As Joan indicated, Genta is taking a different, more targeted approach this time around. After digging around in the old phase III study, Genta says it found a subset of melanoma patients with low levels of LDH, a blood enzyme, who appeared to live longer after treatment with Genasense. Conversely, patients with elevated levels of LDH were not helped at all by Genasense. It's widely believed that melanoma patients with elevated LDH levels tend to have a worse disease prognosis, including shorter survival times. Genta dug even deeper into these Genasense patients with lower LDH levels, eventually separating them into narrower subgroups defined by their specific LDH levels. What the company found was that patients with "low normal" LDH seemed to survive the longest after treatment with Genasense. If there's a scientific rationale for why this is true, I didn't hear Warrell discuss it Wednesday. But the company decided to conduct a second, phase III study of Genasense, this time enrolling only patients with "low normal" LDH levels. The results of that study -- dubbed Agenda -- should be released in the fourth quarter. If the Genasense trial is a success, Genta will displace Vanda Pharmaceuticals ( VNDA) as the most shocking, long shot of long shots winner in biotech this year. After all, Warrell is betting his company's survival on a pivotal study whose hypothesis was derived from a subgroup of a subgroup of a previously failed study. In other words, Genta data-mined a data-mined subset. Stunts like that are never supposed to work in biotech!