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, 438-443. January 15, 1990] Presence of Villin, a Tissue-specific Cytoskeletal Protein, in Sera of Patients and an Initial Clinical Evaluation of Its Value for the Diagnosis and Follow-up of Colorectal Cancers1 B. Dudouet, L. Jacob, P. Beuzeboc, H. Magdelenat, S. Robine, Y. Chapuis, B. Christoforov, G. A. Cremer, P. Pmiillanl, P. Bonnichon, F. Pinon, R. J. Salmon, A. Pointereau-Bellanger, J. Bellanger, M. T. Maunoury, and D. Louvard2 Départementde Médecineinterne [L. J., B. C., G. A. C.J, Clinique Chirurgicale [Y. C., P. Bo.J, and Poste de Transfusion sanguine ¡F.P.], Hôpital Cochin, 27, rue du Faubourg Saint Jacques, 75674 Paris Cedex 14; Départementde Médecineoncologique [P. Be., P. P.], Laboratoire de radiopathologie ¡H.M.], and Service de Chirurgie [R. J. S.], Institut Curie, 26 rue d'Ulm, 75005 Paris; Service de Gastroenterologie, Pr. Y. Le Quintrec, Hôpital Rothschild, 75012 Paris [J. B.]; Pharmacie Hôpital Salpétriére, 83 bd de l'Hôpital, 75013 Paris [A. P. B.]; Institut Gustave Roussy, 39 rue Camille Desmoulins, 94805 Villejuif Cedex [M. T. M.J; and Unité de Biologie des Membranes, Départementde Biologie Moléculaire,Institut Pasteur, 25 rue du Docteur Roux, 75724 Paris Cedex 15 [B. D., S. R., D. L.], France ABSTRACT Villin is an actin-binding protein found in a few normal adult epithelia, namely epithelial cells in the digestive and urogenital tracts. Moreover, villin production is maintained in malignant cells. We assumed that cell lysis and necrosis of solid tumors producing villin might result in villin release into blood. We analyzed the villin content of sera from 788 patients and controls using an enzyme-linked immunosorbent assay. Patients and controls were classified into healthy donors, patients with benign diseases of the gastrointestinal tract, patients with colorectal cancers, and patients with malignant nondigestive diseases. In the panel of sera analyzed, the sensitivity of the assay for colorectal cancers was 50.5%, and its overall specificity for malignant digestive tumors was 94.5%. Results were statistically analyzed comparing each group of sera with each other. We conclude that the presence of villin is indicative of a pathological state in the gastrointestinal tract (/' < 0.001). Finally, we followed villin levels after tumor resections (60 patients). We found that the villin level in sera remains low in remissions but is raised in recur rences. We suggest that the villin assay may have clinical utility as a diagnostic adjunct for adenocarcinoma of the gastrointestinal tract. It may also have some value in monitoring patients with advancing colorectal carcinomas after resection of these tumors. INTRODUCTION Enterocytes, the epithelial cells of the intestinal mucosa, exhibit a functional and morphological polarity. Their apical plasma membrane shows an ordered array of microvilli forming the brush border. Each microvillus is supported by a cytoskeleton based on actin microfilaments. These actin microfilament bundles are associated with a few actin-binding proteins (1). Among them, villin, a M, 95,000 protein, is able to bind actin in a calcium-dependent process. Production of villin and its regulation were previously studied in our laboratory in vivo and in cell cultures (2-4). During ontogenesis of the human diges tive tract, villin is found in immature intestinal cells as early as the eight wk of gestation of the human fetus, before the final stages of histogenesis. Similarly, villin is found in adult undifferentiated crypt cells of the intestinal mucosa. In the adult human, high levels of villin were primarily found in epithelial cells with a well-organized brush border, namely the epithelial cells in the small and large intestine and in cells lining the proximal tubules of the kidney. Received 2/27/89; revised 7/13/89; accepted 9/25/89. 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 grants from the Institut National de la Santéet de la Recherche Médicale(No. 86-7008), by the Fondation pour la Recherche MédicaleFrançaise,by the Association pour la Recherche sur le Cancer (No. 6379), thé CNRS (UA-1149), and thé Ligue Nationale Françaisecontre le Cancer. 1To whom requests for reprints should be addressed. This restricted tissue specificity of villin is also conserved during neoplasie processes. In this respect, it is particularly worth recalling that villin continues to be produced in colorectal tumors at all stages of epithelial cell transformation, but that villin is conspicuously absent from tumors arising from cell types in which it is not normally found (5, 6). These observations are in contrast with the more generalized production of CEA,3 a tumor marker that is routinely assayed for screening and follow-up of patients with tumors of the digestive tract. For example, it is now well established that CEA is found in several normal cells of numerous origins (79). The above mentioned properties of villin prompted us to determine whether it is released into the blood circulation, due to lysis of malignant cells or by means of another unknown mechanism, which could indicate the existence of a lesion in the digestive tract. We therefore investigated the presence of villin in the sera of 788 patients and controls with an ELISA developed for this study in our laboratory. Our initial data showed that the occurrence of villin in human sera corre sponded to the pathological state of the gastrointestinal tract. Moreover, the presence of villin in sera was very often associ ated with a malignant digestive tumor. This is the first report of the presence of villin in human sera. We discuss the value of villin as a diagnostic adjunct for the diagnosis and follow-up of patients with colorectal carcinomas. MATERIALS AND METHODS Clinical Specimens. Age- and sex-matched healthy controls were provided by a serum bank from the Centre National de Transfusion sanguine (HôpitalCochin, Paris, France). The mean age of these donors was 37 ±15 yr. They included a roughly equal proportion of men and women. The mean age of donors from whom sera contained detectable amounts of villin was 39 ±11 yr. Coded serum collections from patients bearing gastrointestinal tumors, adenocarcinoma of other organs, or benign diseases were obtained from three hospitals (Departments of Internal Médecineand Surgery). For this study, the diagnosis and follow-up of patients bearing digestive tumors were carried out with routine clinical and endoscopie examinations. All specimens (healthy donors' sera, benign and malignant disease patients' sera) were coded and collected during the same period of time. All of them were aliquoted and stored in the same conditions at —¿20°C. We have observed that villin is stable in these storage conditions. Provided that the assay was performed on frozen and thawed samples, there was no significant variation in the measured villin concentration. Medical histories of patients were checked to validate this information and to secure addi tional data about the serum withdrawal date and about the ultimate 3The abbreviations used are: CEA, carcinocmbryonic antigen; ELISA, enzymelinked immunosorbent assay; PBS, phosphate-buffered saline; BSA, bovine serum albumin. 438 Downloaded from cancerres.aacrjournals.org on August 3, 2017. © 1990 American Association for Cancer Research. VILLIN, A CYTOSKELETAL PROTEIN IN SERA OF PATIENTS WITH COLORECTAL CANCERS course of the disease. Laboratory assays, such as that for CEA, were determined in parallel on the same sample and were available for correlation with the results of the villin assays. Determination of Villin Levels. Sera were analyzed by an ELISA using a mouse monoclonal anti-villin (BDID2C3) antibody obtained and characterized as described in Ref. 4. This monoclonal antibody recog nizes a well-conserved epitope of villin across species. This epitope is located in an actin-binding domain at the COOH-terminal end of the molecule. Primary sequence analysis of this domain carried out by our group (10) has revealed that the carbo\>-terminal end of villin is villin specific and displays no sequence homologies with other actin-binding proteins nor with any other known proteins currently listed in a protein data library. In addition, cell extracts prepared with various cell lines established from cells not expressing villin in viro, when investigated for villin content with this ELISA, prove to be negative (2, 11). Moreover, if a known amount of exogenous villin were added to these extracts, villin could be quantitatively assayed in these conditions. Altogether, these data illustrate the specificity of the antibody for villin and the lack of cross-reactivity for ubiquitous cytoskeletal proteins, such as gelsolin, tropomyosin, calpactin, actinin, as well as the lack of interference with the assay of structural and cellular proteins unrelated to villin. Polystyrene microtiter plates were coated with 10 Mg/rnl of purified monoclonal anti-villin IgG in SO mivi potassium phosphate buffer (pH 8), using 50 n\ per well. Coating was carried out at 37°Cfor 2 h and then at 4°Covernight. After five washes with 0.1 % Tween 20 in PBS, plates were saturated for 30 min at 4°Cwith 1% (w/w) BSA in PBSTween 20. Serial 2-fold dilutions (from 1/2 to 1/32) of the sera in PBSTween 20-BSA (60 ^I/well) were incubated overnight at 4°C.After five washes with PBS-Tween 20 buffer, the second antibody, a /J-galactosidase-Iabeled polyclonal rabbit anti-villin IgG diluted in PBS-Tween 20BSA, was added, and the incubation was continued for 3 h at 4°C.After five more washes with PBS-Tween 20 buffer, hydrolysis of p-nitrophenyl /3-D-galactopyranoside was determined as described (12) and nitrophenol was measured at 414 nm, using a MK 700 ELISA spectrophotometer (Dynatech, South Windham, ME). Each plate contained a serial dilution of pure porcine villin (a generous gift from Dr. V. Gerke and Dr. K. Weber, Max-Planck Institute, Göttingen,West Germany) diluted in villin-frce human serum as a standard. False-positive assays due to nonspecific binding to the plate were detected by the following assay. Sera containing over 20 ng of villin per ml of serum were incubated with an excess of monoclonal anti-villin IgG (35 Mg/ml) for 3 h at room temperature in order to saturate specific villin binding sites prior to incubation with the plate. Control experiments carried out with dilutions of purified villin added to normal human sera showed a 98% inhibition in these conditions. Sera for which this competition was ineffective were considered as falsepositive and scored as negative. Determination of CEA Levels. To compare villin and CEA as markers from malignant digestive diseases before and after surgery, CEA levels were assayed in some sera with a solid phase enzyme immunoassay (Abbott diagnostic kit). Levels of CEA over 5 ng/nl of serum were considered to be abnormally high. Determination of Sensitivity and Specificity Scores. The sensitivity of a marker, defined here as the ability to identify true-positive cases in patients with coloréela!cancer, is calculated by dividing the number of patients with above-normal concentrations of the marker by the total number of patients with colorectal cancer. The specificity of a marker, defined here as the ability to identify true-negative cases in patients without colonie malignancy, is calculated by dividing the number of patients with normal concentrations of the marker by the total number of healthy donors, patients with benign colorectal disease, and patients with extradigestive disease. Statistical Analysis. The classical x2 test was used to analyze the statistical significance of the presence of villin in sera of patients and controls. RESULTS Sensitivity and Specificity of the Villin ELISA Assay. The ELISA assay for villin designed in our laboratory uses a highly specific monoclonal antibody raised against villin and able to recognize a unique epitope found only in villin (4). The assay for villin used in this study is a sandwich ELISA (see "Materials and Methods"). Fig. 1 shows a typical standard curve, made with villin concentrations ranging from 2 ng/ml to 150 ng/ml. The slope of the curve depends upon the villin-free medium used for the pure villin dilution. The minimal detectable amount is as low as 0.5 ng/ml in PBS and up to 4 ng/ml in some villinfree sera obtained from healthy donors. In all cases, a linear response up to 75 ng/ml was obtained. Such differences may be due to interference by serum proteins and were not investi gated further. Specificity was checked on every serum sample with villin levels above 20 ng/ml by preincubation with an excess of monoclonal antibody against villin as described in "Materials and Methods." Only samples exhibiting a total inhibition after preincubation with the monoclonal antibody were considered to contain villin. Determination of Villin Levels in Sera from Healthy Donors and from Patients with Benign Digestive Diseases and Malignant Digestive and Extradigestive Diseases. The distribution of serum villin concentrations in healthy donors and in patients with benign diseases is shown in Fig. 2. Among the healthy donors (421 cases), 408 displayed serum villin levels below 20 ng/ml, while sera from 13 donors (3%) contained more than 20 ng/ml of villin (Fig. 2A). Among the sera from patients with benign diseases of the digestive tract (132 cases), 17 were found to contain significant amounts of villin (12.9%) (Fig. 2B). Among 19 samples ob tained from patients with duodenal ulcers, only 2 contained levels of villin slightly above the threshold (22 and 46 ng/ml, respectively). Villin assays were also carried out on sera of 20 patients with gastric ulcers: 3 patients had serum villin levels above the threshold. Interestingly enough, one of these patients had an ulcer with a severe dysplasia (50 ng of villin/ml of serum), and the other two had an ulcer with intestinal meta plasia (76 and 107 ng of villin/ml of serum). Among 40 sera from patients with Crohn's disease, 5 patients (12.5%) showed 9 ia7 375 75 150 villin ng/ml Fig. 1. Quantification of villin in sera using an ELISA. A standard curve was constructed using serial dilutions of pure porcine villin in villin-frec human sera (D). This standard curve is compared with a curve obtained for the same dilutions of pure porcine villin in PBS-Tween 20-BSA (•)or of a positive patient serum (A). When an excess of antibody (35 »ig/ml)is added to a serum containing villin, 98% of inhibition of the specific response is observed (O). No villin is detected in a normal human serum ( * ) 439 Downloaded from cancerres.aacrjournals.org on August 3, 2017. © 1990 American Association for Cancer Research. VILLIN, A CVTOSKELETAL PROTEIN IN SERA OF PATIENTS WITH COLORECTAL CANCERS 3000INIINI50«Kl- 3000 300020004mm300g 2000 IMI MO MI£'<>«; '°0 to so2 M >n^ ÖL= 70 60 200S" 1002 70Õ 90^ 6040i MÕÕ 70= 50 «Õ 30S M 20B•.-M»••ülliil*„ii.»Min U^ 30S ...W..•.:•.i• 20>• Fig. 2. A, distribution of serum villin levels in healthy donors. Among 421 healthy donors, 13 (3%) exhibited levels above threshold (20 ng/ml). B, distribu tion of serum villin levels in patients with benign gastrointestinal diseases. For duodenal ulcers, 2 patients of 19 (10.5%) were above threshold (Lane A). For gastric ulcers, among 20 patients, 3 (15%) showed levels above 20 ng/ml (Lane B). For patients with Crohn's disease and hemorrhagic rectocolitis, 5 patients of 46 (10.8%) had villin levels above threshold (Lane C). For colorectal polyposis, one patient of 19 (5%) had levels above 20 ng/ml (Lane D). •¿â€¢..•". t»«••••:.•••i.*.†Fig. 3. Distribution of serum villin levels in patients with colorectal cancers before surgery classified into Stages A and D. Patients (Stage A) with undetectable villin levels (Lane A). Sixteen of 31 (Stage B) patients (51%) exhibited villin levels above threshold (Lane B). Six of 13 (Stage C} patients (46%) had villin levels higher than 20 ng/ml (Lane C). Twenty-six of 48 (Stage D) patients (54%) had values above threshold (Lane D). serum villin levels above threshold (30 to 1280 ng/ml), but none of the 6 patients with ulcerative colitis had a detectable level of villin in their sera. In the case of colorectal polyps, only one patient of 19 had serum villin levels above the threshold (200 ng/ml). This patient had a polyadenoma with a severe displasia. The levels of villin were also determined in 28 sera from patients with liver cirrhosis of various etiologies since this pathological state is often associated with elevated levels of tumor markers, such as CEA. Six sera were found to contain elevated levels of villin (72 to 2000 mg/ml). Four patients in this group had liver cirrhosis with ascites. We found no signif icant statistical difference between each group of patients with benign digestive diseases for the detection of villin in sera (x2 = 3.2; P = not significant). Sera of patients with colorectal cancers (95 cases) were ana lyzed before surgery and at Stages A to D according to a modified Dukes' classification (Stage A, cancer confined to the mucosa; Stage B, cancer penetrating deeper, even through the bowel wall; Stage C, cancer penetrating as in Stage B but with involvement of regional lymph nodes; Stage D, cancer with distant métastases).The three Stage A patients had undetectable serum villin levels (below 2 ng/ml). Among the 31 patients at Stage B, 16 (51%) exhibited elevated serum villin levels (range, 20 to 1250 ng/ml), while 6 of 13 Stage C patients (46%) were positive (range, 20 to 2000 ng/ml). We also found that among 48 patients at Stage D, 26 (54%) had values above threshold (range, from 20 to 3000 ng of villin/ml of serum). When all stages were taken together, 50.5% of samples were above the threshold (Figs. 3 and 4). The slight differences between the groups have no statistical significance (x2 = 0.52, P = not significant). Among 80 patients with various extradigestive cancers, only 4 cases (5%) corresponding to advanced cancers had levels of serum villin above threshold; these cases are: one primary' epidermoid tumor of the esophagus; one breast cancer; one ovary tumor with peritoneal carcinomatosis; and one larynx cancer (Fig. 4). Statistical Analysis—Sensitivity and Specificity Scores. Con trol donors and patients were divided in four groups: healthy 300200HM500«M300 200^ 100Õ 902 »BZ 70I 50 UI " 30v> 20-,*::::::::::::::::::••itA1•1•f-i•*••*•i::::::::':»::::. I II Fig. 4. Distribution of villin concentration in sera from patients with various extradigestive cancers (Lane I) or with coloréela!cancers (Lane II) before surgery (all stages taken together). In Lane I. extradigestive cancers, 4 patients of 80 (5%) had villin levels above 20 ng/ml. This group of patients comprises the following cancers: lung (8); ovary (14); breast (25); uterus (5); prostate (4); melanoma (4); osteosarcoma (3); myosarcoma (5); and a panel of other types of cancers occurring less frequently (12 cases). In Lane II, colorectal cancers, among 95 patients, 48 exhibited villin concentrations above threshold. donors; donors with benign digestive diseases; donors with malignant digestive diseases; and donors with malignant nondigestive diseases. Each group was compared with each other, and the values obtained were subjected to the x2 test. The analysis of the results obtained is summarized in Table 1. Our data show that the occurrence of villin in the serum correlated strongly with a pathological benign or malignant state in the digestive tract. It is particularly noteworthy that, when healthy donors are compared to donors bearing malignant digestive diseases, the statistical parameters x2 and P were found to be particularly significant. Similarly, the comparison 440 Downloaded from cancerres.aacrjournals.org on August 3, 2017. © 1990 American Association for Cancer Research. VILLIN, A CYTOSKELETAL PROTEIN IN SERA OF PATIENTS WITH COLORECTAL CANCERS Table 1 Statistical analysis of villin detection in healthy donors, patients with benign digestive diseases, or those with malignant extradigestive and malignant digestive diseases Healthy donors (3%) and 12.9% of patients with benign digestive diseases exhibited serum villin levels above 20 ng/ml. These two groups are significantly different (x2 = 16.9; P < 0.001). Patients with colorectal adenocarcinoma (50.5%) showed serum villin levels above threshold. This group is significantly different from healthy donors (x2 = 162.81, P < 0.001) and from those having benign digestive diseases (x2 = 36.49; P < 0.001). The group of patients with digestive cancers (colorectal carcinoma) is significantly different from the group of patients with malignant extradigestive diseases (x! = 40.9; P < 0.001). On the other hand, there is no significant difference between healthy donors and patients with extradigestive cancers (x2 = 0.280; not significant). %Healthy Comparison of clinical variables No. of patients' sera with villin levels above threshold (1)* No. of patients' sera with villin levels above threshold (2)' % P16.91 donors (1) and benign digestive (2)Healthy diseases (13/421)3% (17/132)48 12.9% <0.001162.81 donors (1) and malignant diges (2)Benign tive diseases (13/421)12.9% (48/95)48 50.5% «0.00136.49 digestive diseases (1) and malig (2)Malignant nant digestive diseases (17/132)5% (48/95)48 50.5% <0.00140.94 extradigestive diseases (1) and (2)Healthy malignant digestive diseases (4/80)3% (48/95)4 50.5% <0.0010.280 NS° donors (1) and malignant extra(13/421)17 5% (4/80)x2 digestive diseases (2)1313174133% °NS, not significant. *The numbers (1) and (2) correspond to the donor and patient populations listed in column "Comparison of clinical variables". of patients with malignant digestive diseases and patients with malignant nondigestive diseases is also significant (Table 1). From the results reported in this study, we have also deter mined figures for the sensitivity and specificity of the villin as tumoral marker. We found a sensitivity of 50.5% (48 of 95) and a specificity of 94.5% (598 of 633) for villin using this particular panel of sera. A Comparison of Villin and CEA as Markers of Malignant Digestive Diseases before Surgery. Sera obtained from the ma jority of patients with malignant digestive diseases could be compared before surgery for villin and CEA levels (75 cases). All of these patients had colorectal tumors and were classified as being at one of the three Dukes' stages (B, C, or D), as Table 2 Comparison of villin and CEA detection in patients' sera with colorectal cancers before surgery Threshold for villin is 20 ng/ml and threshold for CEA is 5 ng/ml. Totalno. of pa of pa tients' serawith tients' serawith ofpatients'seraanalyzedNo. villinlevels CEAlevels abovethresholdNo. above% threshold Stage B 27 12 of pa tients' serawith villinand CEAlevels aboveó °,No.threshold 37.03 44.44 10 48.578 3161.53 88.5793269.23 17 % 63 Stage C Stage D133561746.15 91.42 POST SUR6ERT FOLLÓN UP defined above. Among the 27 patients of Stage B, 12 (44.5%) exhibited serum villin levels significantly above threshold, and REMISSIONS RECURRENCESVillin 10 (37%) had elevated levels of CEA. When both villin and CEA.. CEA200010001002OIBii Villin CEA were assayed, the sensitivity of the analysis was improved ....*2O••••200010001001 since 17 patients of 27 (63%) were positive for at least one marker. Similarly, among 13 patients at Stage C, 9 (69.3%) were positive for at least one marker as compared with 6 for the villin test above and 8 if only CEA were assayed. At Stage D, 17 of the 35 patients (48.6%) exhibited serum villin levels above 20 ng/ml, and 31 (88.6%) had CEA levels above 5 ng/ ml. When both villin and CEA were assayed, the existence of colorectal cancers was correctly predicted in 32 cases (91.3%). Fig. 5. Villin and CEA levels in patients with colorectal carcinoma in postTaken altogether, these observations suggest that patients in surgery follow-up. Normal levels for villin were fixed at 20 ng/ml, and at 5 ng/ ml for CEA. Remissions and recurrences were determined after clinical exami whom colorectal cancer has escaped diagnosis by CEA analysis nation. Among 60 cases, 49 patients were in remission: 46 of these displayed might be positive for villin. These preliminary data suggest that undetectable villin levels (below 2 ng/ml) and, in 47 cases, the CEA level was also a simultaneous search for villin and CEA in sera may be useful, under threshold. Among the 11 recurrences. 9 were positive in the serum villin test, and 7 had an elevated level for CEA. particularly in the case of patients at Stage B, when tumor exéresisgives a better prognosis. It should be mentioned that the comparison of the figures reported here for villin and CEA before surgery. This included 5 patients positive for both villin measured separately and those obtained when both markers and for CEA. After surgery, 10 patients had remission and were were assayed simultaneously is promising but not statistically negative for both markers. The other two patients were positive significant. Therefore, further analysis including a larger num for both villin and CEA. ber of patients should be performed in the future. These data Among the patients for whom the information for villin and CEA was not available before surgery but available for a postare summarized in Table 2. Villin and CEA Levels during Postsurgery Follow-up. For over surgical follow-up, we found that, among 49 patients clinically 3 to 60 mo, 60 patients bearing colorectal cancers were moni in complete remission, 46 had normal villin and 47 had normal tored after surgery for villin and CEA serum levels. Data for CEA levels (Fig. 5). Villin level remained above threshold in villin and CEA before surgery were only available for 12 pa three cases. Among those, two were positive for both villin and tients. Among those, 7 were positive for at least one marker CEA. The third patient, who had an ascitic cirrhosis, was the 441 Downloaded from cancerres.aacrjournals.org on August 3, 2017. © 1990 American Association for Cancer Research. VILLIN, A CYTOSKELETAL PROTEIN IN SERA OF PATIENTS WITH COLORECTAL CANCERS only one to exhibit normal CEA levels with very high villin values (3300 ng/ml). Tumor recurrence was studied in 11 patients, 9 of whom displayed high villin values (range, from 20 to 2000 ng/ml). Seven of these patients also scored as positive for CEA. Very high villin concentrations (2000 ng/ml) were observed in sera from some cases of cancer recurrence with liver métastases(Fig. 5). The preoperative and follow-up investigations of 7 representative patients are summarized in Fig. 6. Our data demonstrate that, after surgery, villin levels drop and remain below the threshold when a complete remission is clinically confirmed. Moreover in cases of tumor recurrence, villin was observed in the serum in the majority of cases studied. DISCUSSION In this study, we have investigated the occurrence of villin in 788 sera. We show that the presence of villin in human serum can often be correlated with the existence of a malignant tumor in the gastrointestinal tract. In contrast to existing biological assays for colorectal carcinoma, villin concentrations above threshold indicate a pathological state in the gastrointestinal tract in almost 95% of cases. We have focused this first study of villin in human sera on colorectal cancers, since they are more frequent in Western countries than other cancers which might release villin into sera, such as pancreatic cancers, cholangioma, or renal cancers. Indeed, further studies should POST SURGERY FOLLOW UP «ILLIN ng/ml 2000 1000 III months Fig. 6. Serum villin levels in 7 patients with colorectal carcinoma during postsurgery follow-up. Patient I. a case displaying a postoperative residual tumor and métastases.The serum villin level is low but remained above the threshold level. A sharp rise in serum villin levels was detected between 3 and 10 mo after surgery to 1000 ng/ml, whereas the clinical diagnosis of recurrences and métas tases was still negative. At 12 mo, the serum villin level was even higher (2000 ng/ml), and liver métastaseswere clinically diagnosed at this stage. Patient 2, a case similar to the previous one but with no preoperative villin assay available. This patient was bearing a recurrent colorectal tumor with liver métastases.The two patients (Nos. 1 and 2) were not treated by chemotherapy and/or radiother apy. High villin levels (higher than 1000 ng/ml) were observed in both cases. Patients 3 and 4, two patients positive for villin before surgical removal of their tumor; thereafter they exhibited a rapid decrease in villin serum levels. Patient 5, 10 mo after resection of the colorectal tumor, this patient showed a local recurrence of the tumor with a serum villin level slightly above threshold (35 ng/ ml). This recurrence, which was detected by colonoscopy, required readmission for surgery, after which the serum villin concentration decreased and remained below threshold. Patient 6, a case having a history similar to the previous ones but exhibiting a lower increase in serum villin levels in relation to the existence of invasive liver métastases.In contrast to Patients 1 and 2, this patient received a complete course of postoperative 5-fluorouracil and radiotherapy. Patient 7, a case showing complete remission 5 mo and 16 mo after surgery, serum villin levels remaining below threshold. be carried out to evaluate the usefulness of the villin assay for these cancers. The sensitivity score of the villin test is 50%, regardless of the stage of invasion of the tumor. This indicates that the villin assay has a sensitivity equivalent to existing biological assays for colorectal cancers. Thus, taking into ac count the specificity score mentioned above, it would appear that the main advantage of the villin assay in diagnosis as well as for follow-up patients after surgery lies in its high specificity for digestive cancers. The villin test would indeed be primarily advantageous for patients bearing resectable colorectal tumors (such as Stage B tumors). Moreover, significant improvement for all stages of colorectal tumors may be obtained when both CEA and villin are assayed, but the latter proposal remains to be rigorously tested on a larger number of patients. The value in clinical practice of the CEA or of the CA 19.9 tests for the diagnosis or prognosis of digestive cancers is still under debate (13, 14). It is widely accepted, however, that none of the available assays is satisfactory. This is especially true for the early diagnosis of colorectal cancer when tumors are still resectable. The low specificity of these markers can be partly attributed to their wide distribution in malignant and normal cell types (15). On the contrary, the high specificity of the serological assay observed for villin can be easily explained by taking into account the tissue specificity of villin expression in normal or malignant epithelial cells in the gastrointestinal tract. Villin is a nonglycosylated intracellular protein which is pro duced early during embryogenesis in endodermic cells of the primitive gut (2-16). During adult life, villin continues to be produced at high levels in normal intestinal cells. This specific pattern of expression for villin is also maintained during neo plasie processes (2, 5). Thus, if villin is detected in sera of patients, one must suppose that it originates from one of a limited number of tissues. Moreover, our study strongly indi cates that the release of villin into serum is very often associated with the existence of a neoplasie tissue in the gastrointestinal tract. Its entry into the general circulation presumably requires cell lysis. Our results suggest that, even if cell lysis occurs in ulcerative or inflammatory diseases of the large bowel, villin only rarely appears in the blood. This may be due to the fact that, under these circumstances, the blood supply and the cellular environment do not allow the entry of villin into the general circulation. In cases of highly vascularized colorectal adenocarcinomas, rapid necrosis and death of malignant cells are known to occur. Under these conditions one may assume that efficient release of cell content and discharge into the blood are facilitated. If such mechanism explains the release of villin into the blood, one would also expect a direct correlation between the size and stage of the tumor and the presence of villin in the serum. Indeed, our results indicate that there are no significant differences if one compares the percentage of patients positive for villin at Stages B and D. This implies that other factors must be taken into account to interpret our data, such as variability of villin stability in serum and large differ ences in the extent of tumor necrosis among patients. Alterna tively, changes in villin organization in tumor cells may account for the release of villin into the blood. In support of this proposal one should recall the recent observation (6) reporting the localization of villin along the basement membrane in some colorectal adenocarcinoma. The physiopathological conditions leading to the appearance of villin in the blood are indeed important to analyze and remain to be studied in detail. Our current study suggests that villin assays could be espe cially useful to follow the clinical course and diagnosis of recurrence of colorectal cancers in addition to CEA assays that 442 Downloaded from cancerres.aacrjournals.org on August 3, 2017. © 1990 American Association for Cancer Research. VILLIN. A CYTOSKELETAL PROTEIN IN SERA OF PATIENTS WITH COLORECTAL CANCERS display low specificity. Indeed, serum villin levels drop after surgery and remain below threshold upon complete remission. In contrast, recurrence is accompanied by a significant increase in villin level, even before clinical evidence of the recurrence, especially in cases of liver métastases.Moreover, our prelimi nary observations strongly suggest that, in case of recurrence, a high level of villin in serum correlates with a rapid progression of the disease. It would therefore appear that villin may have a predictive value concerning the extent and the prognosis of the disease. However, these important findings require further in vestigation. Since 15% of colorectal carcinomas do not produce CEA (15), this marker cannot be used during follow-up studies of some patients. Our work (5) (confirmed in Ref. 6) has estab lished that villin production is maintained in all colorectal tumors so far studied (44 cases). Therefore, one would expect that the detection of tumor recurrence in patients with a nega tive CEA assay might be established using a villin assay. Our preliminary comparative study on CEA and villin before and after surgery supports this proposal. In spite of the conspicuous production of villin in colorectal tumors, only 50% of patients bearing colorectal tumors at all stages had villin in their sera. We have observed by immunoprecipitation analysis that villin circulating in serum is proteolytically digested (data not shown). Moreover, the pattern of villin digestion by proteolytic enzymes may vary between dif ferent individuals. Thus, the epitope recognized by our mono clonal antibodies may be destroyed; consequently, even if villin is present in the serum, a negative result would be reported. This implies that further developments of our assay and/or new monoclonal antibodies for villin may improve the sensitivity of the test. Other parameters such as protein interference in sera should also be further investigated. Work in progress in our laboratory is aimed at answering these important questions. Most studies on cancer markers are based upon the assump tion that the transformed cell is capable of presenting at its surface new antigens, a dogma which is still widely disputed. Our rationale was the use of a differentiation marker such as villin, a well-characterized constituent of intestinal brush bor ders, that is produced very early during embryogenesis in prim itive endoderm cells. Moreover, the profile of villin gene expres sion appears to be precisely modulated during the onset of the intestinal differentiation program (3, 16). The above findings are consistent with the proposal that molecules displaying these properties may in the future provide a new set of markers in cancer research. Such an approach should be assisted by our increasing knowledge of proteins playing a key role in cellular organization and differentiation. ACKNOWLEDGMENTS We are indebted to J. F. Colombel, M.D., and M. Thioux for providing coded sera, E. Coudrier, Ph.D., and F. Traincard for advi-.e and expert technical assistance, T. Pugsley, Ph.D., for his helpful comments during the preparation of the manuscript, and R. Carréfor typing this manuscript. REFERENCES 1. Mooseker, M. S. Organization, chemistry, and assembly of the cytoskeletal apparatus of the intestinal brush border. Annu. Rev. Cell Biol., /: 209-241, 1985. 2. Robine, S., Huet. C., Moll, R., Sahuquillo-Mérino, C., Coudrier. E.. Zwei baum. A., and Louvard, D. Can villin be used lo identify malignant and undifferentiated normal digestive epithelial cells? Proc. Nati. Acad. Sci. USA, 82: 8488-8492. 1985. 3. Pringault, E., Arpin, M.. Garcia, A., I limimi. J., and Louvard, D. A human villin cDNA clone to investigate the differentiation of intestinal and kidney cells in vivo and in culture. EMBO J.. 5: 3119-3124. 1986. 4. Dudouet. B.. Robine. S.. Huet, C.. Sahuquillo-Mérino,C., Blair, L., Coudrier. E., and Louvard, D. Changes in villin synthesis and subcellular distribution during intestinal differentiation of HT29-18 clones. J. Cell Biol., 105: 359369. 1987. 5. Moll, R., Robine. S., Dudouet. B.. and Louvard. D. Villin: a cytoskelelal protein and a differentiation marker expressed in some human adenocarcinomas. Virchows Arch. B., 54: 155-169, 1987. 6. West, A., Isaac, C., Carboni. J., Morrow, J., Mooseker. M.. and Barwick, K. W. Localization of villin, a cytoskeletal protein specific to microvilli. in human ileum and colon and in colonie neoplasms. Gastroenterology. 94: 343-352, 1988. 7. Moore, T. L.. Kupchik. H. 7... Marcon. N., and Zamcheck, N. Carcinoembryonic antigen assay in cancer of the colon and pancreas and other digestive tract disorders. Am. J. Dig. Dis., 16: 1-7. 1971. 8. Lo Gerfo. P., Krupey. J., and Hansen, H. J. Demonstration of an antigen common to several varieties of neoplasia. N. Engl. J. Med.. 285: 138-141, 1971. 9. Freed, D. L. J., and Taylor, G. Carcinoembryonic antigen in faeces. Br. Med. J.,/.- 85-87, 1972. 10. Arpin, M., Pringault. E., Finidori, J.. Garcia, A., Jeltsch, J. M., Vanderkerckhove, J., and Louvard. D. Sequence of human villin: a large duplicated domain homologous with other actin-severing proteins and a unique small carboxy-terminal domain related to villin specificity. J. Cell Biol.. 107: 17591766, 1988. 11. Coudrier. E., Kerjaschki, D., and Louvard, D. Cytoskeleton organization and submembranous interactions in intestinal and renal brush borders. Kidney Int., 34: 309-320, 1988. 12. Guesdon. J. L.. Thierry, R., and Avrameas, S. Magnetic enzyme immunoassays for measuring human IgE. J. Allergy Clin, luminimi 61: 23-27, 1978. 13. Kuusela, P., Jalanko, H., Roberts, P., Sipponen, P., Mecklin, J. P., Pitkänen, R., and Mäkelä. O. Comparison of CA 19-9 and Carcinoembryonic antigen (CEA) levels in the serum of patients with colorectal diseases. Br. J. Cancer., 49: 135-139, 1984. 14. Moertel. C., O'Fallon, J., Vay, L., Go. V.. O'Connell, M.. and Thynne, G. The preoperative Carcinoembryonic antigen test in the diagnosis, staging, and prognosis of colorectal cancer. Cancer (Phila.), 58: 603-610, 1986. 15. Goldenberg, D. M., Sharkey, R. M., and Primus, F. J. Immunocytochemical detection of Carcinoembryonic antigen in conventional histopathology spec imens. Cancer (Phila.). 42: 1546-1553, 1978. 16. Maunoury, R.. Robine. S., Pringault. E., Huet. C., Guenet, J. L., Gaillard. J. A., and Louvard, D. Villin expression in the visceral endoderm and in the gut anlägeduring early mouse embryogenesis. EMBO J., 7: 3321-3329, 1988. 443 Downloaded from cancerres.aacrjournals.org on August 3, 2017. © 1990 American Association for Cancer Research. Presence of Villin, a Tissue-specific Cytoskeletal Protein, in Sera of Patients and an Initial Clinical Evaluation of Its Value for the Diagnosis and Follow-up of Colorectal Cancers B. Dudouet, L. Jacob, P. Beuzeboc, et al. Cancer Res 1990;50:438-443. Updated version E-mail alerts Reprints and Subscriptions Permissions Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/50/2/438 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.