Biotech Values

[Democritus_of_Abdera]

Re: PSDV Iluvien Data conference call...

Dew, The PSDV conference call (of Dec 23) was focused on data analysis... In my opinion, the presentation was not very helpful, a much better presentation of the data and analysis can be found in Alimera’s Prospectus starting at pg 70 (see http://www.sec.gov/Archives/edgar/data/1267602/000095012309072887/g20643a1sv1za.htm ).

In the CC, however, Paul Ashton did address the statistical issue that you and Turtlepower have been discussing on this board (#msg-44823244 etc). To quote the CC starting at about minute 15...

Lets look at the trial results using the modified all randomized treatment group. This is the same number of patients as the previous analysis, the ART method, but with more of the two year data imputed. This is from patients who had protocol violations such as intravitreal kenalog or avastin, etc. By this method of analysis, in trial A, 21.4% of patients receiving the high dose experienced an improvement in visual acuity of 15 or more letters, 22.6% of the patients in the low dose gained 15 or more letters compared to 12.6% of the patients in the sham. These results compute to a p value of 0.026 for the high dose, 0.057 for the low dose. Now in all of these trials we have been using the Hochberg-Bonferroni statistical test. This test requires that both treatments should have a p value of less than 0.049. And if one dose misses that test, the other p value has to be less than 0.0245. So in this analysis we just missed statistical significance in trial A.

In trial B 29.3% of patients who were treated with Iluvien high dose and 29.7% of patients who were treated with Iluvien low dose experienced a visual acuity improvement of more than 15 letters. This is compared to 13.3% of patients in trial B who received the sham insertion. These results compute to a p value of 0.005 for the high dose and 0.004 for the low dose. So here we met statistical significance for both doses.

The combined data from both A and B by modified ART showed 26.7% patients in the high dose, 26.1% of patients in the low dose gained 15 or more letters compared to 13% of controls and this was highly statistically with a p of less than 0.001. That’s the pooled data using that modified ART. As I said earlier this is perhaps the least conservative method and potentially the most flattering of all the analyses.