Sunday, October 16, 2011

Exclusive Interview with Mangrove Partners' Nathaniel August

I am excited to bring my blog visitors an exclusive interview with a very special guest, Mr. Nathaniel August, founder of Mangrove Partners ( As many of you know, Mangrove has recently taken an 8% stake in Celsion over the last few months ( Mr. August was kind enough to answer some questions I had for him (via email). See below for his responses:

Siavoche: Briefly, describe the origins of Mangrove Partners and your investment philosophy?

NA: Mangrove Partners is the Investment Manager for The Mangrove Partners Fund, LP, a limited partnership fund for accredited investors. We focus on an identified set of investment opportunities that we believe are likely to be mispriced because we can get a systematic edge as a result of either information asymmetry or investor behavior. Our portfolio tends to be concentrated in our top ideas and to have relatively little market exposure as a result of our short investments.

Siavoche: When was the first time Celsion Corporation “popped on your radar”, and what immediately caught your interest in the company?

NA: We began researching Celsion in May of this year after we purchased a large position in Oncothyreon. At the time, we had been introduced by some good friends to the Oncothyreon investment thesis, which is, in essence, that you can make an educated investment in the outcome of an event-based clinical trial if you know when people are enrolled in the trial and the number of events needed to stop the trial. This information can tell you about the behavior of the entire trial population. Since the behavior of the trial as a whole is just the sum of the control and treatment arms, if you can calculate the behavior of the trial as a whole and estimate through research the behavior of the control arm then you can solve for the behavior of the treatment arm. This was the essential insight behind the Oncothyreon investment thesis, and we decided to try to broaden it out so that we could look in a systematic fashion for trials where we could apply this analytical framework. Since this framework is crude, we decided to look for companies where we believed that an unusually large difference existed between the control and treatment arms and where a successful trial would have a very large impact relative to the size of the company. By focusing on these qualitative aspects, we felt we could further tilt the expected value of the investment in our favor. Ultimately, we chose to make large investments in Oncothyreon, Aveo, and Celsion on the basis of this framework.

Siavoche: Why do you think the company does not receive more attention from Wall Street (i.e., analyst coverage by big banks), and only recently has caught the attention of more mainstream biopharma journalists (such as Adam Feuerstein)?

NA: I think the company has a chicken and egg problem - Celsion needs to be larger to get attention from Wall Street but it can't grow larger without getting the attention. Most professional investors refuse to invest in microcap companies, yet it's often difficult for a microcap company to even reach small cap size without some results - revenue, earnings, trial outcomes, technological breakthroughs, etc. For Celsion, the Phase 3 data, whenever it comes, has the potential to have that impact. As regards media attention, I think that the media loves to focus on catalysts and events. Since Celsion is close to an event, it has started to get some media attention. If the trial continues to the final endpoint and we have another 2 years to wait, Celsion will probably be orphaned by the media again until we get to within a few months of the readout on the trial.

Siavoche: We know Celsion has gone through somewhat of a metamorphosis since their days as a device company. What are your impressions of the current management team, and the strategic direction outlined by the CEO?

NA: I think the current management team has taken a number of positive steps and are clearly above average relative to most management teams, but they're not perfect. On the positive side, they have conducted what appears to be a well thought out clinical trial, secured an SPA and fast-track status, and have a promising partnership with Philips. On the negative side, I am concerned that they are too conservative with regards to the circumstances under which they will unblind and may be too tied to maintaining the SPA rather than focusing on correctly powering the trial. These concerns drove our three recommendations to management: (1) that they split the endpoints of the trial between PFS and OS and report on PFS while leaving OS blind in the event that PFS is statistically significant at the interim look, (2) that they reduce the number of events to end the trial to 256, and (3) that they report the hazard ratio at the interim look if they choose to continue the trial to 380 events. The response that we received was that our first two requests would invalidate the SPA and that our third request was effectively an unblinding of the trial. We disagree about the importance of an SPA, because we think that the FDA has shown through its actions that it will not be bound to making poor decisions based solely on a SPA. In essence, we think that the SPA is of minimal value - the same trial will fail or succeed on its merits with or without the SPA. As regards our third request, we can point to many examples of companies that have given some results at the interim look, most notably Dendreon.

Siavoche: Mangrove has recently outlined a 65% chance for an interim halt for overwhelming efficacy in the HEAT trial. This obviously clashes with management’s assessment of a halt being “highly unlikely”, though such statements from the company are probably expected to air on the side of conservatism. What specifically makes Mangrove so confident at the upcoming interim look? We also know Griffin Securities just released a note recently on 10/14/11 with an expectation for the trial to continue to 380 events. Please explain your thoughts and why Mangrove has such a starkly different assessment.

NA: Based on some very recent research of ours, we believe that the standard O'Brien-Fleming framework for a single interim look in a single trial calls for a p-value of approximately 0.001. With this as a starting point, we hired a biostatistician to run simulations of the HEAT trial with various different median times to PFS in the control arm, the known enrollment curve, and 190 events in June, July or August, 2011. With these inputs, we could simulate the median PFS in the treatment arm and we could generate 99.9% confidence intervals around this median. We then sensitized these simulations for several different median times to PFS in the control arm. When the confidence interval we generated had a lower bound greater than the median PFS we assumed in the control arm then our simulation showed, by definition, statistical significance at a p-value of 0.001. I am pretty confident that we're going to see success at any median PFS in the control arm at or below 18-months. Unfortunately, this does not mean that the company will unblind, because they have made some comments to investors that they are self-imposing a new requirement that OS also show a "trend." To date, we have been unable to discern how they define a trend in OS. To add further confusion, I have since received some contradictory messages from management, including a recent telephone call where they seemed to backtrack on the need for a trend in OS. All of this leaves me confused, so while I am very positive on the ability to show PFS at a statistically significant level, I am confused on OS requirements and have added a big dose of maybe to my percentages - hence the 65%. As regards Griffin, I have left a message for their analyst and hope to get a call back. I also called the CEO on his office number and somehow accidentally woke him up in Asia - sorry Mike! This is a long way of saying that I don't know what Griffin is basing this assertion on other than maybe just the comments at the recent Rodman and Renshaw conference.

Siavoche: What is Mangrove’s stock price target if the trial is stopped for overwhelming efficacy at the interim analysis? What sales/revenue assumptions are you making? How does your price target, change, if at all, if you include inevitable off-label use in colorectal liver mets?

NA: First of all, I think off-label use in colorectal liver mets will be minimal without a Phase 2b study because it will be too difficult to get reimbursement without this study. We also discount the ability of the company to generate sales in Asia because it has historically been difficult to sell expensive drugs in China, Taiwan, etc and because the Japanese trial sites have been suspended. Accordingly, we look only at US and EU sales for HCC and I think it's conservative to see $150 million of sales in these geographies. At 2x revenue (again conservative), this is $300 million in market cap or about $10 per share after some additional dilution. All of these assumptions are very conservative, but it's a good place to start.

Siavoche: Is it your opinion that the company’s SPA with the FDA will require statistical significance on both primary AND secondary endpoints (namely, overall survival) at the interim for an efficacy halt? In an email exchange I had with Dr. Borys, he gave me the impression that PFS would be the primary driver, but OS would have to look directionally ok, but not necessarily statistically significant.

NA: I believe that the SPA requires statistical significance on only one of the two endpoints. Your email exchange is symptomatic of the mixed messages being sent by the company regarding the necessary conditions to stop the trial at the interim look.

Siavoche: If the DMC makes a recommendation for the trial to continue to 380 events, we know Mangrove believes 380 would not happen until the middle of 2013. On the one hand, a recommendation to continue by the DMC is a blessing that Thermodox is showing efficacy, an important fact given the small size of the Phase 1 data and different endpoint. On the other hand, you and others have brought up dilution concerns. How do you think the market will reconcile these two realities?

NA: My guess is that the market is already pricing in a continuation to 380 events and that this accounts for the ~30% pullback we've seen in the stock. I continue to believe that 380 events will not be reached until the second half of 2013 at the earliest.

Siavoche: In the event the trial continues to 380 events, Mangrove has recommended that Celsion consult with the FDA to reduce the number of events needed for the top-line analysis. I would argue this might put their SPA at risk for invalidation. Do you have any specific analogs (i.e. company examples) where such a change has been made in a late stage trial?

NA: We believe that negotiating with the FDA to reduce the number of events has no impact on the SPA, because the FDA always has the right to simply say that Celsion cannot change the SPA. Personally, I am of the belief that the SPA is of limited value and that SPAs are, in general, overrated because the FDA's final decision will always rest on the merits of the trial and the data as presented - not based on what a company has negotiated for in their SPA. As regards changing the number of events, Oncolytics recently received a SPA for an adaptive trial where the final number of enrolled patients, endpoint, etc will not be known until well after the trial starts. Similarly, Merck KGaA's START trial received a change to its SPA that allowed them to drop some enrolled patients and add others. It's not that uncommon to see changes made to an existing SPA.

Siavoche: We know management is committed to signing a 2nd license deal to commercialize Thermodox, possibly to include all geographies, although the company has signaled they might want to self-commercialize in the US. Please give us your thoughts on the following pertaining to a second deal:

a. Potential partner?

NA: I'm not sure, but for HCC it would need to be someone with very good coverage of emerging markets and Italy (is that redundant? - joke). Maybe Sanofi?

b. Potential geographies covered?

NA: As per the above, the major geographies for HCC are the Asian countries and Italy. I would tend to advise against partnering for colorectal dominated geographies until the phase 2b data is in. I hope this will be a relatively short trial because colorectal mets are so aggressive.

c. Potential deal terms (i.e. upfront payment, royalties), and more specifically, what type of deal terms Mangrove would like to see?

NA: Obviously, I would prefer that the company get as much as possible. It would be nice to get the cash they will need to get to cash flow breakeven from a partner rather than from equity and it would also be nice for the company to preserve the rights to sell where a colorectal indication is likely to be the primary use until there is better colorectal data which they will be able to use for negotiating a partnership in this indication.

Siavoche: What are your thoughts of the HEAT enrollment pause in Japan, and the company’s recent statement that their partner Yakult would be starting an entirely new study of Thermodox in Japan?

NA: The enrollment pause is quite concerning. I'm not sure what effect it will ultimately have, but it will not be positive. The best outcome is that it ends up being irrelevent. A new trial is helpful and, hopefully, any trouble that the company has as a result of the pause can be addressed by this new trial.

Siavoche: How do you interpret the company’s recent relocation to New Jersey?

NA: I think the stated purpose was to help the company recruit salespeople. I am personally skeptical of the ability of any small biotech company to build a salesforce from scratch and would therefore prefer that Celsion partner Thermodox in all geographies, including the United States. The relocation to New Jersey would appear to be a step in the opposite direction and is therefore concerning.

Siavoche: In closing, is there anything else you would like to add? The floor is all yours.

NA: I think that it is important to remember what we can learn from Delcath - both positive and negative. On the positive side, we learned that targeted drug therapies that deliver a potent dose to the liver have a real chance at showing dramatic results. In fact, the Delcath trial had some of the best statistical results I have ever seen. On the negative side, we need to be cautious about the ability of a small company to navigate the process of going from a successful trial to a successful drug. Delcath has not only failed to get FDA approval, but its self-commercialization strategy in Europe is uncertain. So on the one hand Delcath would tell us to be optimistic about the trial results, but on the other hand it, would tell us that there's a lot of potential value a partner can add in both the approval and commercialization stages.

Siavoche: Mr. August, thank you once again for your time. On behalf of all Celsion shareholders, I thank you so much.

NA: You're welcome.

My Assessment/Takeaways

First and foremost, I want to reiterate how grateful I am to Mr. August for taking the time to answer my questions. Clearly, his calendar must be quite busy, so again, I very much appreciate it. Transparency in the investment world means everything, as I am sure you will all agree, and for that, we should applaud Mr. August.

One area where I will have to agree to disagree with Mr. August is with respect to modifying any aspect of the trial to accelerate the timing of the next top-line analysis, assuming there is the need for one after the upcoming interim analysis. I don't know, the FDA has become increasingly unpredictable as of late, and while Mr. August did cite some examples of companies that were able to successfully modify their SPA's, I just don't think Celsion has the necessary size/clout to get such a thing done. On top of that, with the Japan "issue" (or non-issue, depending on how you want to look at it), I would think the rest of the trial needs to be 100% squeaky clean from the agency's vantage point. But, to his point, modifying the statistical analysis plan per his suggestions could mean the difference between waiting an additional 1 year versus 2 years for "final" data. This all the more places the stakes for the interim analysis that much higher.

The other interesting comment Mr. August made was with respect to securing reimbursement for off-label colorectal liver mets use. I happen to disagree here, and as I put on my previous biopharma reimbursement consulting cap, it is actually quite easy to secure off-label use in oncology. To Mr. August's point, you do need some data, fortunately for Celsion, more than half of their Phase 1 patients had secondary liver mets, including colorectal origin. Believe it or not, even that small of a sample is enough to get compendia listing, supporting off-label utilization. And, to add, if Thermodox is working as well as we think it might be in HCC, oncologists and interventional radiologists will be screaming at payers to cover off-label use, so physician pressure will be hard to overcome. To quote many of the managed care respondents I used to interview while I was a consultant, "no insurance company wants to be on the cover of the NY Times for denying access to life-saving cancer drugs." So, if we factor in more aggressive sales assumptions to include off-label CRLM use, I think it is fair to say his $10 target in the event of overwhelming efficacy might very well be conservative. Of note, the CRLM population is primarily in the western world, where Thermodox is likely to have a more aggressive pricing strategy as well. Make no mistake, that [CRLM] is a huge market for Celsion, as I have pointed out before.

In terms of self-commercializing in the US market, and the potential rationale for the company's move to New Jersey as a means of recruiting sales force talent, I have a slightly different view as well. I think Mr. August brings up a fantastic point that many biotechs try to "go it alone" in terms of self-commercializing, only to later on discover how expensive and costly such an endeavor is, and there are several examples. But, in the case of Celsion and liver cancer in the US, I truly think a small, lean, targeted sales force would be sufficient. At most, I would think ~100 reps is all that is needed to target potential high/med prescribers across the country (again, putting on my prior consulting cap and recalling other analog companies aiming to commercialize specialty products. At ~$150K/annually for each with benefits, that still does add up to a cool $15M annual expense). And, within high unmet need oncology therapy areas, the drug and the data ultimately sells itself. HCC is not a crowded space like, say, rheumatoid arthritis. Options are very limited, and Thermodox will stand besides Nexavar and doxorobucin eluting beads (use in TACE) in an otherwise very "empty" space.

Without knowing the details behind the model developed by Mangrove's team, I would also just caution readers that it is very difficult to simulate the HEAT trial, as I am sure Mr. August would agree. My own interactions with other very statistically savvy CLSN investors confirm this as well. There are a lot of things to consider, including:
  • Average time from enrollment to treatment- There could be at least a 2-3 week gap there, since patients are not likely to get treated the second they are enrolled.
  • Review times- Recall that patients come back in for follow-up after months 1,3,5,7,9,12 and then, every 3 months after that. This would have to be accounted for in the model somehow.
  • Percentage of treatment failures in each arm (i.e., the number of patients who could not achieve a complete ablation after 2 attempts within approximately the first month and half)
In closing, I think Mangrove's actions speak very loudly (talk about putting your money where your mouth is). The company has aggressively bought CLSN stock and options in the last couple of months. As Mr. August mentioned, the rationale for doing so is completely hinged on the premise that the interim will lead to a halt for overwhelming efficacy. Mangrove hired a biostatistician to simulate the trial based on publicly available enrollment data, and using sensitivity analyses to approximate the timing of the 190th event (which probably happened anywhere from Jun-early Sep of 2011). According to him, if the control arm comes in anywhere under 18 months, there is a great chance for an efficacy halt, of course, assuming that hitting statistical significance on the primary endpoint is all that is needed (it sounds like conflicting statements from management give him some pause about this issue). Assuming the above model-specific caveats are accounted for in the statistical simulation, we could very well be en route to an early NDA. In terms of the control arm, my own research, as many of you have seen, points to a median PFS time of anywhere from 12-15 months, and I remain extremely confident in that range. On top of that, the company very recently has confirmed their expectations for the control arm to be running at a median of 12 months.

I don't know about you, but I certainly feel the tension in the air as we approach the interim analysis. Exciting times, to say the least.

As always, feel free to leave questions or comments. I will combine any potential questions I receive and try to circle back with Mr. August for responses.