Here are some things to look out for;
- TMZ refractory GBM is a very very tough disease with MOS in the range of 5-6 months
- In the phase I portion (content of abstract) patients were separated into 3 dosing cohorts
- From the abstract it is not clear if any dosing cohort responded more than another cohort. ie are all of the responses in the high dose cohort
- The preclinical work that supported this trial made it clear that it was the combination of TMZ with radiation and Indoximod that triggered complement deposition (innate immune system activation) on the tumor
- From the abstract it is not clear if all of the responders received this full combination or not
- If the radiation + chemo_IDO inhibitor combination is established in this poster as destroying tumors and delaying disease progression then it could have much wider implications beyond this trial. IDO inhibitors such as Indoximod could render conventional chemo-radiation treatments more efficacious if they are given in a setting in which IDO is blocked.
- Since IDO activation drives tumor angiogenesis and metastasis, it will be interesting to see if any delay or halting or tumor progression occured with Indoximod alone in the absence of radiation.
- In the phase II portion patients are slit into groups that receive avastin and sterotactic (high intensity) radiation with indoximod.
Since half of the trial events havn't yet occured and the average time in trial is 36 months, there clearly is hope. There is nothing deceiving about the math.
Your post is highly offensive to patients like Toad in this trial and their caregivers. I just dont understand how anyone would want to say something like you did.
So we have a combined 3-year survival of 50%. Typical 3-year survival rates in SR PC are 25-30%. Even if its 35% this time, thats a 65% 3-year survival in the trial arm, which after decades of no change in long-term survival, would be a game changer.
Making control and trial the same, it takes 36 months median to drive the milestones through both February 2014/2015 dates (based on all of the publically available enrollment figures). So 36 months is the average time for all patients in the trial.
Maybe this is a good starting point. Do you agree with the above?
Actually the Bull thesis is supported by math and science. On message boards, twitter, SA etc its totally acceptable to boast ignorance of science and proudly claim incompetence in mathematics. It was this ignorance that got shorts completely blind-sided by last years IDO and Ebola deals. I expect the same to happen again shortly. When exactly, who knows - my GUESS is the data is currently in the hands of the FDA.
There are many twitter generalist biotech commentators who finish their posts with "Hmmmm".
Is that 1 "m" for every time they couldn't be cothered doing their own DD?
Please fit the Feb 2014/Feb 2015 222/333 milestones into CTRL+Trial MOS = 24 months and CTRL+Trial 3 year survuval = 30%.
You will be off by 1 year!
Thats right, under the scenario above 333 events occurs in Feb 2014, rather than Feb 2015. 460 events would have occured by Feb 2015.
As Richard Dawkins says, its socially acceptable to boast ignorance of science and proudly claim incompetence in mathematics. So go ahead.
No. Opdivo was approved with 30% of patients crossing the 12 month threshold. As of February 2015, 34% of IMPRESS patients have crossed the 30 month threshold. Also IMPRESS is way overpowered for the redn of death method. So if the redn of death method is used, both hazard ratio and P value should not be a problem if this method is performed.
Bottom line, FDA and NLNK need to do what is right for patients. One poster a while ago noted that enough time has past to see a statistical benefit. With the new statistical approach the FDA is advocating for IO trials, that is 100% true in this case.
Can the FDA and NLNK find a solution to this dilemma?
Can the FDA on one hand approve IO drugs at record pace with one hand, while on the other let the trial continue when it dosent need to?
The fact that no announcement has been made suggests involvement of the FDA, but this should not necessarily be viewed positively and there is no clarity on which alternative methods NLNK have, or will decide to use.
Of course this is a moot point if the trial hits the required threshold using logrank analysis.
The outcome matters a lot for patients in both arms of the trial.
If the FDA and NLNK cannot find a solution and revert to the status quo that will result in statistcial reduncancy, all for the sake of a SPA, then unfortuantely the best interests of the patients are not being considered.
NLNK has a SPA for IMPRESS that calls for log-rank analysis for IMPRESS -not ideally suited for IO trials.
Under this method the trial could possibly continue at this interim analysis if the MOS of the CTRL group about 23 months. This is because of the dampening effect that censoring plays when log-rank is used.
If a continuation occurs then the trial will exist in a state of statistical redundancy. With only 7-10 events occurring between 333 and 444 events, the trial will continue much longer than it needs to – all for the sake of keeping a SPA intact. The 444th event may not occur until April 2016, when in reality it could pass the statistical hurdle under the SPA in April 2015. A continuation to 444 events will be 6 years after the first patient was enrolled. The first censored patient at this point will be 32 months – way past the MOS of any CTRL. Statistical redundancy.
When the SPA was created, there were no alternative statistical methods for IO trials. Now there are.
IO trials are now being designed and stopped early using an alternative statistical method (reduction of risk of death). This alternative method considers the number of patients that are alive past a certain time and favors immunotherapy trials. The relevant metric here is long-term survival. The combined 3 year survival in the IMPRESS trial is ca. 50%. CTRL 3 year survival is only in the 30% range, implying a significant long-term survival benefit in the trial arm which is easily detected by the reduction of death method.
However, this alternative method is not part of the NLNK IMPRESS SPA. NLNK could be concerned that any change in the SPA could be viewed negatively by the FDA.
Meanwhile PC is an unmet need. PC shows low PDL-1 expression, CAR-T does not appear to be effective on solid tumors and survival rates are essentially unchanged over decades. The cupboard is bare.
Can the FDA and NLNK find a solution to this dilemma? next post..
Be aware that I am not an investment adviser or a professional trader. I do not have a track record in biotech investing and have only focused my investing attention on this single company. Everyone should do their own DD.
Again, you just have to scratch slightly deeper. In the IMPRESS trial they have already covered 70% of the centers in the U.S. that carry out surgical resections. Thus, they will be able to to handle the commercial roll out with their own commercial team, so do not need to partner in the U.S., but obviously will have to in Europe/Japan.
Citing a small patient sample size of the 300M cell group, the company has only published the blended trial results.
There have been skeptics over the last few years who have questioned why NLNK would only publish the blended results, when the basis of the IMPRESS trial is the 300M dose for 12 months. Surely, if the 300M cell group imparted a benefit, they would want to advertise it?
To get to the answer you have to review the history of the trial. NLNK commenced the Phase 2 trial by administering high-risk patients with the 300M dose of the algenpantucel-L vaccine (n=26 patients). None of the first 20 patients enrolled in this dosing regimen had died after 1 year. In fact, only one patient died within the first 12 months under the 300M cell dosing regimen. This single patient was very high risk and had a tumor around four times (12 cm) the average size of surgically resected pancreatic tumors. As a point of reference, over the past 20 years in the U.S., typically 30-40% of surgically resected patients would have died after 1 year. In an effort to simplify vaccine administration (the vaccine is given in 50M cell injections), NLNK then enrolled more patients under a lower dosing schedule (100M, n=43), however these results were not as favorable as those observed under the 300M cell group. Because of the provocative results observed from the 300M cell group, NLNK contacted the FDA, and a Phase 3 trial was designed under SPA.
From the above, a couple of important things to note. Firstly, the 300M cell group result was the best outcome in a clinical trial for surgically resected pancreatic cancer ever. Due to these provocative results, a phase 2b dosing trial which was planned never eventuated.
Combined 3 year survival in the IMPRESS trial is 50%. Given that historical 3-year survival rates in SRPC are typically at best 30%, this implies a 3 year survival in the trial arm above 50%, which would be a game changer for patients and their families.
"MOS for the Phase 2 was only 24 months, thats not good enough to get halted at this interim"
- The MOS of the entire trial currently stands at 36 months.
" IF the 300 million cell cohort did better than the 100 million cell cohort, what CEO in their right mind would not break them out and show investors? "
- They have. Initial survival benefit for the 300M cell dose group was 96% (1-year survival) relative to the RTOG-9704 trial (70% 1-year survival) and the 100M cell dose group (82% 1-year survival).
"Not only that, but why would they double the duration of vaccinations from 6 to 12 months for IMPRESS if they had already achieved a 96% 1 year survival with 6 months of dosing?"
- Because the literature supports prolonged prime/boost regimes.
"The discussion about changing the data analysis at this juncture is a huge red flag"
- Quite the opposite. FDA Approved Opdivo (nivolumab) 3 weeks ago for the treatment of patients with previously treated metastatic squamous NSCLC. A closer inspection reveals the FDA used "reduction in the risk of death" as the approval criteria in this trial, which favors the tail of the KP curve, which is a characteristic of immunotherapy. In fact my modelling predicts that if this method is used at this interim, a CTRL MOS of 27 months or higher is required to NOT stop the trial. The odds of success are much higher than what most believe.
"as is the fact the trial has taken much longer to reach the number of events than originally projected"
- There can often be investor concerns surrounding significantly delayed event rates in clinical trials serving as precursors to negative outcomes. However, this has previously occurred in very heterogeneous patient populations (such as melanoma) in which there was little historical evidence to substantiate a meaningful level of comfort around projected patient OS times, which is not the case in the IMPRESS trial. OS in SR PC has not changed dramatically in 30 years.
1. On Wednesday last week the FDA approved Opdivo (nivolumab) for the treatment of patients with previously treated metastatic squamous NSCLC.
2. The FDA used "reduction in the risk of death" as the approval criteria in this trial. Reduction in the risk of death is a metric that is relevant for immunotherapy trials because it focuses on the tail of the KP curve, where the greatest benefit typically lies. This is different to the log-rank method being used in the NLNK trial. Reduction in the risk of death is essentially the number of patients alive at a specified point in time (in the case for Opdivo, 12 months). See my twitter acct. (nhyujm6nhyujm6) for results of the trial. It is the BLUE arrow that compares the reduction in the risk of death that was used for statistical comparison that was used for FDA approval.
3. NLNK is in discussions with the FDA about using alternate statistical methods that favors the tail of KP curves. Based on the timing of the Opdivo approval, the alternate method is likely to be the one described above.
The FDA have now approved an immunotherapy drug using the reduction in risk of death statistical method for BMY (at an interim analysis point) that favors long-term survival. Thus, they are likely to allow NLNK to apply the same analysis method at the 333rd event interim analysis point. Note that the average time in the NLNK trial is around 3 years (implying long-term survival benefit). 5-year survival rates in PC are currently 5-6%.
This also has significant implications for NLNKs other late-stage immunotherapy trials. The Opdivo trial enrolled just over 200 patients (enrollment began 2012) in the PhIII trial vs NLNKs 722 (enrollment began 2010). The reduction in risk method requires less patients to demonstrate clinical benefit, and thus is cheaper and faster.
In case anyone is wondering, this is positive.