Six of ten, or 60 percent, of the lung cancer patients randomized to placebo entered the study with stage IV disease -- the most aggressive and advanced stage of lung cancer. By comparison, five of 13 lung cancer patients, or 38 percent, randomized to the 3 mg enobosarm treatment had stage IV disease. The placebo lung cancer patients were a lot sicker and therefore had greatly reduced physical (muscle) function compared to the enobosarm patients. This difference easily explains the enobosarm benefit on the stair climb power test observed in the phase IIb study. Looking at the Lancet Oncology paper again, 39 percent of patients enrolled in the phase IIb study had colon cancer -- the second-largest group -- yet here the playing field was tilted against enobosarm. Of the colon cancer patients randomized to placebo, 53 percent had stage IV disease, compared to 75 percent with stage IV disease in the 3 mg enobosarm arm. Now we know why GTx claims enobosarm "works" in lung cancer but not in colon cancer. The actual data show this efficacy signal to be a mirage. Let's take a closer look at the data on the co-primary endpoints of the enobosarm phase IIb study. Here's the chart depicting change in total lean body mass, taken from the Lancet Oncology paper:
Not much to say here except increasing lean body mass is relatively easy to do, but won't get a cachexia drug approved by the FDA. Note the p values in the chart compare each arm against itself (baseline vs. Day 113.) It's also noted in the fine print that the increases in lean body mass between the 1 mg vs. placebo arms and 3 mg vs placebo arms were also statistically significant. (p values of 0.006 and 0.041, respectively.) This chart, again from the Lancet Oncology paper, depicts data for the second co-primary endpoint -- change in stair climb test at day 113: I'm going to focus discussion on stair climb power since that's the test being used in the phase III studies. Again, there was a statistically significant improvement in both the 1 mg and 3 mg enobosarm arms when each is compared against its own baseline. (p values of 0.0008 and 0.0006, respectively.) The change in the placebo arm was not statistically significant. But notice there is no mention of statistical significance or p values for a comparison between the enobosarm patients and those treated with placebo with respect to stair climb power. Had this inter-group difference been statistically significant, GTx would have mentioned it. Remember above when I said GTx manipulated the phase IIb data using mean values (instead of medians) to make enobosarm look more effective than it really is? Here's how that works. First, this is how the authors of the Lancet Oncology paper describe the observed muscle function benefit: Absolute changes in stair climb power represented a mean of 18.0% (SD 31.1) improvement compared with baseline for enobosarm 1 mg and 21.7% (65.7) for enobosarm 3 mg (vs placebo, 4.8% SD 23.2 ).