(Total Views: 207)
Posted On: 09/12/2017 11:31:23 AM
Post# of 72436
I thought top line and final results were the same; wrong. Below is something I found that sums up what each means and FYI the drug they were talking about is NFLD (which is immaterial to this discussion). Thought maybe one or two here might have had the same misconception as me re: top line vs final results
Re: Definition of Topline Results
"the top line is the analysis of the statistics that show whether the end points were met or not. If a finding of Non Inferiority is at least demonstrated, we live to fight another day. If a finding of Statisical Superiority is demonstrated, a whole new world is available. If neither is demonstrated, NFLD is history."
Correct but does not tell the whole story. Top line results usually refers to a simple raw calculation based on the kaplan-meir curves to calculate the p-values and the hazard ratios assuming that the data is perfectly representative of the effect we are trying to find. If topline results are favorable and showing statistical significance, then NFLD should fly into the stratosphere, especially if statistical significance is demonstrated for superiority.
However,the story does not really end there because there are always imbalances in a trial and there are always secondary endpoints and subgroups that can be analyzed. A complete result analysis would entail looking at all the subgroups (age, health status before, etc...) as well as the multivariate that create an imbalance, and making a cox regression analysis that takes all these factors into account and produces results which, hopefully, identify strictly the effect of Polyheme Vs other uncontrolled factors.
Typically, once data-lock has occurred, top-line results should be available within a few days from that. As for the complete analysis, that could take a few months to sort out. Timing wise for Polyheme, I would expect top-line results to be available in early fall on the basis that we need to wait the required 30 days before starting the clean-up for our data lock. Then you assume 30 days for data lock, so topline results could be released as early as October. As for the full analysis, I would expect that to occur sometime in 1Q 07 with hopefully a BLA submission sometime in 2Q 07. If everything is on track, and as expected we get priority review, then we could have an FDA decision by the end of 07.
Note that even if topline results do not demonstrate statistical significance for superiority and non-inferiority, the NFLD stock will suffer but will not be history until the full analysis is completed. Often (as was the case with DNDN) topline results may fail but a cox-regression analysis with the right multivariate may tilt the balance in favor of the effect and produce statistically significant results. Whether the FDA will consider that in a favorable light will depend on many factors including the originally agreed upon SPA and the overall strentgh of the data, as well as safety concerns.
Personally, I am betting that non-inferiority will be statistically significant of the bat and that superiority will also show a strong effect, which may or may not be statisticall significant for topline results but will probably be statistically significant when an appropriate cox regression anlysis is performed. I also believe that Polyheme will be approved by the end of 07 but I am not sure at this point how wide the label will be. Hopefully we will get guidance on all these issues at the Aug 8 cc but I am getting more and more optimistic with time and I plan to add to my substantial position today...
Return To Message Return to Message
Re: Definition of Topline Results
"the top line is the analysis of the statistics that show whether the end points were met or not. If a finding of Non Inferiority is at least demonstrated, we live to fight another day. If a finding of Statisical Superiority is demonstrated, a whole new world is available. If neither is demonstrated, NFLD is history."
Correct but does not tell the whole story. Top line results usually refers to a simple raw calculation based on the kaplan-meir curves to calculate the p-values and the hazard ratios assuming that the data is perfectly representative of the effect we are trying to find. If topline results are favorable and showing statistical significance, then NFLD should fly into the stratosphere, especially if statistical significance is demonstrated for superiority.
However,the story does not really end there because there are always imbalances in a trial and there are always secondary endpoints and subgroups that can be analyzed. A complete result analysis would entail looking at all the subgroups (age, health status before, etc...) as well as the multivariate that create an imbalance, and making a cox regression analysis that takes all these factors into account and produces results which, hopefully, identify strictly the effect of Polyheme Vs other uncontrolled factors.
Typically, once data-lock has occurred, top-line results should be available within a few days from that. As for the complete analysis, that could take a few months to sort out. Timing wise for Polyheme, I would expect top-line results to be available in early fall on the basis that we need to wait the required 30 days before starting the clean-up for our data lock. Then you assume 30 days for data lock, so topline results could be released as early as October. As for the full analysis, I would expect that to occur sometime in 1Q 07 with hopefully a BLA submission sometime in 2Q 07. If everything is on track, and as expected we get priority review, then we could have an FDA decision by the end of 07.
Note that even if topline results do not demonstrate statistical significance for superiority and non-inferiority, the NFLD stock will suffer but will not be history until the full analysis is completed. Often (as was the case with DNDN) topline results may fail but a cox-regression analysis with the right multivariate may tilt the balance in favor of the effect and produce statistically significant results. Whether the FDA will consider that in a favorable light will depend on many factors including the originally agreed upon SPA and the overall strentgh of the data, as well as safety concerns.
Personally, I am betting that non-inferiority will be statistically significant of the bat and that superiority will also show a strong effect, which may or may not be statisticall significant for topline results but will probably be statistically significant when an appropriate cox regression anlysis is performed. I also believe that Polyheme will be approved by the end of 07 but I am not sure at this point how wide the label will be. Hopefully we will get guidance on all these issues at the Aug 8 cc but I am getting more and more optimistic with time and I plan to add to my substantial position today...
Return To Message Return to Message
![](/m/images/thumb-up.png)
![](/m/images/thumb-down.png)