ImmunoCellular (NYSEMKT:IMUC) presented an update on the Phase II results of ICT-107 in glioblastoma multiforme or GBM at ASCO in which the authors concluded that pre-specified analyses of MGMT and HLA subgroups (I'll try to explain this later) demonstrated meaningful clinical improvements for ICT-107 patients. The company went on to say that it believes these new findings form the basis for planning a Phase III trial. It plans to discuss its registration strategy with European and US regulators in 3Q, 2014. There is a good possibility of a Phase III study beginning in 1H, 2015. Topline results from this trial could be available in late 2018 or 2019 by my estimates.
This note lays out some initial thoughts on this new data to help investors decide if this is a sound decision; however, mine is not the last word and there is more to be learned from the data. IMUC has suggested that results in two sub-groups prospectively identified in this trial could be the basis for regulatory approval if Phase II results hold up in a larger Phase III trial and I concur. These sub-groups are HLA-A2 positive unmethylated MGMT patients who I estimate are 18% of the broad GBM population and HLA-A2 positive methylated MGMT patients who are 13% of the broad GBM population.
There will be a great deal of discussion about this data and whether it is indeed the basis for going forward with a Phase III trial. The Phase II trial failed to reach its primary endpoint of median overall survival in the broad GBM population. Some investors will accuse IMUC of data dredging to come up with positive results in sub-groups although IMUC states they were prospectively defined in the statistical analysis. Also, the sub-groups were so small that with one exception, the data, which has encouraged IMUC is based on numerical trends in progression free survival and overall survival and not statistical significance.
I think that IMUC will elect to go forward with a Phase III trial and regulatory agencies will concur. I think that it could begin enrolling patients in 1H, 2015, and we could probably see topline data in late 2018 or 2019. The gating factor on undertaking the Phase III will likely be the ability of IMUC to attract enough capital to begin the trial and this may not be easy. Also, there could be concern about the ability to enroll patients as it is likely that Bristol-Myers Squibb's (NYSE:BMY) nivolumab and Merck's (NYSE:MRK) similar drug lambrolizumab, will be starting Phase II trials in GBM and there will be fierce competition to enroll patients.
Although I view the data as positive, I am not moving to a Buy on the stock and have a neutral opinion. The long time to topline Phase III results, financing uncertainties and potentially an intense competition for enrolling GBM patients are issues that are just too daunting. On a more general basis, I find these results encouraging and add further evidence that dendritic cell cancer vaccines have clinical benefit in at least sub-groups of the glioblastoma patient base.
ImmunoCellular and its dendritic cell cancer vaccine ICT-107 were written off as failures in December of 2013 after the Phase II trial in newly diagnosed glioblastoma failed to reach statistical significance on the endpoint of overall survival. However, failure in a clinical trial does not always mean that the drug is ineffective. Trial design can sometimes be the cause of a trial failure that can be corrected and lead to success in subsequent trials.
The recent experience with Acadia Pharmaceutical's (NASDAQ:ACAD) pimavanserin is an example of this. Two Phase III attempts failed which caused the stock after the second failure to drop to $0.67 in November 2011. Then a well-designed Phase III trial based on knowledge gained from the two previous trials resulted in a successful outcome and the stock is currently trading at $20.65. I would caution that the Acadia example is far from the norm. However, it does show that it is possible to use information gained in a failed trial to design a new trial that goes on to be successful.
There is and should be appropriate investor skepticism when a trial fails. This means that management's hypothesis for how the drug works or how active it is in a disease state is wrong and a new hypothesis must be formed. Also, biotechnology managements have so much vested in a developmental drug that they are very reluctant to give up on it because of emotional or financial reasons (they could lose their jobs); they will often perpetuate development of a drug that ultimately is doomed to fail. It is always a hard judgment for investors to make on whether there is a reasonable basis for continuing development of the drug following disappointing trial results.
One important issue for me in considering whether IMUC should pursue a Phase III trial is that I think the Phase II trial of ICT-107, was incorrectly presented by previous management as a potential registrational trial. It was suggested that it might even be stopped at an interim analysis for efficacy. As a result, expectations were extremely high that the trial would reach its primary endpoint of overall survival and a BLA would be filed. The trial more properly should have been presented as an exploratory trial that was intended to provide a signal of biological activity and to gather information needed to plan for a Phase III trial. Had it been represented in this context, investors might well have deemed Phase II results as encouraging or successful. By the way, this is how current management presented the Phase II trial.
I think that there has been a strong consensus view of investors that ICT-107 has failed and any plan of management to do a Phase III trial would be a hopeless, quixotic adventure that would only waste shareholders' money, what little remains. It will be a very difficult challenge for The company to raise this amount of money or even the money needed to begin a Phase III or attract a partner to help in the conduct of the trial. There is a very serious financial risk.
IMUC believes that the data that it has shown in methylated and unmethylated HLA-A2 positive patients is exciting and defines two target populations for a Phase III trial. In order to provide some perspective, let me briefly describe what HLA-A2 positive and methylation of the MGMT gene refers to.
The Importance of the MGMT Gene
The class of chemotherapy drugs called alkylating agents treats cancer by damaging the DNA of cancer cells and blocking their ability to replicate. This is the mode of action of temozolomide, which is part of the standard of care for treating newly diagnosed glioblastoma. The MGMT gene encodes a DNA repair enzyme that can block the activity of alkylating agents such as temozolamide.
If the MGMT gene is active, the damage caused by temozolomide is rapidly repaired. In glioblastoma, the MGMT gene can be inactivated by methylation of its promoter region, which controls gene expression. If the MGMT gene is methylated and thereby inactivated, there is a better response to temozolomide because the tumor has no means to repair the DNA damage. Hence MGMT methylation is a favorable prognostic factor for glioblastoma in the setting of radiation and temozolomide.
The importance of methylated MGMT is shown by looking at trials in which temozolomide was used as part of standard of care. Patients with methylated MGMT had median overall survival of about 22 months and the unmethylated MGMT group had median overall survival of 13 months. Methylated MGMT conveys a very significant advantage for survival. According to IMUC, about 1/3 to 1/2 (I am using the average 42% for some subsequent calculations) of GBM patients have methylated MGMT.
ImmunoCellular recorded the MGMT status of 90% or more of patients in the Phase II trial of ICT-107.
Histocompatibility is a term that is used in organ transplantation to describe the ability of a donor's tissue or organ to be accepted by a recipient (to be matched). There is a group of genes present in all animals called the major histocompatibility complex (MHC); in humans this is referred to as human leukocyte antigens (HLA). The more HLA proteins shared between a donor and recipient, the better the potential outcome of the transplantation.
Human leukocyte antigens are proteins located on the surface of white blood cells and other tissues in the body. Their central role is to recognize the difference from "self" tissue and antigens displayed to the immune system from foreign invaders such as bacteria, viruses and parasites and launch an immune response. Differences can trigger an immune response against antigens. In the case of ICT-107, the human is the recipient of dendritic cells loaded with six peptides that mimic tumor antigens. Because of this, ICT-107 is only histocompatible with certain HLA types. This product design makes ICT-107 only compatible with patients with HLA-A1 and HLA-A2 proteins on their white blood cells. Without this match antibodies will be formed to attack and destroy the dendritic cells of ICT-107.
What Percentage of the Glioblastoma Population are Comprised of Unmethylated and Methylated HLA-A2 Patients?
IMUC issued a press release in December 2013 on the topline results of the Phase II trial and now an update on those results. In those releases, the company did not explicitly state the size breakdowns of the HLA-A2 positive unmethylated and methylated patient groups, which would be nice to know. However, it did give enough information to estimate these numbers so that we can come up with rough estimates. My estimates are summarized in the following table:
|Estimates of Sub-Groups of HLA-A1/ A2 Positive and Methylated and Unmethylated Sub-Groups in the Phase II Trial of ICT-107|
|Number of Patients||% of GBM population|
|ICT 107 + SOC||SOC||Sub-total|
|Unmethylated HLA-A2 patients||29||15||44||17.9%|
|Methylated HLA-A2 patients||21||11||32||13.0%|
|Unmethylated HLA-A1 patients||18||10||28||11.1%|
|Methylated HLA-A1 patients||13||7||20||8.0%|
|Source: SmithOnStocks Estimates|
Here is the information provided in the press releases that I used to make the estimates shown in the above table:
· There were 81 patients on ICT + SOC in the trial and 43 on SOC.
· IMUC says that HLA-A2 patients are about twice as prevalent in the general population as HLA-A1 patients. I am using a 62%/32% breakdown of HLA-A2 to HLA-A1.
· IMUC says that 1/2 to 2/3 of patients have unmethylated MGMT. I am using a ratio of 58%/42% for unmethylated/methylated MGMT status.
· I am assuming that 50% of the overall number of glioblastoma multiforme populations is either HLA-A1 or HLA-A2 positive. In the trial, 278 patients had to be screened to obtain 124 patients (45% of those screened) for randomization. Most of the 55% screened but not randomized, were due to not being HLA-A1 or HLA-A2 positive. I assume that 50% of the GBM populations are HLA-A1 or HLA-A2 positive.
The number of patients who were on ICT-107 plus SOC that were HLA-A2 positive and had unmethylated MGMT can be calculated with the formula:
(number of patients treated with ICT-107 plus SOC) x (percentage of HLA-A2 patients in the trial) x (percentage of unmethylated patients). In numerical terms this is 81 x 62% x 58% or 29 patients.
The number of HLA-A2 positive methylated patients on ICT 107 + SOC is 81 x 62% x 42% or 21 patients.
The number of HLA-A1 positive unmethylated patients on ICT + SOC is 81 x 38% x 58% or 18 patients.
The number of HLA-A1 positive methylated patients on ICT + SOC is 81 x 38% x 42% or 13 patients.
Assuming that the SOC groups break down the same way we can calculate the breakdowns by multiplying the number of ICT + 107 patients by the ratio of 43/81 to find the number of patients on SOC in each sub-group.
Finally, I calculate how prevalent these sub-groups are in the general glioblastoma population by multiplying each group by 50%.
I realize that I am throwing around a lot of numbers that can be confusing. However, this reflects the way that IMUC presented the data. They did not show the above breakdowns but just gave approximate information. I suspect that in the future they will present these breakdowns but for the present I have had to go through these mathematical gymnastics in order to get a perspective on the sub-groups they are talking about in which they presented data.
Sub-Groups Analyses Are Based on Per Protocol Patients
Further exasperating those of us who would like more precision, the data reported on sub-groups was based on per protocol treatment. There were 124 patients randomized in the trial, which is called the intent to treat or ITT population. The company reported on 117 per protocol or PP patients, which were defined as patients who received at least 4 injections. With such small numbers, the difference in ITT and PP patients might be noticeable but I have no way of estimating this. In doing this, IMUC may be presenting the results in a more favorable light as regulatory judgments are made on ITT results. Encouraging data was presented on two sub-groups: methylated HLA-A2 positive and unmethylated HLA-A2 positive patients. Here are the results from the press release.
The Sub-Group of HLA-A2 Positive Patients with Unmethylated MGMT
In the per-protocol (PP) analysis of data from HLA-A2 patients with unmethylated MGMT (SmithOnStocks estimates there were 29 patients on ICT 107 plus SOC in this group and 15 on SOC):
· The control and treated median OS times were 11.8 and 15.8 months, respectively, indicating about a 4-month or 33% numeric survival increase for treated patients (HR=0.612, log-rank p-value=0.175);
· The median PFS times for control and treated patients were 6.0 and 10.5 months, respectively, indicating about a 4.5-month or 75% numeric PFS increase for treated patients (HR=0.758, log-rank p-value=0.442);
· There were also signs of a potential long-term survival benefit for ICT-107-treated patients, with 21% of treated patients still alive compared to only 7% of controls. By SmithOnStocks estimates 23 of 29 patients died on ICT 107 plus SOC as compared to 14 of 15 on SOC.
This group was too small to expect to achieve statistical significance, but the numerical trend in overall survival and progression free survival were numerically better for ICT plus SOC patients. I am not sure how to read that OS was shorter at 4.0 months than PFS at 4.5 months since death constitutes progression and OS should be expected to be longer than OS. There must be some quirk due to small numbers, but this is puzzling on the surface.
That OS and PFS trend in the same direction and thereby support each other is encouraging. In oncology, a 4.0 month improvement in OS is generally considered a significant advance in an aggressive cancer like GBM. These results if replicated in a larger Phase III trial could be the basis for regulatory approval. By my estimates, this sub-group accounts for 18% of the overall GBM population.
The Sub-Group of HLA-A2 Positive Patients with Methylated MGMT
In the PP analysis of data from HLA-A2 patients with methylated MGMT (SmithOnStocks estimates there were 21 patients on ICT 107 plus SOC in this group and 11 on SOC):
· The control and treated groups had still not reached median survival times as of the time of data analysis, with the majority of patients still alive (65% of treated compared to 57% of control patients; SmithOnStocks estimates that 7 of 21 patients on ICT 107 plus SOC died as compared to 5 of 11 on SOC.
· However, the median PFS times for control and treated patients were 8.5 and 24.1 months, respectively, indicating about a 15.6-month or 184% statistically significant PFS increase for treated patients (HR=0.259, log-rank p-value=0.005).
This data shows the dramatic effect of methylated MGMT in prolonging longer PFS and presumably longer OS in the ICT plus SOC and SOC groups. That PFS reached statistical significance in this sub-group, which I estimate as 21 patients on ICT plus SOC versus 11 on SOC is impressive.
The effect of methylation is very much brought out by comparing the deaths and death rates in the unmethylated and methylated groups.
I would expect that if these results could be duplicated in a larger Phase III trial that it would be the basis for regulatory approval. This HLA-A2 positive methylated MGMT group is roughly 13% of the GBM population.
The company plans to have an end of Phase II meeting with the FDA at the end of summer to discuss the design of a Phase III trial. In Europe, there will be advice meetings with regulators from the UK, Germany and Netherlands to discuss the possible design of a Phase III. They will be meeting with the EMA in 2H, 2014 to discuss the possible design of a Phase III trial based on information learned from the Phase II.
The endpoint of the trial will be overall survival. They could start planning the Phase III trials late this year and begin the trial in early 2015. There might be a reason to delay the start of the Phase III if they see something in Phase II that justifies waiting for more data to present to the regulators.
The company had a cash burn of $2.6 million in 1Q, 2014 and ended the quarter with $25 million of cash. The balance of the year could see a comparable cash burn so that the IMUC could end the year with $17 to $18 million.
The Phase III trial could not start before 2015. The design of the trial has not yet been determined, but a possibility is that it could be 300 to 500 patients and could cost $30 to $50 million and take about four years to complete. This will be dependent on the design of the trial.
Disclosure: I am long BMY. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it. I have no business relationship with any company whose stock is mentioned in this article.
Editor's Note: This article covers one or more stocks trading at less than $1 per share and/or with less than a $100 million market cap. Please be aware of the risks associated with these stocks.