Amarin AMRN W
July 28, 2006 - 11:29am EST by
2006 2007
Price: 2.40 EPS
Shares Out. (in M): 0 P/E
Market Cap (in $M): 190 P/FCF
Net Debt (in $M): 0 EBIT 0 0

Sign up for free guest access to view investment idea with a 45 days delay.



DISCLOSURE:  We and our affiliates are long Amarin (AMRN), and may long additional shares or sell some or all of our shares, at any time.  We have no obligation to inform anybody of any changes in our views of AMRN.  This is not a recommendation to buy or sell shares. 



Amarin (AMRN) is a drug development company that 9-12 months away from completing and releasing the results of a Phase III trial for their Huntington’s Disease drug, Miraxion, that could dramatically transform the company over the next few years from $2.50 a share to a likely valuation of $18 with a possible valuation of more than $50 dollars per share in a favorable scenario. 


What makes Amarin attractive is that is the second time at bat for their drug application and we can statistically study the past to predict the likelihood of approval this time around.  On the other hand, Miraxion’s active agent is similar to over the counter (OTC) Fish Oil supplements and combined with a checkered company history and no US analyst coverage, it’s easy to pass on the idea, masking an excellent and researchable risk reward for those willing to dig a little deeper.  These key factors have led to an incredible mis-pricing and an investment that presents a compelling opportunity as we explain below. 



Huntington’s disease (HD) Primer:

HD is a a neurodegenerative (brain wasting) lethal, & genetic orphan disease (not common) affecting 30k patients in the US with 150-200k patients at risk.  Typical onset occurs around the age of 30 and death is certain within 10-20 years of onset.  HD destroys cognitive abilities and motor functions, leading most victims to die of complications such as breathing or choking. 


Huntington’s is believed to be caused by abnormal protein functions in the brain that break down neural communication.  This is measured by the CAG repeat score – essentially the level of abnormal cell repetition in the brain, or the biochemistry disruption.  The CAG score can be accurately measured with the normal US population having a CAG repeat score of 20-30x while HD patients have at least 36x and can go as high as 128x.  The only treatments available today offer little remedy and can only mask the side effects of the disease.  The market for HD treatment is ~ $2.5B market in the US. 



Drug Profile:

Miraxion, or Eicosapentaenoic Acid (Ethyl-EPA) is an essential fatty acid of the Phospholipids family.  Phospholipids are responsible for maintaining membrane structure, protein functioning, cellular signaling, and gene regulation.  As skeptics note, Miraxion is a purified and controlled form of the omega 3 fatty acid that can be found in OTC Fish Oil supplement pills at GNC etc.  Miraxion is believed to help the patient by improving some of the motor functions & cognitive processes that are quickly eroded by Huntington’s, though it is not a miracle drug. 


1st Time at Bat. A History of Miraxion Drug Trials:

In 2000 Miraxion went through a series of trials that were completed in 2003.  Initial results showed an ambiguous benefit overall and the primary endpoint, Total Motor Score 4 (TMS-4) improvement over placebo, was not met with statistical significance in the“intent to treat” (ITT) patient group (135 patients), or the total trial population.  The FDA approves or rejects New Drug Applications solely on the results against the primary endpoint.  Although Miraxion did not meet the primary endpoint, benefits were shown in subsets of the patient population.  Importantly, Miraxion was shown to be remarkably safe and well tolerated.   


There are plausible explanations for why the results were sub par and why the current trials have a much better shot at producing a statistically significant result.  To start, a large number of patients in the ITT group did not follow trial procedures and jeopardized results.  Some examples include:  9 patients failing to take the full dosing regimen, 9 taking prohibited drugs during the trial that interact with Miraxion, skewing results against placebo, and 20 patients waited too long to get screened for final results.  If you back out these patients who did not follow protocol, you are left with the Per Protocol (PP) group of patients who followed the guidelines accordingly.  As would be hoped, The PP group, (83 patients) scored much better than the ITT group and came very close to reaching statistical significance.  Statistical significance in pharma land is considered P value < .005. 


Interestingly, after the trial was published, an even more thorough analysis of the data was performed that broke the patients down into both PP/ITT groups and by ranges in the CAG score.  This breakdown showed an even more profound result, with the PP group in the less than 45x CAG scoring camp displaying remarkably positive and statistically significant results P = .006.  The map of scores per group are presented below and an explanation of the significance will follow:


Group Type          Patients N =          CAG Score            TMS-4 % improvement     Statistical Sig.       Length


ITT – all                135                         All                           na                                           No                           6 mos

ITT                         67                           < 45                        na                                           P = .025                 6 mos

ITT                         67                           < 45                        15.20%                                  P = .029                 12 mos


PP - all                    83                           All                           na                                           No, but close         6 mos

PP                           44                           < 45                        22.7%                                    P = .006, Yes!       6 mos    




The map above shows that overall the trial looked weak on primary endpoints but that there was a significant and meaningful response in the subset of patients who followed the protocol and had a below 45x score on their CAG testing.  We do not want to exaggerate the improvement, 15-20% is good, but not incredible, especially given the fact that motor function is improved but the HD progression is not actually reversed.  Regardless, it is an improvement in a disease where every benefit counts in the patients view, and a benefit they are willing to pay to experience.  Since we now have the data for their first time at bat showing a subset that scored well enough to pass FDA standards, two questions must be asked to decide if the historic data can be statistically analyzed to predict the outcome of a future trial. 


1) Do the results make logical sense given what is currently known about Huntington’s Disease, or is this a random response in a small group (n=44) of patients?


2) Is this an over fit of the data (dropping out all the patients till a certain outcome appeared), or is an actual positive response occurring? 


Our research suggests the results are logical and the data was not over fit:



1) Do different outcomes for each CAG scoring group make sense?

Much research shows that the CAG score is directly related to age onset of the disease and to some degree, the severity of onset and how rapidly the brain function is attacked.  The following CAG scores are rough estimates of CAG score vs. age onset:

CAG score  /  ~ Age onset of HD

20-30 =  normal population score and no onset of HD

36-39 =  60 years

45 = 37 years

55 = 20 years

65+  = childhood onset, extremely severe


It is believed that with a high CAG score and early onset comes a more severe onset of the disease and a more abnormal brain chemistry.  Experts we worked with believe that the Miraxion results makes sense scientifically, as a severely impaired brain might not respond well to treatment, where as a moderately impaired brain could have a more dramatic improvement.  This is made even more evident when looking at the primary endpoint of TMS-4, which measures motor function, cognitive abilities, behavior, and functional abilities.  A severely impaired brain experiencing rapid onset of HD might degenerate too quickly for a positive response to be seen. 


2) Has the data been over fit or doctored?

Given the sheer magnitude of drop outs and the proportion of dropouts to per protocol patients, on the surface the Miraxion trials do seem to suffer from doctoring or over fitment.  For example, a recent trial we have studied at a large pharma co. had 1000 patients and less than 20 drop outs.  Given the unique circumstances of Huntington’s disease the drop outs seem reasonable.  1) Roughly half of Huntington’s Disease patients are already taking supplements including Fish oil.  To go 12 months without taking your supplement (if you believe you’re getting placebo sugar pills) would easily compel a patient to break his/her protocol every now and then.  Frequent blood tests of patients can show if a placebo patient is taking a EPA related supplement that would easily skew placebo to control group results. Since patients were outside of a controlled setting like a hospital room given the long duration of testing @ 1 yr, it makes sense that some dosing was skipped or neglected.  Lastly, not being tested in a reasonable period of time for final results can be rationally explained, as a rare disease like Huntington’s makes population density harder to come by.  Some patients were located hours from the testing center and out of reasonable reach, as was the case here. 


Skeptics at heart ourselves, we also decided to consult with doctors who were very close to the original trials and have analyzed the data independently. It was their opinion that the data had not been over fit and that there was indeed a strong response for a subset of patients. 


It should be noted that there are blockbuster drugs today whose mechanisms of action (MOA) are not well understood.  Blockbuster Psychiatric drugs for instance, working through similar mechanisms as Miraxion in the brain, are still not well understood after decades of use.  Sometimes, with a promising compound, it’s simply a matter of testing enough patient populations till the desired result can be demonstrated in a test and accepted by the FDA.  We think that statement has some application here. 


2nd Time at Bat.  Application of old facts to new trial:

After analyzing the historical trials and the plausible reasons for results it is then possible to take the old results as factual evidence and apply it to a framework for predicting the probability of future approval.  Since we have now seen that the subset of CAG < 45 of patients yielded a strong and positive result, the current trial design must focus on enrolling a high number of low scoring CAG group patients to produce a similar result.  The problem with testing patients prior to enrollment and only accepting CAG < 45 patients is that it would be viewed as coercive and not all patients want to have their CAG score tested.  Therefore, the following changes have been to the enrollment criteria to enrich the trial with the better responding patient pool. 


Comparing the original trial enrollment criteria to the new trial patient enrollment criteria (which can be found at it is clear that there are 4 new items/changes all aimed at targeting a lower scoring CAG patient mix, including:


1) Modified age eligibility to minimum of 35

2) Total Functional Capacity (TFC) score exceeding 7  

3) Dystonia score less than 2

4) Chorea score greater than 2


Most importantly, management claims that in close preparations with the Huntington’s Study Group, who is running the current trial, Amarin was given access to their database of thousands of Huntington’s Disease patients.  Using these parameters against that database of patients medical scores and background, yielded a 90% probability of the patient having a low < 45 CAG score!  Although we believe their claims, we studied the enrollment criteria with professionals to see how much factual evidence there is for an enriched patient population of low scoring CAG counts in the new trial. The findings backed management’s claim.   


Higher age onset combined with only accepting stage 1 and stage 2 HD patients should naturally exclude high CAG scoring patients that are less likely to see a response.  This age was bumped up from “at least 30” in the last trial. Referring back to our age onset vs. CAG score should demonstrate how age and CAG score can enrich the trial population to below 45 CAG.  Under a few assumptions, the available statistical baseline information in the TREND trial base line frequencies for TFC can be statistically analyzed for the current trial.  With the new parameter of greater than 7 TFC score, it fits that 66% of the patients will fit the lower CAG group, using patient scoring and breakdowns from the previous trial.  While the multivariate statistical data is not available for the Dystonia score and the Chorea score to conduct a similar analysis, all of the literary material suggests that the scoring on these eligibility criteria will further improve the odds of enriching the trial with low CAG score patients.  We don’t think a 90% enrichment can be far off from the truth. 


If 90% of the patients are in the low CAG group, further statistical analysis of the old trial results would suggest that the current trial has between a 95 and 74% chance of passing the FDA requirements.  These two scores are based on results from all 6 original testing centers and 5 of the 6 testing centers, respectively, and using two statistical methods of deduction.  (Note - the publication only provides statistical info to run this analysis on the ITT group, so this doesn’t even take the PP group scoring breakdown into account, which is only released by the co. and not in a fashion that allows for a statistical analysis to be run.  Therefore, higher values would be likely).  For accuracy, we backed out one of the 6 centers, where it was noted in the trial publication the lead physician had a financial interest in Amarin, though we do not believe foul play is involved.  Unfortunately, statistics are not the real world so this # is naturally lower, but high nonetheless.



Other Issues Masking the Opportunity:


Fish Oil vs. Patented Drug, Pricing Power, small end market

One major point masking the opportunity is that skeptics believe the availability of similar OTC compounds will set a low ceiling on the Miraxion pricing.  Our conversations with doctors and understanding of the landscape suggest otherwise. First, Fish oil and Miraxion are simply not the same thing.  The ingredient might be synonymous but the purification process and formula used by Amarin strips away components of natural fish oil that hinder the effectiveness.  Second, HD is a life threatening disease and we’re hard pressed to believe all but a fraction of patients would risk shaving time and comfort from their lives to save a few thousand dollars per year.  If that wasn’t enough, Miraxion should receive insurance reimbursement, further lowering the actual cost to the consumer.  A private equity contact we spoke with who is in the neutraceutical business mentioned other situations where drugs have been approved with rival OTC supplement forms and have been able to receive reimbursement and pricing power of more than 10x the OTC version in non-life threatening applications – what matters is that the FDA says this is the approved formula that is produced in a controlled environment. 


Discussion with Doctors made the case even stronger.  Virtually none of the doctors we consulted with would suggest or approve of their patients taking the OTC version if a FDA approved drug is available.  There is litigation risk, health risk from Mercury and unregulated production, and the risk of the overall benefit experienced.  It’s common to go against the “doctors orders” but not likely in this situation.    


Although the low CAG scoring patients represent 70% of the population 30k diagnosed patients in the US, we also believe the high scoring patients are likely to take the drug, despite what could be restrictive labeling (from the focused trial population).  Again, the HD patient is in such need that they are willing to take almost anything that provides a benefit. 




Given the binary risk we look at this investment using a probability tree.  It should be noted that with a small, well-informed, and close knit population of Huntington’s disease patients the marketing expense will be tiny, yielding a 75% EBIT margin.  Effectively, the Huntington’s study group and the 60 odd centers running the trial will act as a “marketing staff.”  International marketing efforts will be headed by their partners, with AMRN taking a royalty cut of 20% on EU sales and allowing Amarin mgmt. to focus on the US.  Mgmt estimates the US HD opportunity at 250-500m in sales and of similar size in EU.




EBIT & 75% , fully taxed, EPS @ 20x multiple, 1.5 yr present value adjusted


1)       Major HD use + good off label use

Annual treatment price:             15K

Patients:                                        57k

Sales:                                             851m

Value per share:                           $88

Probability:                                   %5


2)       Strong HD use + medium off label

Annual treatment price:             12K

Patients:                                        36k        

Sales:                                             437m

Value per share                            $45

Probability:                                   15%


3)       Good HD use + little off label

Annual treatment price:             10K

Patients:                                        26k

Sales:                                             256m

Value per share                            $26

Probability:                                   23%


4)       Moderate HD use + scant off label

Annual treatment price:             5k

Patients:                                        11k

Sales:                                             54m

Value per share                            $6

Probability:                                   17%


5)       Unsuccessful drug, trades to .75 cents & able to raise cash / try again / pursue other drugs

Probability:                                   40%



All in, the probability adjusted value per share is about $18 dollars, but we believe the most likely scenario is #3, at $26 per share.  We think longer term there is a 60% chance of approval and a 40% chance of rejection. 


Competing Drugs:

There are a number of drugs approved today as therapeutic treatments for HD, similar to Miraxion, but all of them have severe side effects and are not heavily used.  These include Haldol, Profenazine, and potentially, Tetrabenzanine, if it gets approved in the US (approved in the EU and only generates 5-10m in sales, approvable letter in the US right now).  The first two sell for extremely cheap prices, although have severe side effects sometimes much worse than HD itself, but can have potent results in those cases where the side effects do not occur.   Doctors expressed more interest in prescribing Miraxion if it is approved over these other compounds given the safety to efficacy profile. 


Amarin has a history that has helped create the opportunity: 

Amarin is an Elan spin out from a few years back.  The history is so ugly that Amarin’s reputation has been long tarnished.  Elan sold a drug to AMRN with seller financing (getting high on your own supply) which immediately went generic.   They stuffed the channel so badly that sales went briefly negative.  Look at a 5 year chart to get a sense of what happened to the Elan stock price if you don’t remember the depth of their mis-dealings, which were not limited to Amarin.  While the ex-Elan scientists at AMRN continue to be well regarded in Ireland the top management and finance team are viewed with skepticism. All that said, investors don’t think or presume that there is any fraud here, it’s more of a show me story. 


We have met with the CFO and CEO on numerous occasions and believe they are honest individuals with integrity.  Though we admit not even the secret service can spot a prevaricator much more than more than 50% of the time, Rick Stewart, CEO, is of the caliber you’d expect at a company 10x the size of Amarin, and our lengthy discussions with him have given us comfort with his intentions.  There is also considerable inside ownership, with nearly 1/3rd the float tied up with insiders.   The pedigree of management, however speckled by former affiliations, is top notch given the size of the company.   


Other points of consideration:

~100 patients from the original trial are still taking Miraxion (not paying), nearly 5 years later.  We spoke to one of the doctors who has more than 5 such patients and he was enthusiastic about the benefit they’ve experienced.


A similar EPA based drug has been sold by Mochida Pharmaceuticals in Japan for 15 years under the name Epadel for roughly $3,000 annual treatment cost.  More than two million people have taken it since launch, mostly for fatty lipid / cholesterol related artery hardening diseases.  If you assume the first 3 years were penetration growth, that leaves normalized annual sales of around 500m to 600m.  We have not been able to verify this with Mochida personally.


Miraxion has a Special Protocol Assessment from (SPA) from the FDA.  This means that the FDA has spent considerable energy in guiding the company through the new trial design using original trial results in hopes of constructing a design that will maximize the probability of success.  We note that the odds of approval in these situations are much more likely than standard phase III trials.  In fact, Management claims that the FDA was so sure in the setup of their new trial, they actually requested some higher CAG score patients be included in the mix so that they can provide more general labeling.  The Huntington’s disease community wants this drug on the market and is pressuring the FDA to approve it.  This will have an impact on the FDA’s decision given it’s extreme safety profile.  It seems there is little holding back the FDA from approving this short of a total miss on the primary endpoint.   


Miraxion is being pursued in other neurological disorders/diseases including a rare form of depression.  We think there is opportunity in these indications and that could help offset the binary outcome of this trial.  We personally know psychiatrists who use fish oil supplements to treat rare forms of depression that do not respond to standard treatments.  They have had periodic success. Depression is a huge market potential, but more a prop on floor value. 


There were 6 trial centers participating in the original trial, 5 of which were independent and one which had financial interests in Miraxion.  The results for both center populations are broken out and while there are differences, there does not appear to be a pattern of biased results and we ran the statistical analysis using both sets and came up with different, yet still encouraging result.    


Miraxion has orphan drug status and will get 7 years of exclusivity in the US – no generics in the EU for 10 years.  This is special treatment the FDA gives to companies who pursue smaller market drugs. 


Ample cash to fund trials / development of ~ 33m.  A probable milestone payment from Valeant of up to 18m.  Chance to sell depression rights of Miraxion for around 8m upfront today with % future sales royalty.  Amarin is holding out to get better data and better terms on a deal for the depressiona application.  + other indications reprsent a nice floor value for the business if it were to quit operations and sell off the assets.   


Even without a specific CAG group breakdown the PP group was close to statistical significance. 







Every drug trial has a considerable amounts of risk.  Even the most bulletproof designs and expectations still fail to meet FDA requirements.


Recently, mgmt has been more open in their outlook / timeline for the trials.  For conservatism sake they believe this current trial will receive an “approvable” letter.  It is our understanding that the approvable letter would only relate to drug labeling.  I.e. – Miraxion only useful in low CAG scoring patients, which would limit rapid market penetration as patients seek to determine their own CAG scores etc.  We note that this is part of the reason why the FDA has actually requested that some higher scoring CAG patients be “let into” the trial, as management has communicated to us before, so that they can avoid restrictive labeling. It seems that this will be a temporary hurdle rather than a longer term nail in the coffin and might be centered on having more safety information (most large trials have 2 or 3 efficacy studies and 1 safety study.  Amarin only has 2 efficacy studies).  It was noted that Mochida Pharmaceuticals, the markets of Epadel in Japan, have agreed to let Amarin use their 300 patient safety study of EPA in patients.  This may or may not be an acceptable medium.  In light of this, the short term voting characteristics of Mr. Market are never easy to predict and this could present volatility if not priced into the market expectations.  We are still working through the significance of this information and may or may not be making a big deal out of what is truly as conservative statement by mgmt.   


Miraxion trials are being run in both the US and EU.  Typically, at least two different trials are needed to yield positive results against a primary endpoint in Phase III testing for the FDA to give the thumbs up.  It is easy to get comfortable with the US testing, the EU is more out of reach and presents more risk in our opinion.  However, both countries have robust requirements. If the US is positive and the EU fails, still a chance of getting approval but would likely need to run one more test either in the US or in the EU.  Likely that the co. had to run the test in the EU and US to ensure a quick patient enrollment – as the strict screening did make it more difficult to find patients for enrollment. 


We could see another large difference between ITT and PP patients due to the difficulty of managing HD patients for such a prolonged period of time that could throw the end results.  We would expect that the patients have all been explained the importance of following procedure but self interests can always conflict.  I.e – thinking you’re on a placebo and supplementing your regimen with OTC fish oil and clouding the comparison vs. the control group.


Eventual approval for Miraxion in HD
Deal with Pharma co. for depression application
    show   sort by    
      Back to top