From the conference call/presentation I saw the comparison to the 2004 trial. It was a little different than what we saw before in other presentations. The numbers they quoted in the presentation were for a sub-set of the patient population: 113 patients with "fully resected GBM". In this sub-set the overall survival was 18.8 months.
Now the current trial is for patients "following resection" and phase I results are usually described for the fully resected patients. So it's hard for me to tell if the phase II population is fully resected or not.
If 18.8 is the number we should see for the control arm then if the placebo is the same as the active arm I calculate a mean time to 32 events of 24.13 months (24 month media) with a 1.3 month standard deviation. That means that if the results came out today (Based on 18.8 months OS), there is about a 50-50 chance of a survival advantage and probably not enough of an advantage to warrant skipping phase III.
For a 3 month advantage I calculate 25.29 months (next month) to reach 32 events with a 1.44 month standard deviation.
For a 6 month advantage I calculate 26.29 months (April) to reach 32 events with a 1.53 month standard deviation.
If we follow phase I results and have 38 month OS, I calculate 29.9 months to 32 events (almost August). This also leads to 44.6 months to 64 events which is maybe December 2014 (maybe sooner due to some modeling quirks I haven't fixed yet for long term survival).
The question is: is the 18.8 month survival the number to use? Is this trial limited to fully resected?
(For a 6 month advantage I calculate 26.29 months (April) to reach 32 events with a 1.53 month standard deviation.)
Why are you calculating beyond the March 2013 timeframe? Management has already said that 32 death will come in the 1st QTR in the last presentation! They would seem to be in postion to make a "best guess". I doubt if they would throw this out there on a whim?
Thanks for taking the time to do some analysis. I am new here. I really have not done any analytical research here. However, I appreciate your efforts. Can you point me in the direction of verified patient enrollment data. Maybe you could just shoot it up on a post since you have obviously done some research. Thanks. In the meantime I will continue doing some snooping around. GL
"Fully resected" GBM is a nebulous concept at best as there are no clear histologic tumor margins and tumor cells are widely infiltrative even into histologically normal tissue. That's why GBM is virtually incurable by resection alone. Surgical resection palliative at best. It's an attempt to debulk ~98% of the tumor.
True, that is why patients were randomized by age in this trial. Age seems to be an important issue. When I looked at their phase 1 trial results it seemed to me that the age group below 50 did really well. However, age group 50-60 was not much left behind so it is encouraging. They had one gentleman who was 79 years old and he passed away relatively soon.
Phase 1 trial was represented by all age group maybe a little younger than the current trial average age. I had the phase 1 trial population in an excel by sex, age, survival etc.
You make a good point about survival and age but I would also point out that the amount of data has an impact on the complexity of your models. With limited data you have to keep your models pretty simple (not too many parameters in play) or have a very good theoretical/physical basis for your model. I know that seems less than satisfactory many times but that is the way it goes.
Basically you are correct, there are assumptions in my survival curves that could be wrong and so feel free to discount my estimates however you like if you feel like hedging some of that risk.
Yes, if they are giving us the 18.8 number on the presentation, which they are, I think we should use it. Nevertheless, the trial is not limited to fully resected. The Phase I trial had 75% fully resected. The benefit of being fully resected is that there are fewer cells to attack and we have been told that the success of ICT-107 is much of a numbers game, specifically the number of attacking T cells compared to the number of tumor cells and cancer stem cells.
This is taken directly from the clinical trials website.:
The proposed phase 2 study is a randomized, double blind, controlled study of the safety and efficacy of ICT-107 in newly diagnosed patients with glioblastoma multiforme (GBM) following resection and chemoradiation. The phase 1 clinical trial demonstrated safety and promising efficacy in a small, open-label study. The purpose of this study is to provide information from a larger, controlled clinical trial. Patients must be newly diagnosed with GBM and not yet received chemoradiation. Patients will have had tumor resection, magnetic resonance imaging (MRI) and tumor assessment prior to enrollment into the study. Post surgical treatment consists of 6 weeks of chemotherapy (TMZ) and radiation followed by a washout period. After Screening and informed consent, patients will undergo apheresis at the study site for collection of peripheral blood mononuclear cells (PBMCs). Apheresis product will be sent to a central site where monocytes will be purified and cultured into dendritic cells (DC). DC will be pulsed with synthetic peptides that correspond to immunogenic epitopes of tumor antigens. The pulsed dendritic cells will then be aliquoted and frozen before shipping back to the site. Patients will have the autologous DCs reinfused intradermally. A control group will receive unpulsed autologous DC. Patients will be randomized by age in a 2:1 ratio to ICT-107 or control.Patients will receive at least four intradermal injections of the ICT-107 vaccine and additional vaccine during a maintenance phase. The primary objective is to compare overall survival (OS) and progression free survival (PFS) in patients when treated with ICT-107 versus Control.
1.Confirmed, initial diagnosis of GBM. Patients must be newly diagnosed with GBM and not yet received chemoradiation.
2.≥ 18 years of age
3.HLA-A1 or HLA-A2 positive
4.KPS score of ≥ 70%
Baseline hematologic studies and chemistry profiles must meet the following criteria:
Hemoglobin (Hgb) 9.9 g/dL total granulocyte count than 1000/mm3 platelet count 100,000/mm3 blood urea nitrogen (BUN)
some of the above was cutoff. go to clinicaltrialsdotgov to see the rest. In any case, let try to get more than one data point and not just an 18.8 month OS number at the median. The only other decent suggestion that I can make is that we use the numbers that sfusurgen gave us, the get an equation for the best fit line, then modify the slope of the curve to shift to an 18.8 month OS. That should work well.
One other question - I just listened to NWBO's presentation (after I asked my event question above) and the CEO discusses the "enrollment" into the IMUC P2 trial and her take was that IMUC only has 124 enrolled in the trial vs 278 (she said that the 278 is the number screened) This would make the 32 and 64 events make more sense... Does anyone know the facts about this?
bdoglo - what shape survival curve did you use for the control? I would like to see your numbers if you could post them.
I'm not done yet with my model but one thing to note is that with immune therapies there usually is a very long tail and this is where the real potential benefit is in this trial so there is a real good chance that the 32nd event will not correctly predict the true median survival rate. If you look at other immune therapy trials - the initial deaths do not get much benefit (maybe their immune system is not as healthy so benefit is not that great) but those that do make it past the first 12 months or so have a very high probability of extended survival. (This is one reason there has been a call to look at the average vs the median...)
BTW - the question that I asked yesterday about the 32 and 64 events - thanks for the info but why determine the median survival off of only 64 events when the original plan was to have 200+ participants... esp since they should know about the "long tail effect" with immune therapies - yes the more events, the longer the trial but you also get better data in this type trial.
The "long tail effect" is a valid point. I also mentioned that the immune system needs more time to respond.
My only issue with that is that if we get a mediocre interim result in the next two months which is not robust enough to stop the trial, we might drag into mid 2014 or later in case of a continuation of the trial. I know we have anough cash currently but if we are in mid 2014 cash would need to be raised to fund the new trials, opex etc.
Having said that pps should reflect a better read by then but what if not and dilution becomes an issue ? Based on a potentially mediocre interim result, nothing is guaranteed.
I do hope to have a strong interim reading but give a decent probability to the "long tail effect" as well.
+ overall market can make it to 2008 lows till mid 2014...
bdoglo - phishes3 - gaborrcz -
I'd like to thank you guys for your time, effort, and contributions.
It's nice to see 18.8 mo. used to calculate the control arm. As this comes from data nearly 10 yrs. old, would bumping control to 19 mo. be considered unreasonable ?
As to the 38 mo. OS in the phase I, that is from 16 patients. My experience with bio investing , an increase in the N for this phase II trial from 16 to 84 , the OS number will decrease. Could 30 or 32 months be used to calculate the vaccine arm in this trial ?
- The CEO saying it would be nice to hear from the DSMC soon, this IS a futility look at data, - while no one here believes ICT-107 has no benefit, Mr. Yu and co. will breathe easier getting past an important hurdle in this trial. - Of course we don't mind waiting till April to read " Continue to conclusion." ; )
Lastly, 64 events is approximately 50% of total enrollment. Final read out of START trial was 705 events, 50% of enrollment.- Maybe a standard for oncology trials ???
Thnks again, no replies required, just putting some thoughts out.
For the control curve I approximated the 15 mo OS curve and just stretched out time by 18.8/15 and interpolated. Here are my 18.8 mo OS curves numbers for month 1, 2, ...
0.984, 0.968, 0.952, 0.936, 0.920, 0.904, 0.877, 0.845, 0.813, 0.781, 0.749, 0.717, 0.685, 0.653, 0.621, 0.589, 0.557, 0.526, 0.494, 0.466, 0.437, 0.408, 0.379, 0.351, 0.322, 0.308, 0.295, 0.283, 0.270, 0.257, 0.244, 0.231, 0.219, 0.206, 0.193, 0.180, 0.168, 0.159, 0.156, 0.152, 0.149, 0.146, 0.143, 0.140, 0.136, 0.133, 0.130, 0.127, 0.124, 0.120, 0.117, 0.114, 0.111, 0.108, 0.104, 0.101, 0.098, 0.096, 0.096, 0.096
don't worry about the last few numbers, I just held the survival constant at the end instead of extrapolating but those numbers are so far out in time that they don't really figure into the 32 event calculation.
As far as how they came up with 64 in the first place, I have no clue. But they are not determining anything directly from 64 events. Rather that is their trigger to unblind the trial. Once the trial is unblinded they can calculate their numbers from whatever data they have. If the trial goes very well this is a potential "problem" because they could be blinded for a long time. You would have thought they would have had a secondary trigger based on so many months from the close of enrollment or something, but they don't.
Thanks for the info bdoglo.
I think it's better to use this 18.8 month just to be more conservative. This number would also be the max we have seen so far (e.g. UCLA info). I would assume this trial is for fully resected but somewhere I saw something like a 75% fully resected figure, I would need to check. Maybe someone from this field can shed more light on this.
Again, we better go through Q1 without any annoucement...I also do not understand their math why a Q1 interim would be a good thing. I read somewhere that the immune system needs more time to respond to these kind of vaccines...however, if that is the case we might spill over to 2014....
if there is an announcement end of Feb what would be the survival advantage (months) ?
A couple scenarios I worry about from the financing perspective:
1. We get interim results right on schedule and the signal for efficacy is not strong enough to skip Phase III. This would mean if there was approval, it might happen much further down the road.
2. There is a strong signal for efficacy that significantly delays the trial reaching 64 events and the NDA is significantly delayed because they will be waiting on results (it's hard to complain about this case though since it means really good news in the future for many people)
The 75% was for the phase 1 trial i.e. 75% was fully resected. I would assume that fully resected makes one live longer but most probably this is not that clear cut.
I would just add that patients get 6 shots in the current phase 2 trial instead of the 3 shots in phase 1 and these shots are much stronger. Based on this you would expect an even more robust result.