Phase 1 is a very important clinical trial. Once a drug is past that it can go right into phase2 for any number of diseases. That is particularly important with PBT434 which could prove to have utility over several neurodegenerative disorders.
When you are looking at medical disorders, the latest or most relevant date to predict success could be from yesterday or five years ago, referenced by the number of citations credited in most scientific peer reviewed papers. The past price performance is not important in predicting future stock performance. Important right now to me is the science (IP and preclinical to support a credible clinical program) and cash in bank. The huge drop was caused by the result of a hopelessly underpowered P2 clinical trial.
ITM did some great work there and here is a repost just in case you missed:
[I've modified the conclusions and added some more information on mean stratification. The revised entire document was re-posted on google drive and the link is provided below. The revisions were made based upon suggestions made by others.
After the data were issued to the ASX I looked at it in a lot of detail. I noticed that there appeared to be at least one outlier in the placebo group that was an ApoE4 carrier that did not respond as expected. The following is a portion of the analysis that I did of the data presented in Masters slides. In my opinion this shows that potentially the problem with only one participant the IMAGINE placebo group reversed outcome of the placebo group and therefore made it look like the treatment group did not reach statistical significance. When this point is removed the primary endpoint of the trial is achieved. This is meant to be an additional line of thought for the data analysis.
The document in its entirety is posted here:
https://drive.google.com/file/d/0B7cR4RREFFpYdWVuaHhjVWhqWWM/edit?usp=sharing
Analysis
Based on the information presented in the July 17, 2014 announcement 42 participants were enrolled in the IMAGINE Trial. The baseline SUVR of the trial participants was reportedly elevated at an average of approximately 2.5 versus the 1.7 SUVR targeted for the trial. It was determined that 12 of the 15 participants in the placebo group carried the ApoE4 allele (3 non-carriers) and 21 of the 27 participants in the treatment group carried the ApoE4 allele (6 non carriers). Therefore, approximately 77% of the trial participants carried the ApoE4 allele.
The figure above displays the change in SUVR burden versus the baseline SUVR of the placebo group observed in the IMAGINE trial. The dotted line represents the envelope of results obtained in the AIBL study, a longitudinal study of a similar population of Alzheimer’s patients (Villemagne, et. al., 2011).
Visual inspection of this plot suggests that 3 ApoE4 carriers of the placebo group participants lie outside of the envelope (red circle). All exhibited improvement (or less) amyloid burden at end of the trial. A result that is highly unexpected since ApoE4 carriers generally worsen at quicker rates than non-carriers. The company did not report how many copies of the gene the carriers have or whether the carriers also have other ApoE alleles that may also affect disease progression. In some cases, the presence of ApoE alleles may counteract the effect of disease progression (i.e. ApoE2 allele). A similar group of “outliers” were also identified visually on the plot of hippocampal atrophy rates which skew the placebo group as shown on the box and whisker plot below. It is not known if these are the same participants that plot within the red circle on the above figure but they are ApoE4 carriers in both cases.
Successive plots were generated removing first all of the points identified as potential outliers on the table above. The slope of the best fit line for the placebo group reversed showing worsening of the placebo group over the course of the trial. Then the points were added back in one at a time until only the last outlier highest outlier was removed. Each time, the slope of the best fit line for the placebo group reversed showing worsening of the placebo group over the course of the trial which is expected. Since this data point represents an ApoE4 carrier, and you would not expect improvement in a gene carrier that did not receive treatment, the analysis carries more weight than a straight statistical analysis would in our opinion. The graph is shown below:
Additionally, the slope of this line approximates the slope of the line observed for the AIBL study data shown in Figure 3 and shown on the above Figure 11. Based upon this information (one outlier removed) the results achieve the primary endpoint of a statistically significant reduction in the levels of beta-amyloid plaques in the brains of prodromal/mild Alzheimer’s disease patients, as measured using PiB-PET SUVR.
Conclusions
• The comparison of the treatment and placebo group hippocampal atrophy rates for the treatment group was approximately half that of the placebo group as reported by the company.
• The spread of data points for the hippocampal atrophy presented for the placebo group unstratified for ApoE4 carrier status span the entire rates of the treatment group when up to 3 potential outliers are included. The removal of the outliers compresses the placebo group and causes less overlap with the treatment group that would result in higher level of statistical significance. It is possible that, depending on the cause of the outliers, removal of outliers from the treatment group would result in less overlap between the placebo and treatment group data and thereby result in higher statistical significance for treatment group.
• One data point for the SUVR change over the IMAGINE trial, the greatest outlier, when removed causes a reversal in slope of the placebo group stratified for ApoE4 carrier status, and indicates worsening over the course of the trial.
• Removal of the one outlier point for the SUVR data causes the slope of the placebo group regression line to closely approximate the slope of the AIBL study results for the MCI group based upon 38-month readouts.
• Upon removal of the single outlier, the primary endpoint of a statistically significant reduction in the levels of beta-amyloid plaques in the brains of prodromal/mild Alzheimer’s disease patients is achieved as measured using PiB-PET SUVR.
• It is possible that there is more than one outlier in the placebo group based upon the visual analysis.
• When 3 outliers are removed and the data are stratified the mean of the placebo group becomes -0.004 and with only one outlier removed its -0.05. A marked change on the mean with removal of the outliers.
• When the placebo means were stratified for baseline SUVR, the additional stratification of ApoE4 carriers did not have a marked effect. However, when the single outlier was removed in the SUVR >2.5 group, the mean was adjusted to -0.017 versus -0.068 for ApoE4 stratification alone. The mean was relatively unaffected by SUVR stratification when three outliers were removed from both SUVR stratification groups.
• With removal of the SUVR outlier, PBT2 reached a statistical significance for the IMAGINE data set.
This analysis, combined with the outliers observed in the hippocampal atrophy box and whisker plots raise several questions:
1. Is it possible that there is a misread on the SUVR measurements on this data point?
2. Or some other issue (mix up in medication)? If this is the case then it follows that one or more persons in the treatment group may have gotten the placebo in which case the treatment group statistics should improve with removal of that outlier. No treatment group analysis was done as part of this work to determine if there are outliers.
3. Do the placebo group participants carry any other ApoE alleles that might affect the disease progression and thereby the IMAGINE trial results? Villemagne et., al. (2011) found there was a gene dose-dependent relationship between the number of ε4 alleles and PiB SUVR increases, with ApoE ε4 heterozygotes and homozygotes showing a significantly higher PiB SUVR increase (0.05 and 0.09 vs 0.02 SUVR/year, respectively, p< 0.003) at follow-up when compared to those with no ε4 allele present (see Figure 3 below taken from Villemagne, et al, 2011). Is it possible that this played a part in the IMAGINE trial?
4. Are there other genes that affected the results of the placebo group (i.e. BDNFmet)? ]
The kicker is that as underpowered as it was the street probably would not have accepted a positive outcome.