Sydney Welt¹², Gerd Ritter², Achim A. Jungbluth², Nancy E. Kemeny¹, Lloyd J. Old^2 
1Memorial Sloan-Kettering Cancer Center, New York, NY; ²Ludwig Institute for Cancer Research, New York, NY

The A33 Antigen of Colorectal Cancer: A Model for the Development of Therapeutic Antibodies

The A33 antigen is a 43kD cell-surface glycoprotein expressed in 95% of colorectal cancers and in colonic mucosa. Radioimmunolocalization studies quantitatively demonstrated high levels of A33 antibody targeting to colorectal cancer tissue. A humanized IgG1 version (huA33) of mouse A33 antibody activates immune effector function and in a phase I-II study huA33 was found to have modest anti-tumor activity. In earlier studies we observed a higher than expected proportion of patients responded to a regimen of carmustine, vincristine, 5-FU and streptozotocin (BOF-Strep) when administered after completing a huA33 treatment protocol. Reports that the combination of chemotherapy and antibody-directed immune effector function enhances tumor cell lysis date back to 1975 (Ohanian and others). Based on these reports and our clinical observation, a phase I study of escalating doses of huA33 + BOF-Strep chemotherapy was initiated. Sixteen patients with advanced colorectal cancer resistant to at least two chemotherapeutic regimens containing 5-FU and irinotecan were entered into four dose levels (5, 10, 25 and 40 mg/m2). Three patients requiring radiotherapy and one requiring surgery were removed early. Of the 12 patients evaluable for toxicity, one had grade 3 and one had grade 4 neutropenia and one had grade 3 thrombocytopenia. These adverse events were related to the chemotherapy. No significant toxicities specifically related to the combination of immunochemotherapy were identified. Of the 12 patients, seven developed anti-huA33 activity (HAHA+) by Biacore analysis. In two cases of high HAHA+ activity, huA33 treatment was discontinued early; one patient developed fever, chills, rash and pruritus. HAHA+ has been shown to correlate with accelerated antibody clearance and lack of tumor targeting. Of the 11 patients completing one treatment cycle, three had radiographic evidence of partial responses lasting for 7.5, 5.5 and 10.5+ mos and serum CEA declined from 7845.5 to 528, 130.4 to 20.1 and 273.8 to 0.9 ng/ml, respectively. One patient had a mixed response (4.5 mos with a CEA decline from 40.0 to 24.9 ng/ml) and one patient had stable disease (9 mos). HuA33 dose escalation is continuing. A new humanized IgG1 A33 antibody has been constructed to overcome the high frequency of HAHA induction (58.3%), which accelerates antibody serum clearance, blocks tumor targeting and consequently abrogates tumor response. Given these limitations with the current A33 antibody, the observation that major responses can be obtained in 3/11 patients who have progressed on 5-FU and irinotecan based therapies is encouraging and warrants further investigation of new A33 antibodies.