Amarin: Today's Market Is Underestimating The Probable Success Of REDUCE-IT

Dec.16.13 | About: Amarin Corporation (AMRN)

Amarin (NASDAQ:AMRN) is a biopharmaceutical company with one drug, Vascepa, on the market since January 2013. Vascepa is an ultra pure omega-3 fat acid product comprising not less than 96% EPA. It is actually marketed and sold as an adjunct to diet to reduce triglyceride levels in adult patients with severe (TG > 500mg/dl) hypertriglyceridemia (= the MARINE indication). In addition, Amarin is developing Vascepa for the treatment of patients with high triglyceride levels (TG > 200 mg/dl and <500 mg/dl) who are also on Statin therapy for elevated LDL-C levels. This indication is referred to as mixed dyslipidemia or the ANCHOR indication.

In February 2013, Amarin submitted a supplemental New Drug Application (sNDA) to the FDA seeking approval of Vascepa for the ANCHOR indication. The PDUFA date for this sNDA is Dec. 20, 2013. On Oct. 16, 2013, the FDA convened an advisory committee to review Amarin's sNDA for the ANCHOR indication, and the advisory committee voted 9-to-2 against recommending approval of the ANCHOR indication. It is probable that the FDA will not approve the ANCHOR indication on Dec. 20, 2013, and will wait for the outcome of the ongoing study (the REDUCE-IT study) before granting approval. The final results of this REDUCE-IT study are expected at the end of 2017.

Before the panel of Oct. 16, 2013 took place, the market capitalization of Amarin was already above $1 billion. Clearly, the approval of the ANCHOR indication (36 million people in the U.S. have TG between 200 and 500 mg/dl) would be a much bigger opportunity for Amarin than just the MARINE indication (3.8 million people in the U.S. have TG > 500 mg/dl). Following the negative outcome of the Adcom of October 2013, the share price of Amarin fell. The market anticipates that Amarin will not have enough cash to finance the REDUCE-IT study, and therefore that it will never obtain approval for the ANCHOR indication.

In this article, I intend to show that there is a significant probability that the REDUCE-IT results could come in early 2016 instead of late 2017. This would allow Amarin to conduct the REDUCE-IT study to completion without additional cash. I propose to demonstrate this in three steps:

  1. Analyze the cash position of Amarin and the cash burn
  2. Calculate the probability of an early and positive completion of the REDUCE-IT study
  3. Discuss the efficacy of Vascepa

1. Cash Position

Amarin had a cash position of $226 million at the end of Q3 2013. I expect that it will end Q4 2013 with $198 million. The management has guided that they will use less than $80 million cash for operations in 2014. If they burn $80 million per year, starting with $198 million at the end of Q4 2013, the cash would be sufficient to run operations until Q2 2016.

2. Statistical Analysis of the REDUCE-IT Study

REDUCE-IT design

Amarin has designed the study in order to detect a 15% relative risk reduction (RRR) in the primary cardiovascular endpoint with 90% power. Based on a placebo event rate of 5.2% per year, it is expected that a minimum of 1,612 primary efficacy events endpoint and approximately 7,990 patients are needed during a median follow-up of 4 years.

90% power, what does it mean?

Suppose I have a large bag containing two million white balls. First I divide the balls into two bags, each containing 1 million white balls. Then, I withdraw 52,000 (5.2%) white balls from the first bag and I replace them with red balls. I do the same with the second bag. I withdraw 44,200 (4.42% = 5.2%*(1-15%)) white balls and I replace them with green balls.

Now, I will draw 1,000 balls from the first bag and 1,000 from the second. Am I sure to draw more red balls than green balls? No. If I repeat the same operation 10,000 times, I will have more red balls than green balls on average for a majority of times. However, if I draw only once, I am not sure this will be the case. The probability is more than 50%, but it is not 100% unless I draw the whole population. Only if I draw 1 million balls from each bag can I be 100% sure to find more red balls than green balls.

This illustrates the problem of biotech companies. They have to show the efficacy of their product with one study, but, of course, they cannot test the whole population. They have to calculate a priori the best number of patients to enroll. If they enroll too many patients, it is too expensive and a waste of time. If they enroll too few patients, the power of the study may not be high enough.

In our example, mathematical formulas exist to calculate 1) the probability of finding more red balls than green balls if I draw a specific number of balls from each bag, and 2) the number of balls that I have to draw from each bag in order to have 80% probability, 90% probability (= 90% power) -- to find more red balls than green balls. Likewise, in clinical studies, formulas exist to calculate the probability to show the specific effect of a drug with a specific number of patients, the number of patients that must be enrolled in order to have 80% probability, 90% probability (= 90% power) -- to show the efficacy of the drug.

REDUCE-IT numbers

  • The patients will be randomized to be treated either with the placebo (statin only) or with Vascepa (statin + Vascepa)
  • The placebo event rate is 5.2% per year
  • There is a 15% RRR (relative risk reduction) for Vascepa compared to the placebo
  • We also have the enrollment data to this date from slide 71 presented by Amarin during the Adcom of October 2013.

Click to enlarge images.

Click to enlarge

This enrollment data will allow us to calculate the timing of the primary cardiovascular events.

Check of My Calculations

Based on the data, I have calculated that the number of patients needed to show a 15% relative risk reduction for Vascepa compared with placebo with a power of 90% is 31,590 (2 groups of 15,795 patients). This would imply 1,519 events. The number 31,590 corresponds to the "7,990 patients during a median follow-up of four years" mentioned by Amarin. While the 1,519 events that I have calculated do not exactly match the 1,612 events mentioned by Amarin, I consider this close enough to validate and continue my calculation. Based on the enrollment data, I expect that the 1,612 events will be achieved in December 2017.

The Interim Analysis

Amarin has mentioned that there will be an interim analysis when 60% of the events are achieved. When will this happen? What is the probability that the outcome would be positive enough to stop the trial earlier?

We need 967 events (60% of 1,612 events). With only 967 events instead of 1,612, the power of the study is no longer 90%. According to my calculation, the power would then be 71%. This means that if Vascepa works to reduce the cardiovascular (CV) events by 15% (relative risk reduction = RRR of 15%), then 967 events would be enough to show this with a probability of 71%. This would happen in January 2016.

The following is also interesting to consider. Imagine that Vascepa is more effective at reducing the CV events by 15% and instead it reduces them by 20% or more. At that point, what would be the power of the test with 967 events? The table below shows what the power of the study at the interim results would be depending on the efficacy of Vascepa. It also gives the expected timings.

Efficacy of Vascepa = RRR%





January 2016



February 2016



February 2016



March 2016

Click to enlarge

We can see that if the efficacy is 19%, then the power with 967 events is 90%, which is the same power of the study with 1,612 events and 15% efficacy. If the efficacy is 25%, the power would be 99%, and the probability of stopping the study at the interim analysis would then be very high.

3. Efficacy of Vascepa

  • Is Vascepa effective to reduce cardiovascular events?
  • If Vascepa is effective, how big is the efficacy?

Since 1999, a few clinical trials have been conducted to evaluate the effect of omega-3 Fat Acids on cardiovascular outcomes. None of these studies was significant enough to answer our question about effectiveness. The studies were either open label studies, or the drug was not the same as Vascepa (EPA+DHA or niacin or fibrates instead of pure EPA for Vascepa). Moreover, in all studies, the population was not as much at risk (TG levels around 150 mg/dl) as the one studied in REDUCE-IT (TG levels > 200 mg/dl).

It is therefore difficult to draw conclusions from these studies. I would like to quote the FDA's reviewer; this quote (p. 87) comes from the FDA briefing information published two days before the Adcom and make three pertinent comments.

Reviewer comment: In the overall population, only JELIS, an open-label trial utilizing 1.8 g EPA daily, demonstrated a positive treatment outcome when added to a low-dose statin regimen. All four trials, including JELIS, did not demonstrate a cardiovascular treatment benefit across baseline TG levels, which may be due to the fact that study populations did not exhibit very high levels of TG at baseline (mean TG 150 mg/dL in JELIS, median TG 162 mg/dL in ACCORDLipid, median TG 161 mg/dL in AIM-HIGH, median TG 125 mg/dL in HPS2-THRIVE). Subgroup analyses from JELIS, ACCORD-Lipid, and AIM-HIGH suggested that patients with elevated TG and low HDL-C might experience a greater potential treatment benefit with additional lipid modifiers to a statin regimen; however, the available HPS2-THRIVE subgroup analyses do not seem to support this hypothesis. Unfortunately, none of these trials were specifically designed to recruit and investigate patients with moderate hypertriglyceridemia without low HDL-C; therefore, these results are hypothesis-generating and require validation.

First of all, the studies that have already shown significant cardiovascular benefits were those with a drug (pure EPA for JELIS) or a population (subgroup of ACCORD-Lipid with TG >204 mg/dl and HDL < 34 mg/dl and subgroup of AIM-HIGH with TG>200 mg/dl and HDL < 32 mg/dl showed a reduction of cardiovascular events of 31% and 36%, respectively) that are the closest to the ones of the REDUCE-IT study.

Second, when the studies showed cardiovascular benefits, the magnitude of the improvement was 19% in JELIS, 31% in ACCORD-Lipid and 36% in AIM-HIGH, respectively. These results were shown during the Adcom (slide No. 24 from Amarin's presentation). If Vascepa is effective in showing cardiovascular benefits, it is reasonable to expect that the magnitude of the effect would be in the range of 20% to 35%.

Finally, the Adcom of October 2013 did not ask about the safety of Vascepa or about its efficacy of reducing triglycerides. The FDA was already confident about those things. Their implicit inquiry was to know if one can predict that the results of REDUCE-IT will be positive by looking at the results of the past studies, or if one must wait for the final results of REDUCE-IT? In my opinion, a panel of statisticians would have been more qualified to answer this question than a panel of MDs. Unfortunately, there was only one biostatistician in the panel. The result of the vote was nine "nos" and two "yeses." Who voted yes? The president and the biostatistician.


Is Vascepa effective in reducing cardiovascular events? If so, then it is reasonable to think that the efficacy could be greater than 15%. This would bring a positive outcome in early 2016 at the interim analysis. In this case, Amarin would have enough cash and no need of new financing. Therefore dilution of the shares would not be necessary. What would Amarin be worth then? $2 billion? More than that? Today's market capitalization of Amarin is $300 million. If you divide the latter by the former, you have the probability of success that is priced today. Based on my previous arguments, it is clear that today's market significantly underestimates the probable success of REDUCE-IT.

Disclosure: I am long AMRN. 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.