A copy of the Spring study protocol dated May 15, 2003 (available on the PLOS One web site) describes the primary endpoint as assessing the difference in pain reduction between 4 ml Ampio and 4 ml placebo and 10 ml Ampion and 10 ml placebo. There is no mention of pooling data on doses to assess the primary endpoint.
But there's also a separate statistical analysis plan dated Aug. 8, 2013 (also available via PLOS One) which includes pooled Ampion dose data as a part of the primary endpoint analysis.
Ampio completed the Spring study in June 2013 and announced results on Aug. 14. Does this mean the statistical analysis plan was implemented after the study was completed and days before the results were announced? Strange.
From the PLOS-One paper, here's what the pain reduction endpoint on the pooled Ampion doses looks like graphically:
Notice there's no statistical difference between Ampion and placebo at 10 weeks. Only a sharp uptick in the pain scores for placebo patients between weeks 10 and 12 saves the day for Ampio and turns the study positive. The ability of Ampio to reproduce this result in the second phase III study -- not yet completed -- should be a concern.
The PLOS-One paper does break out response by Ampion dose. There doesn't appear to be a dose response, meaning the 4 ml Ampion works better than the 10 ml Ampion. No explanation for why that may be so. The paper also does not include statistics for the individual doses, so it's not clear if 4 ml Ampion, on its own, is statistically superior to placebo.
Ampio claimed both doses of Ampion on their own were statistically significant in a Nov. 4, 2013 press release. But if that's true, why not mention in the published paper?
The baseline characteristics of patients enrolled in the Ampion study resemble those enrolled in the Sanofi (Genzyme) study of Synvisc which formed the basis for that drug's approval. The Ampion study also enrolled a group of patients with more severe osteoarthritis. Last November, I wondered about the discontinuation rate in the study but the PLOS-One paper shows dropouts were minimal. All points to Ampio.
Before Ampio conducted the Spring study, a smaller study of Ampion was run in Australia. That study, using the same pain reduction endpoints, failed, according to the Spring study protocol:
Efficacy: Overall pain (as assessed by the pain numerical rating score) and WOMAC scores were reduced post-dose for each of the treatment groups for the duration of the study (p < 0.05), except placebo at Day 84. There were no statistically significant differences between changes from baseline in pain NRS or WOMAC scores for patients who received Ampion (with betamethasone/lignocaine or with betamethasone/saline) compared with patients who did not receive Ampion (saline with betamethasone/lignocaine) and between patients who received Ampion compared with patients who received saline.
Why worry about an old study conducted in Australia? Because in two press releases dated Jan. 31, 2012 and April 30, 2012 Ampio said results from the Australia study of Ampion were positive.
Ampio justifies calling the Australian study a success by using patients treated with Ampion as their own control i.e.. comparing pain scores at baseline and later in the study. The real analysis compared Ampion to placebo and found no difference.
Spin baby, spin!
We wait for the results from the next Ampion phase III study.
-- Reported by Adam Feuerstein.
Follow Adam Feuerstein on Twitter.