Corporate presentations at healthcare conferences conducted by Wall Street financial firms don’t often reveal much that the Street doesn’t already know. So, I didn’t expect Greg Schiffman, Dendreon’s (NASDAQ:DNDN) chief financial officer, to reveal much in the way of new information when he made a presentation at the Lazard Healthcare Conference on November 15, 2011.
Importantly, however, near the end of his presentation, Schiffman mentioned that in 2013, the company will initiate a trial with Provenge in earlier stage prostate cancer, a trial for men who are hormonally sensitive (not castrate resistant, but where the disease has metastasized). In particular, he stressed that the placebo arm in this trial would be a true placebo ... that is, one that did not include frozen Provenge, or Frovenge, as it is known. As such, he said, the company anticipated a much higher overall survival, or OS, advantage than what had been observed in the pivotal Provenge trials.
Recall that in the case of the pivotal Phase III Provenge trials, the overall survival advantage was a median of 4.1 months (at 50% survival, 25.8 months in the Provenge arm vs. 21.7 months in the placebo arm). The new study not only would be the first time that the company has run a survival-based trial in an earlier stage of the disease, but also one using a pure placebo. Of and by itself, this is extremely important. But let’s examine further the point regarding his statement that the company anticipated a much higher OS advantage than what was observed in the Provenge trials.
What’s this all about? Well, some readers may recall that in the pivotal Phase III Provenge trials performed to obtain FDA approval, Frovenge - or APC8015F, as it also is known - was given to more than two-thirds (80%, according to Schiffman) of the placebo arm. As concluded by Gomella, et al.* in a 2011 ASCO poster presentation, “(NYSE:P)ost-progression treatment with APC8015F may have extended survival of subjects, potentially reducing the magnitude of survival difference observed between sipuleucel-T and controls in randomized controlled trials.” Put another way, the life-extension value of Provenge may have been significantly underestimated by the manner in which the trial was designed. (You can view the entire poster here.**)
By how much was it underestimated? Well, inspection of the figure below (Figure 3 on the poster) shows that the segment of the control arm that did not receive Frovenge had a median survival time of 12.7 months while the segment of the control arm that did receive Frovenge had a median survival time of 23.6 months. This implies that the "true" overall median survival advantage to Provenge should be considerably longer if it is determined in a trial for which the control arm is NOT given Frovenge. In fact, one might reasonably argue, on the basis of these data, that the "true" overall, median survival advantage in treating end stage prostate cancer with Provenge should be something on the order of 12.7 months, or more than one year (25.4-12.7=12.7 months).
This might explain on an intuitive level why end stage PCa sufferers such as Eduardo Garcia lived more than nine years after being treated with Provenge (read an interview with Garcia at the six-year mark here), though such an outcome is certainly not outside the realm of statistical probability.
The above notwithstanding, and to play the Devil’s advocate (if I don’t, I’m sure someone else will), the issue with statistics such as these is that, without a bona fide randomized trial, there may be biases in the data that never can be accounted for by any regression analysis.
For example, it isn't difficult to imagine that the patients who did not receive Frovenge in the pivotal Phase III Provenge trials were sicker to begin with, and this may have been the reason doctors hastened them into more aggressive treatments with Taxotere or other treatments. So, it’s logical to assume that those patients would naturally have died earlier compared to the ones who were healthier and could take Frovenge. In turn, that would mean that, to some extent, the analysis cited above might have been comparing healthier and sicker patients and not necessarily treated and untreated patients.
If this were the case - that is, the study was biased by the treatment given to the sicker patients - then the analysis performed by Gomella, et al. would be suspect. Regardless, I believe the trend observed by the authors is real but the benefits may not be of the magnitude cited. Perhaps there’s another way to estimate what the median survival time for a pure placebo group such as what might be found in a trial such as that Dendreon is about to undertake.
The pivotal Phase III Taxotere trial (here) used to obtain FDA approval for that treatment had median survival times of 16.5 and 18.9 months for the placebo and treatment arms, respectively. If we take the 16.5-month median survival time for placebo patients and compare it with the 25.8-month median survival time for treated patients in Dendreon’s IMPACT trial, we obtain something on the order of a nine-month median life extension difference. This still is a little over twice the 4.1-month median life extension difference upon which Provenge was approved, but clearly shorter than the 12.7 month median life extension difference suggested by the data of Gomella, et al. Regardless, what’s important is the fact that the true median life extension measure for Provenge, measured against a pure placebo, almost certainly is well in excess of the 4.1 months upon which FDA approval was based.
The inescapable conclusion is that the Phase III trial designs almost "did in" the results of the trails and put the approval of Dendreon’s revolutionary immunotherapeutic treatment "at risk." As David Miller of Biotech Stock Research would say, “The trials (almost) failed the drug.” That said, however, a reviewer of this article prior to submission noted that Frovenge was allowed because (1) the early trial design was focused on obtaining approval on progression, not survival (the former being faster and less expensive), and (2) the availability of Frovenge was seen as an inducement for getting patients enrolled quickly. That is, they would get some treatment, regardless of how they were randomized. Remember, when Dendreon began running its Phase III trials, Taxotere had not yet been approved either.
Another interesting piece of information gleaned from Schiffman’s presentation included the fact that the corporation intends to initiate a randomized, Phase II sequencing study involving Provenge and abiraterone acetate (Zytiga; from Johnson & Johnson (NYSE:JNJ)) in 2013. The study, for which the protocol already has been approved, will involve 60 patients and roughly 15 sites. Co-medication has been long recognized as a viable means by which to enhance the treatment regime for various diseases (e.g., AIDS/HIV). It should be no different when it comes to prostate cancer.
The co-medication trial involving Provenge and Zytiga will comprise two parts: One in which the two treatments are given simultaneously and one in which Zytiga is given 60 days following the last infusion of Provenge. Why this protocol?
The co-application of Provenge and Zytiga, and the order in which the two should be given, raises interesting questions. In the asymptomatic/minimally symptomatic mCRPC setting for which Provenge is indicated, the National Comprehensive Cancer Network, or NCCN (NCCN; log-in required), has given Provenge a category 1 rating, the highest evidence level recommendation based upon its three Phase III clinical trials. By contrast, and in Provenge’s "space" - the asymptomatic/minimally symptomatic mCRPC setting - Zytiga has a category 2B rating, the second lowest rating, based upon small single arm Phase II trials. This has not stopped some physicians from prescribing Zytiga off-label to PCa sufferers in an estimated 22% of cases where patients probably should be treated with Provenge first. Current thinking is that prescribing Zytiga first will delay medication with Provenge. The reason for this stems from the belief that if Zytiga is given first, a patient’s system must first be cleansed of prednisone before Provenge can be given, something that can take at least 60 days.
The issue of sequencing Provenge and Zytiga also has been addressed by Eric Small in the Journal of Clinical Oncology. Small writes: “The practical dilemma of the appropriate sequence of use of the two new noncytotoxic agents (sipuleucel-T and abiraterone) is being addressed by trials that are under development. For now, given the broader window of applicability of abiraterone and the longer time required to develop an immune response with sipuleucel-T, if both agents are to be used, it seems reasonable to administer sipuleucel-T first with abiraterone after additional disease progression. Biomarkers to help define the optimal use of immunotherapy are needed.”
Given the above, we now see why one segment of Dendreon’s new co-medication trial will delay treatment with Zytiga by two months; that is, for the purpose of determining the impact of the time delay on the efficacy of the combined treatment.
Regardless, given its NCCN No. 1 rating and the recommendation of the Journal of Clinical Oncology, Provenge appears to be the "foundation of care" upon which oncologists and urologists can confidently build a "best practice" treatment regime in the asymptomatic/minimally symptomatic mCRPC setting.
Dendreon jumped on Wednesday, November 16 - in fact, the stock was halted for a brief period - after a leader in the Large Urology Group Practice Association, a 10-year-old organization that represent nearly 25% of U.S. urologists, said in a Goldman Sachs conference call that his practice has already placed seven patients on Provenge and has identified 100-200 eligible patients for the drug as well. Further, this practitioner said it was his belief there could be a logarithmic (perhaps "exponential" is a more familiar term (Cohen)) increase in uptake of the drug in 2012 and that there was a need for urologists to embrace Provenge to optimize patient care as they could otherwise risk losing the patients to others who might provide the treatment. The stock immediately jumped to more than $8, was halted, and upon reopening, drifted down to close at $7.51 Wednesday afternoon. The stock closed above $8 on Thursday.
An alternative explanation for the sharp jump in share price on Wednesday, interestingly enough, appeared in the comments section that can be found at the end of an article I wrote earlier this month (Dendreon, Merck's Hepatitis C Drug: Future Prospects Look Encouraging). The comment of interest, written by a poster using the pen name "slowtime," posits that the run to $8+ in the stock’s price on Wednesday was instigated by a major buyer who triggered a short squeeze of sorts. Whatever the reason, it once again shows that DNDN is volatile, and surprises are more the rule than the exception.
The daily chart, courtesy StockChart.com, is below. It shows the issue rising in relative strength and continuing to climb back toward the 50-day moving average. The weekly chart is more encouraging, showing the stock coming out of its oversold condition, with the MACD about to turn positive.
Finally, as noted by Adam Feuerstein, several major hedge funds have reduced or sold their entire positions in Dendreon, including Cohen's SAC Capital Advisors, which dumped 8.3 million shares during the third quarter. However, noted Feuerstein, "Baker Brothers Advisors gobbled up 2.6 million shares during the September quarter. Camber Capital Management also initiated a new Dendreon position with a 2 million-share purchase." Other hedge fund buyers were cited as well.
*L. G. Gomella, C. Nabhan, J. B. Whitmore, M. W. Frohlich, D. J. George; The Kimmel Cancer Center at Jefferson, Philadelphia, PA; Advocate Lutheran General Hospital and Oncology Specialists, S. C., Park Ridge, IL; Dendreon Corporation, Seattle, WA; Duke Cancer Institute, Durham, NC
**Reproduced with the permission of Dr. L. G. Gomella, personal communication, November 17, 2011
Additional disclosure: I am long DNDN and will not alter my position within 72 hours of the time of publication of this article.