Thanks Summer. I enjoy your explanations even when I don't fully understand them. I should have continued taking statistics courses as it might have steered me away from the boring legal career I opted for instead. I guess I can read up online.
I too have been adding to my position since Thursday's CC. I now have more shares than I want to be holding when the final results are announced. Cormack said in the CC that they're in discussions with Teva regarding whether they can say when the interim efficacy analysis will be. Hopefully they'll provide some guidance. If they don't, then given your estimate that the likely 375 events will occur in Q3 I'll probably sell most of my newest shares around the middle of that quarter (maybe sooner, if the stock climbs into the high teens). I plan to buy puts against the shares I keep, but it certainly would be nice if the interim succeeded and made that unnecessary.
Thanks Summer. It's too bad we missed out on your dating analogy post because I'm sure I would have followed that a little better. I understand your first and second points and the part about Teva being tight-fisted but I get a little lost when you talk about z-scores and linearity. My fault for never taking more than one statistics course, long ago.
If I understand you correctly, the 70% probability that 011 is not placebo applies if there is a linear boundary but you suspect that the reality is somewhere between OBF and linearity, which I assume gives us a probability that's less than 70%. If we nevertheless assume the probability is 70%, can you estimate how likely it would be that the trial would succeed on the interim efficacy analysis? Or is that impossible because you don't know what criteria would be used in that analysis?
eightbitking, I too bear some ugly SNUS-shaped scars but I think you're putting too much emphasis on the billing sequence in the PR. In the CC Cormack both opened and closed the presentation by discussing the fact that they had successfully gotten through both futility analyses, and he talked about it a couple of times during the presentation. He said that both OGXI and TEVA are blinded to the exact data so I don't think there's anything to suggest that he was pessimistic as to the upcoming interim efficacy analysis or, if that fails, the final analysis. At least we now know that a leak of a futility failure wasn't behind the stock's recent weakness, and there's at least a possibility of the trial being halted early if the interim efficacy analysis succeeds. I was very encouraged by the CC.
Obviously, the possibility of a SNUS disaster hasn't been eliminated but it's been reduced. Hopefully Summer can shed a little light on how much. But recall that SNUS didn't have two futility analyses before presenting final data. It may have had one combined futility/efficacy analysis but it's hard to know for sure because it never made an announcement.
Summer, I have to second ubuy's comments and congratulate you for pulling off another coup. It appears that the second futility analysis was indeed completed in the past few weeks just as you predicted. Anything you care to say about how this affects the odds for success in the interim or the final would be much appreciated. In any case, we're all very lucky to have you posting here since otherwise we'd be completely in the dark as to what's going on.
If you Google "Oncogenex: A Cancer Biotech Worth Betting On" you'll find an April 11 SA interview with Scott Cormack. At the end there is the following exchange:
"Sam: You just increased your authorized share count from 25m to 50m. I think an offering is coming soon. Am I wrong?
"Scott: Our present cash carries our operations into 2015, which is approximately a year past when we would expect to release the top line results for the Synergy Phase III trial. So, an additional financing is not required to achieve our objectives. In our proxy, we are requesting an increase to our authorized shares because we have only 25 million shares authorized, which is substantially fewer than our peers. Further, approximately 17 million of those shares have been issued. This is important because, for example, in the event that Synergy is successful, it may be desirable to do a stock split to align our authorized shares more closely with our peers. If we wanted to do a financing, we have enough unissued shares now that we could do a sizable financing so that is not the reason."
In other words, he is claiming that they're increasing the authorized shares to allow a potential split if SYNERGY is SUCCESSFUL (for which the odds just increased somewhat due to passing the futility analyses). In that event, the milestones and royalties they'd get from Teva would make it unnecessary to raise cash for quite awhile -- maybe ever, if they partner 427. If SYNERGY fails, they will need to do a financing, but since they have enough cash to last into 2015 I would think they'd wait at least until they get interim results from the Pacific trial of 427 +/- Zytiga, which they hope to report at ASCO GU in February 2014.
But it may very well be that the proxy proposal to increase the authorized shares has contributed to the recent selloff.
BTW the stock seems to be recovering a little. It was up on pretty good volume the day they announced initiation of the 427 P2 trial in lung. And it was up this morning on a couple of block trades before succumbing, on lighter volume, to the market's selloff. I'm leery of predicting but maybe the big seller of the past few weeks is done.
Well put, eightbitking. I too have a hard time reconciling the fact that according to Summer there is a 90%+ probability that the P2 results were attributable to 011's activity rather than chance with my feeling that the odds are only a little better than even. Even one of the more bullish analysts -- I forget which -- recently estimated the odds for SYNERGY at 60/40. They are probably concerned about the lack of evidence of tumor response. Yet if Cormack is right about the unreliability of x-rays in measuring tumors that have spread to the bone, we can't even be sure there was no tumor response. And, as IR told Summer, if 011 acts to combat resistance to docetaxel, tumor response is less relevant than PFS, on which the evidence is more encouraging if still murky.
The SNUS trauma may be contributing to my fearfulness also. Yet SNUS never had a randomized P2.
With regard to IR's comment to Summer that in the P2 011 seemed to slow progression until the treatment was stopped per protocol after 10 cycles: I've confirmed with IR that in SYNERGY the protocol allows physicians, at their discretion, to continue treatment until progression even if that means treating for more than 10 cycles. I was told that the origin of the 10-cycle limit is that it's the norm for docetaxel.
It would be nice if the company just said when they will occur or have occurred. I realize they don't want to possibly have to announce that the interim missed, and maybe they don't want people to "play" the futility analyses or the interim. But the stock is already acting as though the interim will miss and maybe some big investors have figured out when the analyses will occur and are playing them. At this point I don't really see what they'd have to lose by being more open.
Well, I've give my take. IR's explanation is reasonable but at this point it's basically just a theory. I don't think tumor shrinkage is essential or is necessarily a better predictor of survival than lack of progression, but what it does do is provide tangible evidence of the drug's activity since tumors don't usually shrink on their own. Lack of progression can only be measured by comparing how rapidly the treatment group progresses compared to the control group. Here, there was some evidence of slower progression in the 011 group but the sample size was small and the effect wasn't sustained after the treatment was stopped.
So basically I still think, based on the OS advantage in the P2 trial, that the odds are slightly in 011's favor in SYNERGY. I'll feel better if it passes the second futility analysis but if it then doesn't succeed in the interim analysis I'll probably buy some puts ahead of the final analysis.
Thanks for posting that Summer. Their explanation regarding how it's hard to measure tumor shrinkage once prostate cancer metastasizes to the bone shows that they're not being inconsistent when they tout tumor response as one of the favorable outcomes in the P2 trial of prednisone +/- 427. In the 427 trial, the patients did not yet have metastatic disease.
With regard to the last paragraph of Susan Specht's email, I vaguely recall Cormack saying at one of the CCs that one of the few differences in the protocol between the P2 and P3 trials for 011 is that in the P3 trial they will not be discontinuing treatment at 10 cycles if the patient had not yet progressed.
Box, Cormack has said a couple of times (most recently at the end of the Q&A during the Q4 CC) that there will be two futility and one interim efficacy analysis. He wouldn't say whether any of those analyses have already occurred and he said there would be no announcement unless the trial is stopped.
Summer has said that when she has time she'll explain how a futility analysis is done. I'm interested in that, as well as in getting her take on whether if SYNERGY passes both futility tests we should draw any encouragement regarding the efficacy or the final analysis. I hope she's able to find the time.
Thanks Summer, I'll look forward to your description of how a futility analysis works. When you find the time to write that you may want to post it in a new or different thread as it's getting almost impossible to find messages in this one.
I found a Neopharm PR relating to the PRECISE trial that said the required p-value for their interim analysis was .005 vs. .048 for the final analysis. If we can expect 011 to face a similar bar in its interim that may be a lot to ask. If you're able to say something about that too I'd appreciate it.
Summer, question for you. If, despite the stock's behavior of late (which I hope is just a fund unloading for reasons of its own), SYNERGY survives the second futility analysis, what does that mean for the likelihood of success at the interim or the end of the trial? Obviously it means that achieving statistical significance is still possible but does it mean it's probable? Is it a matter of just continuing the trend shown at the time of the second futility analysis as additional events occur, or would the trend in favor of the 011+DOC group have to improve?
For that matter, could the second futility and interim analyses occur at the same time, that is, when they look at the data can they evaluate it for both futility and efficacy? Or would the design more likely be to hold off on the interim until more events have occurred?
tnk, Cormack said in 2009 and has repeated several times since that Teva's committed investment was over $200 million in upfront payments and funding of the 3 trials. It's not a minimal amount by any standard, though I agree that too much weight shouldn't be assigned to it because big pharmas have often entered into partnerships for products that blow up in P3.
I still think you're assigning too much weight to the absence of proof of tumor shrinkage. I refer you again to slide 59 of the ASCO presentation. No one in either group showed a complete response; 5 in the 011+DOC group and 6 in the DOC only group showed a partial response; 20 in the 011+DOC group and 12 in the DOC only group had stable disease; and 1 in the 011 +DOC group and 4 in the DOC only group had progressive disease. These are very small samples but to the extent they suggest anything it's that there's virtually no difference in tumor shrinkage but that the 011 group tends to delay tumor growth, which may be as good if not better as a predictor of survival. It's also possible that with better x-ray analysis or a bigger sample you'd find that more of the 011+DOC group had tumor shrinkage (i.e., some of the "stable disease" category might actually be partial responders). As for the fact that the 011 group got more cycles of treatment, the reason was that patients stop getting treatment when they progress and fewer 011 group patients progressed. Check out slide 72 and also slide 74, which as eightbitking has said shows that the number of cycles didn't determine the survival advantage.
Slide 58 does show that PSA declines were greater and increases less in the 011 group as Cormack said. Factor in the pain data from the second-line trial, which I think deserves more than zero weight, and you have several factors in addition to the 6.9-month OS advantage that suggest 011 may work. Not a sure thing by any means, but probably more likely than my original estimate of 50%.
tnk, as I said above, Cormack has said there were signs of both progression and PSA benefit in the P2 trial. He said this in the Q&A to a Stifel presentation in 2011 which is no longer on the website but for which you can find someone's notes by Googling "OncoGenex $OGXI notes from 9/7/2011 Stifel Nicolaus webcast." Those notes paraphrase him as saying "We did see a difference in docetaxel combo PFS curve...5-6 week median difference at the front end of curve (treated at 10 cycles only, 30 weeks)... perhaps one would see a more durable difference if kept treating. Only one treated patient had progression as "best response" vs about 1/4 of control patients. We also observed "months" difference in PSA response. Also, it is very hard to assess progression in prostate, because will often metastasize to bone (can't do 3D imaging like soft tissue)."
There were also signs, in the second-line P2 trial of 011, that it relieves pain. That trial wasn't randomized but the historical comparisons were strong enough for Teva to agree to fund the SATURN trial to study it, a trial that was aborted because, according to OGXI, they couldn't find enough patients who met the stringent enrollment criteria the FDA insisted on (pain being difficult to measure). While pain palliation isn't an endpoint in SYNERGY, and they may not even be monitoring it, if it's real doctors will notice it and it will give them tangible proof (of the kind that's missing in the case of Provenge) of 011's activity. Of course, that assumes 011 will meet the 0S endpoint and be approved.
One last point on statistical significance in the P2 trial. The trial wasn't powered to find it, and the fact that it came so close (P=.06) was an unexpected byproduct of the large OS advantage of 6.9 months. According to OGXI, the FDA has said if SYNERGY succeeds it won't require an additional trial because it considers the P2 trial supportive. SYNERGY, with 1,000 patients, IS powered to find stat sig if it exists
tnk, you seem to be confusing me with Summer. I gave the odds as 50/50 but Summer said "P2 trial data indicates that there is 94% chance that the drug is active and there is more than 90% chance chance that the current design (of the Synergy trial) would reach statistical significance if OGX-011 is as active as P2 data indicates." Since Summer knows at least 100 times as I do about the science and the statistics I will gladly accept her odds. Over the past few years Summer has posted careful analyses of the 011 P2 results as well as the results of trials for other agents such as abiraterone and MDV-3100, has predicted when the interim results on the MDV-3100 P3 were likely to be available, has forecast that the P3 trial for 011 in lung would be in 2nd-line patients with docetaxel (this was when OGXI was still saying it was planning on a 1st-line trial with a different agent), and has as far as I can remember never been wrong about anything. She's never said that the SYNERGY is risk-free but you have to take her generally favorable assessment (assuming there's no halt for futility) seriously.
BTW, whatever your concerns regarding tumor shrinkage, I'm a little surprised you're also concerned that the 011 P2 trial fell just short of statistical significance on survival. I'm no statistics expert but I know that the 6.9-month advantage or even something quite a bit less than that would easily achieve statistical significance if it holds up in the 1000-patient P3 trial, as opposed to the 82-patient P2 trial. The fact that some other agents may have reached statistical significance in P2 doesn't strike me as relevant.
Sorry, scratch my point (1) -- obviously there was some shrinkage in the P2 trial since there were some partial responses, just not more of them in the 011 group. I shouldn't post while I'm on the phone. My conclusion remains the same though.
It's certainly possible that the 011 P2 results were a statistical fluke but that's what the P3 is designed to determine. The reason the 011+DOC group received more cycles of chemo was that fewer of them progressed while on therapy. That could also be a statistical fluke but it's possible that there's something about the way 011 works that causes less progression -- hence the fact that the 011+DOC group had more stable disease. There could be a mechanism -- different from Provenge's and/or Yervoy's and perhaps still unknown -- that enables 011 to extend survival without producing shrinkage.
The fact that the tumors didn't shrink doesn't bother me quite so much for several reasons: (1) docetaxel was approved on the basis that it extended survival by about 2 months, and the patients in this trial were getting docetaxel as well as 011 yet their tumors didn't shrink; (2) Cormack has (reasonably, IMO) said that it's hard to accurately determine the size of prostate-cancer-related tumors on x-ray, so maybe some of the patients counted as "stable disease" actually had shrinkage; and (3) I vaguely recall reading that tumor shrinkage is viewed as less important than it used to be because when the tumor starts to grow again it often grows more aggressively than before; therefore, stable disease is a better prognosticator, or in other words it's more important that growth slows down or stops for awhile than that the tumor shrinks. Sorry I can't remember the source for point 3.
My only point about DNDN was that it succeeded in P3 (according to the FDA) despite lack of proof of shrinkage. Its post-approval marketing problems are due in part to its high cost, which isn't a factor for 011.
We'll have to wait and see but I'm sticking with my 50/50 odds.
I don't know enough to try to explain how a survival benefit could exist without an observable effect on progression, other than to say that DNDN famously succeeded in showing one and there may be others that don't come to mind.
I'm a little rushed but did want to summarize the progression data that OGXI presented at ASCO in 2009. This is from slide 59 of the pdf I found by Googling "Oncogenex ASCO reception key opinion leader panel":
Based on the stable and progressive disease numbers, and near-equivalance on partial response, you could argue that the 011+DOC group fared better.
BTW, slide 58 presents a waterfall plot of PSA response which as I read it does seem to show a small advantage for the 011 group as Cormack claimed.
Hi Summer, I too have been a bit concerned that the pps decline on fairly heavy volume lately might have been due to a leak of the futility analysis. I assumed though that OGXI would have had to disclose that pretty quickly but it didn't occur to me that Teva might know but OGXI might not. As you say, the time lag couldn't be that long so each day of no-news is reassuring, and Cormack will be speaking at an investor conference Friday so if he gives his usual spiel about the trial's progress that will be too.
I gather you think the two futility analyses will precede the interim analysis for efficacy. For some reason I had assumed the sequence would be futility-efficacy-futility. I take it that's wrong?
BTW, after my last post I looked up the P2 trial for 011 in breast cancer and it seems I may have been a bit harsh in saying 011 had no effect. There were 15 patients who received 011 + docetaxel, of whom 5 had an objective response and 9 had stable disease. Because they had predetermined that they would expand the trial only if at least 6 patients had an OR they didn't expand it. There was no control group that got docetaxel only so it's hard to say exactly what 5 ORs + 9 stable disease means but you can't necessarily conclude that 011 had no effect.