic agents. Bull Swiss Acad. Med. Sci., 20: 92-121. 1964. Kinjo. M. Lodgement and extravasation of tumour cells in blood-borne metastasis: an electron microscope study. Br. J. Cancer, 38: 293-301, 1978. Millar, R. C., and Ketchem, A. S. The effect of heparin and warfarin on primary and metastatic tumors. J. Med., 5: 23-31, 1974. Hilgard, P.. and Thornes. R. D. Anticoagulants in the treatment of cancer. Eur. J. Cancer. 12: 755-762. 1976. Skolnik, G.. Alpsten. M., and Ivarsson. L. Studies on mechanisms involved in metastasis formation from circulating tumor cells. J. Cancer Res. Clin. Oncol., 97: 249-256, 1980. Colucci, M., Delaini, F., Vitti, G., Locati, D., Poggi, A., Semeraro, N., and Donati, M. B. Cancer cell procoagulant activity, warfarin, and experimental métastases,¡n:K. Hellman. P. Hilgard. and S. Eccles (eds.). Metastasis: Clinical and Experimental Aspects. Martinus Nijhoff, 1980. Donati, M. B.. Poggi, A., Mussoni, L., de Gaetano, G., and Garattini. S. In: S. B. Day. ci al. (eds.). Hemostasis and experimental cancer dissemination. Cancer Invasion and Metastasis: Biologic Mechanisms and Therapy, pp. 151-160. New York: Raven Press. 1977. Pineo. G. F.. Rogoeczi. E., Hatton, M. W. C., and Brian. M. C. The activation of coagulation by extracts of mucus: a possible pathway of intravascular coagulation accompanying adenocarcinomas. J. Lab. Clin. Med., 82: 255266. 1973. Boggust. W. A.. O'Brien. D. J.. O'Meara, R. A. W.. and Thornes, R. D. The coagulative factors of normal human and human cancer tissue. Irish J. Med. Sci., 477: 131-144, 1967. Dvorak, H. F., Quay, S. C., Orenstein, N. S., et al. Tumor shedding and coagulation. Science (Wash. DC). 212: 923-924. 1981. Hilgard. P., and Whur, P. Factor X-activating activity from Lewis lung carcinoma. Br. J. Cancer. 41:642-643, 1980. Gordon. S. G. A proteolytic procoagulant associated with malignant trans formation. J. Histochem. Cytochem.. 29: 457-463, 1981. Gordon, S. G. Cancer procoagulant. In: Proceedings of Koloman Laki Memorial Symposium on Hemostasis in Cancer, pp. 19-32. Boca Raton. FL: CRC Press. 1987. Hansen. H. I.. Snyder. J. J.. Miller. E. Vandevoorde. J. P.. Miller, O. N., Hiñes.L. R.. and Burns. J. J. Carcinoembryonic antigen (CEA) assay. Human Pathol., 5: 139-147. 1974. Galen, R. S., and Gambino, S. R. Beyond Normality: The Predictive Value and Efficiency of Medical Diagnosis, pp. 76-80. New York: J. Wiley & Son. 1975. Meilicki. W.. and Wierbicki, R. Cancer procoagulant in serum of rats during development of experimental epitheloma. Int. J. Cancer, 45: 125-126, 1990. 6234 Downloaded from cancerres.aacrjournals.org on August 3, 2017. © 1990 American Association for Cancer Research. An Enzyme-linked Immunosorbent Assay for Cancer Procoagulant and Its Potential as a New Tumor Marker Stuart G. Gordon and Barbara A. Cross Cancer Res 1990;50:6229-6234. Updated version E-mail alerts Reprints and Subscriptions Permissions Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/50/19/6229 Sign up to receive free email-alerts related to this article or journal. To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at [email protected]. To request permission to re-use all or part of this article, contact the AACR Publications Department at [email protected]. Downloaded from cancerres.aacrjournals.org on August 3, 2017. © 1990 American Association for Cancer Research.