8570 Background: PHP (Chemosat, Delcath Systems Inc, New York, NY) allows repeated delivery of high dose chemotherapy to the liver while limiting unwanted systemic exposure by extracorporeal filtration of hepatic venous blood. A prospective randomized multicenter study compared melphalan PHP versus BAC in pts with unresectable hepatic metastases from ocular or cutaneous melanoma and showed a highly significant advantage for PHP in terms of the primary endpoint, hepatic PFS, and multiple secondary endpoints. We present an exploratory post-hoc analysis of pts assigned to BAC who crossed over to PHP after disease progression (BAC-to-PHP crossover) versus those who did not (BAC-only) versus PHP-randomized pts. Methods: 93 pts were randomized to PHP (n=44) or BAC (n=49). In the BAC group, crossover to PHP with melphalan was permitted after hepatic disease progression. Up to 6 PHP treatments with melphalan 3.0 mg/kg ideal body weight were given every 4–8 weeks. Results: In the BAC group, 28 of 49 pts crossed over to PHP. Updated efficacy findings (investigator) as of 31 March 2011 are shown in the table. Independent Review Committee results were similar to investigator-assessed efficacy. Safety in BAC-to-PHP crossover pts was similar to that seen in PHP-randomized pts. The most common toxicities in BAC-to-PHP crossover pts were hematological: peri-procedural thrombocytopenia (71%) and anemia (57%), and melphalan-related neutropenia (82%) and thrombocytopenia (79%). Conclusions: Efficacy of melphalan PHP in BAC-to-PHP crossover pts was consistent with that seen in PHP-randomized pts with hepatic metastases from melanoma. Crossover from BAC to PHP after hepatic disease progression led to a median 11-month survival advantage vs BAC alone. [Table: see text]