Jason B. Fleming, MD
Assistant Professor
Division of Surgical Oncology
Department of Surgery


Laparoscopy is an effective tool for the diagnosis and staging of intra-abdominal malignancies; it adds to the information provided by other noninvasive diagnostic modalities and may spare the patient the morbidity of a non-therapeutic laparotomy. When routinely used in the staging of upper gastrointestinal malignancies, we have demonstrated an increase in clinical tumor stage in 19-36% of patients. Furthermore, the presence of malignant cells predicts a significantly decreased overall survival in the affected patients. Molecular analysis of tumor suppressor genes from DNA harvested from epithelial cells within the washings has also been evaluated; however, this data does not appear to correlate with tumor recurrence or patient survival.
Angiogenesis, the formation of new blood vessels from the existing vascular network, is essential for continued tumor growth and metastasis. One of the most potent and specific angiogenic factors is vascular endothelial growth factor (VEGF). Evidence for the pivotal role of this cytokine in tumor angiogenesis includes the observations of increased expression in tumor cells of numerous human cancers. VEGF has been assayed and found to be elevated in ascites associated with gastrointestinal malignancies and the level of cytokine present is associated with patient outcome. Similarly, the level of VEGF expression in primary gastrointestinal tumors is associated with cancer stage and outcome.
The purpose of this proposal is to establish a method of analysis of peritoneal washings that will accurately reflect the biology of the primary tumor and predict patient outcome. Our hypothesis is two fold: 1) Vascular endothelial growth factor (VEGF) can be measured in peritoneal washings obtained during staging laparoscopic examination of patients with upper gastrointestinal epithelial malignancies. 2) The level of VEGF in the peritoneal fluid will reflect the clinical/pathologic stage of the malignancy and predict cancer-related morbidity and mortality.

We will use the following aims to test this hypothesis:

I. Using ELISA techniques soluble VEGF will be measured in peritoneal washings obtained during staging laparoscopy of patients with known or suspected upper gastrointestinal (esophagus, stomach, pancreas, hepatobiliary) malignancy.
II. VEGF levels in peritioneal fluid will be compared with plasma VEGF levels and with VEGF expression from the primary tumor and exfoliated cells harvested from the peritioneal wash specimen.
III. VEGF levels in peritoneal fluid will be correlated with the clinical /pathologic stage of the tumor and the clinical endpoints of cancer-related morbidity and mortality.

If this hypothesis is true, measurement of VEGF from peritoneal washings obtained and staging laparoscopy will provide a rapid and accurate assessment of the biologic character of the primary tumor. This prognostic information can then be used to tailor therapy of upper gastrointestinal malignancies.

HYPOTHESIS

The hypothesis of this proposal is two fold:
1) Vascular Endothelial Growth Factor (VEGF) can be measured in peritoneal washings obtained during staging laparoscopic examination of patients with upper gastrointestinal epithelial malignancies.
2) The level of VEGF in the peritoneal fluid will reflect the clinical/pathologic stage of the malignancy and predict cancer-related morbidity and mortality.

This proposal is based upon the premise that malignant epithelial cells of the upper gastrointestinal tract produce the cytokine vascular endothelial growth factor (VEGF). VEGF promotes local formation of new vascular structures (angiogenesis) that allow for tumor growth to greater than 1mm in size and facilitate hematogenous metastases. Previous studies have demonstrated that at least 50% of gastrointestinal malignancies express VEGF; additional studies correlate VEGF expression gastrointestinal tumor stage and cancer-related patient outcome.
Additionally, this proposal relies upon the ability of diagnostic laparoscopy to provide peritoneal washings suitable for the intended laboratory examination. We believe that our previous experience with cytologic, molecular and protein assay examination of peritoneal fluid supports this approach.

BACKGROUND/PREVIOUS WORK ACCOMPLISHED

Many patients with upper gastrointestinal malignancies present with advanced stage disease precluding surgical exploration; patients without distant metastases identified by preoperative imaging often possess occult intra-abdominal metastases that can be identified by diagnostic laparoscopy. (1) These patients usually do not benefit from a resection of the primary tumor given their high risk of distant recurrence, so the identification of this subset is critical. The concept underlying this approach uses the presence of intra-abdominal peritoneal metastasis as a marker of aggressive tumor biology. The effort of our diagnostic laparoscopy program in gastrointestinal malignancies is the development of biologic assays based upon strong pre-clinical and clinical evidence that can be practically applied to the use of staging laparoscopy of gastrointestinal cancers. The ultimate goal is the development of a more sensitive, sophisticated staging method and biologic therapy delivered via a laparoscopic approach.


The UT Southwestern Experience with Laparoscopy of UGI Malignancies.


In 1999, the Center for Minimally Invasive Surgery at our institution sponsored an IRB-approved protocol designed to evaluate the utility of laparoscopy in the staging and diagnosis of tumors arising from the upper gastrointestinal tract. The additional goals were the establishment a clinical database and the procurement of material for cytological and molecular study. Over an 12-month period 110 patients were consented and enrolled into the study and staging laparoscopy performed immediately prior to a planned open procedure for a proven or suspected malignancy of the esophagus, stomach, pancreas or hepatobiliary tract. The procedural steps performed were carefully outlined to ensure uniformity (see protocol in appendix). Fourteen patients were excluded from analysis because no gastrointestinal malignancy was confirmed, so that 96 patients provided data for study. The laparoscopic procedure was performed successfully in 94 (98%) of patients.
Prospective clinicopathologic data was collected from each enrolled patient. By laparoscopic examination alone 36% of patients were upstaged from the clinical stage as determined by preoperative examination and imaging. The majority of these represented an increase to stage IV disease. The data at the right demonstrates the percentage of patients upstaged by disease site; this data is similar to that reported by other staging laparoscopy series.(2)
In an attempt to improve the sensitivity of the examination and predict cancer related outcome; we performed peritoneal washing using a standard technique of delivery and collection. Of the 86 cases with diagnostic material, 17 (19.8%) were identified as positive with the highest percentage found in the gastric cancer site (43%). Clinical data was collected prospectively as patients were followed and outcomes of disease-free and overall survival recorded. As reported by others, the presence of positive cytology from peritoneal washings predicted a poor survival. (3)
Importantly, we have also performed molecular examination of epithelial cells harvested from the peritoneal washing procedure. Using standard molecular biology techniques we have examined the inactivation of known tumor suppressor genes by a biochemical mechanism known as DNA methylation. The inactivation of tumor suppressor genes is a critical event in the cancer progression, and the progressive inactivation of tumor suppressor genes is believed to reflect the biologic aggressiveness of the primary tumor. The data below is an example of single patient sample that was assayed for inactivation of several tumor suppressor genes. The overall survival data at the right also supports that progressive loss of tumor suppressor genes by methylation may be associated with clinical outcomes in patients with gastrointestinal malignancies. (3)

Angiogenesis: vascular endothelial growth factor in Gastrointestinal Malignancies

It has been established that growth of a solid tumor to 1-3mm in size requires new microvascular ingrowth (angiogenesis) to supply oxygen and nutrients to the malignant cells. Tumor angiogenesis and its promoting growth factors have been associated with larger tumor size, lymph node involvement and poor survival in patients with gastrointestinal malignancies. In comparison to normal mucosa and adenomatous polyps, colorectal cancers appear to express elevated levels of the angiogenic factor, vascular endothelial growth factor (VEGF), the biologic effects of which include regulation of pathologic angiogenesis through enhanced endothelial cell mitogenesis, migration and remodeling of the extracellular matrix, and increasing vascular permeability. (4) Immunohistochemical assessment of VEGF in primary tumor tissue has correlated with outcome in Stage II (node-negative) colon cancer, and increasing proportion and intensity of IHC staining has correlated with progression of disease from adenoma to metastatic cancer. (5,6) Increasing levels of both serum and plasma levels of VEGF have been shown to correlate with stage, lymphovascular invasion, and progression of disease. (7-11) Plasma levels may be more reliable than serum levels because of release of cytokines during clotting, or platelet scavenging of VEGF in maintenance of equilibrium. (7,12) Several studies have demonstrated elevated VEGF levels in ascites associated with gastrointestinal malignancies when compared to those patients with benign ascites; furthermore, subsequent in vitro experiments demonstrated that the VEGF in the ascites fluid stimulated vascular endothelial cell growth and permeability. (13,14)
Based upon this background information, we hypothesize that VEGF will be expressed by many of the upper gastrointestinal malignancies in patients to be enrolled in our laparoscopic examination protocol. VEGF, a soluble cytokine, should be present in the peritoneal washings aspirated from the peritoneal cavity in these patients. If VEGF in the fluid arises from the malignant cells, VEGF in the fluid should correlate with the presence of the protein in the free-floating malignant cells within the washings and in the primary tumor itself. Lastly, the known biology also supports the hypothesis that the presence of VEGF, and its level, will predict the malignant behavior of the primary (stage, development of metastases or ascites) and ultimately the outcome of the patient.


METHODS

Patient Counseling and Informed Consent
Men and women, older than 21 years of age, with no contraindications to surgery and histologically diagnosed or clinically suspected esophageal, gastric, peripancreatic or hepatobiliary malignancy will be informed of the study's risks and benefits and consent obtained. The consent allows for the risks imposed by the operative procedure and for the use of patient clinical data, tissue and fluids for laboratory study (see appendix for IRB-approved consent). As our institution serves a large population of underserved and minority patients who will have complete access to this trial; based upon preliminary data, >50% of the anticipated participants will fall into these categories.

Aim I. Using ELISA techniques soluble VEGF will be measured in peritoneal washings obtained during staging laparoscopy of patients with known or suspected upper gastrointestinal (esophagus, stomach, pancreas, hepatobiliary) malignancy.
Experiment 1. After visual inspection and exposure of the peritumoral region during staging laparoscopy, peritoneal washings will be obtained.
Scientific Rationale: Studies have clearly demonstrated that patients with metastatic gastrointestinal malignancies possess elevated VEGF levels in ascites that are greater than VEGF levels in non-malignant ascites. However, an attempt to measure VEGF from peritoneal washings obtained at laparoscopy has not been reported. Based upon our previous experience with cytokine measurements from intra-peritoneal washings (GEO), we believe that optimal measurements will be obtained from small volume irrigation near the primary tumor. This experiment will establish a standardize approach for the collection of peritumoral peritoneal washings for each upper gastrointestinal disease site. As different sites are to be evaluated, different operative methods will be employed to expose the tumor and obtain the specimen.
.Methods: All patients will receive the standard laparoscopic examination as outlined in Appendix B. Subsequently, specific maneuvers will be performed as needed in an effort to expose the tumor at each disease site (figures below) and a suction/irrigation device used to instill 100cc sterile normal saline in the peritumoral peritoneum. This fluid will be aspirated, collected sterilely and placed at 4oC. An additional 500cc will then be instilled throughout the peritoneum and recovered for cytologic study.
Anticipated results: In prior experience the laparoscopic procedure could be successfully completed in 98% of patients (3); we anticipate similar results in this study. Based upon experience with fluid recovery from prior protocols, we anticipate recovering 50cc of the 100 cc instilled around the primary tumor. This fluid will be used for the ELISA. About 300-500 cc of the remaining fluid will be recovered and sent to pathology for cytologic evaluation and the cytospin protocol.
Potential Pitfalls/Alternative Approaches: This portion of the project relies upon the ability of the laparoscopists to retrieve a peritoneal washing sample that will allow for accurate measurement of VEGF present within the peritoneal cavity. Our ability to expose the tumor will vary with anatomic location and the experience and determination of the laparoscopist. This could bias the ELISA results. Using previously generated laparoscopy data sheets (see appendix) we will record data with regard to the findings and technique. The ELISA data (volume of fluid, VEGF values) will be audited during the trial to determine if low values are due to surgeon-technical factors. If this is the case, we will limit the surgeons who may enroll patients to those who generate the most consistent yield and results.

Experiment 2. After centrifugation of the peritoneal fluid collected, the supernatent will be used for ELISA measurements for the VEGF cytokine.
Scientific Rationale: This is the central experiment to be performed as much of the subsequent statistical analysis will use the data collected at this step. ELISA is the established method by which cytokines have been measured from body fluids, and has been almost exclusively used to evaluate VEGF levels in ascitic fluid from previously published studies.(13,14)
Methods: After centrifugation to remove debris, the supernatents will be stored at -80oC until analysis. These samples will be analyzed for VEGF using a commercially available sandwich ELISA (R & D Systems, Minneapolis, MN). The sensitivity of the assay is 9.0 pg/ml as quoted by the manufacturer. Preliminary evaluation of the assay will include the examination of recovery of recombinant VEGF added to samples, the parallelism of diluted samples and the inter- and intra-assay coefficient of variation. The protein concentration in the peritoneal fluid will be determined using the Bradford dye-binding procedure (Bio-Rad Protein Assay, Bio-Rad Laboratories, Hercules, CA). All samples will be assayed in triplicate.
Anticipated results: Based upon published results, we anticipate the range of VEGF levels in patients with Stage IV disease to be 300-1800pg/ml (median ~ 600 pg/ml). Patients with lower stage disease will have lower VEGF levels although individual patient results will vary.
Potential Pitfalls/ Alternative Approaches: Sampling errors (poor peritoneal washing recovery, operator error) and examination of patients with early (eg. T1N0) tumors may result in very low VEGF levels in the specimen. The extreme sensitivity of the assay will allow for low measurements; however, interpretation of these data may be less reliable (eg are these levels falsely low?). The use of triplicate measures should account for technical errors in the performance of the assay. Simultaneous measurement of plasma VEGF levels (Aim II) will allow for cross-correlation of results as plasma levels are known to parallel peritoneal levels in patients with malignant ascites. Likewise, expression of VEGF from primary tumors has been shown to parallel the level of soluble VEGF malignant ascites. If low VEGF levels are secondary to dilution, we will alter the fluid harvest method and use as little fluid as possible. Statistical analysis will also account of very low to absent levels (see Statistical Section).

Aim II. VEGF levels in peritioneal fluid will be compared with plasma VEGF levels and with VEGF expression from the primary tumor and exfoliated cells harvested from the peritioneal wash specimen.
Experiment 1. After induction of anesthesia and prior to the laparoscopic examination, peripheral blood will be collected (in tubes with EDTA) from patients, processed and the plasma assayed for the presence of VEGF.
Scientific Rationale: It is known the peripheral VEGF levels are elevated in patients with metastastic gastrointestinal cancer; it is also been shown that these levels parallel (although several fold lower) the levels of soluble VEGF in malignant ascites. However, this has not been examined in patients with differing stages of gastrointestinal malignancy. We believe that plasma VEGF levels will parallel the stage of disease and the presence of VEGF in the peritoneal washings of the study patients.
Methods: After informed consent, venous samples will be obtained (two EDTA anticoagulant tubes) after general anesthesia is induced and prior to incision of the skin. The samples will be centrifuged (3,000 rpm) at 4oC for 10 minutes and the supernatants stored at -80oC until analyzed. ELISA will be preformed as in Aim I.
Anticipated results: Levels will be lower than the peritoneal washing assays; however, the levels will reflect the stage of disease.
Potential Pitfalls/ Alternative Approaches: VEGF is released into the circulation from activated platelets as well as circulating tumor cells; therefore, it is important that the samples are collected prior to tissue injury at surgery. This may represent a source of falsely elevated peripheral VEGF levels. If we believe this is a confounding element, a theoretical platelet-derived VEGF level can be calculated using the platelet count and the packed cell volume which can be obtained from a complete blood count (CBC) taken prior to surgery. (12)

Experiment 2. Exfoliated cells in the peritoneal washing will be concentrated, fixed and embedded into paraffin blocks using the Shandon cytospin protocol and the expression of VEGF from the cells assayed by Immunohistochemistry.
Scientific Rationale: In preliminary studies we have identified epithelial cells in 90% of peritoneal washings examined. It is believed that a potential source for VEGF in malignant ascites fluid is exfloliated malignant cells shed from the primary tumor. This experiment will refine the method of cytologic analysis so that immunohistochemical examination of the cells for VEGF expression can be performed.
Methods: Approximately 300cc of fluid will be obtained from the 500cc instilled throughout the abdomen. Using the Shandon Cytospin/Cytoblock System (Thermo Shandon, Pittsburgh, PA), the fluid will be centrifuged, the cells collected, fixed and embedded in paraffin. Sections (4-um thick) will be deparaffinized in xylene, rehydrated and stained with rabbit anti-human antibody to VEGF (Biogenex, San Ramon, CA) at room temperature and a biotinylated secondary antibody using an automated system. The slides will be reviewed by a pathologist (RA) and scored. The expression of VEGF will be assessed according to the percentage of immunoreactive cells per 1000 neoplastic cells as: "positive"=greater than 10% of cells staining or "negative"=less than 10% of cells staining.(5)
Anticipated results: Cells will be evaluable in at least 90% of all cases examined. Of these at least 20% will be clearly malignant. VEGF expression will be positive in greater than 60% of the malignant cells.
Potential Pitfalls/ Alternative Approaches: Benign cells of epithelial and non-epithelial origin may be present in the specimen as there is no method of selecting a pure population. Furthermore, non-malignant inflammatory cells can express VEGF, which may confound the results. However, as demonstrated above, this system allows for excellent morphologic examination of the cells, which should allow differentiation. Immunohistochemisty (cytokeratin, mucin, Ki-67) techniques can also be used to characterize the cells and determine origin and malignant characteristics.

Experiment 3. Patients who undergo resection or biopsy of the primary tumor will have the specimen processed by usual methods and immunohistochemistry for VEGF expression.
Scientific Rationale: Strong association as been demonstrated between the expression of VEGF, as measured by immunohistochemistry, and stage in gastrointestinal malignancies. Most of these patients will undergo at least a biopsy of the primary or metastatic tumor, which can be examined and scored for VEGF expression. An association between VEGF expression in the primary tumor and the presence of VEGF in the peritoneal washings will link VEGF in the fluid and the primary tumor biology and support our central hypothesis.
Methods: Formalin-fixed, paraffin embedded samples will be stained and scored for VEGF as outlined in Experiment 2 of this Aim.
Anticipated results: The VEGF expression pattern in the primary will correlate with that found in the exfoliated cells harvested and with the levels of soluble VEGF in the peritoneal washings.
Potential Pitfalls/ Alternative Approaches: Many patients who are upstaged at diagnostic laparoscopy (~40%) may only have a biopsy of intra-abdominal metastasis performed. It is possible that the expression of VEGF in these metastases may differ from that found in the primary tumor. Nevertheless, our primary endpoint, the relationship between the level of soluble VEGF in the peritoneal washings and the stage of the tumor, should remain unchanged.

Aim III. VEGF levels in peritoneal fluid will be correlated with the clinical /pathologic stage of the tumor and the clinical endpoints of cancer-related morbidity and mortality
Experiment 1: Clinicopathologic data will be collected on each patient prior to laparoscopic examination, and VEGF ELISA and IHC data analyzed with respect to the clinical and pathologic stage of the patient's malignancy.
Scientific Rationale: The most widely used method of predicting the outcome of patients with upper gastrointestinal malignancies is the TNM staging system. However, this system often fails to reflect important biologic differences for tumors within each stage. In patients of the same tumor TNM stage, the presence of VEGF expression has been shown retrospectively to predict a worse patient outcome. Using prospectively collected clinical data we hope to link the presence of soluble VEGF in peritoneal washings obtained and laparoscopy with patient stage and eventually outcome.
Methods: After informed consent, clinicopathologic data will be collected from each study patient. The data will be entered onto a large computer database. Using the TNM staging system for each disease site, the patients will receive a clinical stage based upon preoperative clinical and image findings. A new staging will be performed after the laparoscopic exam. Finally, those patients who undergo tumor resection will receive a pathologic stage. Statistically analysis of the ELISA values and staging data will be performed as outlined below.
Anticipated results: The presence of high levels of VEGF in the peritoneal washings will correlate with those patients upstaged staged to stage IV at laparoscopy.
Potential Pitfalls/ Alternative Approaches: Errors in clinical data collection are always possible. To prevent this, data sheets will be audited monthly and compared against the patient record and pathologic findings. Most of the patients will be followed by the principal investigator in a surgical oncology clinic were the remainder of the clinical data can be collected.

Data Presentation and Comparative Statistical Analysis

ELISA data: This is continuous data and will be presented as medians with the 25th and 75th percentiles. Many of the data will likely not be normally distributed or may be under the detection range of the assay. Accordingly, the percentage of subjects with detectable concentrations of VEGF will be presented for analysis. VEGF concentrations will be compared by using the Wilcoxon Rank Sum test and repeated measures analysis of variance. Spearman rank correlation will be used to examine associations between peritoneal VEGF and plasma VEGF.
VEGF vs Tumor TNM Stage
The primary analysis will be the comparison of VEGF levels in the peritoneal washings, serum and tissue to for patients with metastatic cancer (stage IV) versus the others (stage I-III).
Statistical Power Analysis
Based upon preliminary data we anticipate accruing 90 patients over a 12-month period of this study; we estimate that approximately 50% of these patients will have metastatic disease (stage IV) after staging laparoscopy completed (data at right). In patients with metastatic disease, prior studies have shown VEGF levels to be markedly elevated (median: 597 pg/ml) in comparison to patients without metastases (median: 317 pg/ml). (13) To test the null hypothesis that peritoneal VEGF concentrations are not different between patients with and without metastases, and given these reported differences our proposed study will have a power of 83% to detect a statistically significant (p<0.05) difference in mean VEGF concentrations. This computation assumes that the mean difference is 280 (corresponding to estimated means of 597 versus 317) and that the common within-group standard deviation is 450. The report on which we have based these estimates identified a wide range of VEGF concentrations. Therefore, we conservatively chose a relatively large standard deviation. Should our mean differences be smaller (497 pg/ml versus 317pg/ml, for example) the power will fall to 47%. Similarly, should the variability in VEGF concentrations be greater than estimated (standard deviation of 600pg/ml, for example) the power to detect a difference will be approximately 59%. While we recognize these possibilities, we have provided conservative estimates that indicate our proposed study has sufficient power to identify realistic differences.

RESOURCES
Laboratory
Dr. Fleming will work on the NB8 floor of the Simmons Biomedical Research Building on the North Campus at UTSW. The NB8 floor is a recent (1994), state of the art, laboratory research facility of 30,000 gross and 18,000 nsf comprising the Hamon Center for Therapeutic Oncology Research with eleven principal investigators. Dr. Fleming has dedicated laboratory space, which are fully equipped for modern molecular and cell biologic work. In addition, there is considerable shared core lab space. In addition, there are shared core facilities in the Hamon Center and the Department of Surgery for tissue culture, dark rooms, cold rooms, bacteriologic work and incubators, centrifuges, freezers, computers, a DNA sequencing center on the NB8 floor. Currently there are ~70 personnel working in the Hamon Center. In addition, there are shared facilities the investigators have access to in the Departments of Pharmacology, Cell Biology and Pathology including microscopy, imaging center, darkrooms, high speed and ultracentrifuges, scanning densitometer, phosphoimager, spectrophotometers, liquid scintillation and I125 counters, autoclaving and dishwashing facility.
General Clinical Research Center
The General Clinical Research Center (GCRC), supported by the NIH, provides special resources free-of-charge for the conduct of patient-oriented research. Resources include: 11-bed inpatient facility at Parkland Memorial Hospital, outpatient facility at the Clinical Science Building, a general laboratory, and statistical-computational support. There is no cost for hospitalization of research patients. A modest support is available for ancillary routine laboratory tests.
Clinical
Cancer patients are seen in several in and outpatient facilities used exclusively by UTSW faculty. These included Parkland Memorial Hospital (PMH) inpatient and outpatient clinics, the private Aston Ambulatory Care Center, and private Zale Lipshy University Hospital, all connected together. In addition patients are seen at the large Dallas Veterans Administration Medical Center 12 miles from the main campus. There are a large number of new and returning cancer cases seen each year in these facilities. For example, currently there are ~200 newly diagnosed lung cancer cases and over 200 new cases of breast cancer seen in the combined UTSW clinical operations.

REFERENCES

1 Ramshaw BJ: Laparoscopic Surgery for Cancer Patients. CA Cancer J Clin 1997. 47:327-350.
2 Bieligk SC, Huang H, Fleming JB, et al.: The utility of laparascopy in the staging evaluation and management of upper gastrointestinal malignancies. Abstract presented at the North Texas Chapter of the American College of Surgeons Meeting. Feb, 2000.
3 Bieligk SC, Fleming JB, Feller YM, Minna JD. Detection of aberrant tumor promoter methylation in cells from laparoscopic peritoneal washings of patients with upper gastrointestinal tract malignancies. Proc Am Assoc Can Res, 2001; 42:3792.
4 Takahashi Y, Kitadai Y, Bucana CD, et al.: Expression of vascular endothelial growth factor and its receptor, KDR, correlates with vascularity, metastasis and proliferation of human colon cancer. Cancer Res 1995, 55:3964-3968.
5 Cascinu S, Staccioli MP, Gasparini G, et al.: Expression of endothelial growth factor can predict event-free survival in stage II colon cancer. Clin Cancer Res 2000, 6:2803-2807.
6 Lee JC, Chow NH, Wang, et al: Prognostic value of endothelial growth factor expression in colorectal cancer patients. Eur J Cancer 2000, 6: 748-753.
7 Dirix L, Vermeulen PB, Pawinski A, et al.: Elevated levels of angiogenic cytokines basic fibroblast growth factor and vascular endothelial growth facot in sera of cancer patients. Br J Cancer 1997, 76: 238-243.
8 Hyodo I, Doi T, Endo H, et al.: Clinical significance of plasma vascular endothelial growth factor in gastrointestinal cancer. Eur J Cancer 1998, 34: 2041-2045.
9 Takeda A, Shimada H, Imaseki H, et al: Clinical significance of serum vascular endothelial growth factor in colorectal cancer patients: correlation with clinicopathologic factors and tumor markers. Oncol Rep 2000, 7: 333-338.
10 Andre T, Kotelevets L, Valliant J-C, et al. : VEGF, VEGF-B, VEGF-C, and their receptors KDR, FLT-1, and FLT-4, during the neoplastic progression of human colonic mucosa. Int J Cancer 2000, 86: 174-181. 11 Davies MM, Jonas SK, Kaur S, et al.: Plasma vascular endothelial but not fibroblast growth factor levels correlate with colorectal lever metastases vascularity and volume. Br J Cancer 2000, 5: 1004-1008.
12 George ML, Eccles SA, Turron MG, et al.: Correlation of plasma and serum cascular endothelial growth factor levels with platelet count in colorectal cancer: clinical evidence of platelet scavenging? Clin Cancer Res 2000, 6: 3147-3152.
13 Zebrowski BK, Liu W, Ramirez K, et al.: Markedly elevated levels of vascular endothelial growth factor in malignant ascites. Ann Surg Onc 1999, 6(4):373-378.
14 Kraft A, Weindel K, Ochs A, et al.: Vascular endothelial growth factor in the sera and effusions of patients with malignant and nonmalignant disease. Cancer 1999, 85(1): 178-87.