Expert Financial Analysis and Reporting

Report Overview
Adam Feuerstein launched a withering attack on Agenus (AGEN) and me in his column. The gist of the article is that Agenus is misrepresenting data pertaining to the use of Prophage in glioblastoma in ways that overstate Prophage’s efficacy and I am naively buying into this. It was a very short note on his part without much supporting analysis. This note rebuts most of the points that I believe he was trying to make in his article.

Reality Check on Prophage, Does Current Data Justify Larger Trials
As I interpret his brief note, the gist of his argument is that the data on Prophage is contrived and meaningless and that no credible person would think that Prophage might be effective in glioblastoma. Before I go further, I would point out that my position is that the data on Prophage is encouraging and does justify further development.

I am not guaranteeing that future trials will be successful as no one can do this. There is a high failure rate in oncology in going from what appears to be encouraging phase II trials to larger, more carefully conducted phase IIb and phase III trials. I have not seen any reliable numbers, but as a guess I would think that results seen in significantly more than half of phase II trials done by small companies fail to be replicated in larger studies. Drug development is high risk and I have seen lots of blowups even in huge, well conducted phase III trialsundertaken by major pharmaceutical companies. Feuerstein and I agree on this. However, this does not mean that all trials will fail and with success there can be enormous upside. What I am saying is that there is a reasonable chance of success for Prophage, not certainty.

Feuersteinmaintains that the data does not justify further trials of Prophage in glioblastoma. I will let others judge his credentials and of course we don’t know who else, if anyone, he is relying on. However, we can look at a long list of prestigious institutions that agree with me that the data is encouraging and that larger trials are warranted. And of course his point is already disproved as there is a large phase II trial of Prophage in recurrent glioblastoma involving 222 patients that is being funded by the National Cancer Institute (NCI) and is in progress. Let’s look at how this came to be.

The data that led to this phase II study in recurrent glioblastoma was based on investigator led studies conducted by Dr. Andrew Parsa and not by Agenus. Dr. Parsa, MD, PhD is the Lead Clinical Investigator and Chair of Neurosurgery at Northwestern Memorial Hospital and Northwestern University Feinberg School of Medicine. His earlier studies were not funded by Agenus. They were primarily supported through funding from the American Brain Tumor Association, Accelerated Brain Cancer Cure, National Brain Tumor Society and National Cancer Institute Special Programs of Research Excellence. Dr. Parsa has not received any financial support or expense reimbursement for this work or for consulting activities on behalf of Agenus.

The phase II recurrent glioblastoma trial now underway is only supported by Agenus through manufacturing the product used in the clinical trial. The study is being sponsored by the Alliance for Clinical Trials in Oncology (ALLIANCE), a cooperative group of the NCI. ALLIANCE is actually comprised of three legacy NCI funded cooperative groups: American College of Surgeons Oncology Group, Cancer and Leukemia Group B, and North Central Cancer Treatment Groups. These three groups have been integrated in an effort to develop and conduct more efficient clinical research studies to bring clinical trial results to patients more quickly.

This trial is the largest brain tumor vaccine trial ever funded by the NCI and in order to begin the trial the Prophage data was examined in a rigorous review process. Based on the size and scope of the trial, I estimate that if Agenus were conducting this trial it would cost on the order of $25 million. NCI can do the trial somewhatmore cheaply, but doesn’t provide estimates of trial costs. In any event, it is a pretty big and pretty expensive trial.

I take some comfort in knowing that all of these prestigious institutions support my point of view that the data so far obtained with Prophage justifies progressing to larger phase IIb and possibly phaseIII trials.

Is The OS Data For Prophage in Newly Diagnosed Glioblastoma Meaningful?
Feuerstein’s attack was against data that will be the basis for beginning a phase III trial in newly diagnosed glioblastoma. He did not address in any way the ongoing trial in recurrent glioblastoma. I am not sure why. Perhaps he thinks that encouraging data in a more aggressive form of glioblastoma does not have any significance for potential efficacy in an earlier and less aggressive stage of the disease. We don’t know his thinking on this.

His article was short and not detailed. He hammered home the point that this was a 46 patient trial that was non-randomized and of course had no control arm. In interpreting the results of open label trials it is common practice to look at results obtained from larger trials to determine how the current standard of care performs in the same disease setting. This is the only way to judge if a new drug might be better than standard of care to the extent that a larger randomized trial can be conducted.

There is nothing nefarious about this approach, it is common industry practice. He is quite right in pointing out that there is the potential that the characteristics of the patients in the Prophage trial could be different from that of patients treated with standard of care in different trials. Feuerstein’s arguments can be found in his blog.

His central argument is that the 23.3 months of overall survival seen this far in the Prophage phase II trial in newly diagnosed glioblastoma patients is not encouraging or meaningless when viewed against the results obtained with standard of care in other trials. His thesis is based almost entirely on the premise that the patients in the Prophage trial had their tumors more fully resected than those in the control groups against which it was compared. He concludes that Prophage treated patients are more resectedso that the survival results are meaningless. Let’s look at some data that he did not mention to judge the validity of this argument.

Feuerstein represents that the degree of resection is the only relevant factor in determining median overall survival. I think that all would agree that complete resection is almost certainly going to improve median overall survival if all other things are equal. However, all other factors are never equal and there are other factors such as age and the health of the patient as measured by the Karnofsky performance score (KPS) that can also be very important. He makes no attempt to look at those variables. It is the balance of these factors that is key to the outcome. He also makes no effort to determine how important resection is to median overall survival. Let me take a crack at this.

I have found a phase III trial that prospectively looks at the effect of complete resection Stummer et al. (Neurosurgery 2008).It looked at the outcomes of patients with complete resections versus those with less than complete resection. This showed that median overall survival was 16.7 months for the completely resected group versus 11.8 months for the control group of patients with less than complete resection. Clearly, there is a survival benefit based on complete resection. However, the Prophage data showed 23.3 months of survival in a patient group that was greater than 90% resected. If complete resection were the only factor, according to Feurstein we would have expected 16.7 months of median overall survival. However, this argument like Feuerstein’s argument is incomplete.

There was a subset of 100 patients in the Stummer trial that had complete resection and KPS >90 (they were relatively healthy). In this group, the median overall survival was 17.6 months. Another subset of 45 patients had even better variables that affect outcome. They were completely resected, had KPS>90 and were relatively young with age <60. The median overall survival in this group was 19.9 months. The Prophage patients were about 90% resected, had KPS≥ 70 and the majority were ≥ 55 years of age. Hence the Prophage patients relative to these two groups were less surgically resected, were less healthy as judged by low KPS score and were older. If Prophage had no effect on the disease, one would expect that median overall survival would be less than the 17.6 months in the first group and the 19.9 months in the second group. However, Prophage patients had median overall survival of 23.3 months and as I explain later; the median overall survival with Prophage could improve.

Stummer et al., concludes “Taken together, factors potentially affecting survival such astherapies after resection, KPS score, and NIHSS status werehomogeneously distributed among patients with CR or IR, andthe remaining factors of age and eloquent location wereaccounted for by subgroup analysis and multivariate analysis.Thus, the present analysis provides two highly defined cohortsof patients collected in a truly prospective setting with balanceddistributions of known prognostic factors in the tworesection groups. According to the Oxford Centre of EvidencebasedMedicine (www.cebm.net/levels_of_evidence.asp), sucha cohort study provides Level 2b evidence and, thus, providesthe highest level of evidence yet obtained for a major influenceof complete resectionon survival in patients with glioblastomas.”

Feuerstein refers to the Stupp trial but he does not mention a subset of the Stupp trial that was completely resected (based on surgeon judgment). In this group the median overall survival was 18.8 months. This was a relatively young population in which more than 30% of the population was <50 years old. As a reminder the majority of patients in the Prophage trial were ≥ 55. The KPS in this group was not reported in this subset. Comparing Prophage to this group, the Prophage patients were less resected and were older and the relative health of the two groups is unknown. Prophage would appear to have a less favorable patient group and still showed median overall survival of 23.3 months versus 18.8 months. And again, I would point out that the data with Prophage is maturing and could increase from 23.3 months.

There are a lot of issues in interpreting these numbers. However, it is a strong refutation of his argument that the degree of resection in the Prophage trials can explain all of the improvement in median overall survival. There does appear to be a meaningful real effect of Prophage on median overall survival beyond that which we would expect with a high degree of resection.

Comments on Median Overall Survival
In his comments, Feuerstein gives me the impression that hebelievesthat the 23.3 months of median overall survival that was seen with Prophage in the phase II trials of newly diagnosed glioblastoma is a hard and fast number. If that is his thinking, it is not accurate.

For those not familiar with statistics, the median is the numerical value separating the higher half of a data sample from the lower half. The median of a finite list of numbers can be found by arranging all the observations from lowest value to highest value and determining the middle number, e.g., the median of the series of five numbers 1, 4, 5, 8 and 9 is 5. In this 46 patient trial, there have been 23 deaths. The OS of 23.3 month is the median time of survival for the 23 patients who have died. Eleven have died before 23.3 months, 11 have lived beyond 23.3 months and one died at 23.3 months.

There remain 23 patients in this trial who are still alive. Of these, 9 patients have lived longer than 23.3 months and 14 have not yet reached 23.3 months. The 23.3 month figure can move up or down depending on the time of death of future patients. The recent press release of the Company stated that OS in the trial was 23.3 months which compares to the 23.3 months that was stated in a May press release. Agenus wanted to make sure that there was no deterioration in OS before seeking a meeting with the FDA to discuss a possible phase III trial in newly diagnosed glioblastoma. The constancy in OS of 23.3 months means that there have been no deaths since May 2013.

The question is when will other deaths occur and what effect will that have on OS. We can only wait for the data to mature in order to answer that question. The history of immunotherapy is that it takes a while to take effect. It also seems to be the case that patients who respond to the therapy can experience some very long survivals. This was shown in the Phase I trials of Northwest Biotherapeutics' (NWBO) DC Vax-L and ImmunoCellular's (IMUC) ICT-107 in their Phase I trials. This was also seen with Bristol-Myers Squibb's (BMY) checkpoint inhibitor Yervoy. If this holds true with Prophage, I would expect to see the OS improve as more patients in the trial die. My expectation is that OS will improve from 23.3 months.

Are Cancer Vaccines Bunk, an Historical Perspective?
Feuerstein has a firm belief that cancer vaccines are bunk and have no chance of working. I suspect his views are based on experience with cancer vaccines that we have seen to date. Let me give some perspective on past efforts to develop cancer vaccines and the experience of Agenus with Prophage.

Agenus was founded in 1994; it was a pioneer in immunotherapy and bears the battle scars that are inevitable with the development of potential paradigm changing technologies. Its core technology was based on an understanding of the biological role of heat shock proteins in the adaptive and innate immune response to cancer and infectious disease. The company initially applied this technology to the development of autologous therapeutic cancer vaccines, which are complexes of heat shock proteins with cancer antigens derived from a patient's own tumor tissue.

The heat shock protein cancer program resulted in the development of Oncophage (since renamed Prophage), one of the first therapeutic cancer vaccines. It was progressed into phase III trials in renal cell carcinoma and melanoma. Both trials failed to meet their primary endpoints, causing many investors to give up on the product and the company. The stock suffered and Agenus was left in a financial conundrum, forcing a series of financings at depressed stock prices.

Drug development pioneers get battle scars, and this has certainly been the case with cancer vaccine developers. Many of the pioneers in this space - CancerVax, Favrille, Genitope, Cell Genesis - failed in their trials and went out of business. With the experience of Oncophage and these other pioneer companies, there was a common thread underlying their failures. The conventional way of developing cancer drugs when cancer vaccines first entered clinical development was to test them as single agents against standard of care in advanced cases of cancer. If the drugs were shown to be effective in this setting, they could then be tested in earlier and less severe cases of cancer. With perfect hindsight, this has proven to be exactly the wrong way to develop most cancer vaccines and possibly other drugs based on immune therapy.

As a pioneer in cancer vaccines, Agenus was unfortunately locked into this exactly wrong development scheme. Upon analyzing patient subsets, Agenus was able to suggest that the drug seemed to be quite effective in earlier stage cancers. From a scientific point of view, this was an important observation and suggested that Oncophage was effective when used in earlier stages of cancer. However, regulatory agencies, for very legitimate reasons, will not accept retrospective analysis of patient sub-groups. From a regulatory standpoint, the trials were and remain a failure. A large company would have taken the experience gained in these trials to launch new Phase III trials of Oncophage in earlier stage renal cell carcinoma and melanoma patients. However, as a small developmental stage biotechnology company, this was not an option for Agenus. New trials would have taken several years to perform and Agenus was severely cash constrained; indeed, the Oncophage failure threatened its existence.

The last few years have been ones of struggles for the company financially as it moved from one financing to another and yet never had the financial resources to run the clinical trials needed to test the promise of its cancer vaccines. For a company that was not long on its luck, it found some when a neurosurgeon at the University of California in San Francisco, Dr. Andrew Parsa, became interested in the use of Prophage to treat both newly diagnosed and recurrent glioblastoma.

Dr. Parsa ran investigator led trials in glioblastoma that created data that has led to the start of a 222 patient randomized Phase II study in recurrent glioblastoma that has the potential to be a registrational trial. Both Avastin and Gliadel, which are only modestly effective, were approved for recurrent glioblastoma with studies of similar size so that this trial might be sufficient for registration if successful.

Now, Agenus has the hope that Phase II trials in glioblastoma will lead to success and a potential filing of an NDA for recurrent glioblastoma in the 2015-2016 time frame. Agenus will be meeting with the FDA later this year in order to plan for a Phase III trial in newly diagnosed glioblastoma. This could start in early 2014 and would take about two years to create topline data in 2016. Funding for this trial has not yet been put in place.