Dendreon (Nasdaq: DNDN) April 28, 2009 IMPACT final phase 3 study results conference call transcript with analyst question and answer session:
MITCHELL GOLD (Dendreon CEO and President) Prepared Remarks
GOLD: As you know, two weeks ago we were very pleased to announce that our pivotal phase 3 IMPACT study met its pre-specified primary endpoint showing that Provenge significantly prolonged overall survival in men with advanced prostate cancer.
Earlier today Dr. David Penson, one of the study investigators, presented results of the final analysis in a late-breaking clinical trial session at the American Urological Association’s annual meeting here in Chicago showing that Provenge extended median survival by 4.1 months and increased 3 year survival by 38% compared to placebo.
Dr. Mark Frohlich will take us through the data in more detail.
FROHLICH: Thanks Mitchell. Before we review the results in further detail, I would like to review the IMPACT study design and plan for statistical analysis. IMPACT, also known as D9902b, was a randomized, multi-center, double blind, placebo controlled phase 3 study that enrolled 512 men with metastatic androgen-independent prostate cancer.
The primary endpoint of the study was overall survival. Overall survival is the gold standard endpoint in oncology clinical trials because it is the least biased, least variable, and most clinically meaningful assessment.
The study was stratified (slide 4) for bisphosphonate use, primary Gleason score, and the number of bone metastases. Hazard ratios and p-values were calculated from a Cox model adjusted for PSA and LDH. P-values were two-sided and the log-rank analysis was performed as a sensitivity analysis.
One interim analysis was performed. To meet the primary endpoint the final data had to meet a pre-specified level of statistical significance with a p-value less than 0.043. The major eligibility criteria were (slide 5) metastatic, castrate-resistant (also known as metastatic) prostate cancer, life expectancy of at least 6 months, serum PSA of 5 or greater, castrate level of testosterone, and adequate hematologic, renal, and liver functions, and negative serology for HIV and hepatitis.
Baseline disease characteristics were well-balanced between the treatment arms as shown on slide 6, in terms of age, race, performance status, Gleason score, localization of disease, the number of bone metastases, bisphosphonate use, and prior docetaxel use.
Baseline laboratory values were also well balanced (slide 7) including PSA, prostatic acid phosphatase, alkaline phosphatase, hemoglobin, LDH, and white cell count.
Overall survival was the primary endpoint of the study (slide 8). Survival follow-up at the time of data cut-off was obtained in 99% of patients. Only 6 patients were lost to survival follow-up.
The survival data are very mature with a median follow-up time of close to 3 years. The Kaplan-Meier curve plots the percentage of people still alive in the trial on the vertical (y-axis) as a percentage versus time-from-randomization on the horizontal (x-axis) in months.
The Kaplan-Meier curves separate early and then remain separated for more than four years following randomization. The hazard ratio was 0.775, indicating a 22.5% reduction in the risk of death in the treatment arm. The p-value was 0.032, clearing exceeding the pre-specified level of statistical significance of less than 0.043.
The median survival difference was 4.1 months (slide 9). The median survival in the Sipuleucel-T treatment arm was 25.8 months versus 21.7 months in the placebo arm. The median survival of the treatment arm exceeds that reported in prior randomized trials of men with asymptomatic or minimally symptomatic androgen-independent prostate cancer.
The percentage of patients alive at three years by Kaplan-Meier estimates was 31.7% (for Provenge-treated patients) compared to 23.0% (for placebo), a 38% relative increase in three year overall survival.
We performed several analyses to test the robustness of these survival results (slide 10). To rule out the possibility that the survival benefit was being driven by a particular group of patients, we assessed the treatment effect in multiple population subsets.
As examples, forced slots are shown for ten baseline factors. This includes factors such as laboratory parameters like PSA and LDH and the number of bone metastases. All sub-populations demonstrated a positive treatment effect with a hazard ratio of less than one.
In additional sensitivity analyses (slide 11), the results were consistent with the primary statistical model. Using an unadjusted Cox model and a log-rank test, the hazard ratio was 0.766 with a p-value of 0.023. Adjusting for the use and timing of docetaxel initiation in a time-dependent covariate model, the hazard ratio was 0.763 with a p-value of 0.036. Therefore there was no evidence to suggest that post-progression treatment with chemotherapy affects the interpretation of the survival results.
Furthermore, the hazard ratio for prostate cancer-specific survival was 0.772 with a p-value of 0.036, indicating that the survival results cannot be explained by non-prostate cancer-related deaths.
The secondary endpoint of the trial was time to objective disease progression (slide 12). Consistent with the previous phase 3 trials of Sipuleucel-T in this patient population, we were unable to demonstrate a statistically significant difference in this endpoint. This result was consistent with the data from other randomized studies in advanced prostate cancer where progression has proven difficult to measure reliably and has not correlated with overall survival.
The safety profile of Sipuleucel-T was found to be consistent with that reported in previous studies (slide 13). The most common adverse events observed at a higher frequency in the treatment arm were chills, pyrexia (fever), and headache. The majority of these events occurred within a day of infusion and resolved within one to two days. Most events were mild to moderate in severity with very few grade 3 or 4 events.
The percentage of patients that experienced serious adverse events was comparable between the treatment arms (slide 14) at 24.0% (for Provenge) and 23.8% (for placebo). The serious adverse events experienced in four-or-more patients are listed and reveal no clinically significant differences between the treatment arms.
Of note, only 1.2% of Sipuleucel-T patients were unable to receive all three infusions due to treatment-related adverse events.
The cumulative evidence on the efficacy of Sipuleucel-T is based on three randomized, multi-center, double blind, placebo controlled trials with consistent results (slide 15). The median survival benefits observed have been 4.5 months in D9901, 3.3 months in D9902A, and 4.1 months in the IMPACT study.
The percentage of patients alive at 36 months by Kaplan-Meier method in the treatment arms was 34%, 32%, and 32% respectively. Again, remarkably consistent across these studies.
An integrated analysis of these three studies, comprising a total of 737 patients reveals a hazard ratio of 0.735, a 26.5% reduction in the risk of death with a p-value of less than 0.001.
In summary (slide 16), Sipuleucel-T is the first active immunotherapy to demonstrate an improvement in overall survival for advanced prostate cancer. The treatment effect is clinically meaningful, robust, and consistent across sub-populations.
Sipuleucel-T appears to have a favorable benefit-to-risk profile, a short duration of therapy (only a month), and most importantly, has the potential to create a new treatment paradigm in oncology. Based on the strength of these data and following discussions with the FDA, patients in the placebo arm of the study will now be offered access to Sipuleucel-T and other patients may be eligible for one of our ongoing studies. More information is available on clinicaltrials.gov.
GOLD: (10:44 mark) Our focus now turns toward the regulatory pathway to bring Provenge to patients who currently have few appealing treatment options. As you may know, IMPACT was conducted under a Special Protocol Assessment (SPA) with the U.S. Food and Drug Administration. Because the data successfully exceeded the pre-specified level of statistical significance and met the criteria outlined in our SPA agreement, we plan to file an amendment to our existing BLA (Biologic License Application) in the fourth quarter of this year to seek licensure of Provenge in the United States.
In addition, we are preparing our manufacturing, sales, and marketing operations to ensure we are able to launch Provenge successfully in the United States. We plan to update you further on our commercialization plans and progress in an upcoming analyst day this summer.
The results of the IMPACT trial are unambiguous, (therefore) making this a momentous occasion for prostate cancer patients, the cancer community, and certainly Dendreon. We could not have made the significant clinical advancement possible without so many others. We want to thank the more than 1,000 prostate cancer patients and their families who are willing to participate in our clinical trials. In addition, we are grateful to the clinical investigators and their teams that helped conduct our trials over the past decade.
For years scientists have been trying to harness the immune system to fight cancer. Our comprehensive clinical development program for Provenge has been rigorous and we are extremely proud to be pioneering cancer research that has been shown to offer patients prolonged survival with a favorable benefit-to-risk profile.
As you know, Provenge employs our novel proprietary antigen delivery cassette technology that may be leveraged to develop similar products to treat breast, colon, bladder, kidney, and multiple other types of cancer. The IMPACT results validate this platform technology and the potential for us to extend its benefits across other cancer types.
Today Dendreon, our employees, and the scientific community have made a major step forward in demonstrating that the immune system can be harnessed in the fight against cancer. We hope and sincerely believe that this is just the beginning of a new era in medicine.
(Q&A instructions….)
Questions from JOEL SENDEK of Lazard Capital Markets (13:31 mark)
SENDEK: Given how impressive the data is, especially the clean safety profile, it seems to me that it shouldn’t be difficult for the FDA to move this forward quickly so I’m wondering why you couldn’t just file (for approval) sooner than the fourth quarter? And then the next question is, how quickly can you try and get trials up and running in other cancers?
GOLD: First off Joel, this (IMPACT) is a study that has been going on for almost six years and it’s a fairly large study in advanced prostate cancer. The timing of submitting this application in the fourth quarter has nothing to do with the FDA, it has to do with the company putting together the best amendment that it can for its existing license application.
As we talked about in our call when we announced the top-line results, the amendment to the application will be what’s called a type 2 resubmission. The FDA will have six months from the time we amend the application to review the data that we resubmit.
Regarding the question on when do we have the opportunity to bring other trials on-line, we are pretty excited about the platform overall and as we said in our prepared comments, we believe this data validate our approach to using the immune system to fight cancer. We have two other studies ongoing in different stages of prostate cancer. The NeoACT study in men that are having radical prostatectomy and the other one in a very similar patient population to what we studied in IMPACT.
We are going to be evaluating how we move the program forward and that we plan to update you on at an upcoming analyst day this summer.
SENDEK: Just two quick follow-ups, can you give us any more clarity on how comfortable you are that it’s (the type 2 resubmission) is going to be the fourth quarter as opposed to December 31?
GOLD: You know we aren’t going to get any more granular than that but historically the company has been pretty conservative in its guidance.
SENDEK: Is there any potential for patients to get more than three infusions? Do you have any history of that being associated with any side effects and why did you choose three infusions to begin with?
GOLD: To refresh your memory we had a study called P-11, which was a study in early stage prostate cancer (patients) that had serological recurrence after radical prostatectomy. The initial treatment protocol was basically identical to what we saw in 9902B in that they saw three infusions over one month.
The interesting part of that study is that we actually followed (the patients) out for a period of time, on average it was just over a year, but in some men it was over five years. We took a peak at their immune systems to see what was going on and what we saw there was that there was a persistent immune response against the immunizing antigen, on average at about a year out but which persisted up to five years.
We reboosted a portion of those patients (with Provenge) and what we saw was that we were able to get a significant impact, an increase, in their immune response against the immunizing antigen at that time. So there’s no reason to believe that you couldn’t give additional boosters later on but that was not part of the IMPACT study.
FROHLICH: At the time we designed this study back in 1999 we had looked at protocols in our phase 1/2 program that were looking at giving three doses of Provenge and then a boost at six months.
In our assessment of those data at that time, we saw that the three doses given at 2 week intervals was important to get all patients to a point where they were responding to the therapy but the immune response that we measured was persistent and at six months we couldn’t actually see that the boosting we did at that point provided us any additional advantage so we when designed a phase 3 protocol we stuck with the three doses at 2 week intervals as the protocol that we would follow.
Questions from RENI BENJAMIN of Rodman & Renshaw (17:30 mark)
BENJAMIN: A couple of quick questions starting off with the side effect profile. I believe it was Mark that mentioned that the side effects have no clinical significance. Can you comment were any of them were statistically significant? I would imagine chills and fever and headache were statistically significant. Were any of the other ones statistically significant?
FROHLICH: So the slide that you’ll see on the webcast (slide 13), we list those side effects that were seen more frequently in the treatment arm at the p-value of less than the .05 level. Those were chills, fever, headache, hyperhydrosis (which is just sweating) and… hypertension.
BENJAMIN: Regarding the median survival in the statistical analysis, it looks like in your slide that you used an unadjusted Cox model and log-rank analysis in the first two trials but I noticed in this trial the log-rank analysis on the earlier slides. The unadjusted log-rank analysis was 0.023. Is the unadjusted Cox analysis as well statistically significant and do we have that number?
GOLD: The unadjusted Cox is what you use to generate the hazard ratio that we show there and the p-value is what is derived from the log rank test and we did that as a sensitivity analysis in the IMPACT study (slide 11) but the primary methodology as Mark said in his comments was the Cox-adjusted model where we achieved a p-value of 0.032 and a hazard ratio of 0.775. We did do an unadjusted Cox as well as a long rank test as part of a sensitivity analysis plan.
BENJAMIN: And both of those hit as well? (Please see slide 11 for god’s sake, Ren)
GOLD: Yes.
Questions from AARON RINGS of Wachovia Securities (19:52 mark)
RINGS: I was just wondering if you could discuss the median survival results of the placebo patients that didn’t crossover to Provenge and went directly on to Taxotere?
GOLD: Whenever you look at those types of subgroups you have to be careful that you’re not introducing bias into the equation. A good example of that was a subgroup analysis that we did many years back with a Gleason score subgroup and we thought that patients with a Gleason score subgroup of 7 or below performed better…and that turned out to be a red herring so subgroup analysis as we all know, as perils associated with it.
That being said, we believe that there is a benefit to the salvage arm and as Dr. Kantoff said in his comments today in the media, the true treatment effect of Provenge may in fact be greater than what we’ve demonstrated in this clinical study given that we did allow crossover to cryo-preserved version of the product.
RINGS: Can you just remind us again what we should expect from a partnering standpoint? Are you leaning more towards a global partnership now based on the totality of the data, opportunity, and interest? Are you still looking at maintaining rights in the U.S. and out-licensing ex-U.S.?
GOLD: We believe the data support the utility of Provenge. The proposed label for Provenge would have a survival benefit associated with it and we believe it has a very favorable benefit-to-risk profile. We think those characteristics of this product make it a very appealing commercial opportunity.
Our goal is to keep that product ourselves in the United States and to seek a commercialization partner outside the United States. I think it’s fair to say that this data has attracted a lot of interest but we’re not going to comment on anything else beside that.
Questions from DAVID MILLER of Biotech Stock Research (21:49 mark)
MILLER: You mentioned that you talked to the FDA and that patients who were enrolled into the control arm are going to get crossover. Are they going to crossover to Provenge or to “Frovenge” (frozen version of Provenge)?
GOLD: The patients in the placebo arm will actually crossover and get the genuine bonafide version of the product. They will get Provenge.
MILLER: So those patients who had crossed over to “Frovenge” (frozen version of Provenge) originally, are they going to be able to go back and get Provenge too or just the patients that didn’t crossover to anything?
FROHLICH: We would offer it to all patients on the placebo arm including those who had crossed over and received the version of the product prepared from prior preserved cells.
MILLER: The enrollment criteria of having a PSA below 50; was that new to IMPACT or was that in (9901) and (D9902A)?
FROHLICH: It was (the enrollment criteria) just a PSA greater than 5, not less than 50. There was no restriction on the PSA level.
MILLER: Do you have the medians on the time to progression?
GOLD: It was pretty consistent with what you’ve seen in other studies and (in) this study (it) was about 14 weeks.
MILLER: Can you remember the span of the enrollment dates for 9901, D9902A, and IMPACT?
GOLD: I don’t. No I don’t have those. I have a lot of numbers memorized (laughter) but not those.
MILLER: The (placebo) patients who got no crossover at all, their median (survival) was below the patients who crossed over and got Provenge?
GOLD: That is correct.
Questions from MARK MONANE of Needham & Company (23:49 mark)
MONANE: When I look at the results here, should I be or does your group look at this in comparison to Taxotere (docetaxel) studies or mitoxantrone studies or do you believe that the data should be looked at in and of itself in terms of the regulatory and in terms of the commercialization strategy going forward?
GOLD: I think you have to look at it more from a patient’s perspective. You always have to be careful when you compare products across studies. This study was not designed to be a comparison to Taxotere. It’s (IMPACT) positioning Provenge as a therapy for front line use in men with metastatic androgen-independent prostate cancer.
However from a patient’s perspective, there is only one approved form of treatment (to benefit overall survival) and that is Taxotere. As you know it has a median improvement in survival of 2 ½ months and the three-year survival rates on the updated data is about 18%.
I think from a patient perspective (Provenge) offers an appealing alternative to that, particularly given that it has a very appealing safety profile combined with the fact that, and we shouldn’t overlook this, Dr. Penson said that in his talk and Mark (Frohlich) reiterated it, it’s a huge advantage to the patient that they get a full course of treatment in one month (with Provenge). They are not having to go to one of their physician’s office once every three weeks for an I.V. infusion. They get (Provenge) over one month. They can go on and live their normal life. They can go out and play golf and travel and do things that they want to do. That’s a huge quality-of-life issue in terms of not having to go back into the physician’s office every three weeks.
MONANE: Speaking of Taxotere, I think Dr. Petrylak did a study from 9901 (titled “Defining the Optimal Role of Immunotherapy and Chemotherapy…”) that talked about the use of Taxotere after Provenge infusion and the potential priming, have we looked at that yet and is this dataset a planned analysis going forward?
GOLD: Those are the types of things we will look at as we dig into the data deeper but we have nothing there to share at this time.
MONANE: There didn’t seem to be a big effect on PSA? Is that consistent with the effect on the time to progression? Is that independent? A lot of cancer studies talk about tumor burden, how do you think about the effect on PSA or lack there of?
GOLD: So two things; One is that the study was not designed to measure PSA. There was no PSA measured in these patients after they progressed. Looking at PSA kinetics is not very helpful in this study.
PSA use in androgen-independent prostate cancer is not typically correlative with overall survival. When it is, what you see is it (PSA levels) staying in check or not going higher over a period of, say, two years…not dramatic decreases in PSA. That’s number one.
Number two is that the study was not designed to measure PSA in this patient population.
MONANE: There was some data presented on the quartile effects (75%, 50%, and 25%), it seemed that the longer you lived then the longer you lived (i.e. the survival curves were separating more over time). Is that a survival effect of the healthiest patients or could we make any hypothesis concerning the effect on the tumor burden and the primary disease?
GOLD: To us, what warmed our hearts, was looking at the shapes of these (survival) curves. If you look at the 75th percentile (when 75% of patients in Provenge and placebo groups were still alive), so even early on in the curve, there is still a survival benefit of 4.1 months. They are almost identical to the median. This just shows how consistently the curves are separating across the length of the study.
Immunotherapy has this theory that there is a delayed treatment effect and we see that this does pan out to a certain extent. When you look at the 25th percentile (the tail of the curve), the difference (in survival) is close to six months if I recall.
GEORGE FARMER of Canaccord Adams (28:08 mark)
FARMER: Regarding other endpoints such as pain, there was no difference on pain scores in Provenge versus the placebo arms. Was there any trend in any direction? What does it say about other measurements like as PSA and pain when thinking about treating these advanced patients?
FROHLICH: Pain was not an endpoint in the study after it was amended so we don’t have any data on pain scores to share with you. Just to reiterate what Mitch said about PSA and other surrogate measures, the challenge is that none of those have consistently correlated with overall survival, which is really the gold standard.
By definition, if you are extending overall survival then you must be slowing the progression of the disease. The challenge is really with how we’ve chosen to measure progression, which is an endpoint that has a lot of variability. It occurs very early, too soon for even the therapy to affect it.
FARMER: How do you think physicians going forward are going to be measuring progression? I mean, that’s not going to be a reliable metric in determining whether a patient should crossover onto new therapy or is it?
GOLD: I think from a clinical perspective what patients care about the most is living longer but in a clinical setting, when you see Taxotere being implemented, is typically when the patients become symptomatic, when they are having other issues with their disease.
Provenge we would see in the same position as a frontline treatment in men with metastatic androgen-independent prostate cancer and it certainly opens the field for other treatments later on, and someone speculated that it may increase the use of Taxotere or other agents that come downstream because these patients are living longer and they come into that era with a better performance status than they would otherwise.
PAUL LATTA of McAdams Wright Ragen (30:19)
LATTA: Question on the time to disease progression: it sounded like the data was consistent with what you’ve seen in previous trials. Maybe you could just refresh our memory why in both cases that endpoint didn’t work out?
GOLD: The biggest issue with progression is that it tends to be a difficult endpoint to assess reliably. In this study and in multiple other studies that have been reported, they typically have a high rate of discordance between reviewers on the progression endpoint. One reviewer will say the patient progressed and another reviewer say there was no progression.
The biggest issue with this is that there is a tremendous amount of noise in the endpoint and it ends up being an unreliable endpoint at the end of the day. It’s certainly not the type of endpoint that you’d want to put in as your principle endpoint in a large randomized study because of that noise factor you see with the endpoint.
LATTA: What are they typically measuring?
GOLD: They are either looking for additional radiographic criteria to suggest that the lesion has progressed such as new bony lesions or new soft tissue lesions and sometimes it can be clinical progression.
The endpoint that I think is pretty universally agreed upon now, is that survival is the appropriate endpoint to measure in this patient population.
LATTA: On slide 8, the two survival curves look like they separate real nicely and I assume that they rejoin at the end. That’s just sort of a statistical anomaly as you get close to zero or is that…
FROHLICH: I have two comments to make on that. The first is that all survival curves eventually come together because patients all eventually die.
The point made in the comments is that the survival follow-up is particularly long in this case. If you look at the Taxotere trial in which the median follow-up time was on the order of 21 months. Ours was close to three years so a much longer follow-up.
The other thing is that when you look at the number of patients at risk at the tail end of the curves, there is actually very few patients there so there is a lot of variability in those curves. They are a less reliable than in the earlier parts of curves where you see that dramatic separation.
GOLD CLOSING REMARKS…
