This looks impressive. The 3 mg dose of enobosarm produced an almost 22 percent mean improvement in stair climb power compared to just a 4.7 percent mean improvement for the placebo patients.
The stair climb power data in the chart right above depicts the same benefit in a different way. Note the mean 2.21-watt improvement in stair climb power for placebo patients, which is much smaller than the mean 16.81-watt improvement for the enobosarm 3 mg patients.
But it's impossible to interpret these so-called improvements in muscle function without first knowing the relative strength (or weakness) of each patient group before the study began.
This is where it gets interesting, because the mean baseline stair climb power is not disclosed anywhere in the
Lancet Oncology paper. If there was an imbalance in mean stair climb power at baseline, it could explain away the observed enobosarm benefit.
Sure enough, that's exactly what happened in the phase IIb study. Mean baseline stair climb power -- at the start of the study -- was 46 watts for placebo patients compared to 80 watts for the 3 mg enobosarm patients.
These numbers can be calculated using the data available in the
Lancet Oncology paper.
In other words, the mean muscle functionality of enobosarm patients entering the study was almost twice as strong as the placebo patients. No wonder the enobosarm patients performed better on the stair climb power test over time -- they had a huge headstart.
Take another look at the chart above depicting stair climb power. Notice that in the placebo arm of the study, there's a big difference between the mean value (2.21 watts) and the median value (11.34 watts.) I don't want to get too bogged down into mathematical concepts but this means many of the patients in the placebo arm performed very poorly on the baseline stair climb power test. They're outliers.
By comparison, the mean stair climb power of patients in the 3 mg enobosarm arm (16.81 watts) is higher than the median value (12.84 watts.) In other words, many of the patients here performed really well on the stair climb power test. They're outliers in the other direction.
You should see now why GTx chose to analyze the phase IIb co-primary endpoints using mean values. Including the outliers in the analysis tilts the entire study in favor of enobosarm. The drug's benefit looks impressive on paper but it's a false signal.
Think about it this way: Cancer drug studies use median overall survival for a reason, not mean overall survival.
A larger, more well-balanced study -- presumably like the phase III studies which enroll a total of 600 patients -- will be a more level playing field where enobosarm's true benefit to cachexia cancer patients will be more modest, or disappear altogether, relative to placebo.
Strong evidence of enobosarm's lack of benefit for cancer cachexia patients can actually be seen in the phase IIb study by analyzing the stair climb power endpoint using median values. Calculating results this way minimizes the effect of the outlier patients.
The median improvement in stair climb power for placebo patients: 7.5 percent.
The median improvement in stair climb power for 3 mg enobosarm patients: 8.3 percent.
Almost no difference. Enobosarm is basically a placebo.
This is why the ongoing phase III studies of enobosarm in lung cancer cachexia are likely to fail.
GTx is expected to announced results later this quarter.
-- Reported by Adam Feuerstein in Boston.