Int. J. Cancer (Pred. Oncol.): 74, 540–544 (1997) r 1997 Wiley-Liss, Inc. Publication of the International Union Against Cancer Publication de l’Union Internationale Contre le Cancer DETECTION OF DISSEMINATED TUMOR CELLS IN PERIPHERAL BLOOD OF COLORECTAL CANCER PATIENTS Marc G. DENIS1*, Cecile LIPART2, Joël LEBORGNE3, Paul-Antoine LEHUR3, Jean-Paul GALMICHE4, Michele DENIS1, Erik RUUD5, Alain TRUCHAUD2 and Patrick LUSTENBERGER1 1Laboratoire de Biochimie Spécialisée, Institut de Biologie, Nantes, France 2Laboratoire de Technologie Biomédicale, Institut de Biologie, Nantes, France 3Clinique Chirurgicale II, Hôtel-Dieu, Nantes, France 4Service de Gastroentérologie, Hôtel-Dieu, Nantes, France 5Dynal A.S., Oslo, Norway All cancer staging systems seek to identify clinical and pathological features that can predict outcome or guide therapy. In particular, a non-invasive method for the early detection of disseminating disease would be of great interest. We investigated the use of cytokeratin genes expression to detect blood metastases from colorectal tumors. Epithelial tumor cells were isolated from whole blood using the monoclonal antibody (MAb) BerEP4 and magnetic beads, and detected by reverse transcription-polymerase chain reaction using oligonucleotides derived from the cDNA sequences of cytokeratins 8, 19 and 20. The sensitivity of this assay was determined by spiking SW620 colon carcinoma cells in normal blood. Using cytokeratin 19 expression we were able to detect 1 epithelial tumor cell in 1 ml of whole blood. The clinical applicability of this technique was explored by evaluating patients with a colorectal carcinoma. Epithelial cells were detected in the blood of 12 out of 23 patients, 2 (20%) of 10 with Astler-Coller stage A or B, and 10 (77%) of 13 with stage C or D cancer. In conclusion, this test is a non-invasive, sensitive, and specific assay for detecting circulating epithelial cells in blood. It may be useful for the early diagnosis of disseminating disease, to determine whether the presence of micrometastatic cells at the time of surgery is correlated with an early relapse and for monitoring adjuvant therapeutic trials. Int. J. Cancer 74:540–544, 1997. r 1997 Wiley-Liss, Inc. Metastasis is an important factor that regulates the prognosis of patients with cancer. In the process of metastasis, tumor cells are scattered from the original site, are spread hematogenously, and are arrested in small vessels. These cells must then adhere to the vascular endothelium, migrate into the extracellular space, establish a microenvironment, escape host defense mechanisms, and finally grow into secondary tumors. Detection of such tumor cells circulating in the peripheral blood is of interest to detect micrometastases at an early stage. This is likely to provide clinicians with an important predictive tool with respect to recurrence and metastases and to result in a more appropriate selection of patients for adjuvant therapy. Immunocytochemical procedures have been used to detect micrometastases in bone marrow and lymph nodes. But this technique is tedious for routine analysis of 5–10-ml blood samples. An alternative is to analyze nucleic acid following amplification by means of the polymerase chain reaction (PCR). Based on the high sensitivity of the PCR, in conjunction with tumor specific genetic alterations, a few tumor cells can be detected in a great excess of non-malignant cells. In the case of solid tumors, the p53 tumor suppressor gene and the ras oncogene are the most consistently mutated genes (Hollstein et al., 1991). For instance, p53 mutations have been used to detect bladder tumor cells in urine of patients (Sidransky et al., 1991), and mutated ras gene has been used as a marker of cells from colorectal cancer in stools (Sidransky et al., 1992) and in lymph nodes (Hayashi et al., 1995). However, these single mutations have not been consistently found, and therefore this technique can be applied only to a limited number of patients. Alternatively, the phenotypic characteristics of the tumor cells can be used (Denis et al., 1996a; Pelkey et al., 1996). The detection relies on the tissue specific expression of a gene. This gene has to be expressed in cancer cells but not in cells that are normally present in blood. A high level of expression facilitates detection. This gene can even be expressed in the corresponding normal tissue, assuming that in most cases, normal cells (e.g., melanocytes or colonocytes) will not circulate in blood. For instance, tyrosinase, a key enzyme of melanogenesis specifically expressed by melanocytes (both normal and melanoma cells), is used to detect melanoma cells in blood. The prostatic specific antigen is used to detect prostate cancer cells. Total RNA is prepared from blood. The specific RNA is reverse transcribed and this cDNA is then amplified by PCR. In most cases a two-stage amplification is performed by using nested primers to further increase the detection sensitivity. Amplification of cytokeratins mRNA can be used to detect micrometastases of epithelial tumors in lymph nodes (Gunn et al., 1996; Schoenfeld et al., 1996). Therefore, we investigated the use of cytokeratin genes expression to detect blood metastases from colorectal tumors. In this report, we show that by combining immunomagnetic separation (IMS) to enrich for epithelial cells and nested reverse transcription-polymerase chain reaction (RT-PCR), we can specifically detect a few cells in blood and we present the results of the first clinical specimen investigated. MATERIAL AND METHODS Reverse transcription and polymerase chain reaction The reverse transcription reaction and PCR amplification were performed in a Crocodile III thermal reactor (Appligene, Illkirch, France). Aerosol resistant tips (Stratagene, La Jolla, CA) were used to prevent contamination. Primers were designed from the published sequence for human cDNAs using the Genejockey software (Biosoft, Cambridge, UK). They were selected on different exons of the corresponding gene so that DNA amplified from genomic DNA would be larger than the fragment amplified from cDNA. The sequences of these primers and the length of the amplified DNA fragments are indicated in Table I. The primers for cytokeratin 19 were selected to maximize mismatches between cytokeratin 19 mRNA and the cytokeratin 19 pseudogene sequences (Savtchenko et al., 1988) to avoid amplification of this processed pseudogene. Total RNA was heated at 72°C for 3 min and cooled on ice. It was then combined with 100 pmoles of the 38 outer oligonucleotide, transcription buffer (50 mM Tris-hydrochloride pH 8.3, 75 mM KCl, 3 mM MgCl2), DTT (2 mM), dNTPs (1 mM each), RNasin (50 units; Promega, Lyon, France), and Superscript reverse transcriptase (200 units; Life Technologies, Gaithersburg, MD) in a total volume of 25 µl. Incubation was performed at 42°C for 60 min. Amplification was performed with 2.5 µl of cDNA in a total Contract grant sponsor: Fondation de France, Lisue Départementale (44) Contre le Cancer. *Correspondence to: Laboratoire de Biochimie Spécialisée, Institut de Biologie, 9, Quai Moncousu F-44035, Nantes, France. Fax: (33) 02.40.08.40.82; E-mail: [email protected] Received 27 March 1997; Revised 2 June 1997 HEMATOGENOUS DISSEMINATION OF COLON CANCER TABLE I – OLIGONUCLEOTIDES USED FOR SYNTHESIS AND AMPLIFICATION OF CYTOKERATIN Genes for: Cytokeratin 8 PCR1 PCR2 Cytokeratin 19 PCR1 PCR2 Cytokeratin 20 PCR1 PCR2 1The 58 oligonucleotides 541 cDNA FRAGMENTS1 38 oligonucleotides Size (bp) CAGTTACGGTCAACCAGAGC (exon 1) AGACCCTGAACAACAAGTTTGC (exon 1) CTTGTTCTTGAAGTCCTCCACC (exon 2) CGCCTAAGGTTGTTGATGTAGC (exon 2) 341 160 GTGGAGGTGGATTCCGCTCC (exon 2) ATGGCCGAGCAGAACCGGAA (exon 2) TGGCAATCTCCTGCTCCAGC (exon 4) CCATGAGCCGCTGGTACTCC (exon 4) 433 328 CCAGACACACGGTGAACTATGG (exon 1) TGAAGTATGAGACTGAGAGAGG (exons 2–3) ATGATGACGCCAAGGTTCAGGC (exon 4) ACCTCCACATTGACAGTGTTGC (exon 4) 570 210 nucleotides underlined represent reported differences from the pseudogene for cytokeratin 19. volume of 50 µl containing 13 PCR buffer (10 mM Trishydrochloride pH 9.0, 50 mM KCl, 1.5 mM MgCl2, 0.1% Triton X-100 and 0.2 mg/ml gelatin), 200 µM of each dNTP, 100 pmoles outer primers, and 1 unit Taq DNA polymerase (Appligene). Thirty cycles were then carried out (94°C for 30 sec, 56°C for 45 sec, 72°C for 60 sec) followed by a 5 min final extension at 72°C. For reamplification with the nested primers, 0.5 µl of the first amplification was amplified in a final volume of 50 µl with inner primers as described above. The final products were analyzed by electrophoresis on 2% agarose gels containing ethidium bromide. Immuno-magnetic separation Blood sampling was performed immediately before the beginning of surgery. Blood samples (5 ml) were collected in heparinized tubes and stored at 4°C for a maximum of 2 hr before treatment. The samples were washed 3 times with cold PBS. Magnetic beads covalently coated with the BerEP4 MAb (DYNAL, Oslo, Norway) were then added (4 3 106 beads/ml of blood). Following incubation performed at 4°C for 30 min, cells bound to the beads were retrieved using a magnetic field. The beads were then washed 3 times with PBS. Total RNA was extracted from the cells immobilized on the beads with TRIZOL reagent as described by the manufacturer (Life Technologies). cDNA was synthesized and amplified as described above. RESULTS Colon cancer cell lines We have used a combination of reverse transcriptase and polymerase chain reaction (2 successive amplifications) to synthesize and amplify a fragment of cytokeratin genes (8, 19 and 20). Cytokeratins genes were chosen as target genes as they are supposed to be specifically expressed in epithelial cells. We first analyzed expression of these genes in cultured cells from colon carcinoma (HCT8R, SW948, SW1116, SW48, CaCO2, HT29). A DNA fragment of the expected size (Table I) was detected in all these cell lines after the first PCR amplification with primers corresponding to the cytokeratin 8 or 19 mRNA (data not shown), indicating a high level of expression. The nested amplification was required to obtain the expected DNA fragment in all these cell lines with oligonucleotides corresponding to cytokeratin 20 (not shown), indicating a lower expression than for cytokeratin 8 and 19. In order to determine the sensitivity of the RT-PCR, serial dilutions of total RNA prepared from the colon carcinoma cell line HT29 were performed and processed for cDNA synthesis and nested amplification. A clear signal was obtained with primers corresponding to cytokeratins 8 and 19 with as little as 1 pg of total RNA, potentially detecting the corresponding mRNA from a single colon carcinoma cell (data not shown). Whole blood We then analyzed normal blood samples collected in heparinized tubes. Total cellular RNA was extracted using a modification of the original acid guanidinium thiocyanate/phenol/chloroform extraction detailed elsewhere (Denis and Lustenberger, 1993). Approximately 100 µg of total RNA were obtained from 10-ml blood samples. Reverse transcription and amplification were performed. No signal was detected after the first amplification (not shown). The second PCR yielded an amplified product of the expected size in most of the normal samples (Fig. 1), thus making the use of these genes as markers of circulating colorectal carcinoma cells nonspecific and not suitable for this purpose. Immuno-magnetic separation One way to circumvent this lack of specificity was to include an enrichment step based on the use of a specific MAb coupled to magnetic beads. We have used the BerEP4 MAb. This antibody recognizes an epitope on the protein moiety of two 34 and 39 kDa glycopeptides expressed at the surface of epithelial cells in normal and malignant tissues (Momburg et al., 1987). It has been shown to react with all the colorectal tumors analyzed (Latza et al., 1990). This antibody was covalently linked to magnetic beads. The efficiency of separation was first tested by using these beads with normal blood or blood supplemented with SW620 colon carcinoma cells. These cells, as well as all the colon carcinoma cell lines we have tested (SW48, SW620, SW480, HT29, HCT8R, Colo205, ALT-I, ALT-F, ALT-G), were found to express the BerEP4 antigen at their surface as determined by indirect immunofluorescence (data not shown). Following IMS, RNA was extracted from the cells immobilized on the magnetic beads. Nested RT-PCR was then performed. Cytokeratin 8 was detected with normal blood (Fig. 2). By contrast, analysis using oligonucleotides derived from cytokeratins 19 or 20 revealed no signal for normal blood, whereas a strong signal was seen when colon carcinoma cells were added in the sample prior to IMS. As cytokeratin 19 expression was higher than cytokeratin 20 expression in all the cell lines examined, cytokeratin 19 was used in subsequent experiments. In order to determine the sensitivity of this assay, SW620 colon carcinoma cells were spiked in normal blood. As shown in Figure 3, we were able to detect 1 cell in 1 ml of whole blood, i.e., in 5 3 106–107 nucleated cells. This amplification was reverse transcriptasedependent, demonstrating that DNA was amplified from the cytokeratin 19 transcript and not from the processed pseudogene. This was further confirmed by restriction analysis of the amplified fragments using the HinfI restriction endonuclease (data not shown). Controls Blood collected from 26 healthy donors and 16 patients with gastrointestinal diseases (alcoholic hepatitis, diverticulosis, Crohn’s disease, sigmoiditis, gastric ulcers) was tested. The median age was 60 years (range: 34–87), with 25 males and 17 females. All these samples were negative. Colorectal tumor patients The clinical applicability of our technique was explored by evaluating patients with a colorectal carcinoma. Blood samples were collected before they underwent surgical therapy. The median age of the patients was 64 years (range: 33–85) with 11 males and 12 females. The data obtained from these patients are presented in Table II. The clinical staging of these patients was performed DENIS ET AL. 542 FIGURE 2 – Nested RT-PCR detection of cytokeratins 8 (1), 19 (2) and 20 (3) expression following IMS. (A) Blood from a healthy volunteer. (B) Same blood sample supplemented with 100 SW620 colorectal cancer cells per ml of blood. FIGURE 1 – Nested RT-PCR detection of cytokeratins 8 (A), 19 (B) and 20 (C) expression in whole blood. Total cellular RNA was extracted from whole blood of 8 healthy volunteers. Reverse transcription and amplification were performed. The star indicates DNA fragments amplified from genomic DNA. according to Astler and Coller (1954). We also determined the serum concentration of tumor markers carcinoembryonic antigen (CEA) and CA19.9 in these samples. Two patients presenting a tumor no deeper than the submucosa (stage A) were tested. None had epithelial cells detected in their blood. Eight patients of stage B (with invasion to the muscularis propria or to the serosa without nodal involvement) were analyzed. Two samples (from patients 6 and 26) were found to contain epithelial cells. In both cases, tumor markers were normal. Eight patients with histologic evidence of loco-regional lymph node involvement, i.e., stage C of the classification, were analyzed. Six of them (patients 1, 3, 5, 7, 8 and 25) had detectable epithelial tumor cells in their blood. Three of these positive patients (patients 1, 5 and 25) had normal tumor markers. Finally, we analyzed 5 patients with distant metastases (stage D). Four of them had circulating epithelial tumor cells in their blood. Patient 27, presenting with lung metastases, had epithelial tumor cells in his blood and normal tumor markers. DISCUSSION Using an RT-PCR amplification method, the potential of several target genes that could be used to detect colorectal tumor cells in peripheral blood has been evaluated. We initially examined several normal blood samples for CEA, which is commonly used as a tumor marker. A signal was obtained for all the samples (Denis and Lustenberger, 1995). We performed a similar analysis with primers derived from cytokeratin genes, which are, in higher vertebrates, expected to display some specificity for epithelial cells. Again, nested RT-PCR yielded detectable DNA fragments corresponding to all these genes from normal blood RNA. Similar results have been reported for cytokeratin 19 (Krisman et al., 1995). There are 2 possible explanations for this. On one hand, cells of non-epithelial origin expressing these cytokeratin genes (Traweek et al., 1993) might be present in peripheral blood. It has been shown, for instance, that cytokeratins are expressed in smooth muscle cells (Jahn et al., 1993). On the other hand, all the cells from peripheral blood may express scant amount of cytokeratin mRNAs. This may reflect a general process of illegitimate transcription. A similar observation has been reported by Chou et al. (1994), who found FIGURE 3 – Sensitivity of the IMS-RT-PCR technique. IMS was performed with 1 ml of blood from a healthy volunteer (1), 1 ml of blood supplemented with 1 (2) or 10 (3) SW620 colorectal cancer cells. The reverse transcription was performed in the absence (2) or in the presence (1) of reverse transcriptase. that albumin mRNA, previously used as a marker of circulating hepatocytes in hepatocellular carcinoma, could be detected in peripheral blood of most normal subjects and is therefore not suitable for this purpose. The prostate-specific antigen, used to detect micrometastases of prostate cancer in the blood, has also been shown to be expressed in nonprostatic cells (Smith et al., 1995). To increase the specificity of detection, immunomagnetic beads, labeled with an epithelium-specific MAb, were used to isolate epithelial colorectal tumor cells from blood. This antibody has already been used to enrich for epithelial colorectal cells (Hardingham et al., 1993), but the authors used mutation of codon 12 of K-ras to detect these cells following IMS. The major problem is that approx. 50% of colorectal carcinomas do not have a mutated ras gene. Thus the use of this technique is limited. By using cytokeratin 19 gene as a target gene, we do not encounter this limitation. By combining 2 techniques, one based on the detection of a membrane antigen by a MAb, the other on nested RT-PCR, we find that the clinical specificity appears to be high, based on the absence of a false-positive in our control group. The 2 stage A patients tested did not present circulating epithelial tumor cells. By contrast, 4 out of 5 patients with a clinically HEMATOGENOUS DISSEMINATION OF COLON CANCER 543 TABLE II – CLINICAL AND BIOLOGICAL STATUS OF 23 PATIENTS WITH COLORECTAL CARCINOMA Patient Sex/age Staging (Astler and Coller) Metastases Epithelial tumor cells in blood CEA1 (ng/ml) CA19.92 (U/ml) 2 16 9 12 26 6 11 22 23 24 15 25 1 3 5 7 8 19 4 10 20 27 28 F33 M85 F71 F70 M69 F80 M57 M88 F83 F71 F61 F70 M66 M49 F63 M59 M55 M59 M70 F49 F46 M68 F54 A A B1 B1 B1 B2 B2 (local relapse) B2 B2 B2 C1 C1 C2 C2 C2 C2 (local relapse) C2 C2 D D D D D — — — — — — — — — — lymph nodes lymph nodes lymph nodes lymph nodes lymph nodes lymph nodes lymph nodes lymph nodes peritoneum 1 liver liver liver lung liver 2 2 2 2 1 1 2 2 2 2 2 1 1 1 1 1 1 2 1 2 1 1 1 1.2 3.5 0.8 1.1 1.1 5.2 34.7 2.0 12.0 1.5 0.6 1.4 0.6 19.0 1.8 4.0 16.2 1.0 20920.0 21.6 26.2 1.3 101.2 2 12 7 9 5 8 2 11 29 10 25 8 2 22 26 246 150 10 526 106 770 11 3633 1Normal range: 10 ng/ml.–2Normal range: 37 U/ml. established metastatic disease had tumor cells in their blood. This is in agreement with the 5-year survival rates after surgery for these patients, which are .90% and ,5%, respectively (Sinicrope and Sugarman, 1995). The results obtained for stage C patients are also consistent with the published 5-year survival rate, which varies from 50 to 65% for stage C1 to 25 to 45% for stage C2. This clearly shows that an overwhelming majority of these patients have, in addition to established lymph node metastases, potential distant metastases in their blood. Finally, for 2 of the 8 patients who had no detected node involvement (stage B), we were able to identify early a disseminating disease. All cancer staging systems seek to identify clinical and pathological features that can predict outcome or guide therapy. In the case of colorectal cancer, adjuvant chemotherapy is usually applied to stage C patients. In resected lymph-node positive colon cancer, 5-fluorouracil and levamisole have been shown to increase survival rates. By contrast, the optimal treatments for patients with stage B colon cancer (negative pericolic lymph nodes) is unclear. The 5-year survival rate of these patients varies from 80 to 85% (stage B1) to 70 to 75% (stage B2). The clinical problem is therefore straightforward: how can the patients with stage B disease be selected who could benefit from adjuvant therapy after surgery and how can we avoid unnecessary treatment of patients with a low probability of recurrence? Detection of epithelial tumor cells in blood might help in the selection of patients at high risk for metastases in the group of patients who are presently assessed with a good prognosis. Detection of early blood micrometastases in some of these patients might be used to guide therapy. Additional studies to monitor epithelial tumor cells in blood of stage C patients receiving an adjuvant therapy (chemotherapy, radiotherapy, and immunotherapy, either alone or in combination) will also be of critical importance. As mentioned above, adjuvant chemotherapy has been shown to be highly effective for patients with stage C colon cancer. If the circulation is cleared of epithelial tumor cells after treatment, then the procedure described here could also be used to monitor early detection of relapse. Finally, tumor-cell shedding by surgical manipulation, which has been reported for prostate tumors (Eschwège et al., 1995) and melanoma (Denis et al., 1996b) can also be assessed for colorectal carcinomas by using this method. We conclude that IMS combined with a nested RT-PCR is a non-invasive, sensitive, and specific assay for detection of circulating epithelial tumor cells in the blood. This test may be useful for the early detection of disseminating disease and for monitoring adjuvant therapies. Clearly, further studies and longer follow-up are required to establish the prognostic significance of these circulating cells. ACKNOWLEDGMENTS This work was supported by the European Brite Mebioce project (Pr. Bisconte). We are most grateful to Dr. Cassagnau for providing the anatomopathological data. REFERENCES ASTLER, V.B. and COLLER, F.A., The prognostic significance of direct extension of carcinoma of the colon and rectum. Ann. Surg., 139, 846–852 (1954). CHOU, H.C., SHEU, J.C., HUANG, G.T., WANG, J.T. and CHEN, D.S., Albumin messenger RNA is not specific for circulating hepatoma cells. Gastroenterology, 107, 630–631 (1994). DENIS, M., TRUCHAUD, A. and LUSTENBERGER, P., Detection de metastases de tumeurs solides dans la circulation sanguine par PCR. Immunoanal. Biol. Spec., 11, 361–371 (1996a). DENIS, M.G. and LUSTENBERGER, P., A simple method of RNA isolation from large blood samples. Methods Mol. Cell. Biol., 4, 128–130 (1993). DENIS, M.G. and LUSTENBERGER, P., Qualitative low-level internal control for nested RT-PCR. Biotechniques, 19, 906–908 (1995). DENIS, M.G., TESSIER, M.H., DRENO, B. and LUSTENBERGER, P., Circulating micrometastases following oncological surgery. Lancet, 347, 913 (1996b). ESCHWÈGE, P., DUMAS, F., BLANCHET, P., LEMAIRE, V., BENOIT, G., JARDIN, A., LACOUR, B. and LORIC, S., Haematogenous dissemination of prostatic cells during radical prostatectomy. Lancet, 346, 1528–1530 (1995). GUNN, J., MCCALL, J.L., YUN, K. and WRIGHT, P.A., Detection of 544 DENIS ET AL. micrometastases in colorectal cancer patients by K19 and K20 reverse transcription polymerase chain reaction. Lab. Invest., 75, 611–616 (1996). HARDINGHAM, J.E., KOTASEK, D., FARMER, B., BUTLER, R.N., MI, J.X., SAGE, R.E. and DOBROVIC, A., Immunobead-PCR: a technique for the detection of circulating tumor cells using immunomagnetic beads and the polymerase chain reaction. Cancer Res., 53, 3455–3458 (1993). HAYASHI, N., ITO, I., YANAGISAWA, A., KATO, Y., NAKAMORI, S., IMAOKA, S., WATANABE, H., OGAWA, M. and NAKAMURA, Y., Genetic diagnosis of lymph-node metastasis in colorectal cancer. Lancet, 345, 1257–1259 (1995). HOLLSTEIN, M., SIDRANSKY, D., VOGELSTEIN, B. and HARRIS, C.C., p53 mutations in human cancers. Science, 253, 49–53 (1991). JAHN, L., KREUZER, J., VONHODENBERG, E., KUBLER, W., FRANKE, W.W., ALLENBERG, J. and IZUMO, S., Cytokeratin-8 and cytokeratin-18 in smooth muscle cells-detection in human coronary artery, peripheral vascular, and vein graft disease and in transplantation-associated arteriosclerosis. Arteriosclerosis Thrombosis, 13, 1631–1639 (1993). KRISMAN, M., TODT, B., SCHRODER, J., GAREIS, D., MULLER, K.M., SEEBER, S. and SCHUTTE, J., Low specificity of cytokeratin 19 reverse transcriptasepolymerase chain reaction analyses for detection of hematogenous lung cancer dissemination. J. clin. Oncol., 13, 2769–2775 (1995). LATZA, U., NIEDOBITEK, G., SCHWARTING, R., NEKARDA, H. and STEIN, H., Ber-EP4: new monoclonal antibody which distinguishes epithelia from mesothelia. J. clin. Pathol., 43, 213–219 (1990). MOMBURG, F., MOLDENHAUER, G., HÄMMERLING, G.J. and MÖLLER, P., Immunohistochemical study of the expression of a Mr 34,000 human epithelium-specific surface glycoprotein in normal and malignant tissues. Cancer Res., 47, 2883–2891 (1987). PELKEY, T.J., FRIERSON, H.F. and BRUNS, D.E., Molecular and immunological detection of circulating tumor cells and micrometastases from solid tumors. Clin. Chem., 42, 1369–1381 (1996). SAVTCHENKO, E.S., SCHIFF, T.A., JIANG, C.K., FREEDBERG, I.M. and BLUMENBERG, M., Embryonic expression of the human 40-kD keratin: evidence from a processed pseudogene sequence. Amer. J. hum. Genet., 43, 630–637 (1988). SCHOENFELD, A., LUQMANI, Y., SINNETT, H.D., SHOUSHA, S. and COOMBES, R.C., Keratin 19 mRNA measurement to detect micrometastases in lymph nodes in breast cancer patients. Brit. J. Cancer, 74, 1639–1642 (1996). SIDRANSKY, D., TOKINO, T., HAMILTON, S.R., RINGKER, K.W., CEVIM, B., FROST, P. and VOGELSTEIN, B., Identification of ras oncogene mutations in the stools of patients with curable colorectal tumors. Science, 256, 102–105 (1992). SIDRANSKY, D., VON ESCHENBACH, A., TSAI, Y.C., JONES, P., SUMMERHAYES, I., MARSHALL, F., PAUL, M., GREEN, P., HAMILTON, S.R., FROST, P. and VOGELSTEIN, B., Identification of p53 gene mutations in bladder cancers and urine samples. Science, 252, 706–709 (1991). SINICROPE, F.A. and SUGARMAN, S.M., Role of adjuvant therapy in surgically resected colorectal carcinoma. Gastroenterology, 109, 984–993 (1995). SMITH, M.R., BIGGAR, S. and HUSSAIN, M., Prostate-specific antigen messenger RNA is expressed in non-prostate cells: implications for detection of micrometastases. Cancer Res., 55, 2640–2644 (1995). TRAWEEK, S.T., JIU, J. and BATTIFORA, H., Keratin gene expression in non-epithelial tissues. Detection with polymerase chain reaction. Amer. J. Pathol., 142, 1111–1118 (1993).