This is how it is likely to play out. Any comments are welcome.
In earlier data, TNF+Standard-of-care has shown essentially the same survival rate as Standard-of-care for patients that were either treated recently (patients enrolled < 11 months ago) or treated long ago (patients enrolled > 24 months ago). TNF+SOC shows higher survival chances than SOC only in the middle (patients enrolled 11 months – 24 months ago), so we can try to estimate (with several assumptions) on how much better TNF+SOC might be than SOC in that time period, given the fact that 92 event mark was reached in end-Aug.
Using known survival curve for SOC, and looking backwards from end of August 2008 (when 92 deaths were reached),
• Roughly 58 patients had been enrolled < 8 months ago. Assuming 2% trial dropout and 10% death rate for both TNF+SOC and SOC (as shown in earlier trial data), this would have likely led to 5.7 deaths
• Roughly 17 patients had been enrolled 8-11 months ago. Assuming 3% trial dropout and 30%-40% death rate for both TNF+SOC and SOC (as shown in earlier trial data), this would lead to 4.9 likely deaths
• Roughly 43 patients had been enrolled >24 months ago. Assuming 10% trial dropout and 90% death rate for both TNF+SOC and SOC would lead to 36.8 likely deaths
So total dead amongst patients treated recently or long ago = 47.4
Roughly 71 patients were enrolled in middle between 11-24 months ago. Assuming 5% trial dropout, 22 of participating patients would be on SOC and 45 on TNF+SOC. If we assume SOC patients had 75%-80% death rate (very realistic given earlier data), we would need to have likely achieved 55-65% death rate amongst TNF+SOC patients so that all deaths add up to 92.
Based on plugging several iterations of these several assumptions, my best guess is that upcoming data in next few days will show that TNF+SOC achieves 40-45% survival (versus 20-25% survival on SOC alone) in patients that were enrolled between 11-24 months ago, and no statistical advantage either in patients early into treatment (enrolled <11 months ago) or late into treatment (enrolled > 24 months ago). Even though interim data had earlier shown a 8+ month improvement in median survival, the upcoming data will show < 4 month improvement in median survival so GenVec would likely try to use 18 month or 24 month survival rate as a key efficacy end point.
When we discuss patients in the period under 11 months only doing as well as the SOC because the median is the same, we're forgetting something major. The median only consideres those who've passed on.
We still do not know how many passed from the TNFerade group, which constituted 2/3rd the patients, and the SOC group, the other 1/3rd.
In the 92nd event, if these numbers are nearly even, say over 40 deaths from the SOC and 50 or less from the TNFerade, that would put a lot more people from the TNFerade group past the median date who hadn't raised the median because they haven't died yet. Only deaths contribute to where the median is.
When we see the full data at ASCO I believe it will be apparent that more than half of the patients on TNFerade are well beyond 9 months, but of those who passed on, the middle patient to pass was at 9 months.
Let's look at an extreme, say 50 patients from the SOC arm passed, the 25th passed at 9 months. That means only 42 patients from the TNFerade group passed, the 21st being the median, also at nine months. Nothing is said about the fact that of the 50 who passed from the SOC there were only perhaps 65 in the trials, whereas of the 42 who passed on TNFerade there were 130 in the trials. This is not what I expect, the numbers were intentionally skewed to make my point, but even if you reverse them, which is possible, it would then be 42 of 65 and 50 of 130, those numbers may be about right. Even if it were 35 of 65 and 57 of 130 the difference is clear.
I don't know what the numbers are, I think I was a little high with a total of 195 enrolled at the 92nd event, this was only for illustration, but the point is, this is all information we've not seen.
I hate to point this out but there is no "NEW" data to be revealed. We will see the complete analysis from the 92 event data, but there will not be any additional update. All of those results are sealed until the 180+ event.
As I read this post he's talking about ebents after the 92nd event, and we don't get to see that data untiil the fall.
I'm very hopeful that the additional analysis of the 92 event data will be enlightening but the topline has already been revealed.
i had forgotten all about balhara's post
thanks for bringing it back up//
he was very prophetic..
if the data at 184 only mimics.. the 92 event data.. i dont think we are stat sig.. but still getting very close..
I am pretty good at sniffing out BS and I dont smell it here. I think Balhara took the known facts and made the best equation possible. My only concern is I think the dropout rate is probably alot higher than 2%. How would that affect the numbers?
If what was said on the last call was true and the "robustness" of the data had already occurred, you aren't even close.
Don't you remember that the trial was altered to reflect overall survival and not median survival or 24 month survival? This was a huge victory for GNVC since they aren't reporting data at specific time intervals, but specific event intervals.
GNVC does not need to "use the 18 or 24 month survival rates" because the data goes as far back into history as the first patient dosing. In other words, it doesn't matter whether there is a material different in benefit from the the x<11 month patients since you'll see a markedly improved benefit from the TNFerade + SOC arm.
Lastly, please describe why you choose to discount patients enrolled + dosed x>24 months ago as offering no material survival benefit to the top-line numbers. In the case of Colorado man that was "cured", if he is still alive today, a data point could be 24+ months of disease-free survival. Granted, one or two data points won't significantly impact the overall survival, but it does push it slightly to the upside. Every little bit counts in the biotech game.
Just pointing out...yes the man from Colorado did extremely well and likely is past 24 month survival..there will be some on SOC alone the also will have a very positive result.
Both a very small percentage...I do believe TNFerade gives you the best shot at long term survival.
I would not assume that a less than 4 month survival at the 92 look would be bad news. I would not assume much of anything to be honest since we all can tweak the numbers adding a month or two or losing a month.
The presentation of the data will be just as important imho....GNVC management must relay what the data really means and that a positive tract continues.
They also need to solve the cash issue...that is really killing us like a sword at our necks. Hopefully we get a decent partnership deal announced in a timely fashion post data release. This should give us the best indication of what big pharma thinks of our data.
After multiple attempts to dissect your post I must admit that I have no idea how to follow your train of thought to its conclusion.
By the way the primary endpoint is overall survival.
Well...I had to read it 3 time...but I think I kinda get what balhara is trying to say.
Basically he is saying that the differance in survival for TNF+SOC patients doesnt show up until 11 months has past and that after 24 months practically all of those patients that have benefited, have died.
He further extrapolates the the look at 92 events will show a less than 4 month benefit.
Is assumption is that SOC alone hasnt changed.
It may or may not...I see his guess as no better or no worse than anyone elses.
The assumption of a less than 4 month benefit is no great revelation imho either since the trial is powered not to pick up this difference until 330 patients enrolled.
We will know soon.
Both Ph II dosage study and PhIII earlier look show stabilization in survival curve (almost flat horizontal line) after certain number of people die. Before then, survival curves between TNF+SOC and SOC are near identical.
From interim analysis on PhIII data, it seems stabilization is reached after 30% people die. This is impossible to hold out in upcoming data - given my previous calculation, stabilization seems to have been reached after 60-65% people had died otherwise you wont have had 92 events by end of Aug. Once you cross 50%, even if you stabilize the survival rate after that point, it does not change your medial survival. So median survival will show not very significant improvement.