Gastrointestinal (GI) cancers account for the most common cancers. Despite recent advances in GI cancer treatments, the 5-year overall survival rate for these patients remain unacceptable, except for patients who are candidates for metastasis surgical resection. Strategies leading to a decrease of metastatic number and size will contribute to improve the probability to undergo a curative surgical procedure.

Haploidentical Natural Killer (NK) cells can persist and expand in vivo following adoptive transfer and may have a role in the treatment of selected malignancies, since the failure to recognize the appropriate KIR ligand on a mismatched tumor cell can trigger NK cell elimination of that target cell. NK also express an activating Fc receptor that mediates antibody-dependent cellular cytotoxicity (ADCC) and production of immune modulatory cytokines in response to antibody-coated targets. Cetuximab, an IgG1 chimeric monoclonal antibody against colorectal cancers that expressed EGFR (epidermal growth factor receptor), improves overall survival and progression-free survival and preserves quality-of-life measures in patients with colorectal cancer in whom other treatments have failed.

In an attempt to improve the outcome in GI cancers, we will conduct a phase I/II clinical trial assessing NK cell based immunotherapy. Patients with liver metastases related to a EGFR+ GI cancer, previously treated by a standard chemotherapy that did not achieve a complete response or a curative resection of residual metastases will be included in this phase I/II trial supported by the French National Institute of Cancer (INCA, PHRC 2005). This phase I/II study will involve 22 patients. The main objective of this study will be to demonstrate the safety of NK hepatic intraarterial infusion in association with cetuximab. Secondary objectives will include the assessment of the clinical efficacity of this strategy.

Patients will be treated with a conditioning chemotherapy including 60 mg/kg intravenous cyclophosphamide, 25 mg/m2 intravenous fludarabine for 5 consecutive days and cetuximab. A lymphapheresis from an haploidentical related donor will be performed and T cells will be depleted . Allogenic NK cells will then be adoptively transferred by hepatic intraarterial infusion according to a dose escalation protocol (three doses with at least three patients per cohort)to define the dose-limiting toxicity (DLT). Then 10 more patients will receive the recommended NK cell dose.

Dose escalation protocol : 3.10^6 ; 8.10^6 and 12.10^6 cells/kg of recipient body weight.

Inclusion Criteria:

• metastatic gastrointestinal cancer (colorectal, pancreas, small intestine, bile ducts), gastrointestinal stromal tumor or digestive neuroendocrine tumor.
• presence of liver metastases
• Age > 18 and < 65 (male and female)
• ECOG performances status 0 or 1
• EGFR expression confirmed by immunohistochemistry
• allogenic donor with one or more KIR/MHC class I mismatch
• absence of alternative treatment available
• evidence of progressive disease
• written informed consent

Exclusion Criteria:

• malignant secondary disease dated backed < 2 years (exception : in situ carcinoma of the cervix uteri, adequately treated skin basal cell carcinoma)
• hypersensitivity against one of the treatment of this study
• history of cardiac or respiratory failure
• history of auto-immune disease
• renal or hepatic failure
• pregnancy or lactation
• patient with any medical or psychiatric condition or disease which would makes the patient inappropriate for entry into this study.
• EBV serology negative in recipient and positive in donor
• bilirubin greater than 1,5 times upper limit of normal
• ASAT and/or alkaline phosphatase greater than 5 times upper limit of normal in presence of liver metastases.