Introduction: NV1034 is a novel multi-mutated, replication competentherpes simplex virus (HSV) type-I that combines two anticancer strat-egies: 1) Direct tumor oncolysis and 2) Immune modulation. It is anoncolytic virus that also carries the gene for murine GM-CSF. NV1034was compared to NV1020, the parent HSV mutant lacking GM-CSF.

Methods: Antitumor activity against CT26, the murine colorectal car-cinoma cell line was studied. Cytotoxicity and GM-CSF productionwere measured in cells infected in vitro with NV1034 or NV1020 atmultiplicity of infection (MOI) of 1. Efficacy of NV1034 and NV1020 wasthen tested in subcutaneous tumors in syngeneic Balb/c mice. Tumors(5–10 mm) were treated with 13107 plaque forming units of NV1020,NV1034 or PBS (controls) by single intratumoral (IT) injection (n 56–7/group). Tumor volumes were compared by Student’s t-test.

Results: In-vitro cytotoxicity assay demonstrated 96% and 82% tu-mor cell kill at 7 days after infection with NV1034 or NV1020 at an MOIof 1. ELISA demonstrated in-vitro (fig. 1) and in-vivo GM-CSF produc-tion in picogram quantities. Mean tumor volume (mm3) 6 SEM 2 weeksafter treatment was 710 6 203.3, 207.1 6 67.5 and 36.5 6 23 (p 5 0.01vs. control and 0.04 vs. IT NV1020) in the control, IT NV1020 and ITNV1034 groups (fig. 2).

Conclusions: NV1034 and NV1020 are cytotoxic to CT26 cells in-vitro.NV1034 has the ability to produce GM-CSF in-vitro and in-vivo. Effi-cacy of NV1034 for treating experimental colorectal carcinoma isgreater than that for NV1020 when delivered by IT injection. Combi-nation of oncolytic and immune therapy may be more effective intreating colorectal carcinoma.

Strong expression of sialyl-LEx defined by mAb AM-3IS, a prognostic marker after curative resection ofstage II and III colorectal carcinomasBenno Mann1, M.D., Patricia Grabowski2, M.D., Ulrich Mansmann3,Ph.D., Ernst-Otto Riecken2, M.D., Christoph Hanski2, Ph.D., Heinz-Johannes Buhr1, M.D. Departments of Surgery1, Gastroenterology2

and Medical Statistics3, UKBF, Freie Universität Berlin,Hindenburgdamm 30, 12200, Berlin, Germany Phone: 30 8445 2543,Fax: 30 8445 2740, e-mail: mann@ukbf.fu-berlin.de

Introduction: Sialyl-Lex antigen is overexpressed in 90% of colorectalcarcinomas. Binding of sialyl-Lex to E-selectin expressed on endothe-lial cells is supposed to be an initial step of metastatic tumor cellextravasation. The purpose of the present study was to analyze,whether sialyl-Lex expression increases with the progression of colo-rectal carcinoma stage and whether it has an independent prognosticimpact after surgical therapy.

Methods: 190 consecutive patients with complete 5-year follow-updata were analyzed. Sialyl-Lex was detected on paraffin-embeddedsections of the carcinomas by immunohistochemistry using mAbAM-3. A staining score resulting from staining frequency and intensitywas calculated and a cut-off point, separating patients with weak andstrong expression was determined using a matringale residual plot.The Jonckheere-Terpstra-test was used for the staining differences instage I–IV tumors. The predictive value of variables was analyzed withthe log-rank-test in univariate and with a Cox’s regression model inmultivariate analysis.

Results: The percentage of strongly expressing carcinomas increasedwith the progression of UICC stage (stage I 5 10%, II 5 48%, III 563% and IV 5 69%, p,0.0001). In multivariate analysis, strong sialyl-Lex expression showed to increase the relative risk of cancer relateddeath 3.8 fold (95%CI 1.8–7.9, see Table 1). The separate analyses ofall patients at stages II, III and IV revealed a significantly better 5-yearoverall-survival for patients with weakly expressing tumors after cura-

tive resection in stages II and III, but not in stage IV (stage II: 84 vs.55%, p 5 0.0013, III: 84 vs. 36%, p 5 0.00074, IV: 10 vs. 6%, p 50.27).

Table 1: Variables predicting overall-survival in 190 patients withcolorectal carcinoma

Univariate analysis Multivariate analysis

VariableChi-square p - value Variable r.r. 95% CI p - value

pT1-3 vs. pT4 11.5 5 0.0007 pT1-3 vs. T4 1.97 1.1–3.4 5 0.017pN0 vs. pN1 17.0 5 0.00004 pN0 vs. pN1 1.26 0.8–2.1 5 0.37M0 vs. M1 131.9 ,0.000001 M0 vs. M1 17.6 7.1–43.4 ,0.000001sLex 2 vs.1 9.0 5 0.0026 sLex 2 vs.1 3.81 1.8–7.9 5 0.00034

r.r. 5 relative risk; sLex 2 vs. 1 5 weak vs. strong immunohistochemical sialyl-Lex

expression; CI 5 confidence interval.

Conclusions: This data indicate that strong sialyl-LeX expression de-fined by mAb AM-3 is an independent unfavorable prognostic markerfor patients with colorectal carcinoma after curative resection in stagesII and III, whereas it does not influence the prognosis in patients whoalready developed distant metastases.

Human colon cancer cell growth is inhibited byadenoviral delivery of chimeric P27/P16Linda Martin, MD*, Sanjoy Dutta, MD*, Mary Matli, BS*, Salil Patel,PhD#, James McArthur, PhD#, Robert Warren, MD*. *UCSF, SanFrancisco, CA and #Cell Genesys, Foster City, CA. UCSF, Box 0790,533 Parnassus Ave., San Francisco, CA 94143-0790, USA. (415)476-0731

Introduction: Cyclin-dependent kinases play an important role in or-chestrating cell cycle progression. p16 and p27 are cyclin-dependentkinase inhibitors (CDKI) whose loss of function or down-regulation hasbeen implicated in tumor proliferation. Methylation of CpG islands onthe 59end of p16, leading to transcriptional inactivation, occurs fre-quently in colorectal tumors, and low levels of p27 are associated withpoor prognosis in colorectal cancer. Furthermore, reduced p27 ex-pression in metachronous colorectal metastases when compared toprimary tumors suggests that down-regulation of p27 may facilitatemetastatic spread. We have tested the ability of a chimeric p16/truncated p27 gene construct inserted in a replication-deficient adeno-viral vector to inhibit the growth of human colon cancer cells.

Methods: Two human colon cancer cell lines (HCT116 (p53 wt), DLD-1(mutant p53)) were chosen to test our hypothesis. Transduction effi-ciency was determined by flow cytometric analysis for each cell line 48hours after exposure to a replication-deficient adenoviral vector en-coding green fluorescent protein. Each cell line was then treated withreplication-deficient adenovirus encoding p27/p16 or p16, with mocktreatment and null adenovirus as controls. Cell number was deter-mined at the time of transduction (day 0) and 48 hours later. MTTassays were performed at days 0, 2, 4 and 6 to assess proliferation.Proliferative index and apoptosis were assessed by BrDU incorpora-tion and by TUNEL staining of DNA fragments, respectively.

Results: At a multiplicity of infection of 30, transduction efficacy inHCT116 and DLD-1 was greater than 99%. At 48 hours, treatment withp27/p16-Ad led to reductions in cell number of 74% for HCT116 and55% for DLD-1. In comparison, treatment with null-Ad resulted in adecrease in cell number (HCT116 by 47%, DLD-1 by 1%). After 6 days,MTT assays demonstrated further reductions in cell number: 78% inHCT, 64% in DLD-1 (p , 0.0001 for both by ANOVA). The decrease incell number was attributable to both a reduction in proliferative index(HCT116 by 99%; DLD-1 by 97%) and an increase in apoptotic index(19-fold for HCT116 and . 5-fold for DLD-1 compared to null-Adexposure). p16-Ad exposure resulted in cell number decreases inter-mediate to null-Ad and p27/p16-Ad exposure in both cell lines.

Conclusions: Adenoviral delivery of the CDKI chimera p27/p16 dra-matically decreased cell number and proliferative index in two humancolon cancer cell lines. These effects seem to be independent of p53status. Our findings suggest a new approach for gene therapy ofcolorectal cancer.

S69Vol. 191, No. 4S, October 2000 Surgical Forum Abstracts