Expert Financial Analysis and Reporting

Investment Background and Thesis
In my initial report on Celldex, I highlighted four interesting products that the Company has in development. Two of these are in randomized phase II or III trials that could be the basis for regulatory filings: rindopepimut for glioblastoma (GBM) and CDX-011 or glembatumab for triple negative GPNMB positive breast cancer. I am actually more intrigued by two earlier stage products, CDX-1127 and CDX-1135, from the perspective of mechanism of action, but I need to see more data on them before incorporating their prospects into my investment thinking.

I am intrigued by the pipeline but cautious on the stock because of the sharp run-up that has seen the stock increase from $2.78 on January 3, 2012 to $37.83 on October 1, 2013. At the current price of $23.06 it has a market capitalization of $1.9 billion which compares to $121 million back on January 3, 2012. In my view, this is one of the most interesting pipelines in the emerging biotechnology universe. However, I am concerned that momentum investors are controlling the stock. At a more palatable valuation, I likely would be a buyer.

In this report, I have tried to give a basic overview of the technology underlying rindopepimut, clinical data seen so far, current clinical trials and some thoughts on its medical and commercial potential. This report focuses on rindopepimut because of a potentially important data release that is imminent. Rindopepimut is a cancer vaccine that addresses about 30% of GBM patients who express the EGFRvIII mutation (this will be discussed later).

There are estimated to be 8,000 newly diagnosed GBM patients each year in the US and 2,000 recurrent GBM patients. Based on the 30% figure, rindopepimut might be used in 2,400 newly diagnosed and 600 recurrent GBM patients. I am estimating that it will be priced at $100,000 per course of therapy so that in dollar terms these US addressable markets are $240 million and $60 million respectively. The worldwide addressable market would be double the US or approximately $600 million.

Celldex announced on August 12, 2013 that it had completed enrollment in an initial cohort of 25 patients who were refractory to Avastin. Based on preliminary evidence of stable disease, tumor shrinkage and investigator-reported response, the Company decided to add an expansion cohort of approximately 75 patients to better characterize the potential activity of rindopepimut in this refractory patient population.

The results in the 25 initial patients will be reported at the Society for Neuro-Oncology Annual Meeting of November 21 through 24 in San Francisco. The embargo on the abstract will be lifted on Monday, November 11 and available on the SNO website. It has been accepted as an oral presentation and is entitled "ReACT: a Phase 2 Study of Rindopepimut Vaccine (CDX-110) Plus Bevacizumab in Relapsed Glioblastoma." The oral presentation will be made in a session lasting from 10:20 AM to 12:00 PM on Sunday November 24. On Monday, November 25 at 8:30 am EST, management will also hold a conference call to review the data.

The data release relates to 25 patients enrolled in one arm of the phase II ReACT trial in recurrent GBM. Patients with recurrent GBM have the expectation of six to nine months of median overall survival. They can be treated with Avastin, Gliadel Wafer as an adjunct to surgery (seldom used), surgery or just supportive care. These particular 25 patients were part of an arm of ReACT that enrolled patients who did not respond to Avastin. The August 12th announcement of the trial expansion obviously raised the expectation that rindopepimut had produced a clinically meaningful improvement.

The question is what constitutes a clinically meaningful improvement in recurrent GBM patients who are resistant to Avastin. I note that Avastin was approved in recurrent GBM on the basis that it produced tumor shrinkage in about 28% of patients, increased progression free survival by about 4.2 months but had no effect on median overall survival. Could it be the case that rindopepimut also might be approved on the basis of improvement only in progression free survival without an increase in median survival? Remember that these patients have very short survival expectations and have no viable drug option. Because of these factors, it is possible that this narrow indication could be a quick route to approval for rindopepimut.

Of the 600 recurrent glioblastoma patients, we don’t know how many patients would be Avastin resistant. Hence, the addressable patient population is some fraction of 600. While this is a very small patient population, the approval for this narrow indication could be very important commercially. Once approved, it could lead to off-label use in newly diagnosed patients as well. And a s previously noted, I am also expecting that rindopepimut could be priced at $100,000 per course of therapy so that 600 patients represents an addressable market of as much as $60 million in the US and $120 million worldwide.

Celldex has not given much guidance on what to expect from this data release so that the previous reasoning is just a hypothesis. If I am correct in this reasoning, it could lead to an upward move in the stock even though expectations are already high. After, Celldex announced that the data was positive and led them to expand the trial arm by 75 patients, the stock price increased from $20.36 to $37.83 on October 1, an increase of 85% in less than a month. I recognize the potential for a trading move if the data release excites investors as seems likely, but this doesn’t change my cautious stance on the stock.

Cancer Vaccines for GBM
I have done considerable work on cancer vaccines that are being developed by emerging biotechnology companies to treat glioblastoma. I have written extensively on Northwest Biotherapeutics (NWBO) and ImmunoCellular Therapeutics (IMUC) which use a technology based on loading dendritic cells with cancer antigens and Agenus (AGEN) which uses a technology based on heat shock proteins. Rindopepimut uses a still different technology which is discussed later in this report. Each of these products in small open label studies has produced eye opening results. However, these have been largely dismissed by investors as being small and flawed trials. Besides critics add, everyone knows that cancer vaccines don’t work.

My work drew me to analyze Celldex’s cancer vaccine rindopepimut. As with the cancer vaccines of the other companies, it has shown some very promising results in phase II glioblastoma trials. As I took a deeper look at Celldex, I became interested in other products, and investors who are interested in my views on these other products can access them through reports on my website, but for now I will focus just on rindopepimut.

In regard to cancer vaccines, there are two potentially very important data points coming up that should give us a better insight into whether the promising signals of activity that we have seen are validated in larger studies or if the critics are right and cancer vaccines don’t work. These are the rindopepimut results which I have just described and the data readout on the 124 patient phase II trial of IMUC’s ICT-107 in newly diagnosed GBM which should occur late this year or early 2014.

How Do the Four Cancer Vaccines Compare?
Let me anticipate your next question as to how the four vaccines under development compare with each other and how they will be used in clinical practice? The next table contains some of the relevant data on the four cancer drugs in development.

Clinical Trial Data For Cancer Vaccines for Newly Diagnosed Glioblastoma
Drug	Patients Treated	Median Overall Survival for Drug (months)	Median Overall Survival for Standard of Care (months)	Difference in Median Overall Survival (months)
rindopepimut	65	21.8	16.0	5.8
DCVax-L	20	36.0	18.1	17.9
ICT-107	16	38.4	18.1	20.3
Prophage	46	23.3	18.1	5.2
Standard of care	287	14.4	14.4	**

There is a very hot debate on the numbers in the above table because none of these trials were randomized. This means that results can only be compared to results from a different trial. With justification, critics point out those patient populations of different trials can differ widely in their characteristics so that comparing results in one trial to another can be misleading. The defining trial for standard of care in glioblastoma is the Stupp trial that defined temozolomide plus radiation as standard of care in newly diagnosed glioblastoma. The median overall survival in the Stupp trial was 14.4 months.

Comparing results from the Stupp trial to these smaller trials can potentially be misleading. Stupp accepted patients of varying physical condition while those patients included in the four above trials were likelier to be more carefully selected, healthier and expected to have better outcomes. There have been particular questions raised because patients in these smaller trials may have had their tumors more surgically resected than those in in the Stupp trial that could lead to a better outcome. All of these criticisms are valid. I have tried to adjust somewhat for this by using data from a subset of the Stupp trial that was 91 patients who were greater than 90% surgically resected. The median overall survival in this group was 18.1 months as opposed to the 14.4 months of the Stupp trial.

In regard to rindopepimut, the patients in its phase II trial do appear to have been healthier than the Stupp patients. Patients treated with rindopepimut in its ACT-III phase II trial had gross total resection and they were also required to be disease progression free three months after surgery. They were a healthy subset of patients within the glioblastoma patient universe and might be expected to have good outcomes. However, there is another variable to consider and that is that EGFRvIII positive patients may have more aggressive disease. Instead of using the data from Stupp, Celldex has used a retrospective study of comparable patients done by M.D. Anderson that suggested median overall survival in this group with standard of care is 16.0 month.

I hope that I have not confused you too much with all of these numbers, but it should give you a sense of the complexity and uncertainty in trying to draw conclusions from this data which requires some apples to oranges comparisons. I would also discourage readers from looking at this data and suggesting that one cancer vaccine is better than another because of reasons just discussed. To me, the takeaway message is that each of these has produced a signal of meaningful efficacy. Maybe there is more to cancer vaccines than Wall Street has realized.

The critics are somewhat justified in questioning the data presented above, but the results still appear to be striking. In a highly aggressive disease like glioblastoma, a four to five month increase in median overall survival is considered a clinically meaningful advance and the data presented above is suggestive of a meaningful increase. I think that we are seeing a signal that these vaccines do have a very meaningful therapeutic effect and could be a major advance in oncology. However, this has to be confirmed in larger trials.

The Potential Market for Rindopepimut
Let’s assume for the sake of discussion that all of the cancer vaccines are effective in their clinical trials and gain approval. In this event, what might be the potential market for rindopepimut? The mode of action of rindopepimut means that it can only be used in patients who are EGFRvIII positive which is about 30% of patients. ICT-107 because it can only be used in HLA A1/A2 positive patients was applicable bout 45% of patients screened for its phase II trial. DCVax-L can be used in all patients with sufficient tumor mass after surgery to manufacture the product; this is over 95% of patients. Prophage can be used in all patients.

One could probably speculate that rindopepimut in this scenario would seemingly be used in the EGFRvIII patients population either by itself, in combination with one or more of the other vaccines or in sequence with the other vaccines. In the recurrent market, Celldex estimates there are 8,000 newly diagnosed glioblastoma patients each year in the US and 2,000 recurrent glioblastoma patients so that the addressable market for rindopepimut in the US is 2,400 patients and 600 in recurrent glioblastoma. (I note that some reports suggest that there is less EGFRvIII expression in recurrent glioblastoma). As a rule of thumb, the rest of the world addressable market might be about the same as the US.

Recently introduced cancer drugs have been priced extremely high. Medivation’s (MDVN) Xtandi and Johnson & Johnson’s (JNJ) Xytiga have been priced in the $60,000+ range for a course of treatment. Dendreon’s Provenge was priced at $90,000+ and the recently withdrawn Iclusig of Ariad was priced at $115,000. I think that rindopepimut could be priced in the $100,000 range for a course of treatment. It is very important that those EGFRvIII expressing patients in whom rindopepimut might be expected to be effective can be identified before treatment. This means that physicians don’t have to treat all GBN patients without knowing who has the potential to benefit; this targeted therapy allows higher pricing. Also, rindopepimut has a relatively benign side effect profile. With some drugs, treating their side effects rivals the cost of the treatment.

I think that the US addressable market for rindopepimut in the US would be about $240 million (2,400 patients times $100,000) in newly diagnosed glioblastoma and $60 million (600 patients times 100,000 in recurrent glioblastoma). The worldwide potential is about double this. I think that penetration of these markets would be rapid if the results in clinical trials are viewed as clinically meaningful and the products are approved.

Rindopepimut and Its Biological Target
Rindopepimut is an immunotherapy (cancer vaccine) that is designed to create an immune response against a mutation of the epidermal growth factor receptor (EGFR); this is a molecular complex that occurs in the membranes of cells. When proteins such as epidermal growth factor and TNFα bind to the receptor EGFR, they send a signal that to the DNA in the nucleus that causes cells to grow, differentiate and proliferate.

In normal cells, the EGFR receptor is controlled so that it is sometimes turned on and sometimes off. However, in some glioblastomas there can occur a mutation called epidermal growth factor receptor variant III (EGFRvIII). When this occurs, the receptor is permanently activated and this results in rampant, uncontrolled cell division (i.e. glioblastoma).

It is hypothesized that this mutation may also stimulate the growth of neighboring cells through the release of IL-6 even if these cells do not have the EGFRvIII mutation. Moreover, EGFRvIII expressing cells may also release microvesicles (cellular components that are released into the space outside the cell). These contain EGFRvIII which can merge with neighboring cells. The result of all these actions is the uncontrolled cell growth that is glioblastoma.

Rindopepimut is comprised of a peptide that mimics an amino acid sequence that is specifically found in the EGFRvIII molecule. Because EGFRvIII does not occur in normal cells, this is a promising way of stimulating an immune response against cancer cells that express EGFRvIII. The goal is to stimulate an immune response specifically targets EGFRvIII expressing cancer cells and has no effect on normal cells.

Rindopepimut contains the peptide that mimics EGFRvIII that is linked to a carrier protein called keyhole limpet hemocyanin, or KLH. Keyhole limpet hemocyanin is used extensively in immunotherapy as a carrier protein to stimulate a response from the immune system. The peptide/ KLH linkage is administered as an injection along with the adjuvant GM-CSF. EGFRvIII is expressed in 30% of glioblastomas according to estimates provided by Celldex so that rindopepimut would be applicable to about 30% of glioblastoma patients.

Phase II Studies of Rindopepimut in Glioblastoma Were Encouraging
Data from three phase II trials has encouraged Celldex to proceed to a phase III study in newly diagnosed glioblastoma patients and a phase II study in recurrent glioblastoma; the latter while referred to as a phase II could be the basis for approval in this recurrent glioblastoma population. Hence, both trials could potentially be the basis for a regulatory submissions seeking approval.

ACTIVATE was the first phase II study and was conducted at Duke Medical Center and M.D. Anderson. It enrolled 18 patients who could be evaluated. They were treated with surgical resection followed by standard of care (SOC) and rindopepimut. Standard of care was radiation and temozolomide. This was followed by ACT II, essentially an extension of ACTIVATE, that evaluated 22 new patients at the same two centers.

The third study was a phase IIb study called ACT III that was originally randomized to compare rindopepimut plus SOC to SOC alone in over 30 US clinical sites. SOC was defined as temozolomide plus radiation followed by a maintenance regimen of temozolomide. However, this trial was amended in December 2008 because the Independent Data Monitoring Committee determined that the majority of patients in the control arm who were being given SOC were dropping out of the trial because they were able to determine that they were not receiving rindopepimut. Rindopepimut caused an injection site reaction and SOC did not. The decision was made to offer SOC patients rindopepimut with the result that ACT III then became an open label study.

Results of Phase II Studies
In November 2012, Celldex announced three year survival data for ACTIVATE, ACT II and ACT III. The results were very consistent across all three trials:

	Number of patients	Median Overall Survival (months)	Patients alive at three years   (%)
ACTIVATE	18	20.4	33
ACT II	22	20.5	23
ACT III	65	21.8	26

These were open label trials so there was obviously no control arm to compare against. Also, to my knowledge these were the first trials done in which the patient population was entirely made up of glioblastoma tumors that were expressing EGFRvIII. The trial that defined standard of care, the Stupp trial, enrolled all types of glioblastoma patients so that it is probably the case that only 30% of patients in Stupp were EGFRvIII positive.

In order to come up with an historical control group to compare these results to a retrospective analysis of EGFRvIII expression status and associated clinical outcome in the Phase 3 Radiation Therapy Oncology Group's, or RTOG, 0525 study was undertaken. This analysis was conducted by MD Anderson Cancer Center in cooperation with RTOG to provide an assessment of the prognosis for patients with EGFRvIII-positive disease contemporary with the ACT III data.

MD Anderson looked at 17 patients with EGFRvIII-positive disease who were treated at the same time as ACTIVATE was being conducted, but not with rindopepimut. An assumption had to be made that these patients were enrolled three months after diagnosis which creates some statistical uncertainty. Because there was no randomization, there are also issues on how well matched these patients were for degree of resection, age and performance status. In this retrospective study the median overall survival was 12.2 months and the percentage of patients alive at three years was 6%. This compares to 21.8 months and 26% in ACT III so despite the shortcomings with defining the control group, the results looked impressive.

The Radiation Therapy Oncology Group (RTOG) undertook a retroactive study based on case records from M.D. Anderson that looked at 142 patients who were enrolled at about the same time as the ACT III patients. In these 142 patients, median overall survival was 15.1 months and 18% of patients were alive at three years. RTOG then looked more closely at a subset of 29 patients that were more specifically matched to ACT III patients. These patients had similar degrees of gross resection and were about three months post-surgery without disease progression.

I believe that this 29 patient group represents patients with characteristics most similar to ACT-III patients. In this most meaningful control group, median overall survival was 16.0 months and 13% of patients were alive at three years. Remember that the 65 patients in ACT III had median overall survival of 21.8 months and 26% of patients were alive at three years. This suggests that rindopepimut in this disease setting had an advantage of 5.8 months of overall survival.

The reader should bear in mind that patients treated with rindopepimut in ACT-III had gross total resection and because they were disease progression free three months after surgery, were a very healthy group of patients within the glioblastoma patient universe. This was a group that was most likely to have good outcomes.

ACT-IV Phase III Trial in Newly Diagnosed EGFRvIII-Positive Glioblastoma
Celldex initiated ACT-IV, a phase III study of rindopepimut, in December 2011. This is a randomized, double blind study of surgically resected EGFRvIII-positive glioblastoma patients. They are randomized to SOC or SOC plus rindopepimut. In this latter group, rindopepimut plus GM-CSF was given as a priming dose after radiation followed by temozolomide and rindopepimut plus GM-CSF given as maintenance therapy. The primary endpoint is median overall survival.

In ACT-III patients underwent image verified gross total resections. This was intended to minimize the immunosuppressive effects of the tumor mass. As a significant percentage of glioblastoma patients do not receive gross total resection, the efficacy of rindopepimut in less resected patients is unclear. To answer this question, patients with incomplete resections will also be enrolled in ACT-IV and results will be stratified for degree of resection.

ACT IV will enroll up to 440 patients at over 150 centers worldwide to recruit approximately 374 patients with gross total resection to be included in the primary analysis. Celldex expects to complete patient accrual by the end of 2013 and anticipated topline results in the trial in 2H, 2015. The estimated cost of the trial is $60 million.

ReACT: Original Plan for Phase II Trial in Recurrent EGFRvIII-Positive Glioblastoma
Also in December 2011, Celldex initiated ReACT, a phase II study of rindopepimut in combination with Avastin in patients with recurrent EGFRvIII-positive GB. These are patients who have recurred following SOC once or twice. As ReACT was originally designed it was to have 95 patients divided into two groups. The first group was to be comprised of 70 Avastin naive patients and the second group was to have 25 patients refractory to Avastin. The 70 patients in group one were randomized one to one. Patients in one arm received rindopepimut plus Avastin plus GM-CSF. Patients in the second arm were to receive KLH plus Avastin. The second group of 25 patients was refractory to Avastin and they were given rindopepimut plus Avastin plus GM-CSF.

In addition, researchers at Stanford University are conducting an investigator sponsored, pilot trial of rindopepimut in pediatric patients with pontine glioma. Patient enrollment is ongoing for this trial.

New Trial Plan for React
Celldex announced on August 12, 2013 that it had completed enrollment in the initial cohort of 25 patients who were refractory to Avastin. Based on preliminary evidence of stable disease, tumor shrinkage and investigator-reported response, the Company has decided to add an expansion cohort of approximately 75 patients to better characterize the potential activity of rindopepimut in this refractory patient population. The full results in the 25 patients will be reported at the Society for Neuro-Oncology Annual Meeting in November.

The ReACT study will now enroll approximately 175 patients across two groups. Group 1 will be approximately 75 Avastin naive patients randomized to receive rindopepimut plus Avastin plus SOC versus Avastin plus SOC. The other 100 Avastin refractory patients will receive rindopepimut plus Avastin in a single arm study. The primary endpoint of ReACT is 6 month progression free survival rate and secondary endpoints are objective response rate, overall survival and safety and tolerability.

Pfizer Agreement
Celldex entered into an agreement in April 2008, which provided an exclusive worldwide license to Pfizer for rindopepimut. In November, 2010, Pfizer returned all rights to rindopepimut to Celldex which took over responsibility for clinical development. Pfizer originated the ACT III phase II trial of 65 glioblastoma patients while Celldex had earlier done the ACTIVATE trial of 18 patients and the ACT II trial of 22 patients.

 

Tagged as CDX-011, CDX-1127, CDX-1135, Celldex Therapeutics Inc., CLDX, Rindopepimut + Categorized as Company Reports