- Email: [email protected]
Bone marrow micrometastases and gastrointestinal cancer detection and significance
THE AMERICAN JOURNAL OF GASTROENTEROLOGY © 2000 by Am. Coll. of Gastroenterology Published by Elsevier Science Inc.
Vol. 95, No. 7, 2000 ISSN 0002-9270/00/$20.00 PII S0002-9270(00)00991-6
Bone Marrow Micrometastases and Gastrointestinal Cancer Detection and Significance Donal Maguire, M.D., Gerald C. O’Sullivan, M.D., J. Kevin Collins, Ph.D., John Morgan, Ph.D., and Fergus Shanahan, M.D. Department of Surgery and Medicine, Mercy and Cork University Hospitals, and National University of Ireland, Cork, Ireland
ABSTRACT Accurate staging of cancer is important, as the presence or absence of systemic spread determines treatment. The sensitivity of current imaging and biochemical techniques is suboptimal for the detection of minimal residual disease and latent metastases. This results in understaging and potential undertreatment. To improve detection of disseminated epithelial malignancy, immunohistochemical and molecular methods have been employed that search for epithelial cell– specific proteins in nonepithelial tissue. Bone marrow is mesenchymal tissue (that does not normally express epithelial cell components) and represents an accessible window for detection of micrometastatic carcinoma cells. Detection methods for epithelial cell components (cytokeratins, epithelial membrane antigen, carcinoembryonic antigen) include immunohistochemistry, flow cytometry, reverse transcriptase polymerase chain reaction (rt-PCR), and enzyme linked immunoassay (ELISA). Micrometastatic cells in bone marrow are viable, capable of proliferation, resistant to immune attack, and insensitive to s-phase chemotherapeutic agents. Patients with carcinomas of the lung, breast, prostate, or gastrointestinal tract and in whom bone marrow micrometastases are detected have a foreshortened interval to recurrence and impaired survival. Detection of micrometastases deserves serious consideration in treatment protocols, and standardization of methods is now required. (Am J Gastroenterol 2000;95:1644 –1651. © 2000 by Am. Coll. of Gastroenterology)
INTRODUCTION The extent of spread of primary cancer remains the best predictor of clinical outcome and is the major determinant of treatment strategy (1–3). In the presence of disseminated disease, cure by excision of the primary cancer is not possible, so adjuvant systemic therapies are necessary. This implies that rigorous detection of metastatic disease is required to ensure optimal staging and treatment. At present, rigorous detection of subclinical disease is not possible, as available imaging techniques lack the temporal and spatial resolution to discover individual or small groups of disseminated tumor cells (4). Conventional ultrasound, CT, and magnetic resonance scans fail to detect up to 10% of hepatic
metasases and up to 40% of peritoneal recurrences in colorectal cancer patients, whereas in patients with esophagogastric malignancy, ultrasound and CT detects peritoneal or hepatic metastases with a sensitivity of 21% and 47%, respectively (4). Likewise, the sensitivity and specificity of serum tumor antigens is limited (5, 6). Although prognostic markers including DNA content of the primary tumor, oncogene mutations, tumor suppressor genes, and proliferative activity are useful, these fail to provide a direct measure of tumor burden or metastatic spread and do not facilitate repeated study (7). In addition, interval reassessments of disease status are required to monitor treatment efficacy. Therefore, to develop effective therapeutic strategies against minimal residual disease, accurate, reproducible and standardized techniques that can detect and enumerate metastatic cells would be advantageous. Against this background, methods for detecting epithelial micrometastatases in mesenchymal tissue (bone marrow) have been developed and are now the subject of clinical studies (8). In this report we present an appraisal of techniques for detection of bone marrow micrometastases, an overview of the biological properties of the micrometastatic cells, and evaluate their clinical utility including implications for therapy.
MICROMETASTASES AND TUMOR CELL SPREAD The presence of viable micrometastatic deposits in tissues implies dissemination of tumor cells with the capability of independent survival and growth (1, 2). These cells must detach from the primary growth, travel through extracellular matrix and basement membrane into the local microvasculature, survive transit in the circulation, and exit by attaching and extravasating through the endothelium at another site (1, 9). This process involves several interdependent steps regulated by cascades of cytokines, chemokines, growth factors, and matrix metalloproteinases (10 –16). Many cells within the primary tumor are terminally differentiated and do not have the ability to disseminate and grow into metastases (17, 18). In contrast, the finding of malignant cells distant to the primary tumor indicate metastatic phenotype with tumorigenic potential.
AJG – July, 2000
Micrometastases and GI Cancer
1645
Figure 1. Detection of cytokeratin-positive cells within bone marrow by three different techniques: Flow cytometry with dual staining for cytokeratin-18 and propidium iodide (top left); immunocytochemistry with cytokeratin-18 monoclonal antibody and the alkaline phosphatases anti-alkaline phosphatase (APAAP) technique (top right); and reverse transcription polymerase chain reaction (RT-PCR) for detection of mRNA for cytokeratin 19 (bottom). The observed RT-PCR product of 704bp is the predicted product from RT amplification of cytokeratin 19 mRNA. Lane M, 100 bp ladder; lanes 1–7 are patient samples with lanes 1 and 3 testing positively. Lanes 8 –10 represent 1000, 100, and 10 HT29 colon carcinoma cells, respectively.
METHODOLOGY FOR DETECTION OF MICROMETASTASES IN CARCINOMA PATIENTS Bone marrow is derived from the embryonic mesoderm and therefore does not normally express epithelial cell–specific components. Epithelially derived malignancies retain expression of many epithelial specific proteins such as cytokeratins (CKs), carcinoembryonic antigen (CEA), and epithelial membrane antigen (EMA). The finding of these markers in the bone marrow indicate the presence of metastatic tumor cells. Repeated analysis of bone marrow may therefore be a convenient technique for accessing the met-
astatic process and responses to systemically based treatment. Strategies that search for cells bearing these markers within marrow include morphological identification by immunohistochemical staining, antibody-based identification by flow cytometry, reverse transcription–polymerase chain reaction (rt-PCR), and enzyme-linked immunoassay (ELISA) for epithelium-specific proteins (Fig. 1). Potential Bone Marrow Sampling Errors Variable distribution of tumor throughout the body or at different marrow aspiration sites creates the potential for sampling errors, particularly false-negative detection (21).
1646
Maguire et al.
Bilateral ilial crest sampling is superior to unilateral sampling (22), and most studies now include both sites to optimize results with minimal added morbidity or inconvenience (8, 23–25). A study of breast cancer patients with M0 disease revealed that triple sampling (left and right iliac crests and sternum) increased the frequency of detection of micrometastases from 11% to 28% (22). More recently, we found yields of micrometastases from resected rib segments to be superior to those from the iliac crest aspirates in patients undergoing surgery for foregut cancers (26) (O’Sullivan et al., unpublished). The difference between sites was most likely due to enhanced quality of marrow undiluted by blood rather than to site-specific effect. Marrow aspirates are diluted to a variable degree by peripheral blood from venous sinuses in medullary bone that compromises successful detection. The quality of marrow obtained in different studies and by different investigators is therefore likely to vary considerably. This highlights the need for a standardized method of assessment and reporting and eliminate dilution artifact. A proposed approach would be to relate the numbers of detected micrometastatic cells to a standard for marrow cells such as the megakaryocyte (platelet precursor cells found in marrow but not peripheral blood) or stem cell count (blood cell precursor found in marrow). Such standardization is essential for comparable data and ultimately for determining the significance of negative analysis. Markers Used for Detection of Micrometastases The epithelial cell protein markers used in detection of micrometastatic disease include cytokeratins for gastrointestinal, breast, and prostatic malignancies (8, 24 –27), epithelial membrane antigen and mucinous-like carcinoma antigen for breast carcinoma (28 –30), and carcinoembryonic antigen (CEA) for gastrointestinal malignancy (31, 32). The family of cytokeratins comprises about 20 distinct members, with an overall homology of structure, size, and charge (33). These constitute the intermediate filaments of epithelial cytoskeleton and account for almost 85% of the total cellular protein (34). The expression pattern of cytokeratins reflects various pathways of epithelial differentiation. Cytokeratins 8, 18, 19, and 20 are restricted mainly to simple epithelial and are conserved in tumors derived from these tissues (35–37). Control studies of freshly isolated hematopoietic cells from patients without epithelial malignancies show that they are neither stained by pancytokeratin antibodies nor by antibodies specific for cytokeratins 8, 18, or 19 (24, 38). A false-positive rate of ⬍2% has been reported in all studies (8, 23–28, 38). A low level of ectopic cytokeratin expression has been reported in some malignant hematological cells suggesting a potential for disease-induced false-positive analysis (39, 40). Use of cytokeratin 18 in the detection of micrometastases has been validated by several clinical studies (23–27, 38). Comparative study of marrow samples stained using the broad-spectrum monoclonal antibody A45–B/B3, and an antibody directed against
AJG – Vol. 95, No. 7, 2000
cytokeratin 18 (CK18) demonstrated a 50% down-regulation of CK18 expression in micrometastatic cells in bone marrow aspirates from patients with cancers of the breast, lung, prostate, or colorectum (41). Therefore optimal detection of micrometastases should be achieved by examining for more than one type of cytokeratin (given their differential expression in tumor and metastatic cells). The potential value of epithelial membrane antigen (EMA) as a marker of micrometastatic disease is still uncertain. EMA has been reported as a marker for micrometastases in breast cancer patients (28 –30), but attempts to detect this marker in lymph nodes from these patients were associated with an unacceptably high false-positive rate (28). Carcinoembryonic antigen (CEA) has been explored mainly in the detection of gastrointestinal and breast carcinomas. The discriminative value of immunocytochemical staining of bone marrow aspirates for CEA has been disappointing. On the other hand, when messenger ribonucleic acid (mRNA) coding for the CEA protein was examined using molecular techniques (reverse transcription–polymerase chain reaction), seems to be both sensitive and specific for bone marrow micrometastases (42, 43). In these studies, there was no detectable CEA mRNA in marrow from control subjects, whereas single CEA-expressing tumor cells were reliably detected among 2–5 ⫻ 107 normal marrow cells. Methodology for Detection of Epithelial Cell Components Within Marrow Detection of metastatic cells in human bone marrow by light microscopy requires both tumor cell–specific staining and morphological recognition. Because micrometastatic cells are not recognized against the marrow background by routine stains, optimal detection requires prior enrichment of the nucleated cell component of the marrow by density centrifugation and application of cells to adherent slides by cytospin. From 1 to 4 million cells are scanned by light microscopy, which makes the process labor-intensive. Flow cytometric analysis of immunocytochemically stained marrow for tumor cell contamination has the advantage of enumeration (8, 23) and allows for such automation. Marrow samples pretreated with fluorescent labeled monoclonal antibody directed against the epithelial cell component are passed through a detection system that can count total cell numbers and cells with adherent antibody. The validity of this technique has been confirmed in control experiments in which known quantities of cytokeratin-positive cells were added to bone marrow from a patient with no epithelial malignancy (23). Although the technique was sensitive for very low dilutions of tumor cells, levels of ⬎10 metastatic cells per 105 marrow cells is the accurate range. Use of additional monoclonal antibodies to different cytokeratins simultaneously will further increase the sensitivity and efficiency of flow-cytometric analysis of bone marrow micrometastases, as will multiparameter (to look for more than a single entity) rare event analysis (44). This technique,
AJG – July, 2000
which refines conventional flow cytometry by the addition of fluorochrome to stain background cells and particles and is adjusted to remove non–Poisson-distributed events effectively, removes nonspecific fluorescence, auto-fluorescence, background particles, and bursts of events during acquisition from analysis. The potential for false-positive results is thereby reduced and allows for increased sensitivity (44). Molecular techniques have been explored to improve the sensitivity and specificity of detection methods for micrometastatic and residual disease. Detection of messenger ribonucleic acid (mRNA) coding for a specific marker protein by reverse transcription–polymerase chain reaction (rtPCR) may be the most sensitive approach, but awareness of pitfalls in relation to cytokeratin markers of epithelial micrometastases is critical. Essentially the technique detects specific mRNA (codes for the specific epithelial cell protein) and reverse transcribes this to cDNA. This is then amplified by a polymerase chain reaction such that recordable levels may be detected. Pseudogenes for some cytokeratins including CK18 are present in normal marrow cells (42, 45), and amplification occurs in the absence of the specific cytokeratin, resulting in false-positive results. This means that the purity of RNA preparations from marrow must be absolute without genomic DNA contamination. The problem does not apply for all cytokeratins; successful molecular detection of micrometastases in patients with colorectal cancer has been shown using rt-PCR for CK20 message (46). Alternatively, CEA may be better suited to rtPCR detection than cytokeratin (42, 43). Detection of single metastatic cells per 2 ⫻ 107 bone marrow cells has been reported using rt-PCR for CEA message, and this was more sensitive than immunostaining for CEA or cytokeratins (42, 43). A further pitfall of detection of bone marrow micrometastases by rt-PCR relates to the potential for false-positive results because of its exquisite sensitivity. This may arise because of low levels of illegitimate transcription (47), and is particularly likely if there is any contamination with skin epithelia. Enzyme-linked immunoassay (ELISA) has recently been used to detect and to quantify CK19 in lysates of bone marrow aspirates from patients with primary cancers of the breast, stomach, and colorectum (48). Significant concordance was noted between ELISA for CK19 and immunocytochemical analysis for CK18. The technique offers the advantage over immunocytochemistry in that it is rapid, can be automated, and is less observer-dependent, but further studies are required to evaluate sensitivity and specificity. Several studies have also suggested that concurrent fluorescence in situ hybridization (FISH) may be useful to confirm the presence of micrometastases in marrow that tests positive using immunocytochemical methods (49, 50). The most accurate and efficient techniques of micrometastases detection will only be determined by comparative study of the different methods using appropriate markers.
Micrometastases and GI Cancer
1647
These studies, combined with follow-up data from patients measuring disease-free interval and survival are necessary before routine clinical application. Our current practice is to sample iliac crest marrow bilaterally and to take marrow from resected rib at esophagectomy.
BIOLOGICAL PROPERTIES OF BONE MARROW MICROMETASTASES Origin of Micrometastases The possibility that cytokeratin-positive micrometastases might not originate from the primary tumor and theoretically might represent harmless epithelioid reactions within marrow has been addressed. Indirect support of a tumorderived origin of micrometastases was found in patients with prostatic cancer in whom the marrow deposits were found to demonstrate chromosomal aneusomies similar to the primary tumor (51). Also, double staining of marrow from patients with prostatic cancer for cytokeratin and prostatic-specific antigen (PSA), demonstrated a significant concordance rate of expression of PSA between primary prostatic carcinoma cells and marrow micrometastases (52, 53). Collectively, these studies indicate that cytokeratin positive cells found in the marrow of patients with prostatic cancer are cancer cells derived from the primary tumor. There remains the possibility that micrometastatic cells might be transient cells (detached from the primary tumor but nontumorgenic, i.e., unable to establish at a secondary site or unable to escape destruction by the immune system) and may not, therefore, accurately reflect tumor burden in cancer patients. It is likely that some micrometastatic cells are in this category, as a recent study of bone marrow from patients with gastrointestinal cancer before and 6 months after “curative” surgery showed that most patients were able to clear their marrow of metastatic cells (8). However, in several instances, persistence of micrometastatic cells was noted and carried a high risk for development of overt metastatic disease within the subsequent 12–18 months of followup (8). Thus persistent marrow deposits represent residual malignant disease. Viability, Proliferative Potential, and Tumorogenicity The viability and proliferative capacity of bone marrow micrometastases has been confirmed in vitro using combinations of growth factors, especially in the presence of extracellular matrix proteins (54, 55). Cytokeratin-positive cells in the bone marrow of tumor patients also express markers of proliferative activity such as Ki 67 nuclear antigen, receptors for transferrin, and epidermal growth factor (22, 56). In addition, micrometastatic cells from patients with colorectal cancer exhibit measurable levels of nucleolar antigen p120, which is found in the G1 and S phase and is not found in nonmalignant hemopoietic cells (57). We have generated tumor cells lines in culture from rib bone marrow taken from patients undergoing surgery for foregut (esophageal and gastric cardia) cancer, and these
1648
Maguire et al.
AJG – Vol. 95, No. 7, 2000
Table 1. Survival/Relapse of Patients Positive and Negative for Micrometastases Primary Tumor Lung Breast Colorectal Esophagus
Survival/Relapse
Independently Significant
Reference
7.3 vs 35.1 months survival 67 vs 37% relapse at 13 months 33 vs 2% relapse at 2 yr 57 vs 30% relapse by 12–58 months 64 vs 11% relapse at 2 yr
Yes Yes Yes Yes Yes
66 70 69 25 74
cultured cells are tumorigenic when transferred to immunodeficient athymic nude mice (26). Tumorigenicity of bone marrow micrometastases has also been inadvertently uncovered in patients when autologous bone marrow transplants have been used in the treatment of epithelial malignancies (58, 59). Taken together, these data indicate that micrometastatic cells are viable and tumorigenic. Survival of Micrometastatic Cells in a Hostile Environment It would be expected that isolated and small groups of tumor cells in hemopoietic bone marrow should be accessible to immunological attack. However, it seems that these cells can survive for long periods (60). It is likely that metastatic cells deploy the same defensive strategies used by primary tumors to evade the immune system. These include the production of immunosuppressive factors and cytokines to create a local microenvironment of immunosuppression (61). Metastatic tumor cells can also express the Fas ligand on their surface and thereby potentially trigger apoptosis of immune cells if cell– cell contact occurs (62, 63). Binding of leukocytes to tumor cells involves the expression of intercellular adhesion molecules. Altered expression of these in micrometastases may be involved also in immune escape. For example, the prognosis of patients with micrometastatic non–small cell lung cancer was found to be significantly better if the bone marrow micrometastatic cells expressed the adhesion molecule ICAM-1 (64). Metastatic tumor cells may also effectively camouflage themselves by down-regulation of expression of major histocompatibility complex (MHC) class I antigens for presentation of tumor antigens to the immune system and generation of cytotoxic T cells (65). Despite several mechanisms of immune evasion, micrometastatic cells do not always persist as dormant disease and may be eliminated in some patients. In a comparative study of bone marrow before and after surgery for gastrointestinal malignancy, it seemed that those patients who were demonstrated to clear their marrow of micrometastatic disease had the best prognosis (8).
CLINICAL RELEVANCE OF BONE MARROW MICROMETASTASES Prevalence The frequency of micrometastases in bone marrow aspirates of cancer patients, who have no clinical or radiological evidence of metastatic disease is higher than would be
expected. For example, detection of micrometastases in patients undergoing potentially curative resection of colorectal and esophagogastric carcinomas is 88% (23). Detection rates for other common cancers have been reported in the same range or higher (66 – 69). Approximately 50% of patients with recurrent colorectal cancer have marrow micrometastases, and 10% of patients without evidence of overt recurrence by clinical and radiological restaging after “curative” colectomy also test positive for neoplastic cells in the bone marrow (8). These studies support the suspicion that the extent of malignant disease is currently underestimated at diagnosis and follow-up. Prognostic Significance Detection of bone marrow micrometastases is a marker of poor prognosis and has been associated with decreased disease-free survival in patients with non–small cell lung cancer (66, 70), breast carcinoma (69), colorectal carcinoma (8, 25), and gastric cancer (38). Differences in relapse rates for the patients with and without marrow micrometastases are statistically significant (Table 1). The prognostic significance of bone marrow micrometastases in comparison with lymph node metastases was reported in one study of patients with breast cancer, and the results suggested that marrow analysis of micrometastases might become an alternative to axillary node dissection in these patients (71). There is some evidence from a study of bone marrow before and 6 –12 months after surgery that a small number of patients clear their marrow of micrometastatic cells and that these have a favorable prognosis (8). Whether this reflects the biological behavior of the primary tumor or the host immune response is uncertain and requires further study. Therapeutic Implications Micrometastatic spread is undetected by conventional imaging techniques. As this spread influences outcome, patients with bone marrow micrometastases who have no other clinical, biochemical, or radiological evidence of residual disease should receive systemic treatment, provided that an effective therapy is available. High recurrences rates in distant organs (occasionally, after long periods of time), indicate that micrometastatic cells must have been present at the time of initial surgery, persisting in a dormant state for variable periods. Whether such dormant micrometastatic cells are responsive to conventional chemo- or immunotherapy is currently not known, but this is a resolvable question. In our study of bone
AJG – July, 2000
marrow from resected ribs of patients with esophageal carcinoma, we found viable micrometastases even in the group who underwent neoadjuvant chemotherapy (some of these patients had undergone marked pathological response of their primary tumors). This is circumstantial evidence that micrometastases may not respond to conventional chemotherapy. In a randomized, multicenter trial of intravenous monoclonal antibody antiepithelial antibody in patients with Duke’s C colon cancer, there was a significant reduction in the overall death rate by 30% and a reduction in the recurrence rate by 27%, suggesting that micrometastatic cells may be sensitive to immunotherapy (72). In a prospective randomized study of breast and colorectal cancer, patients who received monoclonal antibody therapy responded with a decrease in cytokeratin-positive cells within their bone marrows (73). Finally, the viability and tumorigenicity of bone marrow micrometastases has obvious therapeutic implications when autologous bone marrow transplantation is contemplated for patients with malignant disease. Thus, peripheral blood stem cell harvest may be preferable for these patients, as there is less contamination with viable tumor cells (67).
CONCLUSIONS Conventional methods for the assessment of tumor load understage many patients. The metastatic process is not haphazard and represents a complex process by which cells invade blood vessels, evade the immune system, may remain dormant for prolonged periods, and become re-established as metastases at a distant site. Occult deposits of cells within bone marrow fulfill the criteria required for successful metastases. Bone marrow represents a convenient window on the metastatic process because it is an accessible mesenchymal tissue in which deposits of neoplastic cytokeratin-positive epithelial cells may be identified and quantified by several techniques. Bone marrow micrometastases are viable cells with proliferative and tumorigenic potential. Their presence preoperatively is associated with a poor outcome. Persistence of micrometastases following “curative” surgery indicates minimal residual disease that also carries a poor prognosis, whereas clearance of marrow micrometastases is associated with a favorable outcome. The incorporation of standardized bone marrow analytic techniques into current staging protocols is necessary, as it is currently the best indicator of minimal residual disease and the efficacy of treatment modalities. Identification of patients with systemic metastatic disease or minimal residual disease at intervals is important because such patients may benefit from adjuvant therapy.
ACKNOWLEDGMENTS The investigators are supported in part by the Health Research Board of Ireland, Forbairt Ireland, and by the Cancer Research Appeal Mercy Hospital (CRAM), Cork, Ireland.
Micrometastases and GI Cancer
1649
Reprint requests and correspondence: Fergus Shanahan, M.D., Department of Medicine, Cork University Hospital, Cork, Ireland. Received July 20, 1998; accepted Feb. 25, 2000.
REFERENCES 1. Hart IR, Saini A. Biology of tumour metastases. Lancet 1992; 339:1453–7. 2. Liotta LA, Stetler-Stevensen WG. Tumour invasion and metastases: An imbalance of positive and negative regulation. Cancer Res 1991;51(suppl):5054 –9. 3. Liotta LA, Stegg PS, Stetler-Stevensen WG. Cancer metastases and angiogenesis: An imbalance of positive and negative regulation. Cell 1991;64:327–36. 4. O’Brien MG, Fitzgerald EF, Lee G, et al. A prospective comparison of laparoscopy and imaging in the staging of esophagogastric cancer before surgery. Am J Gastroenterol 1995;90:2191– 4. 5. Posner MR, Mayer RJ. The use of serologic tumor markers in gastrointestinal malignancies. Hematol Oncol Clin North Am 1994;8:533–53. 6. Pandha HS, Waxman J. Tumour markers. Q J Med 1995;88: 233– 41. 7. NIH Consensus conference. Adjuvant therapy for patients with colon and rectal cancer. JAMA 1991;264:1444 –50. 8. O’Sullivan G, Collins JK, Kelly J, et al. Micrometastases: Marker of metastatic potential or evidence of residual disease Gut 1997;40:512–5. 9. Hart IR, Goode NT, Wilson RE. Molecular aspects of the metastatic cascade Biochem Biophys Acta 1989;989:65– 84. 10. Duffy MJ. Cancer metastasis: Biological and clinical aspects. Ir J Med Sci 1998;167:4 – 8. 11. Jiang WG, Hallett MB, Puntis MCA. Motility factors in cancer invasion and metastasis. Surg Res Commun 1994;16:219 –37. 12. Matrisian LM. The atrix-degrading metalloproteinases. Bio Essays 1992;14:455– 63. 13. Albelda SM. Role of integrins and other cell adhesion molecules in tumour progression and metastases. Lab Invest 1993; 68:4 –17. 14. Hynes RO. Integrins: Versatility, modulation and signalling in cell adhesion. Cell 1992;69:11–25. 15. Banks RE, Gearing AJ, Hemmingway IK, et al. Circulating intercellular adhesion molecule-1 (ICAM-1), E-selectin and vascular cell adhesion molecule-1 (VCAM-1) in human malignancies. Br J Cancer 1993;68:122– 4. 16. Gearing AJ, Hemingway I, Pigott R, et al. Soluble forms of vascular adhesion molecules E-selection, ICAM-1, and VCAM-1: Pathological significance Ann NY Acad Sci 1992; 667:324 –31. 17. Fidler IJ, Hart IR. Biological diversity in metastatic neoplasms: Origins and implications. Science 1982;217:998 – 1003. 18. Fidler IJ. Metastases: Quantitative analysis of distribution and fate of tumour emboli labelled with 125I-2-deoxyuridine. J Natl Cancer Inst 1970;45:773– 82. 19. Schlimok G, Funke I, Holzmann B, et al. Micrometastatic cancer cells in bone marrow: In vitro detection with anticytokeratin and in vivo labelling with anti-17-1-A monoclonal antibodies. Proc Natl Acad Sci USA 1987;84:8672– 6. 20. Gatter KC, Alcock C, Heryet A, et al. Clinical importance of analysing malignant tumours of uncertain origin with immunohistochemical techniques. Lancet 1985;1:1302–5. 21. Mansi JL, Berger U, Easton D. Micrometastases in bone marrow patients with primary breast cancer: Evaluation as an
1650
22. 23. 24.
25. 26. 27.
28. 29.
30. 31. 32.
33. 34. 35. 36. 37.
38.
39.
40.
41.
Maguire et al.
early predictor of bone metastases. Br Med J 1987;295: 1093– 6. Schlimok G, Riethmuller G. Detection, characterization and tumorgenicity of disseminated tumor cells in human bone marrow. Sem Cancer Biol 1990;1:207–15. O’Sullivan GC, Collins JK, O’Brien F, et al. Micrometastases in bone marrow of patients undergoing “curative” surgery for gastrointestinal cancer. Gastroenterology 1995;109:1535– 40. Schlmok G, Funke I, Holzmann B, et al. Micrometastatic cancer cells in bone marrow: In vitro detection with anticytokeratin and in vivo labelling with anti-17-1A monoclonal antibodies. Proc Natl Acad Sci USA 1987;84:8672– 6. Lindemann F, Schlimok G, Dirschedel P, et al. Prognostic significance of micrometastatic tumour cells in bone marrow of colorectal cancer patients. Lancet 1992;340:685–9. O’Sullivan GC, Clarke A, Stuart R, et al. Micrometastases in esophagogastric cancer: High detection rate in resected rib segments. Gastroenterology 1999;116:543– 8. Pantel K, Aignherr C, Kollermann J, et al. Immunocytochemical detection of isolated tumour cells in bone marrow of patients with untreated stage C prostatic cancer. Eur J Cancer 1995;31A:1627–32. Tsuchiya A, Sugano K, Klmijima I et al. Immunohistochemical evaluation of lymph node micrometastases from breast cancer. Acta Oncol 1996;35:425– 8. Harbeck N, Untch M, Pache L, et al. Tumour cell detection in bone marrow of breast cancer patients at primary therapy: Results of a 3-year median follow-up. Br J Cancer 1994;69: 566 –71. Basu D, Singh T, Shinghal RN. Micrometastases in bone marrow in breast cancer. Indian J Pathol Microbiol 1994;37: 159 – 64. Neumaier M, Gerhard M, Wagener C. Diagnosis of micrometastases by the amplification of tissue-specific genes. Gene 1995;159:43–7. Gerhard M, Juhl H, Kalthoff H, et al. Specific detection of carcinoembryonic antigen expressing tumour cells in bone marrow aspirates by polymerase chain reaction. J Clin Oncol 1994;12:725–9. Moll R, Franke WW, Schiller DL, et al. The catalog of human cytokeratins: Pattern of expression in normal epithelia, tumours and cultured cells. Cell 1982;31:11–24. Fuchs E. Keratins as biochemical markers of epithelial differentiation. Trends Genet 1988;4:277– 81. Badder BL, Frank WW. Cell type-specific and efficient synthesis of human cytokeratin 19 in transgenic mice. Differentiation 1990;45:109 –18. Stasiak PC, Lane EB. Sequence of cDNA coding for human cytokeratin 19. Nucleic Acids Res 1987;15:10058. Stasiak PC, Purkis PE, Leigh IM et al. Keratin 19: Predicted amino acid sequence and broad tissue distribution suggest it evolved from keratinocyte keratins. J Invest Dermatol 1989; 92:707–16. Schlimok G, Funke I, Pantel K, et al. Micrometastatic tumour cells in bone marrow of patients with gastric cancer: Methodological aspects of detection and prognostic significance. Eur J Cancer 1991;27:1461–5. Knapp AC, Franke WW. Spontaneous losses of control of cytokeratin gene expression in transformed non-epithelial human cells occurring at different levels of regulation. Cell 1989;59:67–9. Jarvinen M, Andersson L, Virtanen I. K562 erythroleukamia cells express cytokeratins 8, 18 and 19 and epithelial membrane antigen that dissapear after induced differentiation. J Cell Physiol 1990;143:310 –20. Pantel K, Schlimok G, Angstwurm M, et al. Methodological analysis of immunocytochemical screening for disseminated
AJG – Vol. 95, No. 7, 2000
42. 43.
44. 45. 46. 47. 48.
49. 50.
51. 52.
53. 54. 55. 56. 57. 58.
59.
60. 61.
epithelial tumour cells in bone marrow. J Haematotherapy 1994;3:165–73. Neumaier M, Gerhard M, Wagener C. Diagnosis of micrometastases by the amplification of tissue-specific genes. Gene 1995;159:43–7. Gerhard M, Juhl H, Kalthoff H, et al. Specific detection of carcinoembryonic antigen expressiong tumour cells in bone marrow aspirates by polymerase chain reaction. J Clin Oncol 1994;12:725–9. Hussain M, Kurkuraga M, Biggar S, et al. Prostate cancer: Flow cytometric methods for detection of bone marrow micrometastases. Cytometry 1996;26:40 – 6. Traweek ST, Liu J, Battifora H. Keratin gene expression in non-epithelial tissues. Am J Pathol 1993;142:1111– 8. Burchill SA, Bradbury MF, Pitman K, et al. Detection of epithelial cancers in peripheral blood by reverse transcriptasepolymerase chain reaction. Br J Cancer 1995;71:278 – 81. Chelly J, Concordet JP, Kaplan JC, et al. Illegitimate transcription: Transcription of any gene in any cell type. Proc Natl Acad Sci USA 1989;86:2617–21. Hochtlen-Vollamar W, Gruber R, Bodenmuller H, et al. Occult epithelial tumour cells detected by an enzyme immunoassay specific for cytokeratin 19. Int J Cancer 1997;70:396 – 400. Little VR, Lockett SJ, Pallavicini MG. Genotype/phenotype analyses of low frequency tumour cells using computerized image microscopy. Cytometry 1996;23:344 –57. Muller P, Weckermann D, Reithmu¨ller G, et al. Detection of genetic alteration in micrometastatic cells in bone marrow of cancer patients by fluorescence in situ hybridization. Cancer Genet Cytogenet 1996;88:8 –16. Pallavicini MG, Cher ML, Bowers EE, et al. Chromosomal aneusomies in prostate cancer micrometastasis. Proc Am Acad Cancer Res 1995;36:69. Riesenberg R, Oberneder R, Kriegmair M, et al. Immunocytochemical double staining of cytokeratin and prostate specific antigen individual prostatic tumour cells. Histochemistry 1993;99:61– 6. Gallee MP, Van Der Korput HA, Van Der Kwast TH, et al. Characterization of monoclonal antibodies raised against prostatic cell line PC-82. Prostate 1986;9:33– 45. Braun S, Pantel K. Biological characteristics of micrometastatic carcinoma cells in bone marrow. Curr Top Microbiol Immunol 1996;213:163–77. Pantel K, Dickmanns A, Zippelius A, et al. Establishment of micrometastatic carcinoma cell lines: A novel source of tumour cell vaccines. J Natl Cancer Inst 1995;87:1162– 8. Pantel K, Schlimok G, Braun S, et al. Differential expression of proliferation-associated molecules in individual micrometastatic carcinoma cells. J Natl Cancer Inst 1993;85:1419 –24. Braun S, Pantel K. Biological characteristics of micrometastatic carcinoma cells in bone marrow. Curr Top Microbiol Immunol 1996;213:163–77. Shpall EJ, Stemmer SM, Bearman SI, et al. New strategies in marrow purging for breast cancer patients receiving high-dose chemotherapy with autologous bone marrow transplantation. Breast Cancer Res Treat 1993;26(suppl):519 –27. Ross AA, Miller GW, Moss TJ, et al. Immunocytochemical detection of tumour cells in bone marrow and peripheral blood stem cell collections from patients with ovarian cancer. Bone Marrow Transplant 1995;15:923–9. Riethmu¨ller G, Johnson JP. Monoclonal antibodies in the detection and therapy of micrometastatic epithelial cancer. Curr Opin Immunol 1992;4:647–55. O’Sullivan GC, Corbett AR, Shanahan F, et al. Regional immunosuppression in esophageal squamous cancer: Evi-
AJG – July, 2000
62. 63.
64.
65.
66. 67.
68.
dence from functional studies with matched lymph nodes. J Immunol 1996;157:4717–20. O’Connell J, O’Sullivan GC, Collins JK, et al. The Fas counter attack: Fas-mediated T cell killing by colon cancer cells expressing Fas ligand. J Exp Med 1996;184:1075– 82. Bennett MW, O’Connell J, O’Sullivan GC, et al. The Fas counterattack in vivo: Apoptotic depletion of tumor-infiltrating lymphocytes (TIL) associated with Fas ligand (FasL) expression by human esophageal carcinoma. J Immunol 1998; 1600:5669 –75. Passlick B, Izbicki JR, Simmel S, et al. Expression of major histocompatibility class I and class II antigens and intercellular adhesion molecule-1 expression on operable non-small cell lung carcinomas. Eur J Cancer 1995;30:376 – 81. Pantel K, Schlimok G, Kutter D, et al. Frequent down regulation of major histocompatibility class I antigen expression on individual micrometastatic carcinoma cells. Cancer Res 1991; 51:4712–5. Cote RJ, Beattie EJ, Chaiwun B, et al. Detection of occult bone marrow micrometastases in patients with operable lung carcinoma. Ann Surg 1995;222:415–23. Ross AA, Miller GW, Moss TJ, et al. Immunocytochemical detection of tumour cells in bone marrow and peripheral blood stem cells collections from patients with ovarian cancer. Bone Marrow Transplant 1995;15:929 –33. Pantel K, Aighnerr C, Kollermann J, et al. Immunocytochemical detection of isolated tumour cells in bone marrow of
Micrometastases and GI Cancer
69.
70.
71.
72.
73.
74.
1651
patients with untreated stage C prostatic cancer. Eur J Cancer 1995;31A:1627–32. Cote RJ, Rosen PP, Lesser ML, et al. Prediction of early relapse in patients with operable breast cancer by detection of occult bone marrow micrometastases. J Clin Oncol 1991;9: 1749 –56. Pantel K, Izbicki JR, Angstwurm M, et al. Immunocytoclogical detection of bone marrow micrometastases in operable non-small cell lung cancer. Cancer Res 1993;53:1027–31. Diel IJ, Kaufmann M, Costa SD, et al. Micrometastatic breast cancer cells in bone marrow at primary surgery: Prognostic value in comparison with nodal status. J Natl Cancer Inst 1996;88:1652– 8. Riethmu¨ller G, Schneider-Gadicke E, Schmiegel W, et al., German Cancer Aid 17-1A Study Group. Randomised trial of monoclonal antibody for adjuvant therapy of resected Duke’s C colorectal carcinoma. Lancet 1994;343:1177– 86. Schlimok G, Pantel K, Loibner H, et al. Reduction of metastatic carcinoma cells in bone marrow by intravenously administered monoclonal antibody: Towards a novel surrogate test to monitor adjuvant therapies of solid tumours. Br J Cancer 1995;31A:1799 –1803. Thorban S, Roder JD, Nekarda H, et al. Disseminated epithelial tumour cells in bone marrow of patients with esophageal cancer: Detection and prognostic significance. World J Surg 1996;20:567–72.
Vol. 95, No. 7, 2000 ISSN 0002-9270/00/$20.00 PII S0002-9270(00)00991-6
Bone Marrow Micrometastases and Gastrointestinal Cancer Detection and Significance Donal Maguire, M.D., Gerald C. O’Sullivan, M.D., J. Kevin Collins, Ph.D., John Morgan, Ph.D., and Fergus Shanahan, M.D. Department of Surgery and Medicine, Mercy and Cork University Hospitals, and National University of Ireland, Cork, Ireland
ABSTRACT Accurate staging of cancer is important, as the presence or absence of systemic spread determines treatment. The sensitivity of current imaging and biochemical techniques is suboptimal for the detection of minimal residual disease and latent metastases. This results in understaging and potential undertreatment. To improve detection of disseminated epithelial malignancy, immunohistochemical and molecular methods have been employed that search for epithelial cell– specific proteins in nonepithelial tissue. Bone marrow is mesenchymal tissue (that does not normally express epithelial cell components) and represents an accessible window for detection of micrometastatic carcinoma cells. Detection methods for epithelial cell components (cytokeratins, epithelial membrane antigen, carcinoembryonic antigen) include immunohistochemistry, flow cytometry, reverse transcriptase polymerase chain reaction (rt-PCR), and enzyme linked immunoassay (ELISA). Micrometastatic cells in bone marrow are viable, capable of proliferation, resistant to immune attack, and insensitive to s-phase chemotherapeutic agents. Patients with carcinomas of the lung, breast, prostate, or gastrointestinal tract and in whom bone marrow micrometastases are detected have a foreshortened interval to recurrence and impaired survival. Detection of micrometastases deserves serious consideration in treatment protocols, and standardization of methods is now required. (Am J Gastroenterol 2000;95:1644 –1651. © 2000 by Am. Coll. of Gastroenterology)
INTRODUCTION The extent of spread of primary cancer remains the best predictor of clinical outcome and is the major determinant of treatment strategy (1–3). In the presence of disseminated disease, cure by excision of the primary cancer is not possible, so adjuvant systemic therapies are necessary. This implies that rigorous detection of metastatic disease is required to ensure optimal staging and treatment. At present, rigorous detection of subclinical disease is not possible, as available imaging techniques lack the temporal and spatial resolution to discover individual or small groups of disseminated tumor cells (4). Conventional ultrasound, CT, and magnetic resonance scans fail to detect up to 10% of hepatic
metasases and up to 40% of peritoneal recurrences in colorectal cancer patients, whereas in patients with esophagogastric malignancy, ultrasound and CT detects peritoneal or hepatic metastases with a sensitivity of 21% and 47%, respectively (4). Likewise, the sensitivity and specificity of serum tumor antigens is limited (5, 6). Although prognostic markers including DNA content of the primary tumor, oncogene mutations, tumor suppressor genes, and proliferative activity are useful, these fail to provide a direct measure of tumor burden or metastatic spread and do not facilitate repeated study (7). In addition, interval reassessments of disease status are required to monitor treatment efficacy. Therefore, to develop effective therapeutic strategies against minimal residual disease, accurate, reproducible and standardized techniques that can detect and enumerate metastatic cells would be advantageous. Against this background, methods for detecting epithelial micrometastatases in mesenchymal tissue (bone marrow) have been developed and are now the subject of clinical studies (8). In this report we present an appraisal of techniques for detection of bone marrow micrometastases, an overview of the biological properties of the micrometastatic cells, and evaluate their clinical utility including implications for therapy.
MICROMETASTASES AND TUMOR CELL SPREAD The presence of viable micrometastatic deposits in tissues implies dissemination of tumor cells with the capability of independent survival and growth (1, 2). These cells must detach from the primary growth, travel through extracellular matrix and basement membrane into the local microvasculature, survive transit in the circulation, and exit by attaching and extravasating through the endothelium at another site (1, 9). This process involves several interdependent steps regulated by cascades of cytokines, chemokines, growth factors, and matrix metalloproteinases (10 –16). Many cells within the primary tumor are terminally differentiated and do not have the ability to disseminate and grow into metastases (17, 18). In contrast, the finding of malignant cells distant to the primary tumor indicate metastatic phenotype with tumorigenic potential.
AJG – July, 2000
Micrometastases and GI Cancer
1645
Figure 1. Detection of cytokeratin-positive cells within bone marrow by three different techniques: Flow cytometry with dual staining for cytokeratin-18 and propidium iodide (top left); immunocytochemistry with cytokeratin-18 monoclonal antibody and the alkaline phosphatases anti-alkaline phosphatase (APAAP) technique (top right); and reverse transcription polymerase chain reaction (RT-PCR) for detection of mRNA for cytokeratin 19 (bottom). The observed RT-PCR product of 704bp is the predicted product from RT amplification of cytokeratin 19 mRNA. Lane M, 100 bp ladder; lanes 1–7 are patient samples with lanes 1 and 3 testing positively. Lanes 8 –10 represent 1000, 100, and 10 HT29 colon carcinoma cells, respectively.
METHODOLOGY FOR DETECTION OF MICROMETASTASES IN CARCINOMA PATIENTS Bone marrow is derived from the embryonic mesoderm and therefore does not normally express epithelial cell–specific components. Epithelially derived malignancies retain expression of many epithelial specific proteins such as cytokeratins (CKs), carcinoembryonic antigen (CEA), and epithelial membrane antigen (EMA). The finding of these markers in the bone marrow indicate the presence of metastatic tumor cells. Repeated analysis of bone marrow may therefore be a convenient technique for accessing the met-
astatic process and responses to systemically based treatment. Strategies that search for cells bearing these markers within marrow include morphological identification by immunohistochemical staining, antibody-based identification by flow cytometry, reverse transcription–polymerase chain reaction (rt-PCR), and enzyme-linked immunoassay (ELISA) for epithelium-specific proteins (Fig. 1). Potential Bone Marrow Sampling Errors Variable distribution of tumor throughout the body or at different marrow aspiration sites creates the potential for sampling errors, particularly false-negative detection (21).
1646
Maguire et al.
Bilateral ilial crest sampling is superior to unilateral sampling (22), and most studies now include both sites to optimize results with minimal added morbidity or inconvenience (8, 23–25). A study of breast cancer patients with M0 disease revealed that triple sampling (left and right iliac crests and sternum) increased the frequency of detection of micrometastases from 11% to 28% (22). More recently, we found yields of micrometastases from resected rib segments to be superior to those from the iliac crest aspirates in patients undergoing surgery for foregut cancers (26) (O’Sullivan et al., unpublished). The difference between sites was most likely due to enhanced quality of marrow undiluted by blood rather than to site-specific effect. Marrow aspirates are diluted to a variable degree by peripheral blood from venous sinuses in medullary bone that compromises successful detection. The quality of marrow obtained in different studies and by different investigators is therefore likely to vary considerably. This highlights the need for a standardized method of assessment and reporting and eliminate dilution artifact. A proposed approach would be to relate the numbers of detected micrometastatic cells to a standard for marrow cells such as the megakaryocyte (platelet precursor cells found in marrow but not peripheral blood) or stem cell count (blood cell precursor found in marrow). Such standardization is essential for comparable data and ultimately for determining the significance of negative analysis. Markers Used for Detection of Micrometastases The epithelial cell protein markers used in detection of micrometastatic disease include cytokeratins for gastrointestinal, breast, and prostatic malignancies (8, 24 –27), epithelial membrane antigen and mucinous-like carcinoma antigen for breast carcinoma (28 –30), and carcinoembryonic antigen (CEA) for gastrointestinal malignancy (31, 32). The family of cytokeratins comprises about 20 distinct members, with an overall homology of structure, size, and charge (33). These constitute the intermediate filaments of epithelial cytoskeleton and account for almost 85% of the total cellular protein (34). The expression pattern of cytokeratins reflects various pathways of epithelial differentiation. Cytokeratins 8, 18, 19, and 20 are restricted mainly to simple epithelial and are conserved in tumors derived from these tissues (35–37). Control studies of freshly isolated hematopoietic cells from patients without epithelial malignancies show that they are neither stained by pancytokeratin antibodies nor by antibodies specific for cytokeratins 8, 18, or 19 (24, 38). A false-positive rate of ⬍2% has been reported in all studies (8, 23–28, 38). A low level of ectopic cytokeratin expression has been reported in some malignant hematological cells suggesting a potential for disease-induced false-positive analysis (39, 40). Use of cytokeratin 18 in the detection of micrometastases has been validated by several clinical studies (23–27, 38). Comparative study of marrow samples stained using the broad-spectrum monoclonal antibody A45–B/B3, and an antibody directed against
AJG – Vol. 95, No. 7, 2000
cytokeratin 18 (CK18) demonstrated a 50% down-regulation of CK18 expression in micrometastatic cells in bone marrow aspirates from patients with cancers of the breast, lung, prostate, or colorectum (41). Therefore optimal detection of micrometastases should be achieved by examining for more than one type of cytokeratin (given their differential expression in tumor and metastatic cells). The potential value of epithelial membrane antigen (EMA) as a marker of micrometastatic disease is still uncertain. EMA has been reported as a marker for micrometastases in breast cancer patients (28 –30), but attempts to detect this marker in lymph nodes from these patients were associated with an unacceptably high false-positive rate (28). Carcinoembryonic antigen (CEA) has been explored mainly in the detection of gastrointestinal and breast carcinomas. The discriminative value of immunocytochemical staining of bone marrow aspirates for CEA has been disappointing. On the other hand, when messenger ribonucleic acid (mRNA) coding for the CEA protein was examined using molecular techniques (reverse transcription–polymerase chain reaction), seems to be both sensitive and specific for bone marrow micrometastases (42, 43). In these studies, there was no detectable CEA mRNA in marrow from control subjects, whereas single CEA-expressing tumor cells were reliably detected among 2–5 ⫻ 107 normal marrow cells. Methodology for Detection of Epithelial Cell Components Within Marrow Detection of metastatic cells in human bone marrow by light microscopy requires both tumor cell–specific staining and morphological recognition. Because micrometastatic cells are not recognized against the marrow background by routine stains, optimal detection requires prior enrichment of the nucleated cell component of the marrow by density centrifugation and application of cells to adherent slides by cytospin. From 1 to 4 million cells are scanned by light microscopy, which makes the process labor-intensive. Flow cytometric analysis of immunocytochemically stained marrow for tumor cell contamination has the advantage of enumeration (8, 23) and allows for such automation. Marrow samples pretreated with fluorescent labeled monoclonal antibody directed against the epithelial cell component are passed through a detection system that can count total cell numbers and cells with adherent antibody. The validity of this technique has been confirmed in control experiments in which known quantities of cytokeratin-positive cells were added to bone marrow from a patient with no epithelial malignancy (23). Although the technique was sensitive for very low dilutions of tumor cells, levels of ⬎10 metastatic cells per 105 marrow cells is the accurate range. Use of additional monoclonal antibodies to different cytokeratins simultaneously will further increase the sensitivity and efficiency of flow-cytometric analysis of bone marrow micrometastases, as will multiparameter (to look for more than a single entity) rare event analysis (44). This technique,
AJG – July, 2000
which refines conventional flow cytometry by the addition of fluorochrome to stain background cells and particles and is adjusted to remove non–Poisson-distributed events effectively, removes nonspecific fluorescence, auto-fluorescence, background particles, and bursts of events during acquisition from analysis. The potential for false-positive results is thereby reduced and allows for increased sensitivity (44). Molecular techniques have been explored to improve the sensitivity and specificity of detection methods for micrometastatic and residual disease. Detection of messenger ribonucleic acid (mRNA) coding for a specific marker protein by reverse transcription–polymerase chain reaction (rtPCR) may be the most sensitive approach, but awareness of pitfalls in relation to cytokeratin markers of epithelial micrometastases is critical. Essentially the technique detects specific mRNA (codes for the specific epithelial cell protein) and reverse transcribes this to cDNA. This is then amplified by a polymerase chain reaction such that recordable levels may be detected. Pseudogenes for some cytokeratins including CK18 are present in normal marrow cells (42, 45), and amplification occurs in the absence of the specific cytokeratin, resulting in false-positive results. This means that the purity of RNA preparations from marrow must be absolute without genomic DNA contamination. The problem does not apply for all cytokeratins; successful molecular detection of micrometastases in patients with colorectal cancer has been shown using rt-PCR for CK20 message (46). Alternatively, CEA may be better suited to rtPCR detection than cytokeratin (42, 43). Detection of single metastatic cells per 2 ⫻ 107 bone marrow cells has been reported using rt-PCR for CEA message, and this was more sensitive than immunostaining for CEA or cytokeratins (42, 43). A further pitfall of detection of bone marrow micrometastases by rt-PCR relates to the potential for false-positive results because of its exquisite sensitivity. This may arise because of low levels of illegitimate transcription (47), and is particularly likely if there is any contamination with skin epithelia. Enzyme-linked immunoassay (ELISA) has recently been used to detect and to quantify CK19 in lysates of bone marrow aspirates from patients with primary cancers of the breast, stomach, and colorectum (48). Significant concordance was noted between ELISA for CK19 and immunocytochemical analysis for CK18. The technique offers the advantage over immunocytochemistry in that it is rapid, can be automated, and is less observer-dependent, but further studies are required to evaluate sensitivity and specificity. Several studies have also suggested that concurrent fluorescence in situ hybridization (FISH) may be useful to confirm the presence of micrometastases in marrow that tests positive using immunocytochemical methods (49, 50). The most accurate and efficient techniques of micrometastases detection will only be determined by comparative study of the different methods using appropriate markers.
Micrometastases and GI Cancer
1647
These studies, combined with follow-up data from patients measuring disease-free interval and survival are necessary before routine clinical application. Our current practice is to sample iliac crest marrow bilaterally and to take marrow from resected rib at esophagectomy.
BIOLOGICAL PROPERTIES OF BONE MARROW MICROMETASTASES Origin of Micrometastases The possibility that cytokeratin-positive micrometastases might not originate from the primary tumor and theoretically might represent harmless epithelioid reactions within marrow has been addressed. Indirect support of a tumorderived origin of micrometastases was found in patients with prostatic cancer in whom the marrow deposits were found to demonstrate chromosomal aneusomies similar to the primary tumor (51). Also, double staining of marrow from patients with prostatic cancer for cytokeratin and prostatic-specific antigen (PSA), demonstrated a significant concordance rate of expression of PSA between primary prostatic carcinoma cells and marrow micrometastases (52, 53). Collectively, these studies indicate that cytokeratin positive cells found in the marrow of patients with prostatic cancer are cancer cells derived from the primary tumor. There remains the possibility that micrometastatic cells might be transient cells (detached from the primary tumor but nontumorgenic, i.e., unable to establish at a secondary site or unable to escape destruction by the immune system) and may not, therefore, accurately reflect tumor burden in cancer patients. It is likely that some micrometastatic cells are in this category, as a recent study of bone marrow from patients with gastrointestinal cancer before and 6 months after “curative” surgery showed that most patients were able to clear their marrow of metastatic cells (8). However, in several instances, persistence of micrometastatic cells was noted and carried a high risk for development of overt metastatic disease within the subsequent 12–18 months of followup (8). Thus persistent marrow deposits represent residual malignant disease. Viability, Proliferative Potential, and Tumorogenicity The viability and proliferative capacity of bone marrow micrometastases has been confirmed in vitro using combinations of growth factors, especially in the presence of extracellular matrix proteins (54, 55). Cytokeratin-positive cells in the bone marrow of tumor patients also express markers of proliferative activity such as Ki 67 nuclear antigen, receptors for transferrin, and epidermal growth factor (22, 56). In addition, micrometastatic cells from patients with colorectal cancer exhibit measurable levels of nucleolar antigen p120, which is found in the G1 and S phase and is not found in nonmalignant hemopoietic cells (57). We have generated tumor cells lines in culture from rib bone marrow taken from patients undergoing surgery for foregut (esophageal and gastric cardia) cancer, and these
1648
Maguire et al.
AJG – Vol. 95, No. 7, 2000
Table 1. Survival/Relapse of Patients Positive and Negative for Micrometastases Primary Tumor Lung Breast Colorectal Esophagus
Survival/Relapse
Independently Significant
Reference
7.3 vs 35.1 months survival 67 vs 37% relapse at 13 months 33 vs 2% relapse at 2 yr 57 vs 30% relapse by 12–58 months 64 vs 11% relapse at 2 yr
Yes Yes Yes Yes Yes
66 70 69 25 74
cultured cells are tumorigenic when transferred to immunodeficient athymic nude mice (26). Tumorigenicity of bone marrow micrometastases has also been inadvertently uncovered in patients when autologous bone marrow transplants have been used in the treatment of epithelial malignancies (58, 59). Taken together, these data indicate that micrometastatic cells are viable and tumorigenic. Survival of Micrometastatic Cells in a Hostile Environment It would be expected that isolated and small groups of tumor cells in hemopoietic bone marrow should be accessible to immunological attack. However, it seems that these cells can survive for long periods (60). It is likely that metastatic cells deploy the same defensive strategies used by primary tumors to evade the immune system. These include the production of immunosuppressive factors and cytokines to create a local microenvironment of immunosuppression (61). Metastatic tumor cells can also express the Fas ligand on their surface and thereby potentially trigger apoptosis of immune cells if cell– cell contact occurs (62, 63). Binding of leukocytes to tumor cells involves the expression of intercellular adhesion molecules. Altered expression of these in micrometastases may be involved also in immune escape. For example, the prognosis of patients with micrometastatic non–small cell lung cancer was found to be significantly better if the bone marrow micrometastatic cells expressed the adhesion molecule ICAM-1 (64). Metastatic tumor cells may also effectively camouflage themselves by down-regulation of expression of major histocompatibility complex (MHC) class I antigens for presentation of tumor antigens to the immune system and generation of cytotoxic T cells (65). Despite several mechanisms of immune evasion, micrometastatic cells do not always persist as dormant disease and may be eliminated in some patients. In a comparative study of bone marrow before and after surgery for gastrointestinal malignancy, it seemed that those patients who were demonstrated to clear their marrow of micrometastatic disease had the best prognosis (8).
CLINICAL RELEVANCE OF BONE MARROW MICROMETASTASES Prevalence The frequency of micrometastases in bone marrow aspirates of cancer patients, who have no clinical or radiological evidence of metastatic disease is higher than would be
expected. For example, detection of micrometastases in patients undergoing potentially curative resection of colorectal and esophagogastric carcinomas is 88% (23). Detection rates for other common cancers have been reported in the same range or higher (66 – 69). Approximately 50% of patients with recurrent colorectal cancer have marrow micrometastases, and 10% of patients without evidence of overt recurrence by clinical and radiological restaging after “curative” colectomy also test positive for neoplastic cells in the bone marrow (8). These studies support the suspicion that the extent of malignant disease is currently underestimated at diagnosis and follow-up. Prognostic Significance Detection of bone marrow micrometastases is a marker of poor prognosis and has been associated with decreased disease-free survival in patients with non–small cell lung cancer (66, 70), breast carcinoma (69), colorectal carcinoma (8, 25), and gastric cancer (38). Differences in relapse rates for the patients with and without marrow micrometastases are statistically significant (Table 1). The prognostic significance of bone marrow micrometastases in comparison with lymph node metastases was reported in one study of patients with breast cancer, and the results suggested that marrow analysis of micrometastases might become an alternative to axillary node dissection in these patients (71). There is some evidence from a study of bone marrow before and 6 –12 months after surgery that a small number of patients clear their marrow of micrometastatic cells and that these have a favorable prognosis (8). Whether this reflects the biological behavior of the primary tumor or the host immune response is uncertain and requires further study. Therapeutic Implications Micrometastatic spread is undetected by conventional imaging techniques. As this spread influences outcome, patients with bone marrow micrometastases who have no other clinical, biochemical, or radiological evidence of residual disease should receive systemic treatment, provided that an effective therapy is available. High recurrences rates in distant organs (occasionally, after long periods of time), indicate that micrometastatic cells must have been present at the time of initial surgery, persisting in a dormant state for variable periods. Whether such dormant micrometastatic cells are responsive to conventional chemo- or immunotherapy is currently not known, but this is a resolvable question. In our study of bone
AJG – July, 2000
marrow from resected ribs of patients with esophageal carcinoma, we found viable micrometastases even in the group who underwent neoadjuvant chemotherapy (some of these patients had undergone marked pathological response of their primary tumors). This is circumstantial evidence that micrometastases may not respond to conventional chemotherapy. In a randomized, multicenter trial of intravenous monoclonal antibody antiepithelial antibody in patients with Duke’s C colon cancer, there was a significant reduction in the overall death rate by 30% and a reduction in the recurrence rate by 27%, suggesting that micrometastatic cells may be sensitive to immunotherapy (72). In a prospective randomized study of breast and colorectal cancer, patients who received monoclonal antibody therapy responded with a decrease in cytokeratin-positive cells within their bone marrows (73). Finally, the viability and tumorigenicity of bone marrow micrometastases has obvious therapeutic implications when autologous bone marrow transplantation is contemplated for patients with malignant disease. Thus, peripheral blood stem cell harvest may be preferable for these patients, as there is less contamination with viable tumor cells (67).
CONCLUSIONS Conventional methods for the assessment of tumor load understage many patients. The metastatic process is not haphazard and represents a complex process by which cells invade blood vessels, evade the immune system, may remain dormant for prolonged periods, and become re-established as metastases at a distant site. Occult deposits of cells within bone marrow fulfill the criteria required for successful metastases. Bone marrow represents a convenient window on the metastatic process because it is an accessible mesenchymal tissue in which deposits of neoplastic cytokeratin-positive epithelial cells may be identified and quantified by several techniques. Bone marrow micrometastases are viable cells with proliferative and tumorigenic potential. Their presence preoperatively is associated with a poor outcome. Persistence of micrometastases following “curative” surgery indicates minimal residual disease that also carries a poor prognosis, whereas clearance of marrow micrometastases is associated with a favorable outcome. The incorporation of standardized bone marrow analytic techniques into current staging protocols is necessary, as it is currently the best indicator of minimal residual disease and the efficacy of treatment modalities. Identification of patients with systemic metastatic disease or minimal residual disease at intervals is important because such patients may benefit from adjuvant therapy.
ACKNOWLEDGMENTS The investigators are supported in part by the Health Research Board of Ireland, Forbairt Ireland, and by the Cancer Research Appeal Mercy Hospital (CRAM), Cork, Ireland.
Micrometastases and GI Cancer
1649
Reprint requests and correspondence: Fergus Shanahan, M.D., Department of Medicine, Cork University Hospital, Cork, Ireland. Received July 20, 1998; accepted Feb. 25, 2000.
REFERENCES 1. Hart IR, Saini A. Biology of tumour metastases. Lancet 1992; 339:1453–7. 2. Liotta LA, Stetler-Stevensen WG. Tumour invasion and metastases: An imbalance of positive and negative regulation. Cancer Res 1991;51(suppl):5054 –9. 3. Liotta LA, Stegg PS, Stetler-Stevensen WG. Cancer metastases and angiogenesis: An imbalance of positive and negative regulation. Cell 1991;64:327–36. 4. O’Brien MG, Fitzgerald EF, Lee G, et al. A prospective comparison of laparoscopy and imaging in the staging of esophagogastric cancer before surgery. Am J Gastroenterol 1995;90:2191– 4. 5. Posner MR, Mayer RJ. The use of serologic tumor markers in gastrointestinal malignancies. Hematol Oncol Clin North Am 1994;8:533–53. 6. Pandha HS, Waxman J. Tumour markers. Q J Med 1995;88: 233– 41. 7. NIH Consensus conference. Adjuvant therapy for patients with colon and rectal cancer. JAMA 1991;264:1444 –50. 8. O’Sullivan G, Collins JK, Kelly J, et al. Micrometastases: Marker of metastatic potential or evidence of residual disease Gut 1997;40:512–5. 9. Hart IR, Goode NT, Wilson RE. Molecular aspects of the metastatic cascade Biochem Biophys Acta 1989;989:65– 84. 10. Duffy MJ. Cancer metastasis: Biological and clinical aspects. Ir J Med Sci 1998;167:4 – 8. 11. Jiang WG, Hallett MB, Puntis MCA. Motility factors in cancer invasion and metastasis. Surg Res Commun 1994;16:219 –37. 12. Matrisian LM. The atrix-degrading metalloproteinases. Bio Essays 1992;14:455– 63. 13. Albelda SM. Role of integrins and other cell adhesion molecules in tumour progression and metastases. Lab Invest 1993; 68:4 –17. 14. Hynes RO. Integrins: Versatility, modulation and signalling in cell adhesion. Cell 1992;69:11–25. 15. Banks RE, Gearing AJ, Hemmingway IK, et al. Circulating intercellular adhesion molecule-1 (ICAM-1), E-selectin and vascular cell adhesion molecule-1 (VCAM-1) in human malignancies. Br J Cancer 1993;68:122– 4. 16. Gearing AJ, Hemingway I, Pigott R, et al. Soluble forms of vascular adhesion molecules E-selection, ICAM-1, and VCAM-1: Pathological significance Ann NY Acad Sci 1992; 667:324 –31. 17. Fidler IJ, Hart IR. Biological diversity in metastatic neoplasms: Origins and implications. Science 1982;217:998 – 1003. 18. Fidler IJ. Metastases: Quantitative analysis of distribution and fate of tumour emboli labelled with 125I-2-deoxyuridine. J Natl Cancer Inst 1970;45:773– 82. 19. Schlimok G, Funke I, Holzmann B, et al. Micrometastatic cancer cells in bone marrow: In vitro detection with anticytokeratin and in vivo labelling with anti-17-1-A monoclonal antibodies. Proc Natl Acad Sci USA 1987;84:8672– 6. 20. Gatter KC, Alcock C, Heryet A, et al. Clinical importance of analysing malignant tumours of uncertain origin with immunohistochemical techniques. Lancet 1985;1:1302–5. 21. Mansi JL, Berger U, Easton D. Micrometastases in bone marrow patients with primary breast cancer: Evaluation as an
1650
22. 23. 24.
25. 26. 27.
28. 29.
30. 31. 32.
33. 34. 35. 36. 37.
38.
39.
40.
41.
Maguire et al.
early predictor of bone metastases. Br Med J 1987;295: 1093– 6. Schlimok G, Riethmuller G. Detection, characterization and tumorgenicity of disseminated tumor cells in human bone marrow. Sem Cancer Biol 1990;1:207–15. O’Sullivan GC, Collins JK, O’Brien F, et al. Micrometastases in bone marrow of patients undergoing “curative” surgery for gastrointestinal cancer. Gastroenterology 1995;109:1535– 40. Schlmok G, Funke I, Holzmann B, et al. Micrometastatic cancer cells in bone marrow: In vitro detection with anticytokeratin and in vivo labelling with anti-17-1A monoclonal antibodies. Proc Natl Acad Sci USA 1987;84:8672– 6. Lindemann F, Schlimok G, Dirschedel P, et al. Prognostic significance of micrometastatic tumour cells in bone marrow of colorectal cancer patients. Lancet 1992;340:685–9. O’Sullivan GC, Clarke A, Stuart R, et al. Micrometastases in esophagogastric cancer: High detection rate in resected rib segments. Gastroenterology 1999;116:543– 8. Pantel K, Aignherr C, Kollermann J, et al. Immunocytochemical detection of isolated tumour cells in bone marrow of patients with untreated stage C prostatic cancer. Eur J Cancer 1995;31A:1627–32. Tsuchiya A, Sugano K, Klmijima I et al. Immunohistochemical evaluation of lymph node micrometastases from breast cancer. Acta Oncol 1996;35:425– 8. Harbeck N, Untch M, Pache L, et al. Tumour cell detection in bone marrow of breast cancer patients at primary therapy: Results of a 3-year median follow-up. Br J Cancer 1994;69: 566 –71. Basu D, Singh T, Shinghal RN. Micrometastases in bone marrow in breast cancer. Indian J Pathol Microbiol 1994;37: 159 – 64. Neumaier M, Gerhard M, Wagener C. Diagnosis of micrometastases by the amplification of tissue-specific genes. Gene 1995;159:43–7. Gerhard M, Juhl H, Kalthoff H, et al. Specific detection of carcinoembryonic antigen expressing tumour cells in bone marrow aspirates by polymerase chain reaction. J Clin Oncol 1994;12:725–9. Moll R, Franke WW, Schiller DL, et al. The catalog of human cytokeratins: Pattern of expression in normal epithelia, tumours and cultured cells. Cell 1982;31:11–24. Fuchs E. Keratins as biochemical markers of epithelial differentiation. Trends Genet 1988;4:277– 81. Badder BL, Frank WW. Cell type-specific and efficient synthesis of human cytokeratin 19 in transgenic mice. Differentiation 1990;45:109 –18. Stasiak PC, Lane EB. Sequence of cDNA coding for human cytokeratin 19. Nucleic Acids Res 1987;15:10058. Stasiak PC, Purkis PE, Leigh IM et al. Keratin 19: Predicted amino acid sequence and broad tissue distribution suggest it evolved from keratinocyte keratins. J Invest Dermatol 1989; 92:707–16. Schlimok G, Funke I, Pantel K, et al. Micrometastatic tumour cells in bone marrow of patients with gastric cancer: Methodological aspects of detection and prognostic significance. Eur J Cancer 1991;27:1461–5. Knapp AC, Franke WW. Spontaneous losses of control of cytokeratin gene expression in transformed non-epithelial human cells occurring at different levels of regulation. Cell 1989;59:67–9. Jarvinen M, Andersson L, Virtanen I. K562 erythroleukamia cells express cytokeratins 8, 18 and 19 and epithelial membrane antigen that dissapear after induced differentiation. J Cell Physiol 1990;143:310 –20. Pantel K, Schlimok G, Angstwurm M, et al. Methodological analysis of immunocytochemical screening for disseminated
AJG – Vol. 95, No. 7, 2000
42. 43.
44. 45. 46. 47. 48.
49. 50.
51. 52.
53. 54. 55. 56. 57. 58.
59.
60. 61.
epithelial tumour cells in bone marrow. J Haematotherapy 1994;3:165–73. Neumaier M, Gerhard M, Wagener C. Diagnosis of micrometastases by the amplification of tissue-specific genes. Gene 1995;159:43–7. Gerhard M, Juhl H, Kalthoff H, et al. Specific detection of carcinoembryonic antigen expressiong tumour cells in bone marrow aspirates by polymerase chain reaction. J Clin Oncol 1994;12:725–9. Hussain M, Kurkuraga M, Biggar S, et al. Prostate cancer: Flow cytometric methods for detection of bone marrow micrometastases. Cytometry 1996;26:40 – 6. Traweek ST, Liu J, Battifora H. Keratin gene expression in non-epithelial tissues. Am J Pathol 1993;142:1111– 8. Burchill SA, Bradbury MF, Pitman K, et al. Detection of epithelial cancers in peripheral blood by reverse transcriptasepolymerase chain reaction. Br J Cancer 1995;71:278 – 81. Chelly J, Concordet JP, Kaplan JC, et al. Illegitimate transcription: Transcription of any gene in any cell type. Proc Natl Acad Sci USA 1989;86:2617–21. Hochtlen-Vollamar W, Gruber R, Bodenmuller H, et al. Occult epithelial tumour cells detected by an enzyme immunoassay specific for cytokeratin 19. Int J Cancer 1997;70:396 – 400. Little VR, Lockett SJ, Pallavicini MG. Genotype/phenotype analyses of low frequency tumour cells using computerized image microscopy. Cytometry 1996;23:344 –57. Muller P, Weckermann D, Reithmu¨ller G, et al. Detection of genetic alteration in micrometastatic cells in bone marrow of cancer patients by fluorescence in situ hybridization. Cancer Genet Cytogenet 1996;88:8 –16. Pallavicini MG, Cher ML, Bowers EE, et al. Chromosomal aneusomies in prostate cancer micrometastasis. Proc Am Acad Cancer Res 1995;36:69. Riesenberg R, Oberneder R, Kriegmair M, et al. Immunocytochemical double staining of cytokeratin and prostate specific antigen individual prostatic tumour cells. Histochemistry 1993;99:61– 6. Gallee MP, Van Der Korput HA, Van Der Kwast TH, et al. Characterization of monoclonal antibodies raised against prostatic cell line PC-82. Prostate 1986;9:33– 45. Braun S, Pantel K. Biological characteristics of micrometastatic carcinoma cells in bone marrow. Curr Top Microbiol Immunol 1996;213:163–77. Pantel K, Dickmanns A, Zippelius A, et al. Establishment of micrometastatic carcinoma cell lines: A novel source of tumour cell vaccines. J Natl Cancer Inst 1995;87:1162– 8. Pantel K, Schlimok G, Braun S, et al. Differential expression of proliferation-associated molecules in individual micrometastatic carcinoma cells. J Natl Cancer Inst 1993;85:1419 –24. Braun S, Pantel K. Biological characteristics of micrometastatic carcinoma cells in bone marrow. Curr Top Microbiol Immunol 1996;213:163–77. Shpall EJ, Stemmer SM, Bearman SI, et al. New strategies in marrow purging for breast cancer patients receiving high-dose chemotherapy with autologous bone marrow transplantation. Breast Cancer Res Treat 1993;26(suppl):519 –27. Ross AA, Miller GW, Moss TJ, et al. Immunocytochemical detection of tumour cells in bone marrow and peripheral blood stem cell collections from patients with ovarian cancer. Bone Marrow Transplant 1995;15:923–9. Riethmu¨ller G, Johnson JP. Monoclonal antibodies in the detection and therapy of micrometastatic epithelial cancer. Curr Opin Immunol 1992;4:647–55. O’Sullivan GC, Corbett AR, Shanahan F, et al. Regional immunosuppression in esophageal squamous cancer: Evi-
AJG – July, 2000
62. 63.
64.
65.
66. 67.
68.
dence from functional studies with matched lymph nodes. J Immunol 1996;157:4717–20. O’Connell J, O’Sullivan GC, Collins JK, et al. The Fas counter attack: Fas-mediated T cell killing by colon cancer cells expressing Fas ligand. J Exp Med 1996;184:1075– 82. Bennett MW, O’Connell J, O’Sullivan GC, et al. The Fas counterattack in vivo: Apoptotic depletion of tumor-infiltrating lymphocytes (TIL) associated with Fas ligand (FasL) expression by human esophageal carcinoma. J Immunol 1998; 1600:5669 –75. Passlick B, Izbicki JR, Simmel S, et al. Expression of major histocompatibility class I and class II antigens and intercellular adhesion molecule-1 expression on operable non-small cell lung carcinomas. Eur J Cancer 1995;30:376 – 81. Pantel K, Schlimok G, Kutter D, et al. Frequent down regulation of major histocompatibility class I antigen expression on individual micrometastatic carcinoma cells. Cancer Res 1991; 51:4712–5. Cote RJ, Beattie EJ, Chaiwun B, et al. Detection of occult bone marrow micrometastases in patients with operable lung carcinoma. Ann Surg 1995;222:415–23. Ross AA, Miller GW, Moss TJ, et al. Immunocytochemical detection of tumour cells in bone marrow and peripheral blood stem cells collections from patients with ovarian cancer. Bone Marrow Transplant 1995;15:929 –33. Pantel K, Aighnerr C, Kollermann J, et al. Immunocytochemical detection of isolated tumour cells in bone marrow of
Micrometastases and GI Cancer
69.
70.
71.
72.
73.
74.
1651
patients with untreated stage C prostatic cancer. Eur J Cancer 1995;31A:1627–32. Cote RJ, Rosen PP, Lesser ML, et al. Prediction of early relapse in patients with operable breast cancer by detection of occult bone marrow micrometastases. J Clin Oncol 1991;9: 1749 –56. Pantel K, Izbicki JR, Angstwurm M, et al. Immunocytoclogical detection of bone marrow micrometastases in operable non-small cell lung cancer. Cancer Res 1993;53:1027–31. Diel IJ, Kaufmann M, Costa SD, et al. Micrometastatic breast cancer cells in bone marrow at primary surgery: Prognostic value in comparison with nodal status. J Natl Cancer Inst 1996;88:1652– 8. Riethmu¨ller G, Schneider-Gadicke E, Schmiegel W, et al., German Cancer Aid 17-1A Study Group. Randomised trial of monoclonal antibody for adjuvant therapy of resected Duke’s C colorectal carcinoma. Lancet 1994;343:1177– 86. Schlimok G, Pantel K, Loibner H, et al. Reduction of metastatic carcinoma cells in bone marrow by intravenously administered monoclonal antibody: Towards a novel surrogate test to monitor adjuvant therapies of solid tumours. Br J Cancer 1995;31A:1799 –1803. Thorban S, Roder JD, Nekarda H, et al. Disseminated epithelial tumour cells in bone marrow of patients with esophageal cancer: Detection and prognostic significance. World J Surg 1996;20:567–72.