Guanylyl cyclase C: a molecular marker for staging and postoperative surveillance of patients with colorectal cancer

Transcription

1 Review For reprint orders, please contact Guanylyl cyclase C: a molecular marker for staging and postoperative surveillance of patients with colorectal cancer Glen S Frick, Giovanni M Pitari, David S Weinberg, Terry Hyslop, Stephanie Schulz and Scott A Waldman CONTENTS Colorectal cancer staging Guanylyl cyclase C Postoperative surveillance for disease recurrence Guanylyl cyclase C reverse transcriptase PCR detects metastatic colorectal tumor cells in blood Conclusion Expert commentary Five-year view Key issues References Affiliations Staging patients with colorectal cancer defines their prognosis and therapeutic management. Unfortunately, histopathology, the current standard for staging, is relatively insensitive for detecting occult micrometastases and a significant fraction of patients are understaged and, consequently, undertreated. Similarly, current approaches to postoperative surveillance of patients with colorectal cancer detect disease recurrence at a point when interventions have little impact on survival. The detection of rare cells in tissue, for accurately staging patients, and in blood, for detecting disease recurrence, could be facilitated by employing sensitive and specific markers of disease. Guanylyl cyclase C (GCC), the receptor for the diarrheagenic bacterial heat-stable enterotoxin, is expressed selectively by cells derived from intestinal mucosa, including normal intestinal cells and colorectal tumor cells, but not by extragastrointestinal tissues and tumors. The nearly uniform expression of relatively high levels by metastatic colorectal tumors suggests that GCC may be a sensitive and specific molecular marker for metastatic colorectal cancer cells. Employing GCC reverse transcriptase PCR, occult colorectal cancer micrometastases were detected in lymph nodes that escaped detection by histopathology. Moreover, marker expression correlated with the risk of disease recurrence. Similarly, GCC reverse transcriptase PCR revealed the presence of tumor cells in blood of all patients examined with metastatic colorectal cancer and, in some studies, was associated with an increased risk of disease recurrence and mortality. These observations suggest that GCC reverse transcriptase PCR is a sensitive and specific technique for identifying tumor cells in extraintestinal sites and may be useful for staging and postoperative surveillance of patients Author for correspondence Thomas Jefferson University, Division of Clinical Pharmacology, 32 South 0th Street, 70 Main, PA 907, USA Tel.: Fax: scott.waldman@jefferson.edu KEYWORDS: cancer recurrence, circulating cancer cell, colorectal cancer, disease staging, guanylyl cyclase C, lymph node, molecular marker, postoperative surveillance, real-time detection, reverse transcriptase PCR Expert Rev. Mol. Diagn. 5(5), (2005) Colorectal cancer is the third most common neoplasm in the USA [ 4]. It is the second leading cause of cancer-related mortality, and is responsible for 0% of cancer-related deaths in the USA. The overall mortality rate of newly diagnosed large bowel cancer approaches 50%, with mortality reflecting metastases [3,5 7]. Approximately 20% of patients have unresectable disease at presentation, and approximately 33% develop metastases during the course of their disease. Surgery continues to have the greatest impact on colorectal cancer survival. Curative surgery, intended to remove all clinically detectable tumor, is most successful in the treatment of early disease. However, while surgery removes detectable tumor, residual micrometastases may result in relapse. Recurrence rates vary widely, from 3% for disease limited to the mucosa to more than 50% for tumors that have spread to regional lymph nodes [3 8]. Overall, approximately 50% of patients who have undergone surgery with curative intent suffer recurrent disease. In these patients, locoregional recurrence occurs in / Future Drugs Ltd ISSN

2 Frick, Pitari, Weinberg, Hyslop, Schulz & Waldman approximately 30% of cases, while distant recurrence occurs in approximately 80%. Factors other than stage that suggest a poor prognosis include histologic grade, number of positive lymph nodes, and venous and lymphatic invasion [3]. In the context of the high proportion of patients who develop recurrent disease following definitive surgical management, accurate staging at the time of surgery may identify patients at greatest risk for developing recurrence who could most benefit from adjuvant chemotherapy. Similarly, detecting recurrent disease at the earliest possible stage, such as by identifying rare circulating tumor cells, may permit medical or surgical intervention at a point where the disease remains amenable to treatment. This review will examine the utility of guanylyl cyclase C (GCC), a sensitive and specific molecular marker of normal and malignant intestinal epithelial cells, in staging and postoperative surveillance of patients with colorectal cancer. Colorectal cancer staging Current methods of staging The single most important determinant for predicting survival from colorectal cancer is the presence of cancer cells in regional lymph nodes [3,6,9]. The presence or absence of metastatic disease in regional lymph nodes is a central component of all colorectal cancer staging schema. Cancer stage at diagnosis not only determines prognosis, but also guides selection of patients to receive adjuvant chemotherapy [3,5,7]. It is woth noting here that the accuracy of staging is highly dependent upon adequate sampling of lymph nodes from surgical specimens, and the number of lymph nodes analyzed is a prognostic variable for patient outcome [0]. Staging of colorectal cancer is based on preoperative testing (e.g., laboratory studies and imaging), operative findings and histopathology of excised tissues. Several staging paradigms have been developed since the original publication of Dukes criteria for rectal cancer staging and all include a histopathologic description of tumor invasion at the primary site, tumor cell dissemination to regional lymph nodes, and metastases to distant anatomical sites [3,4,6,9]. In the Dukes system, stage A disease is confined to the mucosal lining, without penetrating the lamina propria of the intestine. In stage B disease, the tumor has invaded through the lamina propria into the bowel wall to a variable depth. In stage C disease, tumor cells have metastasized outside the bowel and into regional lymph nodes. In stage D disease, tumor cells have metastasized to distant sites, most frequently the liver (80%) and lung (20%). Role of chemotherapy in the treatment of colorectal cancer Beyond prognosis, stage determines whether patients will receive chemotherapy. Treatment of patients with established metastatic (stage D) colorectal cancer prolongs disease-free survival but does not improve overall survival and more than 5% of patients with distant metastases to lung, liver, bone or brain survive for over 5 years [5]. Integration of 5-fluorouracil (5-FU), an inhibitor of thymidylate synthetase, and leucovorin, which increases the affinity of 5-FU for its target enzyme, increases the median disease-free survival of patients with metastatic disease from 6 to 2 months [5,]. Addition of the topoisomerase I inhibitor irinotecan (FOLFIRI) or the platinum compound oxaliplatin (FOLFOX) increases survival to 4 6 months [5,2 5]. Moreover, incorporation of all three agents into the treatment regimen, or the addition of monoclonal antibodies to epidermal growth factor receptors (cetuximab) or vascular endothelial growth factor (bevacizumab) increases survival to 20 months [5,6,7]. In contrast to patients with distant metastases, fluoropyrimdine therapy increases the overall survival of patients with regional lymph node metastases from 50 to 60% in the adjuvant setting [8,9]. While adjuvant therapy improves overall survival of patients with regional metastases, it is not typically administered to patients with tumor confined to the bowel wall (stages A and B) as clinical trials have demonstrated no consistent improvement in survival [3,5,4]. Inaccurate staging implies inaccurate prognosis and may result in withholding potentially beneficial therapy. Improved diagnostic accuracy that selects subsets of early-stage patients at increased risk for developing disease recurrence would direct the administration of adjuvant chemotherapy to patients who could derive the greatest benefit. Limitations of current approaches to staging patients with colorectal cancer Histopathology remains the mainstay of colorectal cancer staging. However, the imprecision of conventional microscopic examination reflects several methodologic limitations [3]. For example, it can be difficult to identify single, or even small clumps, of tumor cells in typical microscopic sections, and the sensitivity of this technique has been estimated at one cancer cell in 200 normal nucleated lymph node cells [3,20 23]. Also, it is not unusual for only one section from each lymph node to be reviewed by the pathologist, generating a substantial sampling error. This approach can omit more than 99.9% of each specimen from histopathologic review. The limitations of current methodologies are most evident when the frequency of postresection colorectal cancer recurrence is considered. Dukes A and B lesions, by definition limited to the bowel wall with no histologic evidence of extraintestinal spread, should be amenable to complete surgical excision. However, recurrence rates of 30% for Dukes A and 50% or greater for Dukes B lesions have been reported in some series [3 8]. In the case of Dukes C lesions, where all detectable tumor is removed, including involved extraintestinal tissue, recurrence rates of over 60% have been described. Similarly, the variability in reported recurrence rates for stage B disease likely reflects the mixture of true stage B lesions as well as stage C or D lesions undetected by histopathology. This suggestion is supported by rigorous re-review by serial sectioning and immunohistochemistry of histologically negative lymph nodes from stage B patients, which upstaged approximately 30% of patients to stage C disease [3,20 22]. 702 Expert Rev. Mol. Diagn. 5(5), (2005)

3 GCC as a molecular marker for colorectal cancer Alternative techniques for staging patients with colorectal cancer Immunohistochemical staining of lymph nodes Immunohistochemical staining of proteins expressed by epithelial cells permits visualization of tumor cells undetected by standard histologic techniques in tissue sections [3,20 22]. Proteins most commonly employed to identify occult micrometastases include cytokeratin and carcinoembryonic antigen (CEA). In some series, immunohistochemical identification of occult micrometastases correlated closely with the prognosis of patients For example, in one series, lymph nodes from Dukes B patients subjected to immunohistochemistry using antibodies specific for cytokeratin revealed tumor cells in 30% of patients [24]. In fact, stage B patients who were node negative by cytokeratin staining exhibited a survival rate after surgery that was similar to patients with stage A disease. In contrast, stage B patients who were node positive by cytokeratin staining exhibited survival rates that mimicked patients with stage C disease. However, while immunohistochemical staining is a relatively sensitive technique for detecting colorectal micrometastases, this technique has not uniformly correlated with patient prognosis. In 5 reported series of patients staged employing CEA or cytokeratin immunohistochemistry, studies revealed that the identification of occult micrometastases did not provide prognostic information useful for patient management [2,25,26]. Lack of prognostic utility of these markers may, in part, reflect their limited specificity, since they can be identified in cells other than colorectal cancer, potentially yielding false-positive results. Furthermore, the potentially large sampling error inherent to histopathologic analysis described in the previous section also applies to immunohistochemistry. Although immunohistochemistry improves the detection of micrometastases, it is prohibitively labor intensive and economically unfeasible to apply to all lymph nodes for routine colorectal cancer staging. Mutant allele-specific nucleic acid amplification Studies employing nucleic acid amplification-based detection (PCR) of K-Ras and p53 gene mutations have re-examined histopathology-negative lymph nodes for occult metastases [27,28]. In patients with mutations in K-Ras or p53, these analyses detected deposits of colorectal tumor cells in histopathologynegative lymph nodes of more than 50% of patients. Patients who were node negative by these amplification techniques exhibited 5-year survival rates after surgery that were substantially better than patients who were node positive by amplification. As with cytokeratin staining, these impressive results underscore the importance of detecting micrometastases in patients However, it is unlikely that amplification of K-Ras or p53 mutations will be widely employed. Fewer than 60% of patients with colorectal cancer exhibit K-Ras or p53 abnormalities, limiting the applicability of this technique. Also, the specificity of amplification of these mutations is limited since these same genetic mutations are found in other tumors and some nonmalignant conditions. Finally, mutations in those genes associated with colorectal cancer are heterogeneous. Therefore, primary tumors from patients must first be evaluated to determine the presence and type of genetic mutation, and then lymph nodes would subsequently be analyzed by amplification techniques individualized for each specific mutation. Nucleic acid-based amplification of molecular markers Qualitative reverse transcriptase (RT)-PCR employing CEA-specific primers was used to assess micrometastases in fresh lymph nodes of 26 Stage II (Dukes B) colorectal cancer patients [23]. CEA RT-PCR detected micrometastases in lymph nodes of more than 50% of patients that were node negative by histopathology. Patients who were CEA RT-PCR positive exhibited 5-year survival rates of 50%, compared with those who were CEA RT-PCR negative, who exhibited survival rates of 9%. Similar results were obtained in a study of 64 colorectal cancer patients employing paraffin-embedded lymph nodes [2]. Moreover, quantitative RT-PCR for CEA identified occult micrometastases in lymph nodes of patients with colorectal cancer [25]. However, in several studies, CEA RT-PCR exhibited low specificity, with detection in tissues from patients without cancer, possibly reflecting illegitimate transcription or ectopic epithelial cells [29 3]. Moreover, CEA is expressed by less than 80% of colorectal cancers, potentially limiting its sensitivity. Furthermore, CEA is expressed by normal intestinal tissues and other tissues involved in pathologic processes, such as inflammation, contributing to a high false-positive rate in normal lymph nodes [32]. Finally, CEA RT-PCR appears to be less sensitive for detecting the presence of colorectal cancer cells compared with other molecular markers in side-by-side comparisons [29 3,33,34]. In similar studies, RT-PCR employing primers specific for cytokeratin-20 (CK20) was employed to detect micrometastases in lymph nodes from patients with colorectal cancer [35]. In these studies, two fresh peritumoral lymph nodes from 85 patients with colorectal cancer without lymph node metastases by histopathology were subjected to CK20 RT-PCR. Of significance, CK20 RT-PCR identified micrometastases in lymph nodes that correlated closely with the risk of developing recurrent disease. Again, while these observations suggest that CK20 RT-PCR may be useful for identifying clinically significant occult micrometastases [36], other studies have not confirmed these results [37,38]. The above discussion highlights the potential utility of combining a specific molecular marker and a powerful amplification technique such as RT-PCR to identify clinically important occult micrometastases in the management of patients However, they underscore the importance of sensitivity, with uniform association with diseased tissue, and specificity, with expression limited only to that diseased tissue, in defining the utility of markers. Guanylyl cyclase C GCC is a transmembrane receptor selectively expressed in apical membranes of intestinal mucosa cells from the duodenum to the rectum in normal placental mammals [39]. While studies 703

4 Frick, Pitari, Weinberg, Hyslop, Schulz & Waldman have suggested that GCC might be expressed in kidney [40], pancreas [4], lung [42] and biliary epithelium [43], these observations have not been corroborated [44 47]. Activation of GCC by the endogenous peptides guanylin and uroguanylin results in the intracellular accumulation of the second messenger cyclic GMP [39]. GCC is also the receptor for bacterial heat-stable enterotoxins (STs), major etiologic agents that cause secretory diarrhea [39]. STs induce secretion by binding to GCC on mucosal cells of the intestine. ST GCC interaction also elevates intracellular cyclic GMP, which mediates intestinal secretion. Of significance, human colon carcinoma cell lines also express GCC in vitro [48]. Receptors on colorectal cancer cells bind ST with an affinity comparable to that of native intestinal cells [45 48]. Also, ST stimulates guanylyl cyclase activity and increases cyclic GMP in these cells in a fashion similar to that observed in native intestinal cells. Furthermore, ST alters electrolyte transport in these cells similarly to the effects of this toxin on normal intestinal mucosal cells [49,50]. Since GCC is selectively expressed by normal digestive mucosa and colorectal cancer cells compared with extragastrointestinal tissues or tumors, this protein fulfills the criteria for a cell surface receptor with high specificity for colorectal tumors in extraintestinal Table A. Tissues and tumors in which guanylyl cyclase C expression was detected by reverse transcriptase PCR. Tissue source Intestinal tissues Small intestine Colon Rectum Intestinal tumors Primary colon carcinoma Primary rectal carcinoma Primary ileal adenocarcinoma Ileal carcinoma metastatic to lymph node Cloacogenic carcinoma of the rectum No. of specimens evaluated - Colorectal cancer metastatic to Lung Liver Lymph nodes Ovary Peritoneum Kidney Stomach No. of patients from which specimens were obtained Note: Guanylyl cyclase C reverse transcriptase PCR was conducted using µg of total RNA from each tissue, as described in [46]. These data reflect analyses of over 330 specimens from more than 280 patients sites. These characteristics suggest that GCC may be uniquely suited for identifying metastatic colorectal cancer cells in lymph nodes and blood. GCC mrna is selectively expressed by human colorectal tumors Total RNA isolated from surgical samples was subjected to RT- PCR employing GCC-specific primers. A single GCC-specific amplicon was detected in 20 samples of tissue derived from intestinal mucosa from 00 patients, including specimens of normal colon and rectum, primary colorectal tumors, and colonic adenocarcinomas metastatic to various sites (TABLE A) [45 47]. In contrast, mrna for this receptor was not detected in 25 samples of extraintestinal tissues and tumors obtained from 83 patients (TABLE B) [45 47]. These data demonstrate that expression of GCC is highly specific and confined to tissues derived from intestinal mucosa in humans. In addition, they highlight the high sensitivity of this marker, with expression of GCC in approximately 00% of colorectal tumors examined. GCC RT-PCR identified occult metastases undetected by routine histopathology in lymph nodes of patients with colorectal cancer Fresh lymph nodes were dissected from bowel resections from 33 patients, of whom 28 were diagnosed with colorectal adenocarcinoma, two had ulcerative colitis, two had adenomatous polyps and one had diverticulitis [5]. A total of 68 lymph nodes, ranging from one to 20 per patient (4.5 ± 2.9), were obtained and bisected prior to formalin fixation. One half was analyzed by GCC RT-PCR, while the other was analyzed by histopathology. Metastatic colon carcinoma cells were identified in 26 of 68 (5%) of these lymph nodes by histopathology. GCC expression was not detected by RT-PCR in lymph nodes from patients with ulcerative colitis, polyps or diverticulitis. In contrast, GCC expression was detected in all 26 lymph nodes containing histopathology-confirmed tumor cells obtained from ten patients with stage C disease (00% concordance). Importantly, GCC expression was detected in 8% (2 of 68) of lymph nodes from patients with stage B disease, where, by definition, histopathology was negative in these cases. These data suggest that the clinical staging of four of 5 (27%) patients might be upgraded from stage B to stage C, reflecting the detection of micrometastatic disease employing GCC RT-PCR. Moreover, the ability of GCC RT-PCR to detect occult micrometastases in otherwise uninvolved lymph nodes from patients with colorectal cancer has been confirmed [20,22,38,52]. Occult micrometastases detected by GCC RT-PCR in lymph nodes of patients with colorectal cancer is associated with an increased risk of developing recurrent disease The results presented above demonstrate that GCC RT-PCR detects micrometastases in lymph nodes that escape identification by histopathology. However, the ability to detect a small number of cancer cells may not correlate with an increased risk of disease recurrence in patients. Therefore, the association of 704 Expert Rev. Mol. Diagn. 5(5), (2005)

5 GCC as a molecular marker for colorectal cancer GCC mrna in lymph nodes with disease recurrence was examined in stage B colorectal cancer patients presumed cured by surgical resection [33]. A total of patients free of disease for over 6 years following surgery, and who were considered cured, formed the control group, while ten patients who had developed recurrence under 3 years following surgery formed the case group. Total RNA was extracted from paraffin-embedded lymph nodes from these patients and analyzed by RT-PCR using GCC-specific primers. GCC-specific amplicons were not detected in reactions employing RNA from lymph nodes of patients who were free of recurrent disease at least 6 years post resection. In contrast, GCC-specific amplicons were detected in reactions employing RNA from lymph nodes of all patients who developed recurrent colorectal carcinoma up to 3 years post resection. Analysis of CEA by RT-PCR has also been suggested as a sensitive and specific marker for colorectal cancer micrometastases [23]. Thus, total RNA extracted from lymph nodes from the above patients was also analyzed by RT-PCR using CEA-specific primers. CEA-specific amplicons were not detected in reactions employing total RNA from patients free of recurrent disease. In stark contrast to results obtained with GCC, CEA-specific amplicons were detected in reactions employing total RNA from only one of ten patients who developed recurrent disease [33]. Thus, detection of GCC mrna in regional lymph nodes correlated closely with disease recurrence in patients with stage B colorectal cancer and appears to be a more sensitive marker for recurrence than CEA. Indeed, the odds ratio for mortality associated with GCC mrna expression in lymph nodes was 6.5 (95% confidence interval: ) [33]. These data support the suggestion that expression of GCC detected by RT-PCR analysis may be a sensitive and specific method for detecting clinically important micrometastases in lymph nodes of patients undergoing staging for colorectal cancer. Clinical translation: GCC RT-PCR for staging patients with colorectal cancer The foregoing discussion suggests that GCC is a highly sensitive and specific marker for detecting metastatic colorectal cancer, particularly in combination with a powerful amplification technique such as RT-PCR. Indeed, these data form the basis for an ongoing multicenter clinical trial examining the utility of quantitative RT-PCR of GCC for staging patients In this trial, approximately 000 patients with colorectal cancer will be enrolled at the time of colectomy. Lymph nodes obtained for staging will be bisected and the resulting halves submitted for routine histopathology and GCC quantitative RT-PCR, respectively. Results obtained by histopathology and RT-PCR will be correlated with the clinical outcomes of patients. It is expected that this trial, which will analyze approximately 0,000 20,000 lymph nodes, will define the sensitivity and specificity of GCC RT-PCR compared with routine histopathology for detecting micrometastases and the clinical significance of this staging technique. In the event that GCC RT-PCR exhibits greater sensitivity and specificity Table B. Tissues and tumors in which guanylyl cyclase C expression was not detected by reverse transcriptase PCR. Tissue source No. of specimens evaluated No. of patients from which specimens were obtained Extraintestinal tissues Adrenal 8 8 Thyroid Breast 2 Lung 7 7 Ovary 5 4 Spleen Thymus Testis 2 2 Uterus Brain 8 6 Parathyroid Gall bladder 3 3 Lymph node 36 8 Fallopian tube 3 2 Esophagus 9 9 Liver 4 4 Lymphocytes Pancreas 2 9 Stomach 3 3 Bone marrow Prostate 8 Extraintestinal tumors Thymoma Hepatoma 4 3 Seminoma 3 3 Lymphoma Renal cell carcinoma 3 3 Breast carcinoma 3 3 Prostate carcinoma 2 Cholangiocarcinoma Esophageal squamous cell 7 7 carcinoma Rhabdomyosarcoma Melanoma Ovarian carcinoma 5 5 Uterine carcinoma - Pancreas Neuroendocrine tumor 2 2 Islet cell tumor Carcinoma Lymphoma Non-Hodgkin (spleen; liver) 2; 2; Hodgkin (spleen) Lung Squamous cell carcinoma 2 Adenocarcinoma Note: Guanylyl cyclase C reverse transcriptase PCR was conducted using µg of total RNA from each tissue, as described in [46]. These data reflect analyses of over 330 specimens from over 280 patients

6 Frick, Pitari, Weinberg, Hyslop, Schulz & Waldman compared with histopathology for detecting clinically significant micrometastases, an intervention trial will be initiated to examine the utility of this technique for identifying patients most at risk of developing recurrent disease who could benefit from receiving adjuvant chemotherapy. It is noteworthy that the utility of GCC-based quantitative RT-PCR for identifying clinically significant occult micrometastases has been examined in a small cohort of 32 colorectal cancer patients whose lymph nodes were free of disease by histopathology [37]. In the study, there was significant overlap in the quantity of GCC transcripts identified in lymph nodes from patients without colorectal cancer (true negatives) and those containing colorectal cancer metastases identified by histopathology (true positives), wherein some true negative lymph nodes contained approximately 0,000 copies of GCC/µg of total RNA (baseline expression is approximately 00 copies/µg of total RNA). These observations, which stand in contrast to results obtained in the ongoing clinical trial where there is a clear separation in the GCC transcripts quantified in true positive and true negative lymph nodes, might reflect contamination of nodes by intestinal epithelial cells during dissection. Indeed, contamination of nodes by epithelial cells has been observed as an important source of false-positive signals in studies of molecular markers detected by RT-PCR in colorectal cancer patients [35]. Moreover, in that study, GCC levels in lymph nodes of the four patients with lymph nodes free of disease by histopathology, but who developed recurrent metastatic colorectal cancer, were indistinguishable from levels in nodes of the 28 patients who were free of obvious recurrences [37]. In that context, the study sample is too small to test hypotheses concerning the utility of a molecular marker to discriminate patients at increased risk for disease. In order to have sufficient power to detect differences in the relative risk of node-negative patients for recurrent disease, a cohort of 84 patients with 3 recurrences would be required, presuming equal sample sizes in RT-PCR-negative and -positive groups. This is based on the assumption of a 9% recurrence rate in RT-PCR-negative patients, a 36% recurrence rate in RT-PCR-positive patients, 80% power, no loss to follow-up and a two-sided test. In contrast, using comparison of median differences as in [35], the cohort reported had sufficient power to detect only much larger differences. Postoperative surveillance for disease recurrence Current paradigms for postoperative surveillance of patients with colorectal cancer The goal of postoperative surveillance is to detect disease recurrence at the earliest time, to maximize the possibility that intervention will be clinically beneficial [4,53]. Indeed, it has been estimated that up to 40% of patients with recurrent disease might experience an increase in overall survival if surgical intervention could be applied at an appropriately early stage. All strategies for postoperative surveillance stress regular follow-up employing clinical, radiologic, endoscopic and laboratory data to facilitate earlier detection of disease. Prompt identification of tumor recurrence is important for several reasons. First, metastatic cancer cells appear to be more sensitive to chemotherapeutic agents after resection. Also, initiating treatment at an early stage permits eradication of tumor prior to the development of chemotherapy-resistant clones. In addition, assuming that recurrence is initially localized, early detection and treatment may prevent dissemination. Furthermore, micrometastatic tumors are particularly well suited as targets for chemotherapy due to their increased accessibility compared with clinically evident solid tumors [8]. The reduction in mortality observed in some colorectal cancer patients undergoing intensive postoperative surveillance highlights the potential of improved early detection of disease recurrence [54 58]. Limitations in current postoperative surveillance programs Current surveillance protocols have not been uniformly successful in improving the overall survival of patients with recurrent colorectal cancer [4,53,59,60]. The components of a postoperative surveillance program, together or separately, are insensitive for detecting most early recurrences. The first major hurdle to overcome is the absence of a sensitive, responsive marker for early, potentially treatable, recurrent disease. Moreover, improving therapies in the face of recurrence make the concept of early detection even more attractive. Limitations of CEA for postoperative surveillance While all follow-up programs emphasize the need for multiple modalities, the cornerstone of colorectal cancer postoperative surveillance in practice is serial measurement of serum CEA levels [53,59,6,62]. Periodic postoperative CEA determinations may detect recurrent disease at a surgically treatable stage, while increasing CEA levels may be associated with an increased risk for disease recurrence, tumor progression, tumor burden and mortality [53,59,6,62]. However, the sensitivity and specificity of CEA to detect recurrent colorectal cancer early remain limited since [3,4]: CEA is produced by less than 80% of colorectal tumors CEA is generated by other tumors and some normal extraintestinal tissues Conditions other than colorectal cancer can cause elevated serum CEA levels Overall, the sensitivity and specificity of this test for recurrence are 36 75% and 62 99%, respectively [4]. Although the test detects tumor recurrence in the liver, it is less sensitive in detecting local, regional or pulmonary metastases [4,53,63,64]. In most cases, CEA fails to detect recurrent disease at a point where cure can be improved [4,53,63,64]. Tumor cells in blood as markers of early recurrent disease Some recurrent colorectal tumors are amenable to curative resection if identified sufficiently early [65 70]. Also, early detection of tumor recurrence would permit identification of patients who could receive maximum benefit from adjuvant chemotherapy [4,5]. In addition, as colorectal cancer therapy 706 Expert Rev. Mol. Diagn. 5(5), (2005)

7 GCC as a molecular marker for colorectal cancer evolves, novel therapies will likely have the largest impact if applied early in the course of recurrent disease [4,62]. Thus, the earliest possible detection may permit better outcomes with less treatment-related morbidity and mortality. Solid tumors shed cells into the circulation, which are accessible for sampling. These observations suggest that early recurrence could be detected by identifying circulating cancer cells. Cells from solid tumors migrate into blood vessels and enter the circulation where they are carried to distant sites [7]. Indeed, intravasation is a principal process underlying tumor metastases. However, the metastatic process is generally inefficient and most viable tumor cells (>99.99%) in the circulation are destroyed. Nevertheless, even an inefficient process such as intravasation, if it recurs sufficiently often, will establish metastatic tumors that may ultimately result in the death of the patient. While intravasation is central to tumor metastasis, this process also results in the presence of cancer cells in blood, a compartment that is readily accessible for sampling. Up to 0.02% of cells in solid tumors are shed into the circulation each day [4,7,72]. For tumor volumes of approximately ml ( 0 9 cells), the lower limit for reliable detection of metastases by standard imaging modalities [73], this represents approximately 00,000 cells in the circulation each day ( 20 tumor cells/ml of whole blood). Furthermore, the number of circulating cancer cells appears to be directly proportional to the volume of the tumor shedding those cells. While these values may vary for different tumors and different anatomical sites, the accessibility of these tumor cells for sampling underscores the importance of identification of these cells in monitoring disease recurrence. Hence, intravasation of tumor cells into the circulation presents a unique opportunity to identify recurrent disease early in cancer patients. One limitation to this approach is a sensitive and specific method for identifying rare circulating tumor cells in blood. Nucleic acid-based amplification methods may be ideally suited for this application, due to the high sensitivity and specificity inherent in the technique. However, application of these molecular techniques to identify rare circulating tumor cells to detect early disease recurrence has been limited by the availability of markers that are specific for tumor, but not normal, cells with low baseline expression in blood. Immunocytochemistry employing antibodies to epithelial markers has been used to identify circulating tumor cells [8,74]. However, this technique detects approximately one cancer cell in 0 6 normal mononuclear cells and, consequently, may not have sufficient sensitivity to reliably detect rare circulating cancer cells. In addition, epithelial cell proteins employed as markers for these techniques are expressed not only by cancer cells, but also by normal epithelial cells in a variety of tissues and disease states, thereby decreasing the specificity of this approach [75]. More recently, amplification of nucleic acids, employing PCR for DNA-based targets and RT-PCR for mrna targets, has been employed to detect rare tumor cells in body tissues and fluids. This approach is highly sensitive, detecting approximately one cancer cell in 0 7 mononuclear blood cells, highly specific, reflecting transcript-specific amplification and detection, and quantitative. Employing these techniques, circulating tumor cells were detected in patients with colorectal cancer, which, in some studies, correlated with increased risk of disease recurrence and/or mortality [8,75 87]. Although PCR-based approaches are highly sensitive, they are limited by the specificity of the marker employed. This reflects, in part, a paucity of available tumor-specific markers: most studies of tumor cell detection by PCR employ epithelial cell markers or genetic markers related to carcinogenesis. As discussed in previous sections, some epithelial cell markers, such as CEA or CK20, can be expressed by normal, abnormal and neoplastic cells, thus limiting their specificity. However, it is noteworthy that in some studies, CEA and CK20 have been sensitive and specific for identifying colorectal tumor cells in the circulation of patients with active primary and/or metastatic tumors, and these cells can be associated with a risk of disease recurrence [75,76,82 87]. Similarly, genetic markers, such as mutations in p53 or K-Ras, may not be expressed uniformly by all tumors of a specific type, limiting their application. In addition, detection of tumor cells by RT-PCR may be limited by ectopic expression, in which some tissue- or tumor-specific transcripts are expressed nonspecifically at low levels in cellular components of blood [30,34,74]. Although these transcripts appear to be nonfunctional, they contribute to a baseline signal in patients that can limit the sensitivity of detection of tumor cells [34,76,88]. Guanylyl cyclase C reverse transcriptase PCR detects metastatic colorectal tumor cells in blood Preliminary studies have compared the expression of GCC mrna in blood from patients with various stages of colorectal cancer, nonmalignant pathology of the intestine, extraintestinal malignancies, and normal healthy volunteers by nested RT-PCR employing low levels of mrna (<.0 µg of total RNA; TABLE 2) [34,46,76]. GCC mrna was not detected in blood from normal subjects, nor from patients with adenomatous polyps, inflammatory bowel disease, perforated small bowel, extraintestinal malignancies, Dukes A colon carcinoma, or some patients with Dukes B or C colon carcinoma. In contrast, GCC-specific amplicons were obtained with blood from some patients with Dukes B and C, and all patients with stage D colorectal cancer. Comparisons of multiple epithelial cell markers have demonstrated that GCC is one of the most sensitive and specific markers of circulating colorectal cancer cells [34,76,88]. In addition, detection of GCC mrna by RT-PCR is an index of metastatic colorectal cancer cells in the circulation [34,76 8]. Furthermore, detection of GCC mrna by RT-PCR correlated with the risk of tumor metastases and mortality in patients with colorectal cancer [77,80]. Improving the signal-to-noise ratio of RT-PCR in blood significantly increases the specificity of GCC Low-level illegitimate transcription complicates the utility of many epithelial cell markers for detecting circulating tumor cells [30,34,74,89]. Thus, false-positive results could potentially lead to errors in staging wherein a stage B patient could be 707

8 Frick, Pitari, Weinberg, Hyslop, Schulz & Waldman Table 2. Detection of colon cancer cells in blood by guanylyl cyclase C reverse transcriptase PCR. Sample type No. of subjects/patients from which samples were obtained GCC mrna not detected by RT-PCR Normal subjects 45 Cholecystitis Pancreatic carcinoma 8 Leukemia Lymphoma Adenocarcinoma of the ovary 3 Endometrial carcinoma Testicular carcinoma Adenocarcinoma of the breast Leiomyoma Adenocarcinoma of the lung 2 Perforated small bowel Gastric carcinoma 2 Inflammatory bowel disease 5 Adenomatous colonic polyps 0 Dukes A colon carcinoma 7 Dukes B colon carcinoma 7 Dukes C colon carcinoma 5 GCC mrna detected by RT-PCR Dukes B colon carcinoma 2 Dukes C colon carcinoma 5 Dukes D colon carcinoma 34 Note: GCC RT-PCR was conducted using a nested protocol and <.0 µg of total RNA, as described in [34]. GCC: Guanylyl cyclase C; RT: Reverse transcriptase. misclassified as stage C, resulting in unnecessary adjuvant chemotherapy. To further examine this issue, mononuclear cells from blood were employed to compare the illegitimate transcription of GCC and other epithelial cell markers [34]. This fraction is of particular use, as it specifically concentrates cancer cells that have metastasized from epithelial tumors, including colorectal cancers. Mononuclear cells from 20 healthy volunteers were examined employing nested RT-PCR and at least.0 µg of total RNA to identify GCC transcripts. All 20 yielded positive results, which reflected low-level illegitimate transcription of GCC by CD34 + cells. Moreover, CD34 + cells illegitimately transcribed many tissue-specific markers including prostate-specific antigen, prostate-specific membrane antigen, CEA, CK9, CK20, MUC and GA Removal of CD34 + cells from mononuclear cells eliminated ectopic transcription of all epithelial markers, including GCC. Similarly, limiting dilutions of total RNA isolated from mononuclear cells also eliminated background expression of GCC in control subjects. Employing limiting dilution, GCC amplicons were detected in mononuclear cell RNA from 24 of 24 stage D colorectal cancer patients employing µg of total RNA. In contrast, GCC amplicons were not detected in any of the 20 samples from healthy subjects. While GCC amplicons were detected in blood from all stage D patients examined, CEA amplicons were detected in only 30% of those patients. These data support the suggestion that GCC is a specific and sensitive marker for detecting colorectal tumor cells in blood and may represent an advance, relative to other epithelial markers such as CEA, for postoperative cancer surveillance. Also, the data suggest that the utility of markers for gastrointestinal and other epithelial malignancies could be improved by application of quantitative RT-PCR methodologies to accommodate background signals contributed by illegitimate transcription in blood mononuclear cells. Clinical translation: GCC RT-PCR for detecting disease recurrence in patients with colorectal cancer undergoing postoperative surveillance The foregoing observations formed the basis for a prospective multicenter clinical trial examining the sensitivity and specificity of GCC quantitative RT-PCR for identifying circulating metastatic colorectal cancer cells and detecting the earliest stages of disease recurrence in patients undergoing postoperative surveillance. In this study, 500 patients with colorectal cancer in surveillance programs are being enrolled and peripheral blood samples obtained at intervals corresponding to their routine physician visits. Samples will be analyzed by GCC RT-PCR and the results compared with those obtained by routine surveillance including history, physical exam, chest x-ray, routine blood chemistries, routine hematologic analyses and CEA. It is anticipated that approximately 5,000 blood samples will be analyzed by quantitative RT-PCR in this trial. The results of routine analyses and GCC RT-PCR will be correlated with the clinical progression of patients, to determine whether GCC RT-PCR increases the sensitivity and specificity of routine surveillance for detecting disease recurrence at the earliest time. In that context, quantitative RT-PCR for GCC has been applied to a limited number of patients Indeed, quantitative RT-PCR for GCC was more sensitive and specific for identifying patients with colorectal cancer, compared with CK9 or CK20 [76]. Moreover, detection of GCC by quantitative RT-PCR in blood 2 weeks following presumptively curative surgery was a poor prognostic indicator for recurrent disease and mortality [77]. In the present large-scale clinical trial, it is anticipated that if GCC RT-PCR improves the sensitivity and specificity of postoperative surveillance, an interventional trial will be initiated to determine whether this technique can identify patients at risk for developing clinical progression of their disease who could benefit from receiving adjuvant chemotherapy. Conclusion There are unmet clinical needs in managing patients with colorectal cancer. Specifically, current methodologies for staging patients by routine histopathology are relatively insensitive for detecting occult micrometastases, and a significant fraction of patients are understaged and, consequently, undertreated. Moreover, current paradigms for postoperative surveillance 708 Expert Rev. Mol. Diagn. 5(5), (2005)

9 GCC as a molecular marker for colorectal cancer detect disease recurrence beyond the point in the evolution of the disease when interventions meaningfully impact survival. The detection of rare cells in tissue, for accurately staging patients, and in blood, for detecting the earliest evidence of disease recurrence, could be facilitated by combining a powerful amplification technology, such as RT-PCR, with highly sensitive and specific markers of disease. One limitation to this approach has been the availability of only epithelial cell-specific markers, which may lack the necessary sensitivity and specificity, and the associated paucity of tumor-specific markers. In that context, GCC is a protein that is selectively expressed in apical membranes of mucosal cells in intestine, but not in other tissues such as lymph nodes, lung, liver and blood. Furthermore, GCC expression has been detected functionally and at the mrna level in all primary and metastatic colorectal tumors examined, regardless of the stage or grade of the tumor. These characteristics suggest that GCC may fulfill the criteria for a near-absolutely specific marker for metastatic colorectal cancer cells. Indeed, GCC RT-PCR is sensitive for identifying clinically important occult micrometastases associated with an increased risk of developing disease recurrence. In addition, GCC RT-PCR can detect clinically important, rare, circulating colorectal cancer cells, which are associated with the risk of disease recurrence and mortality. These observations suggest that GCC RT-PCR may be useful for accurate staging and postoperative surveillance of patients Expert commentary The sensitivity and specificity of current techniques for staging colorectal cancer is inadequate for identifying patients with micrometastases who are at increased risk for developing recurrent disease and could benefit from adjuvant chemotherapy. Similarly, current paradigms for monitoring colorectal cancer patients are relatively insensitive for identifying disease recurrence at a point when intervention can produce clinically meaningful outcomes. In that context, in the field of molecular diagnostics, pairing powerful detection technologies such as RT-PCR with sensitive and specific disease markers, will advance the management of these patients. While a number of disease-associated markers have been identified, GCC has emerged as a highly sensitive and specific marker for metastatic colorectal cancer cells. It is anticipated that quantitative RT-PCR employing highly specific markers, such as GCC or other markers yet to be identified, will become the standard of care for managing patients with colorectal cancer, as well as other tumors, in the future. Key issues While current diagnostic techniques, including standard histopathology and serum carcinoembryonic antigen (CEA), are relatively insensitive for detecting rare tumor cells in tissue and blood, respectively, the presence of those cells profoundly affects the prognosis and management of patients Alternative techniques have been applied to the detection of tumor cells in patient samples, including immunohistochemistry and mutant allele-specific amplification, but these techniques have been inconsistent in their reproducibility or lack generalizability. Amplification of the expression of tissue-specific differentiation markers employing reverse transcriptase (RT)-PCR, which can detect the presence of one tumor cell in ~0 7 normal cells, is emerging as a technique to increase the sensitivity of detecting rare tumor cells in patient samples. RT-PCR has enjoyed only modest success in detecting metastatic colorectal tumor cells in the lymph nodes of patients undergoing staging and in blood samples of patients undergoing postoperative surveillance for recurrent disease. However, markers currently available, including CEA and cytokeratin-20, have recognized limitations, including inconsistent expression by tumors, expression by normal tissues, and expression in pathologic conditions other than colorectal cancer, which restrict their sensitivity and specificity. Guanylyl cyclase C (GCC) is a protein whose expression is normally highly restricted only to brush-border membranes of intestinal epithelial cells from the duodenum to the rectum. Of significance, GCC continues to be expressed following neoplastic transformation, and has been detected in all primary and metastatic tumors, but not in any extragastrointestinal tumors examined to date. These observations suggest that GCC may be a highly sensitive and specific marker of metastatic cancer cells with utility for staging and postoperative surveillance of patients with colorectal tumors. GCC expression, detected by RT-PCR, revealed the presence of occult tumor cells in lymph nodes in all patients examined with colorectal cancer whose nodes were free of tumor cells by histopathology but who developed recurrent disease. Similarly, GCC expression, detected by RT-PCR, revealed the presence of circulating tumor cells in all patients examined who had metastatic colorectal cancer. Thus, GCC may fulfill the criteria for a tumor-specific marker with utility for staging and postoperative surveillance of patients with colorectal cancer. The utility of GCC as a marker for managing patients with colorectal cancer is currently being examined in two large, prospective, multicenter clinical trials