Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
Please provide that quote or link. I listened to her Dec. 15 presentation and there was no mention of a specific time that enrollment completed. She may have said "we recently completed enrollment", but by no means does that mean it was days away from her presentation.
On the FDA website, it clearly states that as soon as the agency makes that teleconference call, the hold is in place, even partial holds.
All any one here has provided is heresay, or "I recall", not one link backing up the fact that enrollment was completed in Oct or November. I have provided NUMEROUS quotes showing that Screening starts at the Pre-Leukapheresis and is not complete until the Baseline Visit. Prior to my posts here, it sounds like everyone here thought screening started at the "evaluation stage" at surgery, which we all now know is not the case.
Also, numerous links show that the baseline visit is 4 - 6 weeks after Leukapheresis, again, not 3 months.
Also, would love to see the link that states that the FDA allowed patients that were still in the screening process to complete screening. From their website, it clearly states the hold or partial hold occurs as soon as they make that first teleconference call.
RKmatters, it states CLEARLY in the form that you provided a link to that Screening starts at the Pre-Leukapheresis visit. It is pretty Black and White. And this stage is 4 - 6 weeks before the baseline visit.
It also states part of the screening takes place at the baseline visit. SO unless you can provide some evidence that the FDA allowed NWBO to complete their screening for patients not yet at the Baseline visit, I am not sure how you can make the case. I can use the Pre-Leukapheresis stage to be even more conservative, however, that still brings us to end of Sept, beginning of October, and provides >15 month PFS.
Screening phase is NOT Evaluation Phase.
Read your link, its states on page 3:
Screening Procedures (Pre-Leukapheresis Visit)
Not at surgery. And with the 30 day FDA notice given over teleconference and then officially given within those 30 days, the screening halt could have been initiated as early as July 2015.
All I am saying is that Sept. 2015 to late Jan 2017 ~ 16 months is a conservative, and realistic minimum amount for the 83 patients that had no tumor progression.
Moreover, FDA states Clinical Hold would have been given 30 days prior, to allow NWBO to provide feedback, answer questions and attempt to possibly lift clinical hold (or in this case, partial clinical hold) within the firs 30 days. This news is given on a teleconference, NOT written, and can take up to 30 days for a written notice. Meaning the screening halt (or partial clinical hold) could have started sometime in late July 2015.
https://www.fda.gov/RegulatoryInformation/Guidances/ucm127537.htm
And you just made my point, surgery is "Pre-screening" NOT Screening. Screening starts at the Pre-Leukapheresis stage which is 4 - 6 weeks from t = 0, not 3 months. And after reading about this 30 days, I am thinking that all vaccines were provided well within Sept 2015.
So unless you have a link, RK isnt correct.....
Press Release States "Over 300 recruited", not just over 300 with 31 more to be enrolled....
Look up the Consent Form that specifically shows the schedule and process.
The screening halt occurred mid August 2015. As per the consent form, Pre-screening occurs at surgery. And yes, it states tumor tissue is taken. However, it isnt until Pre-leukapheresis, that screening begins (AND IT STATES IT IN BLACK AND WHITE), and states that pre-screening occurs prior to this.
I would assume that at the Leukapheresis stage that the decision is made to make the vaccine. Why would NWBO make a vaccine prior to taking any blood work or tests to determine if the patient is a viable candidate. That would be extremely costly.
So if that is the case, then anyone that had surgery in late July would have been able to be part of the trial, as leukapheresis is 1 - 2 weeks after surgery. Therefore that would only leave 4 - 6 weeks before randomization and the first vaccine. Plenty of time to make the vaccine and proving that the last t = 0 would have been in Sept. 2015 not Oct. 2015. However, if the screening was halted at the baseline (which is when additional screening tests are taken), then the latest t = 0 would have been sometime in late August 2016 to early Sept. 2016, which would have meant these patients had surgery in late June.
Please send the link or data that shows 31 more patients were enrolled in the trial after August 22, 2015? Actually, what you are reading is The Street's release stating "just over 300 patients", however, if you look at NWBO's ACTUAL Press Release, on their website, on August 16, 2015, they state:
"Over 300 patients have been recruited for the trial. "
That tells me that as of Aug. 16th, 2015, over 300 patients could be 331, not 301 with 31 just finishing surgery. So again, please show me where there was only 300 patients Aug 16, 2015 with 31 more enrolling over the next several months?
States a Pre-Screening stage and a Screening Stage in the Consent Form.
Pre-Screening occurs at surgery, and doesn't involve any medical tests. The screening process would require some type of blood or biological tests at minimum.
Screening begins at Leukapheresis stage, and is not completed until after the Baseline Visit.
Therefore once screening was halted, even if a tumor sample was provided or taken, that patient wouldn't have started the screening process yet.
No matter, we are talking semantics here, the difference is ~ 4 - 6 weeks if screening is at leukapheresis or 8 - 10 weeks of at surgery. It doesn't change the results very much, you still get >60% of patients > = 20 months OS. AND, remember that only a small number of the 83 PFS free patients, and 98 patients alive at 18 or 20 months, would actually have had surgeries prior to August, more likely June/July and prior. Meaning their actual PFS and OS results are several months greater than if August.
Screening entry occurs at the Pre-Leukapheresis stage NOT at Surgery:
QUOTE:
"After your Pre-Leukapheresis Visit, if you meet the screening entry criteria, and have been declared eligible for the DCVax-L clinical trial."
So the screening process doesn't even start until pre-Leukapheresis, which is 7 - 10 days after surgery, and prior to 4 - 6 weeks of chemotherapy and baseline visit.
So even if you wanted to push the enrollment 2 months from mid August (August 15 I believe it was changed in Clinical Trials, and even then, the halt would have occurred prior to it changing online), you could then state a minimum of 15 months PFS. AND AGAIN, THAT IS ASSUMING ALL 83 PATIENTS STARTED SCREENING IN AUGUST 2015, AND/OR, all 98 patients started the screening process in August 2015 to give them 17 - 18 months OS.
AND...DCVax-L P3 OS results show a minimum of 54% of all 331 patients living ~20 months (Assuming 233 OS endpoints occur in May, or 19 months as of today). This is assuming Optune's control group statistic of ~ 34% SOC patients reach ~20 months. If we assume ICT P2 results, wherein median OS for SOC patients was ~20 months, then there will be ~65% of the 331 patients (98 + 50% of 233 = 214 patients living >=20 months. This would also mean that the overall median value for both placebo and vaccinated groups, meaning the entire 331 patient population, would have a median OS > 20 months.
So here is the kicker. Two scenarios:
1) Placebo group performs worse than the SOC; 16 - 20 months OS. If that is the case, then the majority of the 98 patients are DCVax-L in order to have ~30% of 331 patients with OS > 20 months and PFS > 16 months.
2) If Placebo group performs better than expected, meaning median OS of 110 patients is ~20 months. THEN THAT MATHMATICALLY MEANS THAT THE OVERALL MEDIAN VALUE WILL BE SIGNIFICANTLY GREATER THAN 20 MONTHS. Let's look at the overall median value of Optune's P3 and ICT-107's. Optune's would have been between 16 and 20 months, so < 20 months. ICT-107 saw little difference with SOC ~ 20 months and ICT-107 at ~21 months, so lets say ~20 months for the overall median value.
WITH 98 patients showing an OS >=20 months, and using SOC statistics, between 55% - 65% of the total 331 patients will have SOC > 20 months. Meaning the median value will be well over 20 months.
Why PFS is 17 months and not 15 months -
In the NWBO -DCVax-L P3 Consent to Treat Form, it specifies that the Screening Process occurs at several times during the initial visits. THE FINAL STEPS TO COMPLETE SCREENING TAKES PLACE AT THE BASELINE VISIT. The quoted excerpt below is from the P3 Consent Form:
"Tests done at the baseline visit complete Screening examination. The patient must meet all eligibility criteria verified at screening to be enrolled in the investigational arm of the study."
The baseline visit occurs at week -1. Literally, 1 week and 1 day before enrollment and randomization. Therefore, if screenings were halted in mid August (and most likely done prior to being updated on the Clinical Trials website, no new patients would have been enrolled after Sept 1, 2015. Therefore using Sept 1, 2015 to Feb 1, 2017, we get 17 months. Even if you take away 1 or 2 months, and get 15 months or 16 months, the fact that >35% of the P3 patients will have at least 15 - 17 months PFS as opposed to Optune's results of just over 20% reaching 15 - 17 months.
Statistics do not lie; DCVax_P3 results beat Optune's Phase 3 trial.
I am putting together an article, with all my calculations and results for SA, but wanted to share a couple small tidbits. It goes over some of what I stated before but in more clarity.
- Screening halt occurred mid August. I have the DCVax-L handout with screening and treatment schedule. From this, it states that the screening process occurs right into 1- 2 weeks before enrollment/randomization (t = 0 months). Therefore, if screening was halted in mid-August, no patients was enrolled at t = 0 after the first week of September if not earlier. That means that from Sept 2015 to the end of Jan 2017 (~ 17 months), ~83 patients had no progression or with data-lock at Feb 2017, would have a PFS value of > 17 months. Now this doesn't include the other 248 patients, nor how many of the 83 are control and vaccinated (control has to be less than 50 patients). I'll address that in my article.
Using the Optune P3 GBM trial results, and lumping the 83 patients into the overall DCVax-L population (331 patients), we can do some comparisons. There was a minimum of 25% (83/331) that had a PFS > 17.3 months.
From Optune results (https://www.optune.com/hcp/newly-diagnosed-glioblastoma/efficacy )
50% - 40% - 30% - 20% - 15%
Optune - 7.2 mths - 9 mths - 13 mths - 18 mths - 22 mths
Control - 4.0 mths - 5.0 mths - 8 mths - 13 mths - 15 mths
DCVax-L - 17 mths
AND, we know that there is probably > 83 patients with 17 months PFS since a portion of the other 248 would have had ~ 17 months no progression as well. To be insanely conservative, assuming the Optune control results (15% of 248 patients), ~37 patients. Therefore, one could easily conservatively assume that ~120 of the DCVax-L patients (at minimum) will have PFS results of ~ >=17 months. That's 36% of the entire P3 trial. Optune's treatment arm only saw >20% reach a PFS of 17 months. AND WE KNOW DCVAX-L WILL HAVE ATLEAST 36% of their patients using SOC/placebo results, there is a statistically high probability that DCVax-L's Primary Endpoint PFS, will not only be reached, but would surpass that of Optune.
Mathematically that is the case. p is calculated using the standard deviation and the population when comparing two populations. Therefore, to power or strengthen the test in order to get a better p value (lower p value), you must either increase the population (and after a certain point, this increase does very little, or increase the standard deviation (however this you can not control). What you are testing ultimately is what is the probability that median value of the vaccinated group is different from the placebo group (or more so the probability that they are the same, no difference).
If you have median values that are 2 months apart, you'll need a larger power of patients to show statistical significance that they are different populations (that the vaccine effects the patients). If you have median value difference of 12 months (meaning the vaccinated group has a substantially larger standard deviation), then you would need less patients to show statistical significance. SO.... if you had a 12 month difference, you could have statistical significance before you even fully enrolled the trial, so to speak.
Hope that explains it.
There are numerous trials and products that were approved using PFS as a primary, these are all aggressive diseases. They use PFS as a surrogate endpoint for accelerate approval.
Also, as I mentioned, GBM is a disease where only 2 months increase in OS would be considered a breakthrough. With 331 patients, it would be difficult to get enough early data to provide a p value < 0.03 for a difference of 2 - 4 months.
To show statistical significance for 2 months early on, 50% of the vaccinated patients would need an increase of 12 months or more than the placebo. Couple this with the fact that the placebo patients are living longer as well, it means it would be difficult to show statistical significant efficacy early on.
The fact that over 1/3 are already at 20 months OS though is huge. And that isn't counting the number of patients in the other 2/3 that would have survived > 20 months as well.
Why GBM trials would rarely be stopped early for efficacy and AA.
A breakthrough in GBM treatment is considered an increase in OS of 2 months. Also, GBM is a rare disease, only 14,000 cases a year in the US. SO, either there has to be a HUUUUUUGE (in Trump voice) difference like > 12 - 24 months (rather than 2) or, there would need to be a trial of 1000 of patients to reach statistical significance early on for efficacy.
So unless DCVax-L is the ultimate cure for cancer, there will probably not be an extreme difference in OS to allow for early stoppage while still having a p value below 0.03.
On aside note; Who quoted that?
Stopping early for efficacy is not always a good thing:
"Compelling evidence suggests that trials stopped early for benefit systematically overestimate treatment effects, sometimes by a large amount. Unresolved controversies in trials stopped early for benefit include ethical and statistical problems in the interpretation of results."
What this is saying is that, to achieve a strong p-value, there needs to be enough data points to justify the p-value. Moreover, unless data is showing extreme differences, like 2+ years difference between OS placebo and OS treated group, it is unlikely that a p-value low enough will be achieved early on. The fact that a 2 month OS benefit is a huge leap of success for GBM shows that to achieve a p value low enough for efficacy, there probably needs to be more data, meaning, less likely that a trial for GBM will have good enough data to stop early for efficacy. As of 1 - 2 months ago, there was still over 1/3 of the patients that had not had an OS event, or were at ~>= 16 months OS still. Therefore to say that since this trial didn't stop early for efficacy, is BY NO MEANS HAVE ANY WEIGHT, to justify the trial will fail. If anything, the fact that over 1/3 of the 331 patients will have an OS of > 20 months provides a lot of statistical weight that the trial is showing substantial increase in OS values. At this point, 2/3 of the patients would need to have OS values < 20 months to provide a scenario where there is no efficacy (assuming a median OS of ~20 - 21 months as seen with nearly 200 patients in the ICT-107 P2 Trial.
Was scared there until I saw the news. Another public offering. This will just drop the price down to the 0.20 - 0.30 mark. Still waiting for Phase 3 news. Means I can buy some more at a discount. Its worth the risk in my opinion.
And they used the Wilcoxon Method! Which I used on the Ph1/2 trial results. Nice!
Or is the last 8K related to the $11M due. I think it may have been a way to move out the $11M 6 months but in terms of the two loans (promissory notes) so to speak. So there may not be any news in the next couple days. And that could be why the price fell prior to the 8K news, and then has been moving back up since the 8K release.
Yep, its a deal that shows they are banking everything on positive results. For your sake, lets hope the results are negative, because the short squeeze could bankrupt you!!
Noncovertible promissory notes w/ 6-month maturity. This means that they are banking on positive results from the Phase 3. With 6 of these, it gives them enough to both pay the convertible notes from the Nasdaq delisting, as well as have enough to stay afloat for < 6 months when Phase 3 results come out. And they are non-convertible, so they cant be converted after the fact.
Its a make it or break it. Milner will either be correct, or so so so so wrong!!!
AVII, OMG! The stat sig of the P2 ICT-107 OS results was regarding the mOS of 1.6 months of vaccinated group vs. the control group. No other Hypothesis test! Just that specific Stat Sig was regarding the 1.6 months and if the probability was less than 5% that the median value of the vaccinated group was actually 1.6 months greater than the control group. WOW!!
The actual p-value calculated was p-value=0.643. That means that there was a 64.3% chance that these two groups were from the same population, meaning NO EFFICACY. That is huge. It wasn't even close. That also tells you that there wasn't enough time given to allow the "tail" to significantly lower that p value, as it was soooo high.
Your quote:
"Once that trial failed and used it's allocated alpha, there is no "over time, the stat significance could have actually become
significant "
YES, completely wrong here. ALSO THIS IS WHY THEY ACTUALLY KEPT COLLECTING DATA WITH LIVING PATIENTS!!!
Quote from news PR:
"Patients who have not yet progressed will continue until an appropriate termination point can be determined and the company will continue to analyze the data from the study."
http://news.cancerconnect.com/follow-up-study-shows-ict-107-vaccine-fails-to-extend-survival-in-gbm/
Why would they keep analyzing the data, unless it could eventually improve the p-value? AGAIN! P Value is calculated using the standard deviation, so if the standard deviation continues to widen, it affects the p value.
AVII, WRONG AGAIN! You keep digging yourself into a hole!
PS, I am done trying to correct you, and argue with you. Its a lost cause.
AVII, YOU GOT ME! NO WAIT, YOU DIDNT!!
Sorry, I mistyped, the median OS was 1.6 difference, not 1.8 months. And they even tried to see if that was stat sig and it wasn't, however, the trial was looking at 2 month difference. Moreover, over time, the stat significance could have actually become significant (as more of the tail lived longer, thus increasing the variance). However, for trials, they only look at the data lock events and not any accumulated numbers after the fact, and that is why it is important to have enough events, particularly when you have the long tail as has been seen with immunotherapy.
Oh and yes, I do understand simultaneously comparing/evaluation multiple tests. I think this argument has run its course. I cant force you to understand stats.
AVII - WOW! YET AGAIN!
First off, yes you are incorrect with all your BOLD statements! And they are BOLD!! When including a subset of data, even with similar Median, and similar distribution, IF YOU REMOVE THESE DATAPOINTS, AND 50% OF THE POINTS ARE < OVERALL MEDIAN, AND 50% OF THE POINTS < THEN IT IS A WASH!!!
No matter if it is similar, or dissimilar. SO WRONG YET AGAIN!!.
Lastly, not sure you under stand "Power" either.
AVII Quote:
"Just because a trial is powered to detect a 4 month difference"
No trial is "powered to detect a specific difference, like 4 month as you state above."
AND you say
"Just because a trial is powered to detect a 4 month difference doesn't mean it has to detect/measure a 4 month difference to be stat sig. "
If the median PFS value for the treated group is < than 4 months, it doesn't matter about the Hypothesis Test, there is no way to calculate stat sig of 4 month increase, because THERE IS NO 4 MONTH INCREASE IN MEDIAN VALUES. Reversely, if there was a 7 month difference, one could calculate the statistical significance of 7 months difference as well as 4 month difference.
The power of a statistical test is the probability that the test will reject the null hypothesis when the null hypothesis is false. If the hypothesis isn't even seen with the trial results (i.e. median PFS < 4 month), then there is no statistical significance. And that was what occurred with ICT-107 P2, however, over time, they were hoping the variance would be enough to show stat sig of ~ 1.8 months difference.
There are numerous parts that effect this:
The trial size, the variance, the confidence interval or p value used to evaluate the test. I don't know if you understand all these together.
AVII,
Not sure how I don't know or didn't know any of those things. I actually proved each of them to you. And again, I do understand Log Rank, and used the Wilcoxon Rank Test instead.
And again, your quote " It should not have been a surprise to you (and I don't know why you actually had to put this in a spreadsheet before you reached this dramatic "conclusion") that when you merge two data sets with similar mOS, and similar survival distributions, you will arrive with a combined dataset with similar mOS.
"
IS INCORRECT YET AGAIN! Has nothing to do with them being similar or dissimilar. If you add or subtract the "progressors data", because the MEDIAN is calculated as it is (the middle most point), if you remove 20 low values, and remove 20 high values, even if the dataset has a similar distribution, the MEDIAN VALUE WOULDNT CHANGE. What part of that do you not understand?
I always knew DCVax-L was randomizing their trial, I think you are confusing your argument about that with other people.
Also SS (stat sig), has nothing to do with the 4 month different. One makes a hypothesis test and evaluates it to see if its statistically significant. That means comparing median PFS of two populations being different by 4 month, in that DCVax-L has to show it is better than SOC by at least 4 months. So its not less. Its actually 4 month. Here is the thing, ICT-107 didn't meet the 2 month median OS difference, so it doesn't matter if it reached stat sig or not, it didn't achieve its objective for one to even do a Hypothesis Test to see if it was Stat Sig.
I am starting to think you might not even understand Statistical Significance, etc...
Exactly, meaning AVII is incorrect.
You can not compare two datasets each with their own median values, by just looking at the median values. You have to look at the entire dataset. The median value is the middle most value. So if you have 10 values of 2 and 11 values of 40, the median is 40, but the average is 22.
My point is that by removing the Rapid Progressors & Pseudoprogressors from ICT-107's P2 data, it isn't a Given that the median value will then be higher. AVII's point was that ICT-107's P2 median OS of ~ 20 months & median PFS of ~9 months would be higher if they removed early progressors (both pseudo and rapid), however, I just mathematically proved that isn't the case using the informational arm's values.
By removing these, it does takes less "noise" out of the data, but I do not believe that it would effect the median value by more than a month either way. However, that would have been enough for the P2 results to have been met. If anything, the mOS of the P2 control groups seemed quite high at 20 months, and could have been pushed up 1 month by pseudoprogressors. Also, the larger the trial, the less effect this "noise" has on the median value.
AVII, Not sure if you understand Median or statistics.
You quote:
"It should not have been a surprise to you (and I don't know why you actually had to put this in a spreadsheet before you reached this dramatic "conclusion") that when you merge two data sets with similar mOS, you will arrive with a combined dataset with similar mOS"
Your statement above has SOOOOO much wrong with it regarding statistics. The fact that the median values are similar has nothing to do with it. In fact, you could take a sample of 51 patients with a median value of 10, and with the right distribution, when added to the 119 patients, it would bring up the value.
Take for example 26 patients at 10 months, and 25 patients at 17 months. Now we can see that the difference in median OS is huge, from 17 (119 patients) to 10 (51 info arm patients). Guess what the NEW median value is: 16.76 months. YEP! It better than using the 15 month median value of the information arm's median value of 16.09 months.
Lets take a median value of 2 months, 26 patients with 2 months, and 25 patients with 17 months. Guess what the new median value is for the 119 + 51 patients; 16.76 months! WOW!!! How could that be? I mean the median value of the 51 patients is 13 months less?
Okay, so AVII, would you say you were INCORRECT about your statement above? "that when you merge two data sets with similar mOS, you will arrive with a combined dataset with similar mOS"
So you know now that this statement is completely not true??? Again, if you were speaking of MEAN/AVERAGE value, then yes, it would be true, but again, we are speaking about MEDIAN!!!
I REALLY DO LIKE BEING RIGHT!!! :)
A good comparison of DCVax-LInfo on Rapid Progressors
So I found this publication on rapid progressors:
https://www.ncbi.nlm.nih.gov/pubmed/18854836
It showed that rapid progressors, group that showed GBM cultures with cycle times < 10 days, showed median survivals of 6.5 months. Cycle times > 10 days saw 9 months.
When looking at the chart of 20 rapid progressors, this would correlate (as I mentioned before to the historical values NWBO used of 8.5 - 11 months. Though as a rapid progressor, the median value would be ~ 6.5 - 7 month mark. Looking at the 15 months, this is a significant bump and I believe a good indication that there is efficacy in >50% of patients. Yes, it is only 20 patients, however, because these are 20 rapid progressor patients, the tumor is growing at a much faster, aggressive rate, and therefore, to show double the median value I believe is very significant, because it shows that GBM cell growth had to have been slowed down to over a rate of > 10 days for cell cycle times.
In my mind, the reason immunotherapy for GBM patients is a breakthrough, is that at minimum, it slows down tumor cell regrowth, and obviously in subsets of GBM patients, it not only will slow down tumor cell growth, it can keep it in check from growing at all.
And that is why I believe PFS more and more has become a valuable primary endpoint for highly aggressive forms of cancer such as GBM. By proving that tumor growth/regrowth has been significantly reduced, there is obviously a positive effect of the immunotherapy.
Understanding the ICT-107 trial
First off, I never compared the informational arm to the ICT-107 P2 trial. I did compare the PFS results of the ICT-107 trial to that of DCVax-L which I do think is valid. Particularly with the PFS results, in that the median PFS of 8 - 9 months is what every other trial has seen. Secondly, as ICT-107's P2 trial failed, and IMUC wanted to move it to a P3 still, I take much of their comments about including "Progressed" patients with a grain of salt. Unless I see the data, that shows numerous rapid progressors were taken, one must know that IMUC wants to make a case to move to P3, and by adding some potential P2 trial failure "reasons", they can make the case that a P3 will show positive results. The fact that they screened 2/3 of the patients enrolled, and only 1/3 was included in the trial means there was some heavy screening going on. Also, that along with just as many if not more pseudoprogressors were included in the informational arm, ALONG WITH A PSEUDOPROGRESSOR TRIAL ARM, I DO NOT put much weight on potential "Rapid Progressors included in the ICT-107 arm holding the median PFS and OS values considerably smaller than what DCVax-L's P3 control group will see.
What I do find interesting is how much criticism NWBO and Linda Powers has been given due to their P3 Trial design, with PFS being primary, removing pseudoprogressors and rapid progressors, etc... However, its interesting that IMUC decided to change their trial design to look much more like DCVax-L's. And yes they still have OS as their primary endpoint, however they include OS in MGMT patients as their secondary along with PFS. I think they would have been better to design it with PFS as primary though, or OS in MGMT patients. That way it would have given it more chance to succeed.
JUST TO PROVE AVII WRONG AGAIN...
Okay, so I found all my old data from the 119 SOC patients that I sued when doing my statistical analysis back years ago on the P2 trial. The median value of the 119 was 17.39 months. Now the 119 SOC patients values were taken at diagnosis, so instead of using the 15 months, we use the 18 month dataset. Including these additional 51 patients, the median value actually increases, from 17.39 to 17.75.
Now lets be extra conservative and see what the median value of the 119 would be if we used the 3 month offset values of the 51 patients (PS, I estimated each patients OS from their poster graph), so subtracted 3 month from all 51 patients OS, giving a median of 15, the new median value of the 119 + 51 is 16.09. So yes there is a drop, but it isn't that much considering that we just added 50% of the trial. If I ad used 331 patients, then the median value would have barely been changed yet again. AND AS I MENTIONED, IT IS BECAUSE THEY OFFSET ONE ANOTHER.
DAMN I LIKE BEING CORRECT!!
If you like I can email you my spreadsheet or the numbers if you want to check my work.
AVII, the poster for you:
https://www.nwbio.com/AACR_poster_prolonged_survival.pdf
Here it actually states that the informational arm was not randomized. Therefore, using their numbers and trying to compare them to ICT107 P2 numbers doesn't make sense. Again, as I also do not think IMUC included rapid progressors.
What I think is interesting is the PSEUDO-PROGRESSOR ARM of 32 patients, which was randomized (like the 331 P3 trial).
They pretty much just did a Phase 1 study within this Phase 3, which was very smart. It is randomized, blinded, 2:1 .
Flipper, I don't argue that ICT-107 P2 enrolled rapid and pseudoprogressors, however, I truly doubt they randomized (meaning, they included them in the trial) many of these, particularly the rapid progressors. They enrolled ~278 patients, but only randomized 124 patients. That means there was some intensive screening there between enrolled and randomized patients.
Its on their website:
http://www.imuc.com/pipeline/ict-107
AVII Its 3 month not 4 months, and its obvious NWBO uses from diagnosis in order to compare them to the historical OS values that are all against from diagnosis as well. If they used the Randomized value, there would be a much less difference. However, no matter if its 18 months or 15 months, or 14.75 months (however, if you look back at your initial post it stated 14 months, not 14.75), you are comparing apples and oranges.
It would be like me comparing the subset of MGMT patients with the overall groups median OS. APPLES & ORANGES MY FRIEND!!
AND its pretty obvious from the number of ICT-107 P2 patients enrolled and the number randomized, that there was some serious screening taking place.
AVII - DID IMUC ENROLL RAPID PROGRESSORS???
I happen to think that IMUC didn't, and that the mOS ~20 months is not "DECREASED" because of the inclusion of Rapid progressors.
1), it states "ND-GBM" Newly diagnosed so not recurrent. And ....
2) this quote:
" the trial enrolled 278 patients at 25 centers throughout the U.S. and 124 patients were randomized to standard of care treatment plus ICT-107 or standard of care plus placebo (i.e., control). "
The fact that IMUC enrolled 278 patients, but less than half of them were randomized, tells me there was much more "Cherry-Picking" or "Screening" then one would think. Actually, that even strengthens the case that is the reason for such a high mOS of the control group.
AVII, You were wrong about the medianOS of the 51 being 14 months. Just sayin'
8 rGBM patients were not part of the 51 informational arm.
"A second cohort of patients with rGBM consisted of 8 patients with recurrence following several adjuvant temozolomide treatment cycles. Median OS for these patients is 14.7 months from the time of surgery for first recurrence."
These patients were not listed in the 51. Okay so that is out of the way.
Quote:
"Of note in the trial PFS and OS times are estimated from time-point of randomization, which happens approximately three months after initial surgery, whereas in common clinical practice these are usually calculated from the time of surgery." This is from the P3 DCVax-L
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4514134/
~ 3 months, its the average time, and unless the patients is waiting a month to get surgery, which I don't think is the case, its usually < 1 week or so.
Now I do not know if the rest of those were taken from diagnosis. If anything, that would be a very difficult measure, since they would have to get that information from the patient. What I do know is that the P3 results are from randomization, and this informational arm is not a reflection of what the trial will be. Plus they offset each other, so I do not put much weight on it.
It is impossible to find the difference between the two groups as the data is blinded, however, what I get from all of this is that all studies and/or trials I have seen results on in the past 10 years has shown a median PFS beyond 8 - 9 months. Even the ICT-107 P2 showed a median PFS <9 months. AVII is trying to equate that since ICT-107 included the rapid progressors, their median PFS and OS should actually be significantly higher, however that would also mean pseudoprogressors were included, which I believe would offset the effect.
What tells me that there is an effect is that we know that mathematically, there has to be > 50% of the patients with PFS 16 months or greater. Now say that this is because AVII is correct and the placebo and vaccine groups have no difference, and show a median PFS of 16 months. We know that PFS and OS correlate, and with ICT107's P2, there was a difference of 12 months between PFS & OS, so add 12 months to the 16 months, and you get 28 months for median OS. Give or take 1 -2 months, seeing placebo patients living median of 24 - 28 months would be a breakthrough in GBM survival. For the OS of the trial to be around 20- 22 months, the median time between PFS and OS would need to be 4 or 5 months, however, we are seeing atleast 9 - 12 months. The more and more I crunch the numbers, the more and more the data suggests that overall the results are overly increased to past trials. And yes, there is a chance that the control group showed PFS results as good as 16 months, however, since that hasn't been seen in trials to date, and is considerably lower, I look at this as being a result of vaccine efficacy.
AVII, This doesn't include all 51 patients
First off, the title is on recurrent GBM (rGBM) which is different than the rapid progressors. Recurrent GBM patients wouldn't even be allowed in the screening for the P3, let alone the informational arm. The 51 information arm patients were all pre-screened and were initially included in the trial, but didn't pass the baseline MRI visit on month 2, so where then excluded, and moved to the informational arm.
Recurrent GBM is when you have already removed the tumor and it progresses, and so this second cohort of rGBM (8 patients), would have had a second surgery to remove the tumor and create the DCVax-L vaccine.
As for the 19, it looks like the 19 are part of the 20 rapid progressors, however, still, the overall 51 patients mOS isn't 14 months, as you state, as that 14 is only the 19 patients, not the overall 51 so it doesn't includes the other 31 patients. Even if the 18 months is taken from diagnosis, then from randomization, it would be ~16 months.
Again, as I mentioned, because of how the median is calculated, the majority of these 51 patients would offset one another, which would leave ~ 7 patients between 16 - 19 months (assuming everything has been adjusted to diagnosis) that would effect the median, which, depending on the data (if there is a lot of patients in the 18-20 OS, might not even effect the median value. If it were the mean or average, yes it would for sure effect the value.
Again, I wouldn't use the DCVax-L informational arm to evaluate that the ICT-107 mOS should be considerably higher than 21 months. For all we know, there could have been substantial amount of pseudoprogressors included that pushed up the median OS to ~21 months. I guess what I am saying is that mathematically we do not know.
However, we do know that the DCVax-L P3 trial measures OS, and PFS from randomization, NOT diagnosis, so it is using the same measuring time stamp to ICT-107 P2.
AVII
"The median OS of the group of 51 Information Arm patients as a whole is 18.3 months. About 30% of the patients (15 of the 51) lived beyond 2 years, and most of these patients (12 of the 15) remain alive. "
excerpt from news release, linked below: (so not 14 months)
http://www.prnewswire.com/news-releases/nw-bio-reports-promising-survival-data-in-51-gbm-patients-treated-with-dcvax-l-300056928.html
Using the information above we know
25 patients had OS < 18 months and 26 > 18 months. We also know 16 of the 25 indeterminates lived more than 21 months, and 3 others lived 30.1 months. And since 26 of the overall 51 lived more than 18 months, that leaves 7 patients that lived between 18 and 21 months. So depending on where these OS were, there could be minimal effect to the median value as the rest of the 37 patients all offset one another on each side of 21 months. Again, I don't believe that including these would have dropped the ICT-107 OS results drastically, we are talking about 7 patients between 2-3 months, therefore the median would be 20 instead of 21 months, or so. Therefore one could suggest ICT-107 mOS could have been 22 months instead of 21 months, but also could have had more pseudoprogressions and should have been 19 months. Again, as I said, I do not believe including them would effect the median value as much, but could effect the tail.
Difference between median and average. However, I will show you that if those 51 patients were included in the DCVax-L study, the median value would have gone up, not down, and subsequently the same for ICT-107.
AVII you are correct, the 51 patients have a median OS of 18 months, now that is less than the 20-21 months of ICT-107. However, unless you are calculating the average OS value, these would slightly increase the OS median, and here is why.
Lets assume that for the four groups, all patients have the median OS value for simplicity however it wouldn't matter (and you will see why). So we have:
20 Rapid Progressors all at 15.3 months
5 Unclassified Patients at 9.2 months
25 Indeterminate Patients at 21.5 months
1 Pseudoprogrssor at 30.1 month.
Now if the median OS before adding these was 21 months, the 25 patients with OS < than the median 21 months would be offset by the 25 Indeterminate Patients at 21.5 months, and then the extra pseudoprogressor would then mean the new median would be the next patient with OS > 21 months. So the new median OS would be 21.2 months (or whatever that was).
Anyways, as such, because there is ~ 50-50 split, it would have little to no impact on the median value. On the average value, yes, but not the median value. Hope that helps.
30% of P3(99 patients) have OS>25months
I actually think I may have to accumulate all of this and write an article. But here goes. I found some more good data when investigating GBM Rapid Progressors. The following study showed statistical significance in that cell cycle times may be used as predictor of survival. There were two groups they looked at, rapid progressors, which had tumor cell regrowth in less than 10 days, and then those > than 10 days. The found that in 42 GBM patients, the median OS for these two groups were 6.5 months and 9 months, meaning that after progression, the median patients lived 6.5 - 9 months longer.
Now looking at NWBO's historical literature to compare their 20 rapid progressors, they state comparable SOC values would be ~8.3 - 10.8 months. The rapid progressors were removed after month 2, so adding that to the literature value range above gives the 8.5 - 11 months. Nearly bang on.
Okay, so that said, if we know that ~83 patients had PFS events > or = to 16 months (as stated in numerous of my previous posts). Therefore, if there is a correlation of the length of time from tumor progression to overall survival, we know that the OS of these 83 will be > or = (and most likely greater) to 16+6.5 = 22.5 to 16 + 9 = 25 months. Assuming that since the patients have gone 16 months with no tumor progression, they are not rapid progressors, so they should at least be seeing OS of 25 months. Also that is assuming all 83 patients have tumor progression in Jan/Feb, which will not be the case.
Then again, incorporating historical SOC ratios to the other 248 (again conservative), of 30% living 24 months, that gives us ~50% of the 331 patients having an OS > then 24 months at minimum!
SO I guess what I am saying is, there is a lot of information pointing to an overall OS > 24 months.
https://www.ncbi.nlm.nih.gov/pubmed/18854836