A well thought-out question was asked in the comment section of my last article. It is one that I have pondered long and hard over myself, which led me to research it extensively. The question is, "Where oh where has NWBO's DMC recommendation gone?"
First, it should be noted there is no aggregate efficacy or safety data available to the sponsor of a clinical trial if that trial is blinded.
The only instance in which a sponsor (in this case NWBO) of a blinded trial is unblinded to data is when a DMC has noted a stopping boundary being crossed (futility or efficacy) before the conclusion of a trial, and thus recommends to the sponsor that the trial be stopped and why (although they may not always recommend it). And if the sponsor agrees with the assessment and stops the trial, then and only then are they unblinded to data.
Otherwise, there is no available data to the sponsor, other than what is discussed in "open meetings." These include basic enrollment and treatment information, how each site is performing, etc. To allow a sponsor to view aggregate blinded data during a trial would introduce a type I error (bias) and would seriously jeopardize the validity of the results going forward.
So when someone says "interim data" regarding a blinded trial, they are referring to data the independent DMC views only. If they are implying it is something the sponsor of a blinded trial has ready access to, it is being said out of ignorance or worse (and in the above cases, I refer to the "sponsor" as executives with high public visibility and/or those in continual interaction with patients. Not select employees of the company fire-walled for such purposes).
But let's discuss the apparently missing DMC recommendation based on efficacy data from the first interim analysis of NWBO's Phase III trial.
From Guidance for Clinical Trial Sponsors:
When there is a steering committee, the sponsor may elect to have the DMC communicate with this committee rather than directly with the sponsor.
There are two CROs managing NWBO's Phase III trial (also called "steering committees"). One in the US, and one in Europe. They handle oversight and management of the clinical trial sites and collection of data. They also can serve, and often do serve, as mediators between the sponsor and the DMC, as the FDA Guidance above states. With smaller companies like Northwest Bio, it is almost always the case that a steering committee is appointed to discuss matters directly with the DMC, and not to involve the sponsor directly. To continue:
Interactions between the steering committee and the DMC consist primarily of discussions during "open sessions" (see Section 4.3) of DMC meetings and the communication of recommendations following each DMC review of the trial. More extensive interactions might occur when early termination is being considered.
The above guidance points out: "More extensive interactions (between the DMC and CRO) might occur when early termination is being considered." That is, when a DMC has viewed aggregate data, and is considering a recommendation to halt the trial early (for futility or efficacy), discussions will become more extensive. This will involve unblinding key members of the CRO to aggregate data, to make a determination on how to proceed, and will include extenuating circumstances (to halt a trial affects many variables). To continue:
In almost all cases, a DMC is advisory to the sponsor [or advisory to the steering committee who is advisory to the sponsor]; the sponsor decides whether to accept recommendations to discontinue a trial. FDA will rarely, if ever, tell a sponsor which decision to make. [brackets mine]
In most cases, and for flexibility purposes, a sponsor will design a trial with nonbinding futility and efficacy stopping boundaries. That means the trial could cross a stopping boundary for efficacy, thus resulting in the DMC recommending to the CRO (or at least opening discussions with them concerning it) to stop the trial. But due to the nature of the boundary (nonbinding), the CRO (and as per protocol laid out to the sponsor's wishes before the trial began) may choose, because of various other factors, to continue the trial instead, keeping the DMC recommendation undisclosed to the sponsor (who would be affected by such for obvious reasons, which could jeopardize the trial by introducing bias).
The same could be said of nonbinding futility recommendations. Though, in an early interim analysis of a clinical trial, if the boundary is crossed for futility, the DMC will nearly always recommend a continue anyway (especially if the treatment is an adjunct to SOC), as the data is usually too immature to make that broad of a declaration. And especially for life-threatening diseases, which is always taken into account. This can even be true in cases where OS is slightly worse in the treatment group than control (as in the 2004 corticosteroid CRASH trial).
But what will always cause a DMC to recommend a halt after an IA is toxicity. If the treatment group performs appreciably worse than the placebo group, or many toxic side effects are noted, they will recommend the trial be stopped early. This is a binding stopping boundary.
The other thing to consider is effect size. In an adjunct therapy trial (treatment in addition to SOC vs. SOC), if the treatment is non-toxic, the DMC will usually recommend a continue at the IA, even if it appears futile--unless the sponsor has given them instructions not to (in order to reallocate funds to other drugs in their development pipeline, for instance). Therefore, a futility halt is rare in these cases (and would be very rare for a company like NWBO testing DCVax-L. They would likely express the desire that the trial continue at all costs, unless deemed unsafe, even if appearing futile. And the DMC would oblige their wishes and recommend it--as a continue, if this were the case--again, so long as it's deemed safe). Even with poor or no real effect being seen, the DMC would normally issue a continue in trials like these until the second interim. But here, they haven't.
Getting back to effect size, little or no advantage of "safe treatment + SOC" vs. "SOC" is not harming patients in any way. Therefore, it is ethical to continue and see if the trend changes. The DMC, upon seeing this, would simply recommend "continue" based on efficacy data, with little deliberation (again, unless instructed otherwise by the sponsor). If, however, the effect size is large for "safe treatment + SOC" vs. "SOC," then that is a different thing. Now, it suddenly appears unethical to prevent this treatment from the rest of the patients and the rest of the world. Marked difference in dilemma.
(By the way, one of my main sources in this discussion is the book "Designs for Clinical Trials, Perspectives on Current Issues" by David Harrington. I highly recommend it).
To finish the relevant Guidance quote:
For trials that may be terminated early because a substantial benefit has been observed, however, consideration may still need to be given to the adequacy of data with regard to other issues such as safety, duration of benefit, outcomes in important subgroups and important secondary endpoints.
That was "safety, duration of benefit, outcomes in important subgroups and important secondary endpoints." These are potential causes for additional consideration, even when "substantial benefit" that warrants an early halt to a trial has been seen. This means that if there are no unresolved issues pertaining to "safety, duration of benefit, outcomes in important subgroups and important secondary endpoints," AND there is substantial benefit, a trial may be halted early. Otherwise, resolution needs to be brought to any or all of these topics first.
Regarding safety, the Company stated,
In ten years of clinical experience, there have only been a couple serious adverse events, and we believe these were due to the patient's underlying brain cancer, rather than the DCVax treatment. The broad and rapidly growing body of scientific literature about dendritic cells is consistent with the DCVax clinical experience.
Safety seems "safe" to rule out.
"Duration of benefit" is the primary factor in determining if ORR (objective response rate), which measures tumor response, ultimately leads to clinical benefit. As an aside, this marks one noticeable difference between chemotherapy and immunotherapy. Often, ORR is more dramatic with chemo, yet overall survival has been mixed. Tumor recurrence and metastases after chemo are often more aggressive than before. Immunotherapy is showing in many cases to have moderate ORRs (less of a chemical obliteration of humanity), but more substantial OS than chemos. At any rate, this would not be a factor for extensive interaction between the DMC and CRO in this particular trial (full resection prior to treatment).
"Outcomes in important subgroups" refers to the pseudoprogressive subgroup, mentioned on clinicaltrials.gov, and would broach a whole other long discussion. Suffice it to say that although it could create extensive interactions (between the CRO and DMC), immature results for this subgroup alone (set to enroll 72 total) would probably not hold back an entire trial from being halted early for efficacy. This is strictly a matter of opinion, however. And in my opinion, data on this subpopulation could be ascertained via open label studies subsequent to a halt, and such data could then be the basis for later label extension of DCVax-L. But if they WERE to hold up the entire trial to enable data on this subgroup to mature, it would most likely go to its designed end anyway (as it stands, 4 months away). The same could be said of "important secondary endpoints," namely, OS.
But can this account for the now almost 6 months without any word from the Company other than "the DMC recommendation based on efficacy data remains outstanding?" I think so. The DMC may even have recommended a halt for efficacy to executive members of the CRO, and they, because of protocol that may have insisted a halt not occur before certain provisions are reached, are not releasing it to the Company until these occur.
Another possibility is that the DMC determined that an unscheduled interim look, or two, would give them a view of more mature data that could be used as the basis for a halt for efficacy, without having to wait until the second interim was triggered to do so (or, this could have been discussed and decided between the CRO and DMC--and possibly the FDA/EMA).
Something similar to this was enacted by the DMC in a sunitinib trial. The interim look was scheduled to occur after 130 PFS events, but they unblinded efficacy data at 20, 50, and 73 PFS events instead. These correlated with safety analyses, except for the last one, which was all its own:
6.3.1 Early Study Termination
The DMC met three times to review safety data (May 2008; November 2008; and February 2009), at which times, PFS data was also reviewed (based on 20, 50, and 73 PFS events, respectively). These DMC efficacy reviews were not pre-specified and no alpha was allocated for them in the Statistical Analysis Plan (NYSE:SAP). At the November 2008 meeting, when 50 PFS events had been observed, the DMC stated that "given the marked PFS difference between arms it would be reasonable to consider closure at an earlier time," and decided to meet again in 3 months, instead of 6 months.
In February 2009, at its third meeting, the DMC recommended closure of Study A6181111 based on its review of preliminary safety and efficacy data after only 73 PFS events (28.1% of planned events) had been observed. This recommendation was based on an observed median PFS of 11.1 months on sunitinib versus 5.5 months on placebo with corresponding hazard ratio of 0.397, 95% confidence interval (0.243,0.649), and 2-sided unstratified log-rank test p-value 0.0002. However, no formal alpha spending and statistical stopping boundaries were implemented. The DMC noted that if the study continued as planned and the interim analysis at 130 events was conducted, there would have been a 91% chance (conditional power) of stopping the study assuming that the true HR was 0.649 (upper limit of the 95% CI of the observed HR at 73 events). Thus, the DMC claimed that the study had met its primary endpoint in demonstrating a significant PFS advantage for sunitinib.
The sponsor agreed with the DMC and notified all investigators in March 2009 that the study would be closed and all patients (regardless of whether they had PD or not) should be offered open-label sunitinib on one of two extension studies (A6181078 or A6181114). The final PFS analysis was based on all PFS events reported by April 15, 2009 (the data cutoff date), at which time, there were 171 patients enrolled and 81 PFS events (31.2% of planned events) observed.
So we see a case where a DMC superseded protocol based on "trends in efficacy data," and made the decision to increase interim looks--especially with their final of the three decisions to unblind data 3 months after the second, where they found 73 events had occurred and the same PFS trends. In order to do this, a DMC is supposed to account for additional alpha spend, but in a well-powered trial, that would be insignificant.
Hypothetically speaking, the DMC may have decided to forgo a continue recommendation based on efficacy data in NWBO's Phase III trial. Either because the efficacy stopping boundary was close to being crossed at the first interim look (and so, calling a continue until the second interim would delay what they saw trending to occur before the second interim), or because the pseudoprogression subgroup's data or OS data was too immature in their own right, or some combination of these. They may have decided to add in unplanned interim looks perhaps in March, and then again in June, rather than wait for 88 events. This would be a clear ethical decision. It may also mean a halt could potentially occur at any moment.
In summary, these are two hypotheses:
The DMC is putting off making an offical recommendation until they conduct one or two more unplanned interim analyses, by which time they deduce they will have crossed the efficacy stopping boundary set for this trial. At that time, they may recommend a halt for efficacy.
The DMC recommended a halt for efficacy to the CRO, who, because of protocol, acted in the sponsor's stead, choosing to continue the trial instead, until certain provisions are in place, or data has matured. Thus, no recommendation made its way to the Company. Once these conditions are met, then they may reconvene with the DMC and remit a recommendation to halt for efficacy to the Company at that time.
The third hypothesis involves Accelerated Approval.
A recent FDA guidance states a process by which a company would file an NDA or BLA for Accelerated Approval (AA) using an interim analysis as the basis, while the trial continues to satisfy the post-AA confirmatory trial requirement.
The FDA wants to see a confirmatory trial in progress when granting AA. Why? So many of the companies granted AA drag their feet thus for many years, while taking full advantage of their newly acquired marketing capabilities. The FDA is changing the way this works. It appears they are wont to demand a confirmatory trial (Phase III) in progress before they consider AA. One way to do this is to use the interim data of a larger Phase III trial as the basis for filing a BLA or NDA for AA, while data collects as the trial continues to satisfy the post-AA confirmatory trial requirement:
When it is possible to use a later effect in a trial to verify the effect seen earlier in the same trial that supported accelerated approval, the same clinical trial(s) can be used to support accelerated approval and verify and describe the clinical benefit. In this case, the protocol and the statistical analysis plan should clearly account for an analysis of the surrogate endpoint data to provide support for accelerated approval, with continuation of the randomized trial(s) to obtain data on the clinical endpoint that will be the basis for verifying the clinical benefit. When the same trial is used to support accelerated approval and verify clinical benefit, the data to verify the clinical benefit may be, in some cases, nearly complete by the time of accelerated approval.
This is the path that was followed with Eloxatin:
In 2002, oxaliplatin (Eloxatin) was granted accelerated approval for use in combination with 5-FU/LV on the basis of data from a randomized, three-arm study of oxaliplatin plus infusional 5-FU/LV versus 5-FU/LV alone versus single-agent oxaliplatin in patients with advanced colorectal cancer refractory to first-line treatment with irinotecan and bolus 5-FU/LV (21). The study enrolled 821 patients and had a preplanned interim analysis after 450 patients were enrolled. At the time of this interim analysis, a response rate of 9% (13 of 152) was observed in the combination arm compared with a response rate of 0% (0 of 151) in the 5-FU/LV arm (P = .002). Patients in the combination arm had an approximately 2-month increase in median time-to-progression (4.6 months, 95% CI = 4.2 to 6.1 months) compared with patients in the 5-FU/LV-alone arm (2.7 months, 95% CI = 1.8 to 3.0 months).
The study was not halted early, and was granted AA after data from the interim analysis was used as the basis for their NDA application. The time in which this occurred saw the trial continue, accruing data to verify clinical benefit.
Third hypothesis: The DMC unblinded executive members of the CRO to aggregate data for the purpose of them filing a BLA for AA, according to this trial's protocol. Thus, no continue recommendation would follow.
Now considering German Hospital Exemption:
According to the FDA-PEI Confidentiality Commitment, the FDA is "authorized" to share "non-public, pre-decisional information" with the PEI, as part of "cooperative regulatory activities." Seeing as how they are both regulatory bodies overseeing the same clinical trial, it seems quite germane.
The exact timing of the Hospital Exemption approval also seemed uncanny. Ten weeks into the interim review. Hard to just ignore that detail. Hard also to ignore the 5-year term given, when HE can be granted for less. Why not 1 year, seeing as how the trial would have top line results before then? One hypothesis is they were unblinded to interim data, along with the FDA.
But how would the FDA be unblinded to interim data, to begin with?
(from FDA Guidance):
We recommend that sponsors of trials that could potentially be terminated early for efficacy reasons discuss these issues with FDA prior to implementing the trial, when the statistical monitoring plan and early stopping boundaries are being developed. In these settings, consultation with FDA may provide the sponsor with important information regarding the regulatory and scientific implications of a decision and may lead to better decisions.
Sponsors are encouraged to revisit these issues with FDA when considering DMC recommendations for early termination if new issues have arisen and/or if the regulatory implications of early termination were not adequately clarified at the outset of the trial. [emphasis mine]
The FDA may be unblinded to interim data for clarification purposes (as above), or by way of data used to apply for Fast Track designation, which may be sought "at any point during the drug development process," including following an interim review. The FDA responds in under 60 days to such requests (Fast Track also enables "rolling review," which can speed up the BLA or NDA filing and response time). The Company said it was planning to petition the FDA for Fast Track, but had not done so yet (in 2013).
Moving away from hypotheses for the moment, the DMC (or in this case more aptly called the DSMB) of course has declared the trial is safe to continue, which they delivered to the CRO, and then the CRO to the Company. That is a separate analysis, often conducted alongside efficacy analysis. Which leads me to an aside...
Why do you suppose Linda Powers, CEO of Northwest Biotherapeutics, made the decision to inform shareholders via PR that the recommendation based on efficacy data was still "pending," thenceforth revised to "outstanding?" It was not required of her or the Company to do so. They could simply have stated, "the DSMB has reviewed interim data from our Ph III GBM trial and deemed it safe to continue." There was no need to distinguish between efficacy and safety data recommendations. She was the one to bring this all up, after all. It stunned investors, who joined in chorus: "What do you mean, pending??"
Was this a faux pas? Or a seemingly clever tactic that may have backfired?
Without being of the same mind as the CEO, it is impossible to know, but I will say that it doesn't take very much investigation of the literature (Guidance) and case studies (such as in the book listed above) to reveal that such a scenario, for a trial like NWBO's Phase III, leans much more heavily towards it being due to positive rather than negative developments/data. Again, whether or not this was foresight on the Company's part, I cannot say. But it appears that way. It was intentionally said, yet not required.
The downside to putting out information that you hope is interpreted "correctly," is, of course, when it gets interpreted incorrectly instead. In biotech, every investor is afraid of his own shadow it seems, often seeing things that have no substance and making them into decisive events. It is always the lack of understanding of a thing that causes us to make it into something that it isn't. I see no difference here.
The CEO would perhaps have better served investors by leaving them in the dark on something so easily misconstrued. But I, for one, am glad she didn't.
After researching the topic exhaustively, I can honestly say that the "outstanding" nature of the DMC recommendation only increases my confidence in the Company's sponsored technology, DCVax, developed out of the prestigious UCLA Health Center, and the probability, all things considered, that it is the result of positive rather than negative efficacy data.
So there you have it: 4 hypotheses on the whereabouts of said DMC efficacy recommendation. But why not add a 5th?
The final hypothesis is that the Company, even though trying hard to keep enrollment figures undisclosed, did, in fact, give away enough timeline data to enable one to figure out that an effective current PFS range is being seen in the trial, even when conservatively figured, with reasonable accuracy. I will quote someone more knowledgeable than me on the subject:
DCVax-L Model Positive on Primary Endpoint.
Disclosure: I am long this stock; I have no inside information; I am a clinical researcher.
Hypothesis 1: The Control Patients in the DCVax-L GBM trial should relapse at a rate similar to what was shown in the Stupp trial. The Stupp trial is the best GBM data available and represents the standard of care for GBM patients. In order to compare the Stupp Trial to the current DCVax-L trial control patients, however, one must review the eligibility requirements for each trial. The DCVax-L Trial only allows patients to enter the trial, if "surgical resection with the intent of a gross total or near gross total resection" can be completed. The Stupp Trial contained 17% patient who just had a biopsy only and another 44% of the patients who only had a partial resection. Extent of surgical resection is a major prognostic factor for these tumors and historical reviews have shown the degree which the various surgical procedures can change survival rates. The Median progression free survival (NYSE:PFS) for all patients in the Stupp trial was 6.9 months. When one applies algebra proportions to correct for the differences in prognosis between the trials in terms of surgical extent, the likely median PFS for the control patients in the DCVax-L trial should be around 9 months.
Hypothesis 2: Given that we know the current number of events in the total DCVax-L trial at this time (first interim analysis) is 67, then one can model the control arm PFS to be similar to the corrected Stupp trial results and thus obtain an estimate for the probable number of events required in the control arm of the DCVax-L trial to give a 9 month median PFS. When one does this with a Kaplan Meier plot, one obtains a result of a minimum of 30 events in the control arm with the total number of patients in the control arm of 57. Then one can deduce that the number of events in the experimental arm for the DCVax-L trial would be 67-30 or 37 events.
Hypothesis 3: Using the derived 37 events in the experimental arm, one can model the experimental arm to be similar to the Stupp trial results but, in this case only allow for just 37 events out of 114 total patients in the DCVax-L experimental arm. When one does this with a Kaplan Meier plot, the median PFS for the experimental arm returns > 17.5 month median PFS in the DCVax-L experimental arm. When one applies log-rank and Wilcoxon significance testing to the two curves one obtains a significant result of p=<.0001 and p =.00015 respectively.
My best educated guess is that the DCVax-L trial is likely going to return a positive result on the primary end point given the information that we have currently available.
My best educated guess is he's right.
Disclosure: I am long NWBO. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.