The purpose of this study is: evaluate the safety and activity of administering arsenic trioxide (Trisenox) in the treatment of unresectable or metastatic primary liver cancer, to evaluate the qualitative and quantitative toxicities of this treatment, and to measure the response to treatment and the patterns of failure and survival. The primary response measurements will be the achievement of an objective tumor response, response duration and progression-free survival

Hepatocellular carcinoma (HCC) is one of the most frequently diagnosed malignancies in some regions of the world, particularly Africa and the Asian portion of the Pacific rim. However, it is an uncommon malignancy in the United States, with less than 14,000 cases diagnosed annually. Malignancies of the gallbladder and biliary tree are also uncommon. As with most solid tumors, unless either neoplasm can be resected completely, the prognosis is grim. A variety of agents have been utilized in the therapy of HCC and cholangiocarcinoma, both as single agents and in combination regimens. However, despite response rates which exceed 50% when hepatic arterial therapy is utilized for HCC, response rates for cholangiocarcinoma and for systemic treatment of HCC are considerably lower. Long term survival remains uncommon. For this reason, new therapeutic approaches must be evaluated.

Trisenox (arsenic trioxide) is a newly-approved pharmaceutical grade arsenic compound antineoplastic agent which has demonstrated clear activity in anthracycline- and all-trans retinoic acid-resistant acute promyelocytic leukemia. Trisenox also has US Compendium listing for acute leukemia, chronic myelogenous and lymphocytic leukemias, myelodysplasia, multiple myeloma, and hepatocellular carcinoma (HCC). Similar to the taxanes and vinca alkaloids, Trisenox appears to interfere with microtubule function, which triggers cell differentiation, and induces programmed death, or apoptosis. The mechanism of this is unclear, but appears to involve activation of caspases and the down-regulation of the BCL-2 oncogene. Trisenox also interfere with the function of NF-kappaB, leading to inhibition of cellular proliferative signals, apoptosis, and inhibition of tumor angiogenesis.

Trisenox was shown to be effective in a pivotal trial consisting of 40 subjects, ages 5 to 72, at a nine-institution multi-center trial led by Memorial Sloan-Kettering Cancer Center. With Trisenox, 70% of the subjects achieved a complete remission, and 79% of the complete responders achieved a molecular remission as measured by reverse transcriptase-polymerase chain reaction (RT-PCR). Complete remission was achieved on average within two months after initiation of Trisenox. Sixty-eight percent of subjects who achieved complete remission were still alive and 58% were disease free, at a median follow-up time of 16 months.

To date, an optimal dose and schedule of Trisenox has not yet been defined. This agent has been administered on daily, twice a week, and weekly doses. As there is no evidence that one regimen is clearly superior to any other, for the sake of convenience, participants on this trial will receive Trisenox on the weekly schedule. Administration of chemotherapy on a weekly schedule is commonly utilized across a broad spectrum of regimens and tumor types. On this trial, subjects will receive a dose of intravenous Trisenox 0.35 mg/kg on days 1, 8, 15, and 22 over two hours. Each cycle will be 28 days in length. Subjects will receive two treatment cycles (8 weeks) and then undergo objective radiographic tumor assessments. On this study, subjects may receive up to a maximum of 12 such treatment cycles. As the outlook for subjects with advanced primary liver cancer is so poor, new regimens and treatment strategies must be evaluated.

Inclusion Criteria:

• Signed Informed Consent
• ECOG/Zubrod/SWOG Performance Status < 2
• Life expectancy > 12 weeks
• Male or female' age >18 years Subjects of childbearing potential must be using an effective means of contraception.
• Histologic diagnosis of HCC or cholangiocarcinoma that is locally advanced but cannot be treated adequately by radiotherapy or surgery (more than one lesion, portal vein involvement, or poor hepatic reserve); or metastatic disease or an AFB > 20 w/CT scan consistent with HCC
• Renal function: Serum creatinine < 2.0 mg/dl
• Serum calcium < 12 mg/dl
• Serum electrolytes, including magnesium and potassium, within normal limits

Exclusion Criteria:

• Pregnant or lactating females
• Myocardial infarction or ischemia within the 6 months before Cycle 0' Day 0
• Uncontrolled' clinically significant dysrhythmia
• Known left ventricular Ejection Fraction below the normal limit for UTMB
• History of prior malignancy within the prior five years, with the exception of non-melanoma carcinomas of the skin, and carcinoma in situ of the cervix
• More than one prior chemotherapy regimens for liver cancer (subjects who are receiving antineoplastic agents for non-malignant conditions, such as methotrexate for rheumatoid arthritis, must be off such therapy for at least four weeks prior to receiving the first dose of protocol therapy, and may not receive such therapy while participating in this protocol)
• Receiving any other chemotherapy or cytokine therapy
• Subjects receiving radiation therapy (Trisenox will be held during the administration of palliative or emergent radiotherapy)
• Subjects who have received radiofrequency ablation or hepatic arterial embolization within the past four weeks (patients who have received prior RFA or HAE are otherwise eligible)
• Prior radiotherapy to an indicator lesion unless there is objective evidence of tumor growth in that lesion
• Uncontrolled metastatic disease of the central nervous system
• Prior and on-going Grade 2-4 peripheral neuropathy, as measured by NCI Common Toxicity Criteria version 3.0
• Radiotherapy within the 2 weeks before Cycle 1' Day 1
• Surgery within the 2 weeks before Cycle 1' Day 1
• Any co morbid condition that' in the view of the attending physician' renders the subject at high risk from treatment complications