The integrin alpha 6 beta 4 is a laminin receptor.

In this study, the putative laminin receptor function of the alpha 6 beta 4 integrin was assessed. For this purpose, we used a human cell line, referred to as clone A, that was derived from a highly invasive, colon adenocarcinoma. This cell line, which expresses the alpha 6 beta 4 integrin, adheres to the E8 and not to the P1 fragment of laminin. The adhesion of clone A cells to laminin is extremely rapid with half-maximal adhesion observed at 5 min after plating. Adhesion to laminin is blocked by GoH3, and alpha 6 specific antibody (60% inhibition), as well as by A9, a beta 4 specific antibody (30% inhibition). Most importantly, we demonstrate that alpha 6 beta 4 binds specifically to laminin-Sepharose columns in the presence of either Mg2+ or Mn2+ and it is eluted from these columns with EDTA but not with NaCl. The alpha 6 beta 4 integrin does not bind to collagen-Sepharose, but the alpha 2 beta 1 integrin does bind. Clone A cells do not express alpha 6 beta 1 as evidenced by the following observations: (a) no beta 1 integrin is detected in beta 1 immunoblots of GoH3 immunoprecipitates; and (b) no alpha 6 beta 1 integrin is seen in GoH3 immunoprecipitates of clone A extracts that had been immunodepleted of all beta 4 containing integrin using the A9 antibody. These data establish that laminin is a ligand for the alpha 6 beta 4 integrin and that this integrin can function as a laminin receptor independently of alpha 6 beta 1.

Rat model for carcinogenesis in ureterosigmoidostomy.

A rat model is used to study the carcinogenesis that occurs when urine is surgically diverted into the fecal stream, as in ureterosigmoidostomy. Adenocarcinoma of the colon occurs adjacent to the urine inlet. It is completely prevented by proximal diversion of the feces, implying that fecal carcinogens are activated locally by the urine or the urothelium.

Preoperative and postoperative combination chemotherapy for potentially resectable gastric carcinoma.

BACKGROUND: Median survival of patients with local-regional gastric carcinoma is 10 months. Resection of the primary tumor and regional lymph nodes, with tumor-free margins (curative resection), has been the most effective treatment for local-regional gastric carcinoma. However, median survival of patients with curative resection of gastric carcinoma is 24 months, and the 5-year survival rate is about 20%. A single institution pilot study has established the feasibility of administering two courses of chemotherapy preoperatively and three courses postoperatively. In another study, a 15% pathologically documented complete response (pathologic complete response) has been reported in unresectable gastric carcinoma treated with etoposide, doxorubicin, and cisplatin. PURPOSE: Our purpose was to increase the curative resection rate in potentially resectable gastric carcinoma and to delay or eliminate micrometastases and thus improve survival. We also evaluated clinical and pathologic response to chemotherapy. METHODS: Forty-eight previously untreated patients with potentially resectable gastric carcinoma received a chemotherapy regimen (EAP) consisting of etoposide (120 mg/m2 intravenously over a 2-hour period on days 4, 5, and 6), doxorubicin (20 mg/m2 as a 10-minute intravenous infusion on days 1 and 7), and cisplatin (40 mg/m2 as a 1-hour intravenous infusion on days 2 and 8). Patients received three courses of chemotherapy before resection, and responding patients received two courses postoperatively. Clinical and pathologic response rates, toxicity, patterns of treatment failure, and survival times were assessed. RESULTS: A median of three courses (range, 1-5) of preoperative therapy was administered; six (12%) of the 48 patients had clinical complete response, and nine (19%) had partial response. Forty-one (85%) underwent surgery; 37 (90%) of these 41 (77% of the 48 patients) had a curative resection. There were no pathologic complete responses. Median survival for all patients is 15.5 months (range, 2-29+ months). Therapy was discontinued because of the toxic effects in one patient before surgery and in six patients after surgery. Doses were reduced in 37 patients (77%), mainly because of hematologic toxicity. Nineteen (40%) were hospitalized because of toxic effects, including 15 patients who developed fever with neutropenia. Grade 3 or 4 nausea and vomiting occurred in 15 patients and grade 3 or 4 diarrhea in seven patients. One death was directly related to chemotherapy. CONCLUSIONS: These data support that administration of preoperative and postoperative chemotherapy for local-regional gastric carcinoma is feasible in a multi-institutional setting. Our findings demonstrate that this EAP regimen is modestly active but is associated with substantial toxicity. IMPLICATIONS: Use of preoperative chemotherapy in resectable gastric carcinoma merits further evaluation, but more effective drug regimens will be required before a controlled trial is initiated.

Detection of the tumor-associated glycoprotein antigen (TAG-72) in premalignant lesions of the colon.

We used monoclonal antibody B72.3 to study the expression of the colorectal carcinoma-associated antigen TAG-72 in premalignant colonic lesions with the immunoperoxidase technique. This antigen, which is rarely detectable in the normal colonic epithelium, was expressed in 13 of 19 adenomas with moderate to severe dysplasia and nine of nine cases of inflammatory bowel disease. The antibody reacted with the normal-appearing mucosa adjacent to a carcinoma in 10 of 12 cases, although only eight of the tumors expressed the antigen. The expression of the TAG-72 antigen in the colonic epithelium may be an early marker of malignant transformation.

Relationship between extracellular matrix interactions and degree of differentiation in human colon carcinoma cell lines.

Human colon carcinoma cell lines that vary in their degree of differentiation were examined for their ability to interact with extracellular matrix components. For this purpose, established cell lines were classified on the basis of several criteria that relate to degree of differentiation. These criteria include histology of the original tumor, histology of xenografts, in vitro morphology, and carcinoembryonic antigen expression. On this basis, the cell lines used were either moderately well or poorly differentiated. The poorly differentiated cell lines adhered to surfaces coated with laminin or reconstituted basement membrane extract (Matrigel) to a significantly greater extent than the moderately well differentiated lines with the exception of one moderately well differentiated line that was derived from a highly aggressive signet ring cell carcinoma. In addition, the poorly differentiated cell lines exhibited considerable spreading on laminin and Matrigel after adherence that was not evident for the moderately well differentiated lines. The adherence of these cell lines on fibronectin-coated surfaces did not correlate as well with differentiation although, in general, poorly differentiated cell lines adhered better than moderately well differentiated lines. None of the cells that adhered to fibronectin exhibited the extensive spreading seen on laminin. The specificity of tumor cell interactions with extracellular matrix glycoproteins was examined using synthetic peptides which correspond to sequences within these proteins that are recognized by cell surface receptors. The pentapeptide YIGSR-NH2 significantly inhibited the adherence and spreading of the tumor cell lines on laminin, but not on fibronectin. The peptide RGDS, however, did not inhibit tumor cell interactions with laminin although it did inhibit their interactions with fibronectin. Thus, the interactions of colon carcinoma cells with laminin and fibronectin are probably mediated by separate receptors. Taken together, the data demonstrate that cells derived from colon carcinomas exhibit considerable variation in their ability to interact with extracellular matrix components, and that this variability is related to the degree of differentiation of original tumor.

Aberrant mitochondria in two human colon carcinoma cell lines.

Electron micrographs of CCL237 and FET cells (two slowly growing, differentiated human colon carcinoma lines) revealed enlarged mitochondria with few cristae. Polarographic measurement of respiratory activity in mitochondria isolated from these cell lines was compared to that of CV-1 cells (a normal monkey kidney epithelial line) and MIP101 cells (another human colon carcinoma line), both of which have mitochondria with a "normal" appearance. The respiratory control ratios of CCL237 and FET mitochondria were found to be considerably lower than those of CV-1 and MIP101 mitochondria (approximately 3 as compared to greater than 10, respectively), indicating that in CCL237 and FET mitochondria the processes of substrate oxidation and phosphorylation of ADP are only loosely coupled. In intact cells, differences in radiolabeled tetraphenylphosphonium uptake showed that the mitochondrial membrane potential in CCL237 and FET cells was less than that in CV-1 and MIP101 cells, and that nigericin failed to hyperpolarize the mitochondria of CCL237 and FET cells. In addition, FET mitochondria exhibited significantly lower ADP-stimulated and uncoupled respiratory rates than mitochondria isolated from the other cell types, indicating that in the former, the capacity for oxidative phosphorylation is somehow impaired. Selective toxicity of FET cells was obtained by treatment with 2-deoxyglucose, an inhibitor of glycolysis, suggesting the possibility of exploiting the phenotype of impaired oxidative metabolism for chemotherapy.

Sialyltransferase activity and hepatic tumor growth in a nude mouse model of colorectal cancer metastases.

Sialyltransferase activity (EC 2.4.99.6) was measured in the microsomal fraction of colorectal cancer cell lines using an assay based on the incorporation of [14C]CMP-sialic acid into asialofetuin. In the poorly differentiated lines MIP101 and Clone A, sialyltransferase activity had a Vmax of 0.36 and 0.31 nmol/mg protein/h, respectively, while the moderately differentiated to well-differentiated cell lines HT-29, CCL188, and CX-1 had Vmaxs of 2.46, 1.05, and 1.24 nmol/mg protein/h, respectively. All cell lines tested had a Km of 15.4 (+/- 0.7)(SD) mumol/liter. The better differentiated cells had higher levels of sialyltransferase activity, which correlated with their higher levels of sialic acid and their enhanced ability to form liver metastases in the nude mouse following intrasplenic injection compared to the poorly differentiated cell lines. Treatment of the cell lines with KI-8110, a CMP-sialic acid derivative which prevents incorporation of sialic acid into glycoconjugates, resulted in reduced formation of hepatic metastases by the colorectal carcinoma cell lines in the nude mouse model. It is suggested that reduced sialylation of adhesion molecules such as carcinoembryonic antigen may change the biology of the tumor cell, one consequence of which is the prevention of implantation of the cells into distant sites, resulting in a reduced incidence of metastases.

c-Ha-ras-I oncogene-induced differentiation and natural killer cell resistance in a human colorectal carcinoma cell line.

Natural killer (NK) activity is primarily a peripheral blood function of a lymphocyte population capable of spontaneous lysis of many transformed and metastatic targets. However, NK-susceptible targets tend to be relatively poorly differentiated. We have previously shown that poorly differentiated human colorectal carcinoma are lysed by NK cells. Well-differentiated and chemically differentiated colorectal carcinomas are insensitive to NK lysis. The present study demonstrates that transfection of the c-Ha-ras-I oncogene into a poorly differentiated colorectal carcinoma cell line also renders it NK resistant. This resistance is accompanied by a more differentiated colorectal carcinoma phenotype. Two ras-transfected lines (Clone-A-5 and Clone-A-4) showed a 30-66% decrease in susceptibility to NK lysis as compared to the parental line in standard cytotoxicity assays. The resistance of these transfectants was strictly dependent on expression of the activated p21, the H-ras protein product. Studies to assess the integrity of the initial binding step in NK lysis showed a significant decrease in the ability of these transfectants to form conjugates with fresh NK cells. It is likely that transfection with c-Ha-ras-I has selectively modulated critical NK target recognition structures.

Expression of a Mr 32,000 laminin-binding protein messenger RNA in human colon carcinoma correlates with disease progression.

Cell surface receptors for laminin may play an important role in tumor migration and metastasis. To evaluate laminin receptor/laminin-binding protein expression in human colon carcinoma, surgical specimens of primary colon cancers and liver metastases were examined by blot hybridization of total RNA with a complementary DNA clone which encodes a Mr 32,000 human laminin-binding protein. The mRNA level of the laminin-binding protein was higher in primary colon carcinoma than in adjacent normal colonic epithelium in 20 of 21 cases. In all 6 cases of colon cancer liver metastases, the laminin-binding protein mRNA level was more than 3-fold greater in tumor than in adjacent normal liver tissue. The tumor/normal ratio of this laminin-binding protein mRNA expression in primary colon cancer has significant correlation with Dukes' classification (P less than 0.001). Our results suggest that mRNA expression of the laminin-binding protein may be a marker of human colorectal cancer progression and biological aggressiveness.

Increased expression of human ribosomal phosphoprotein P0 messenger RNA in hepatocellular carcinoma and colon carcinoma.

To search for differentially expressed gene products in selected cancers of endodermal origin, cDNA libraries derived from mRNA in human hepatocellular carcinoma and adjacent grossly normal tissue were generated. From these parent libraries, subtracted cDNA libraries of tumor minus normal and normal minus tumor tissues were constructed. After screening these subtracted libraries by +/- hybridization, a cDNA clone that is overexpressed in hepatocellular carcinoma and encodes the human acidic ribosomal phosphoprotein P0 (P0) was identified. We then evaluated the expression of this phosphoprotein P0 in human colon carcinoma samples. Surgical specimens of primary tumors and liver metastases were examined by Northern hybridization of total RNA with one of 2 32P-labeled P0 probes. The mRNA level of the P0 was greater in primary colon carcinoma than in paired adjacent normal colonic epithelium in 36 of 38 cases; the mean tumor/normal ratio was 2.7 (range, up to 13). The tumor/normal ratio, when plotted against the Dukes' stage of disease, gave evidence for increasing P0 expression with increasing stage of colon carcinoma (P = 0.02). In all 8 cases of paired colon carcinoma metastatic to liver and 2 cases of paired primary hepatocellular carcinoma, the P0 mRNA level was greater in tumor than in adjacent normal liver tissue. The mean tumor/normal ratio was 4.0 (range, up to 11) for the colon cancers metastatic to liver and 4.2 for the primary hepatocellular carcinoma samples. These findings support a common increased expression of selected gene products in different tumors of endodermal origin and suggest that increased P0 expression, in line with certain other ribosomal proteins, may be associated with human colorectal cancer progression and biological aggressiveness.

Gastric and hepatocellular carcinomas do not overexpress the same ribosomal protein messenger RNAs as colonic carcinoma.

The levels of a number of ribosomal protein mRNAs are reported to be increased in human colon cancer. We have assessed whether selected ribosomal protein mRNAs are overexpressed in other gastrointestinal malignancies, namely gastric and hepatocellular carcinomas. Subtracted complementary DNA libraries were generated from paired samples of human (a) colorectal carcinoma minus adjacent normal colonic mucosa and (b) hepatocellular carcinoma minus adjacent normal liver. Screening of approximately 3% of these library clones determined that ribosomal protein mRNAs encoding L18 and L37 (not previously reported) and P0 and S6 were overexpressed in one or the other library. Their complementary DNA inserts were then used as probes to evaluate their expression in a larger number of paired tumor/normal surgical samples of human colonic, gastric, and hepatocellular carcinomas, by Northern hybridization. The mRNA signal was greater in the colonic carcinoma than in paired adjacent normal colonic mucosa in 38 of 42 cases for P0 [tumor/normal (T/N) ratio = 3.0 +/- 0.3, mean +/- SE, P < 0.001] (G. F. Barnard, R. J. Staniunas, S. Bao, K. Mafune, J. L. Gollan, G. D. Steele, Jr., and L. B. Chen, Cancer Res., 52: 3067-3072, 1992), in 25 of 28 cases for L18 (T/N ratio = 3.7 +/- 0.5, P < 0.001), in 27 of 28 cases for L37 (T/N ratio = 5.3 +/- 0.4, P < 0.001), and in 24 of 28 cases for S6 (T/N ratio = 3.1 +/- 0.5, P < 0.01). The level of mRNA overexpression of L18 and S6 did not correlate with the Dukes' stage of disease. In hepatocellular carcinoma samples, using the same four ribosomal protein complementary DNA probes, only P0 mRNA was significantly increased (T/N ratio = 2.8 +/- 0.4, n = 6, P = 0.047). In gastric carcinoma samples, none of these mRNAs was increased (mean T/N ratios = 0.9-1.2, n = 6). Therefore, gastric and hepatocellular carcinomas do not overexpress the same ribosomal protein mRNAs as do colonic carcinoma.

Ubiquitin-ribosomal protein S27a gene overexpressed in human colorectal carcinoma is an early growth response gene.

One of the extension proteins on the carboxy terminus of ubiquitin was reported as the ribosomal protein S27a. We have cloned a gene which encodes this ubiquitin hybrid protein from a complementary DNA library of a human colon carcinoma cell line. Northern blot analysis of surgical specimens from colon cancer patients showed that these messenger RNA levels were higher in tumor tissue than in adjacent normal mucosa. Furthermore, to investigate the role of this novel ubiquitin hybrid gene in cellular growth control, the responsiveness of this gene to serum growth factors was examined. Within 30 min after serum or 12-O-tetradecanoylphorbol-13-acetate stimulation, its messenger RNA expression in rat fibroblast cells (Rat 1) was increased. Nuclear runoff transcription studies showed that the kinetics of induction of this gene is almost identical to that of protooncogene c-jun or c-fos, the known early growth response genes. Thus, this ubiquitin hybrid gene appears to be a novel early growth response gene overexpressed in human colon cancer and warrants further studies in the pathogenesis of colorectal carcinoma.

Prothymosin-alpha mRNA expression correlates with that of c-myc in human colon cancer.

Prothymosin alpha (PT-alpha) is a nuclear protein involved in cell proliferation. Transcription of PT-alpha has been reported to be regulated by the c-myc gene in vitro. We identified PT-alpha as being overexpressed in a human colon cancer minus normal mucosa subtraction cDNA library. Northern blot (messenger RNA) analysis showed that both PT-alpha and c-myc genes were overexpressed in human colorectal cancers compared with adjacent normal tissues. Immunohistochemical studies for PT-alpha and c-myc supported these findings. There was no correlation between PT-alpha or c-myc messenger RNA expression and Dukes' stage of colorectal cancer; or between either of these two and actin messenger RNA expression. There was, however, a significant correlation between the PT-alpha expression and c-myc expression (P < 0.001). These findings support the hypothesis that PT-alpha gene transcription may be associated with, and possibly under the control of, the c-myc gene in human colorectal cancers.

Human ribosomal protein L37 has motifs predicting serine/threonine phosphorylation and a zinc-finger domain.

Ribosomal protein L37 mRNA is overexpressed in colon cancer. The nucleotide sequences of human L37 from several tumor and normal, colon and liver cDNA sources were determined to be identical. L37 mRNA was approximately 375 nucleotides long encoding 97 amino acids with M(r) = 11,070, pI = 12.6, multiple potential serine/threonine phosphorylation sites and a zinc-finger domain. The human sequence is compared to other species.

Nucleotide and deduced amino acid sequence of human ribosomal protein L18.

Ribosomal protein L18 mRNA is overexpressed in human colorectal cancer compared to normal colon tissue. We report the nucleotide sequence of human L18 cDNA derived from a normal colon source. There were no mutational changes in segments of L18 cDNA derived from two tumor sources. The L18 cDNA was 690 base pairs long and predicts a single open reading frame of 564 nucleotides, encoding 188 amino acids with a M(r) = 21,621, it is homologous to rat L18 and Xenopus laevis L14.

Ubiquitin fusion proteins are overexpressed in colon cancer but not in gastric cancer.

A cDNA clone (AF3) encoding the ubiquitin A gene 52 amino acid extension fusion protein (UbA52) was isolated from a subtracted cDNA library of human colorectal carcinoma minus adjacent normal mucosa. In Northern hybridization the mRNA signal for UbA52 was greater in surgical samples of colonic carcinoma (T) than in paired adjacent normal (N) tissues in 24 of 29 cases (T/N = 3.4 +/- 0.5, P < 0.01). An oligonucleotide probe specific for only the 52 amino acid extension confirmed the overexpression of UbA52. In contrast, there was no overexpression of UbA52 mRNA in gastric cancer samples (n = 7, T/N = 1.0 +/- 0.3). The mRNA of several ribosomal proteins, and of another ubiquitin A gene fusion protein, UbA80, with an 80 amino acid extension of ribosomal protein S27a, have been reported to be over-expressed in colon cancer, but not as yet at the protein level. Using rabbit antisera to the ribosomal protein component S27a we demonstrate over-expression of S27a at the protein level in colonic (n = 5), but not gastric (n = 6) carcinomas. Therefore it is likely that both UbA80 and UbA52 are overexpressed in colon cancer, but not in gastric cancer.

Etoposide, doxorubicin, and cisplatin chemotherapy for advanced gastric adenocarcinoma: results of a phase II trial.

PURPOSE: A phase II study of etoposide, doxorubicin, and cisplatin (EAP) therapy in patients with advanced gastric carcinoma was performed in an attempt to confirm encouraging results reported by German investigators using an identical EAP regimen. PATIENTS AND METHODS: Thirty-six consecutive, previously untreated patients with surgically unresectable, measurable gastric carcinoma were treated every 28 days with etoposide (120 mg/m2, days 4, 5, and 6), doxorubicin (20 mg/m2, days 1 and 7), and cisplatin (40 mg/m2, days 2 and 8). A total of 108 courses of treatment was given. RESULTS: Therapy was associated with myelosuppression (median granulocyte nadir, 239/microL; median platelet nadir, 81,000/microL), which reached its maximum 14 days after the start of therapy and necessitated hospitalization after 24 of 108 (22%) treatment courses. Four of 36 (11%) patients died of treatment-related toxicity: three from sepsis and one from hemorrhage. Objective responses were observed in 12 of 36 (33%) patients; three (8%) patients experienced a clinical complete response. The median time to progression was 4 months for all 36 patients and 8 months for the 12 responding patients. Of the 13 patients with localized but unresectable disease, five subsequently underwent surgical resection; only one of these five was rendered disease-free. CONCLUSION: The EAP regimen is highly toxic and results in response rates and survival times no better than those of more easily tolerated treatment programs.

Serial levels of CA 19-9 and CEA in colonic cancer.

The use of serial carbohydrate antigen (CA) 19-9 assays was assessed by comparison with serial carcino-embryonic antigen (CEA) levels on the plasmas of 53 patients with colorectal carcinoma. The patients had all undergone resection for their primary tumors and in six instances subsequent resections for hepatic metastases. Initial CA 19-9 levels were greater than or equal to 37 U/mL in 22 of the 53 patients (41%) and in 68% of the patients with metastatic disease. Similar trends of serial CA 19-9 and CEA levels were found in 79% of the 53 patients. One patient with initially normal CEA levels had elevated CA 19-9 levels from the start. In ten of the 53 patients (19%), serial CA 19-9 levels remained low despite tumor recurrence or progression, and despite increasing CEA levels above 5 ng/mL. The increasing serial CEA trends predicted recurrence in 88% and increasing CA 19-9 trends in 50% of cases, which was increased to 70% by including trends of CA 19-9 levels below 37 U/mL. Following hepatic lobectomy, both serial CEA and CA 19-9 levels decreased rapidly. Used alone, serial CA 19-9 levels did not appear to be as sensitive as standard CEA in this retrospective study of selected patients.

Long-term hepatic arterial infusion of 5-fluorodeoxyuridine for liver metastases using an implantable infusion pump.

Twenty-one patients with liver metastases of various histologies (predominantly colorectal carcinoma) underwent Infusaid pump implantation for long-term hepatic arterial 5-fluorodeoxyuridine (5-FUdR) infusion. Patients received 5-FUdR infusion on a 2-wk cycle alternating with a 2-wk saline--heparin infusion. A dosage of 0.2-0.3 mg/kg/day (average 0.23 mg/kg/day) was infused for a cumulative 5-FUdR administration of 1940 days. Six patients (29%) responded to therapy (five colorectal, one carcinoid); median response duration was 6 mo. Median survival for the treated group was 17 mo from diagnosis of liver metastases and 13 mo from pump implantation. Median survival among the six responding patients was 15 mo from diagnosis of liver metastases and 11 mo from pump implantation. Comparison of survival from the diagnosis of liver metastases of the treated group to ten patients found ineligible for the study by virtue of extrahepatic metastases revealed no significant difference in median (18 mo for ineligible group) or overall survival. However, median survival for the treated group after pump implantation (13 mo) was significantly better than the median survival of the ineligible group after evaluation for this study (4 mo). Toxicities of therapy included fatigue, anorexia, nausea, vomiting, toxic hepatitis, epigastric pain, and diarrhea. No patients died of toxicity, but six patients required hospitalization for management of pain or vomiting. No serious technical complications developed in any patient except separation of the infusion catheter at its junction with the pump in one patient, necessitating pump replacement for continuation of therapy. These survival data suggest identification of new anticancer agents for hepatic arterial infusion.

Human and murine Kupffer cell function may be altered by both intrahepatic and intrasplenic tumor deposits.

The liver is the most common site of hematogenous metastases from colorectal carcinoma. Kupffer cells (KC), which line the hepatic sinusoids, may form the first line of defense against circulating tumor cells. The purpose of this study was to determine the effect of hepatic metastases and intra-abdominal tumor growth on KC binding of human colorectal carcinoma (HCRC) cells. MIP-101, a poorly metastatic cell line, and CX-1, a highly metastatic cell line, were injected intrasplenically into nude mice and KC were isolated by collagenase perfusion at varying intervals after injection. Conditioned media were collected from MIP-101, CCL 188 and CX-1 to determine their in vitro effect on KC function. KC from MIP-101 injected mice (14% liver metastases, 100% splenic tumors) bound a significantly greater number of MIP-101 and clone A cells than CX-1 cells in vitro. KC isolated from mice 5 weeks after CX-1 injection (100% liver metastases) also showed increased binding of MIP-101 and clone A cells compared to CX-1 cells. Similar results were obtained when tumor cell binding to normal human liver KC was compared to binding to KC from human livers from patients with hepatic metastasis from colorectal cancer. In contrast KC obtained from mice 3 weeks after CX-1 injection (44% liver metastases) showed significantly decreased binding of MIP-101 and clone A cells. The conditioned medium from CX-1 cells significantly decreased the in vitro binding of both MIP-101 and CX-1 by KC. These results indicate that the ability of KC to bind HCRC cells (which precedes phagocytosis and tumor cell killing) is a dynamic function and affected by concomitant tumor growth. HCRC cells may alter KC function via the production of specific tumor-derived soluble factors. In order to devise new and more effective therapeutic options in the treatment of liver metastases the nature of this tumor cell-KC interaction must be better understood.