Bearish AMRN Report (part 2)
: by Medical Research Collaborative, LLC (sample pages) https://medicalresearchcollaborative.com/reports 1.5.1 Log-transformed hs-CRP Data
A line in the supplement to the NEJM paper on REDUCE-IT cites the result of an analysis of log-transformed hs-CRP data, which showed a highly significant 21.8% reduction in IPE group, and no change from baseline in the placebo group. The sponsor’s website provides the following explanation:
Extreme outliers due to infections caused by temporary illness or other factors can heavily influence summary statistics of hsCRP, even beyond what is handled by using a non-transformed data approach (e.g., a conventional mean or median on a nominal scale). These individual outlier results can affect a mean or median population measurement in a way that can convey a misleadingly skewed result for the population studied. For this reason, a more reliable log transformation of hsCRP is used to incorporate outlier data appropriately within the context of the entire data set.
However, with a study as large as REDUCE-IT, and with nine separate fasting lipid panels performed on average per subject to ascertain levels, log-transforming biomarker data is unnecessary, and analyses based on such data could be misleading.
The Central Limit Theorem (CLT) states that the sampling distribution of the mean of any independent, random variable will be near normal if the sample size is large enough. Thus, the means of large samples will closely follow a normal distribution, whatever the distribution of the observations themselves. Generally, this is considered true in samples of modest size where skewness and kurtosis are low, but certainly in samples of over 500 subjects when skewness and kurtosis are high.
As the CLT stipulates that for large samples the mean can be approximated by a normal distribution even if the population is non-normal (i.e. data are skewed), then if the sample size is large, it is appropriate to use a t-test. With over 4,000 subjects per arm in REDUCE-IT, and with the added benefit of so many repeat lipid panels performed, untransformed hs-CRP data will follow a normal distribution (mean and median approximately equal). Essentially, REDUCE-IT is over-sized and over-powered when doubling as a biomarker trial, i.e. there were 1,612 major coronary events, but >8,000 biomarker “events.” The percent changes noted in the study are based on median values, which reflect the central tendencies of the robustly sized groups. And because of the large size of each treatment group, the geometric mean and median hs-CRP values will be near equal, and the kurtosis near 3. Skewness will also affect both groups approximately equally, and thus, any subsequent regression to the mean will equally impact both arms.
As the previous citation also enumerates, log-transformation has been shown to have major drawbacks. It can cause right-skewed data to become left-skewed, and may even increase skewness; it can often increase—instead of reduce—the variability of data, whether or not there are outliers; it can lead to inaccurate estimates of the true population mean of the original data; it can cause significant errors in hypothesis testing, and is prone to numerous user errors as well. There is also little value in comparing the variability of original versus log-transformed data because they are on different scales. Log-transformed data therefore cannot usually facilitate inferences concerning the original data, since it shares little in common with the original data. Therefore, an analysis based on classical statistical methods or generalized estimating equations is most appropriate.
Lastly, and as was also seen in JUPITER and CANTOS, hs-CRP values were highly consistent between lipid panel tests over years in both arms of REDUCE-IT (after the initial post-randomization changes had occurred). It has previously been shown that hs-CRP levels are as consistent over time as total cholesterol and blood pressure. This adds to the reliability of interpretations based on observed levels.
The above considerations make drawing conclusions based on log-transformed hs-CRP data from the REDUCE-IT trial potentially misleading. Comparing the percent median change between groups using untransformed hs-CRP data is an accurate and reliable way to analyze these data.
*** 1.7 Confirmation of an Adverse Impact of Mineral Oil on Drug Absorption
The elevation of all atherogenic and inflammatory markers measured in MO placebo group in the REDUCE-IT trial was previously observed to a similar degree in a separate study—the ANCHOR trial, also sponsored by Amarin Corp.
In ANCHOR, over 700 subjects (all of whom were on background statin therapy) were randomized 1:1:1 to either 4 g/d MO arm, 2 g/d EPA arm (which included 2 g/d MO in addition to 2 g/d EPA), or 4 g/d EPA arm. The trial was designed with an extensive lead-in period to stabilize lipid and lipoprotein values before randomization. The median percent changes in parameters from baseline to 12 weeks per group were as follows:
An extensive wash-out and stabilization period should help control against errant swings in markers between the three robustly sized groups (approximately 230 randomized per arm). The lead-in period protocol details from the ANCHOR trial are outlined below:
Eligible patients who wished to participate provided written informed consent, underwent a fasting blood draw, received dietary counseling on implementing the NCEP Therapeutic Lifestyle Changes diet, and initiated either a 4- or 6-week lead-in period depending on whether either a washout of a non-statin lipid-lowering therapy or an adjustment to the background statin was necessary.
Patients who did not require washout of non-statin lipid-lowering therapy: The screening visit occurred at Visit 1 (Week -6). Eligible patients entered a 4-week diet lead-in period and continued on their current dose of statin before the first TG/LDL-C qualifying visit (Visit 2/Week-2). Patients who required a change in their statin dose during the 2 weeks following Visit 1 entered a statin stabilization period so that the statin dose was stable for at least 4 weeks before the first TG/LDL-C qualifying visit (Visit 2/Week -2). At the discretion of the investigator, patients could be switched from a non-study statin to an allowed statin at Visit 1.
Patients who required washout of non-statin lipid-lowering therapy: The screening visit occurred at Visit 1 (Week -8). Eligible patients began a 6-week washout period before the first TG/LDLC qualifying visit (Visit 2/Week-2).
Qualifying period: At the end of either the 4-week or 6-week lead-in period, eligible patients had fasting LDL-C (calculated with Friedewald equation) and TG levels measured at Visit 2 (Week -2) and Visit 3 (Week -1).
After examining the ANCHOR trial data, an FDA reviewer stated the following in the briefing documents:
The changes in lipid and lipoprotein parameters from baseline to Week 12 in the mineral oil placebo group are rather atypical for a trial that included a stabilization period for diet and lipid-lowering therapy, raising the possibility that mineral oil may not be as inert as assumed. If true, the treatment effects observed with AMR101 [Vascepa] may be overestimated.
In his book Dyslipidemia: Pathophysiology Evaluation and Management
, 2015., Dr. Garg writes,
Compared to placebo, LDL-C increased by 3.5% in the COMBOS trial using Lovaza, and decreased by 6.2% in the ANCHOR study using Vascepa 4 g/d. Compared to baseline, however, there were no differences in LDL-C in either trial. In other words, the difference in LDL-C response between Lovaza and Vascepa is due to the different LDL-C responses to placebo. In the Lovaza studies, lipids in the placebo arms were either unaffected by treatment or decreased to a small extent, whereas in the Vascepa studies, lipids increased more than would be expected (e.g. 9% increase in LDL-C and 6% increase in TGs in the placebo arm of the ANCHOR trial). The reason for the differences in placebo responses in unknown. It is possible that the Vascepa study populations became less adherent to lifestyle and/or pharmacologic therapy over the course of the trial. However, the elevation of atherogenic lipids in the placebo arms of the Vascepa trials raises the question of whether the light paraffin oil could have interfered with effectiveness of concomitant therapies.
…in ANCHOR, apoB increased 4 mg/dL on the 2 g/d dose [comprised of 2 g/d EPA + 2 g/d MO], decreased 3 mg/dL on the 4 g/d dose, and increased 7 mg/dL on the placebo. As noted previously, the reason for the consistent and substantial increase in apoB while on placebo is not apparent. Thus, one can conclude that either Vascepa (EPA-only) has apoB lowering effects not seen with EPA + DHA, or more likely, the placebo in these trials (light paraffin oil) was not wholly inert.
Although the most profound elevations in markers were observed in the 4 g/d MO arm, the 2 g/d arm exhibited a muted effect compared to what might be expected from this dose of concentrated EPA. For example, in the ORD study, comparing TAK-085 (a POM3) with EE-EPA, the 1.8 g/d EPA arm (n=195) showed an 11.2% reduction in triglycerides at the end of 12 weeks of dosing. However, in ANCHOR, 2 g/d EPA combined with 2 g/d MO resulted in a more tepid 5.6% reduction in TG. Importantly, mean TG levels at baseline were similar between the two studies (272 mg/dL in the ORD study and 262 mg/dL in ANCHOR).
In fact, the 5.6% reduction in TG in the 2 g/d arm is similar to the reduction seen in the 4 g/d olive oil placebo arm in the ESPIRIT trial, which enrolled subjects of similar background therapy and baseline TG levels as ANCHOR. There, a 5.9% reduction from baseline in TG level was observed in placebo group.
Was concurrent dosing with MO somehow negatively impacting the efficacy of EPA, or perhaps hindering statin absorption, thus attenuating the overall TG lowering effects observed in the ANCHOR 2 g/d group? The same incongruency is observed in other markers as well, for example non-HDL-C was reduced by 5.7% in ORD (1.8 g/d arm), but increased by 2.4% in ANCHOR (2 g/d arm).
From the MARINE trial data (~25% of whom were on background statin therapy), which also tested 4 g/d EPA against 4 g/d MO or 2 g/d MO+2 g/d EPA in subjects with very high triglycerides (>500 mg/dL), an FDA reviewer commented on the wild fluctuations in mean triglyceride levels in the two arms that received MO, while no fluctuation in the 4 g/d EPA arm that received none. This is consistent with the notion that MO causes some form of interference with drug absorption.
Reviewer Comment: Although the Vascepa 2g dose reduced TG, there were wide fluctuations in TG levels. By Week 11, the slight improvements in TG levels achieved at Week 4 were reduced back to almost the Baseline TG. Within one week (from Week 11 to Week 12) the mean percent change in TG changed from 0.20% to -9.78%. Wide fluctuations in TG were also observed in the Placebo group.
The Vascepa 4g dose reached maximum effectiveness by Week 4 and despite a slight decrease at Week 11 and Week 12, showed none of the wide fluctuations seen in Vascepa 2g or Placebo.
Interestingly, in the ANCHOR and REDUCE-IT placebo groups there was a 4.8% and 4.7% increase in HDL-C respectively, both significant. Counter-intuitively, HDL-C has been shown to increase more from a lower versus higher atorvastatin regimen.
A meta-analysis of 32,258 patients of dyslipidemia was done by Philip J. Barter et al. It included 37 randomized studies of different types of statin (Rosuvastatin, atorvastatin, and simvastatin, etc.). Effect of these statins on HDL cholesterol level was assessed in this meta-analysis. They found that increase in serum HDL cholesterol was inversely related to dose of atorvastatin. There was 4.5% increase in serum HDL cholesterol with 10 mg dose while there was a 2.3% increase in serum HDL cholesterol with 80 mg dose of atorvastatin. It was concordant with our study which showed percentage increase in HDL-Cholesterol in the range of 9.52 ± 30.07 and 11.36 ± 28.62 at 3 and 6 months follow-up respectively in 40 mg group. While in 80 mg group, percentage increase in HDL-Cholesterol was 7.74 ± 26.43 and 9.02 ± 27.47 at 3 and 6 months follow-up respectively.
The increase in HDL-C in both placebo groups might therefore be expected if the dose was effectively lowered due to malabsorption. In the MARINE trial, where the vast majority were not on statin therapy, HDL-C was unaffected in either the 4 g/d MO or 2 g/d MO+2 g/d EPA groups.
In the REDUCE-IT trial, highly significant increases in atherogenic markers also appeared in the MO placebo group quickly (at the next blood draw 4 months later) and remained elevated throughout the study. Meanwhile, in those randomized to IPE arm, there was little change (i.e. LDL-C moved down slightly from 85.8 mg/dL at baseline to 83.6 mg/dL at month 4, then back up to 85.3 mg/dL at year 1 and to 85.5 mg/dL at year 2)—except in markers known to be impacted by EPA, such as TG and, to a lesser extent, non-HDL-C.
Regarding the differences in risk reduction seen between subjects on low (HR 1.12), medium (HR 0.76) and high-intensity (HR 0.69; p=0.12, significant for interaction) statins in the REDUCE-IT study, a related phenomenon surfaced in ANCHOR.
Compared to placebo in ANCHOR, IPE 4 g/day significantly decreased hsCRP levels in patients receiving higher (-29 %, p < 0.05) and medium (-23 %, p < 0.01) but not lower-efficacy statin regimens (+4 %, p > 0.05).
However, this was “compared to placebo.” The “decreased hs-CRP” as stated was not due to an observed reduction in hs-CRP levels in IPE group from baseline (-2.4%, p>0.05), but instead due to the highly significant increase in hs-CRP levels in the MO placebo arm. The breakdown is given as follows:
In ANCHOR, the changes from baseline in hsCRP for IPE 4 g/day and placebo in patients treated with atorvastatin were -12 and +31 %, respectively, resulting in a statistically significant placebo-adjusted reduction of 37 % (p = 0.0475). The changes from baseline in hsCRP for IPE 4 g/day and placebo in patients treated with rosuvastatin were -1.2 % and +15.2 %, respectively, resulting in a statistically significant placebo-adjusted reduction of 31 % (p = 0.0217). The changes from baseline in hsCRP for IPE 4 g/day and placebo in patients treated with simvastatin were 0.0 and +13.2 % respectively, resulting in a statistically non-significant placebo-adjusted reduction of 13.6 % (p = 0.0755).
But why would IPE—compared to MO placebo—have such differential effects on CRP, depending on the intensity of the background statin therapy of the subject? We can see that IPE had no measurable impact on CRP levels from baseline in those dosed with it; therefore, whatever impact on CRP is likely coming not from IPE, but from an adverse effect of MO on placebo group subjects.
High-intensity statin regimens produce the largest reductions in CRP levels, followed by moderate and then low-intensity regimens.,  Therefore, if MO attenuated the efficacy of statin therapy, we would expect those on high-intensity statins to show the largest increase in CRP levels, followed by those on moderate-intensity, with the least change in those on low-intensity regimens. And that is what the ANCHOR trial data show.
Regarding the difference in changes in hs-CRP levels between the ANCHOR (+17% at 12 weeks) and REDUCE-IT (+32.3% at year 2) studies, Bonnet et al. (2008) showed that after an initial sharp decrease in CRP levels from statin therapy, high-dose therapy further lowered levels over a 26-week follow-up period; i.e. atorvastatin 10 mg/d produced a 25% decrease in hs-CRP over a 5-week period, which was maintained at 26 weeks (-24.3%), whereas the 80 mg dose produced a 36.6% decrease over 5-weeks, which then continued on to a 57.1% decrease at 26 weeks. Thus, even in this regard, the ANCHOR and REDUCE-IT studies mirror what might happen at different time points if statin therapy was attenuated in one group of randomized high-risk subjects, but not the other.
Having two clinical trials testing the same therapies (IPE vs MO) in back to back succession, reproducing the findings in one other, adds to the reliability of the evidence presented. In this case, the evidence suggests that MO adversely impacted the efficacy of concomitant therapies.
*** 3.2 Critique of Amarin’s U.S. ITC Complaint
Regarding the Complaint brought to the ITC by Amarin against high-purity omega-3 manufacturers (initially dismissed, but being appealed currently), which side-stepped the FDA’s ministerial role on what constitutes a dietary supplement vs. a drug, other authors have previously offered extensive reporting on the issues involved. It is clear to us that high-purity EPA-E was in use as a dietary ingredient well before the Investigational New Drug Application (IND) for Vascepa (no. 102,457 ) was granted. Thus, code 21 USC 321(FF)(3)(B) of the FD&C Act does not preclude high-purity EPA-E as a dietary ingredient in that regard. If Amarin had developed and received an IND for a fully synthetic or otherwise unique EPA analogue (vs. the semi-synthetic prodrug EPA-E), which thereafter was made into and sold as a dietary supplement, they would have a case under 21 USC 321. However, Amarin is arguing more broadly that ethyl ester forms (and thus also rTG forms that use an EE step) of EPA and/or DHA are not dietary supplements as defined by the FD&C Act, and further, that their use historically as such only came after these semi-synthetic versions of EPA and DHA were granted INDs.
Amarin’s Complaint: In August 2017, Amarin filed a complaint with the Commission alleging that certain competitors are falsely labeling or deceptively describing synthetically produced omega-3 products as (or for use in) “dietary supplements” when the products are in fact “drugs” that have not been approved for sale or use in the United States. Appx19–29. (Amarin’s complaint applied only to a small group of synthetically modified products, not to the majority of fish oil dietary supplements.) Amarin alleged that those acts constitute unfair acts or unfair methods of competition under Section 337 of the Tariff Act. Appx24 ¶ 1; see 19 U.S.C. § 1337. Amarin also asserted that those unfair acts violate Section 43(A) of the Lanham Act because falsely labeling or deceptively describing drugs as (or for use in) dietary supplements deceives consumers and others in the supply chain regarding the nature of the product. Appx24 ¶ 1; see 15 U.S.C. § 1125(A)(1). Amarin alleged that its domestic-industry commercial interests were being injured as a result of certain competitors’ false and deceptive representations concerning the nature and characteristics of their imported products. Appx115–126.
To reiterate, FDA defines a “dietary supplement” as that containing one or more “dietary ingredients,” further defined as follows:
As defined in section 201(FF)(1) of the FD&C Act (21 U.S.C. 321(FF)(1)), a “dietary ingredient” is any one of the following:
(A) A vitamin;
(B) A mineral;
(C) An herb or other botanical;
(D) An amino acid;
(E) A dietary substance for use by man to supplement the diet by increasing the total dietary intake; or
(F) A concentrate, metabolite, constituent, extract, or combination of any ingredient described in (A), (B), (C), (D), or (E).
This includes synthetic or semi-synthetic forms of vitamins, minerals and amino acids. However, synthetically or semi-synthetically produced botanicals, and even esterified omega-3s (as well as other esterified nutrients) at some point may not be deemed to be dietary ingredients (FDA hasn’t issued final guidance on this yet). In their natural forms, EPA and DHA are considered essential fatty acid nutrients, fitting in the category: “a dietary substance for use by man to supplement the diet by increasing the total dietary intake…” But in esterified form, they may not be deemed thus.
From the latest FDA-draft guidance on this topic as of 03-2019:
5. If I alter the chemical structure of a dietary ingredient, is the new substance still a dietary ingredient?
It depends. Altering the chemical structure of a dietary ingredient (e.g., creation of new stereoisomers, addition of new chemical groups as in esterification) creates a new substance that is different from the original dietary ingredient. The new substance is not considered to be a dietary ingredient merely because it has been altered from a substance that is a dietary ingredient and, therefore, is in some way related to the dietary ingredient.
In some cases, however, the new substance may independently qualify for one of the dietary ingredient categories listed in section 201(FF)(1) of the FD&C Act. For example, taurine is the end product of the metabolism of the amino acid cysteine. It is thus a metabolite of an amino acid and fits one of the definitions of a dietary ingredient (see 21 U.S.C. 321(FF)(1)(D), (F)). The enzymatic or synthetic processing of cysteine or any other dietary ingredient would be an appropriate method for the manufacture of a metabolite of a dietary ingredient like taurine for use in a dietary supplement...
The language in the above mention seems to suggest that an esterified product is not deemed to be a dietary ingredient, but it is not definitive. It asserts only that the esterified form is a “new substance” and that this “new substance is not considered to be a dietary ingredient merely because it has been altered from a substance that is a dietary ingredient…” It does not explicitly state that this new substance “is not a dietary ingredient,” only that it is a “new substance” that cannot be deemed a dietary ingredient solely (aka “merely”) on the basis that its unaltered form is a dietary ingredient. That does not preclude other rationale for it to potentially be considered a dietary ingredient. As such, some or all esterified products could still fall in line with the requirement to be reported as a New Dietary Ingredient (NDI), which is the main subject matter of this draft guidance document the mention is found in, entitled, “Dietary Supplements: New Dietary Ingredient Notifications and Related Issues: Guidance for Industry.”
As an aside, we wonder here if FDA did not instead mean transesterification  when they wrote “esterification” above, as the latter is also a natural process and frequent result of human metabolism (as is “re-esterification”). In fact, any process (metabolic or otherwise) that results in an ester being made is an ‘esterification process.’ This occurs when, for example, a fatty acid is combined with an alcohol (such as ethanol, glycerol, etc.). Trigycerides are thus fatty acid esters of glycerol, formed as a result of esterification.
The “It depends” mention in the draft guidance above relating to metabolites might therefore still be applicable to esterified forms of EPA and DHA, but not ethyl ester forms, which cannot be achieved as a result of human metabolism (no naturally occurring ethanol). However, so called “monoglyceride omega-3s” would appear to fit the excepted clause., , , , ,  Also, “rTG” forms appear to be applicable, as we will elaborate on forthcoming.
The draft guidance clarifies that if a product was in use before Oct. 1994, it may be “grandfathered in,” and not required to be reported as a “New Dietary Ingredient” (NDI). Yet, the guidance also states that an ingredient must first be defined as a “dietary ingredient” to then be either an NDI or not. Thus, the main question (and it is an open question) is whether or not the FD&C Act precludes ethyl ester forms (and potentially rTG forms that first rely upon a transesterification step) of EPA and DHA being deemed “dietary ingredients.”
Another mention by FDA in a letter to AIBMR Life Sciences brought up in Amarin’s appeal is also relevant.
Based on the information in your submission, it is unclear if "fatty acid esters, derived from anchovy or menhaden oil" which you intend to market under the trade name Provinal™ is a "dietary ingredient" within the meaning of21 U.S.C. 32l(FF)(L). For example, synthetic fish oil fatty acid ethyl esters do not fit within the statutory definition of "dietary ingredient" because they are not constituents of a dietary substance for use by man under section 201(FF) (L)(F). Therefore, FDA cannot determine, at this time, whether your product contains a dietary ingredient that may lawfully be marketed as a dietary supplement.
However, once again we are met with open-ended phrases such as “it is unclear if…” and “FDA cannot determine, at this time, whether your product contains a dietary ingredient (DI)…” FDA clearly acknowledges the ethyl ester forms of EPA and DHA lie outside the statutory definition of a DI, but do not go one step further and outright state they are not DIs. They conclude only it is “unclear” and “cannot be determined at this time.” That at least leaves the door open to a future determination by FDA that EPA-E and DHA-E are DIs.
According to the FD&C Act as interpreted by FDA, synthetic vitamins, minerals, and amino acids are considered dietary ingredients despite their synthetic nature. But that does not mean all other synthetic ingredients are precluded as DIs. For instance, FDA mentions “vanillin” and “cinnamic acid” as synthetic botanical constituents that are considered “dietary ingredients,” due to their long-standing use in food products and very safe track record. The same could possibly be said of ethyl ester omega-3s included in food products for many years now.
In a warning letter to an “ethyl ester creatine” merchant, there is found no mention of the substance itself being deemed as not qualifying as a dietary ingredient, only issues regarding labeling and quality control. If FDA deemed ethyl ester creatine to not be a DI, it would be most straightforward for them to have notified the merchant that they were selling a “drug,” not a “dietary ingredient.” The omission of such is in one respect a concession.
The following from the draft guidance in question may allow reconstituted (aka “re-esterified”) TG (rTG) forms, which return EPA-E/DHA-E to their original TG-form components as found in nature (plus some metabolites) and absent any ethanol, to be considered DIs:
If reagents used during processing are likely to make covalent changes to components in the mixture during processing, you should determine whether the new material is still a dietary ingredient. For example, use of a large amount of an oxidizing acid like sulfuric acid to process a botanical mixture may create a new “semi-synthetic” mixture that is no longer a mixture of components that were present in the original plant. Therefore, the mixture would no longer be a dietary ingredient.
Contrariwise, it could be said that “components that were present in the original plant” that are part of the “new material” are dietary ingredients, despite the semi-synthetic nature of the composition.
A description of rTG is as follows:
DHA and EPA supplements can be given as free fatty acids, natural and reconstituted triglycerides and ethyl esters. Natural fish oil triglycerides (NTG) correspond to 100% triglycerides whereas chemically reconstituted triglycerides (RTG), as defined in the European Pharmacopeia are a mixture of monoglyceride (MG),12 diglyceride (DG) and triglycerides with triglyceride being the main component (>60%).
Nordic Naturals’ products take advantage of a process that yields an even higher percentage of TG in the final composition:
Until recently, Nordic Naturals ?sh oil products contained up to 60% triglycerides (with the remaining 40% comprised of diglycerides and monoglycerides). Now, however, we have perfected the technology that allows us to reassemble 93% of the fatty acids in our ?sh oils into the triglyceride form (with only 7% monoglycerides and diglycerides).
They also provide helpful graphics to illustrate the components and their metabolism:
Thus, the rTG form is a mixture of what exists in the “original” fish oil, with a lesser percentage of the components that are known metabolites as the body breaks down TGs into free fatty acids and monoglycerides, with some diglycerides also present. Re-esterification is part of this natural metabolism process, with ethanol absent in the final composition. Thus, by our reading of FDA guidance, rTG forms of EPA/DHA should pass muster as dietary ingredients, due to being comprised of 1) mostly TG forms, which are present in the “original” nutrient source (i.e. anchovy, sardines), and 2) metabolites of a dietary ingredient: monoglyceride and diglyceride EPA/DHA. The process also need not take place in the human body.
A metabolite that has been synthesized from another dietary ingredient would be a dietary ingredient under section 201(FF)(1)(F) and could be used as a dietary ingredient in a dietary supplement. Although the definition of a metabolite requires human ingestion of the dietary ingredient to increase the production or flux of the metabolite in the human body, it does not require the metabolism to actually take place in a human being during the manufacture of a dietary ingredient. A metabolite may be synthetically produced, provided that the starting material is a dietary ingredient and the production process mimics the metabolic process in the body following ingestion.
The only hitch we can see is the last sentence in the above, which might exclude rTG forms in a very strict reading, as an intermediary transesterification process does occur that breaks off the glycerol “backbone” and causes the resultant free fatty acids to cleave to ethanol. This greatly helps concentrate the EPA and/or DHA present in the crude batch via subsequent molecular distillation. But the final rTG product form contains only natural EPA-TG and/or DHA-TG and metabolites of EPA-TG/DHA-TG. This may be enough to convince the ITC (especially given this is an unfinalized draft guidance), even if it might take an eventual citizen petition to persuade the FDA:
FDA can create an exception to this prohibition by regulation, but only if the agency finds that the use of the article in dietary supplements would be lawful. To date, no such regulations have been issued. The appropriate mechanism to request such a regulation is to file a citizen petition under 21 CFR 10.30.
Whether ethyl ester forms of EPA and DHA (whereby the free fatty acids remain cleaved to ethanol) are dietary ingredients would not be defensible by this argument, however. Yet by extension, it may be. FDA has stated the following with regard to ethyl alcohol in goods or in the manufacture of foods:
Practically and scientifically, pure ethyl alcohol synthesized from natural gas or petroleum products does not differ from that obtained by fermentation with subsequent distillation. Furthermore, foods in which one is used cannot be distinguished objectively from those in which the other is used.
Synthetic ethyl alcohol may be used as a food ingredient or in the manufacturing of vinegar or other chemicals for food use, within limitations imposed by the Federal Food, Drug, and Cosmetic Act, the Alcohol Administration Act, and regulations promulgated under these acts.
Also, the human body metabolizes the consumption of alcohol and fatty acids to make “fatty acid ethyl esters” (FAEE), which in this case cannot then be considered synthetic.
A number of intracellular proteins have been isolated from different sources and shown to catalyze esterification of fatty acids to ethanol (12–14). By aminoterminal-sequence analysis, it was shown that two of these purified FAEE synthases are apparently identical to liver microsomal carboxylesterase ES-10, the predominant carboxylesterase in rat liver (15, 16). Other enzymes such as pancreatic cholesterol ester synthase (17) and pancreatic carboxylester synthase (18) also have been shown to possess some FAEE synthesizing activity. FAEE also can be synthesized by transesterification of ethanol and fatty acyl-CoA, a reaction catalyzed by acylcoenzyme A:ethanol O-acyltransferase (AEAT).
Beyond these observations, there may also be some room for interpretation of the following mention from the current draft guidance:
Synthetic vitamins, minerals, and amino acids are recognized as dietary ingredients because a vitamin, mineral, or amino acid is defined by its nutritional function (its ability to provide nutrients to the human body), not by its state of matter like a botanical.
For example, a synthetically produced amino acid, i.e. L-carnitine, which is often taken as a weight-loss supplement, is considered a DI based upon its ability to provide L-carnitine to the human body, and the identification of L-carnitine as a “nutrient” itself—as opposed to a synthetic botanical such as echinacea, which cannot in and of itself be considered a “nutrient”—and furthermore, “A substance that has been synthesized in a laboratory or factory has never been part of an herb or other botanical and, therefore, is not a dietary ingredient under section 201(FF)(1)(C) of the FD&C Act.”
But EPA-E and DHA-E were “part of” a natural food source, and further, are directly assimilated in and used by the human body as nutrients (essential fatty acids). Therefore, it may be more apt to consider EPA and DHA in a similar category as “vitamins, minerals and amino acids,” or perhaps more succinctly for them all: nutrients. Certainly, the end result is the same—500 mg of EPA-E and 500 mg of NTG EPA will both result in increases in serum EPA levels. It is similar to various forms of vitamin E, in fact, available in natural, synthetic and semi-synthetic forms., 
Furthermore, one could argue that since semi-synthetic and synthetic fats of various kinds have long been used as dietary ingredients, such as hydrogenated oil (common fat bound to hydrogen) and “Olestra” (a sucrose polyester), and because EPA/DHA are fatty acids themselves, that a semi-synthetic modification of these falls under the same category of DI.
Amarin’s goal is to get the Federal Circuit to remand the Complaint back to the ITC, charging them to investigate the matter and rule one way or the other. The ITC must then decide whether to infer from the draft guidance or not, and whether it should be concluded that EPA-E is not a dietary ingredient, but rather, a drug, from its interpretation of the FD&C Act. But that is a tall order considering the open-ended nature of the mentions in the draft guidance, and the major changes FDA has historically made to previous drafts to arrive at its final guidance—even omitting entire sections from draft-only versions. Furthermore, FDA is not even legally bound by its own final guidance in making specific decisions that may at times contradict its own guidance. And, most importantly, Congress has solely charged FDA with the ministerial role of interpreting and applying the FD&C Act.
FDA issued a comment to the ITC on the case, stating:
FDA respectfully submits that the Commission should decline to initiate the requested investigation. As pled, Complainants’ claims—unfair methods of competition under the Tariff Act based on false advertising under the Lanham Act and violations of the Federal Food Drug and Cosmetics Act (“FDCA”)—can succeed only if the Commission finds that Respondents’ products are unapproved “new drugs” rather than “dietary supplements” under the FDCA. The Complaint here is predicated on open questions of law and policy on which FDA has not reached final conclusions. Any such findings by the Commission on those issues may conflict with later determinations by FDA. Further, through the Complaint, Complainants attempt an unlawful private FDCA enforcement action based on Complainants’ allegations, not on FDA’s findings. As detailed below, because Congress has authorized only FDA to initiate FDCA enforcement actions, the FDCA precludes claims that would require the adjudicator to interpret, apply, or enforce the FDCA. For Complainants to succeed on any of their claims, the Commission would have to do all three of those things.
FDA is also working on a master list of those “grandfathered-in dietary ingredients,” including all products that are considered “dietary ingredients” and are exempted from reporting as NDIs by FDA. If ethyl ester forms of omega-3s make it onto the list, then that would absolve manufacturers of the same.
And so, we await the outcome. At worst, supplement manufacturers would be restricted to producing concentrated free fatty acid, natural triglyceride and monoglyceride forms of EPA and DHA (and we think rTG forms would also be allowed), all of which do appear to be better absorbed than ethyl ester forms in any event. Of course, being disallowed to produce EE forms would certainly be a setback to the omega-3 supplement industry, and at least a short-term victory for Amarin. However, we think there is a high likelihood that the Federal Circuit upholds the ITC decision not to take up the case, as the subject matter requiring its ruling lies outside its jurisdiction, and abides only with the FDA. UPDATE
The Federal Circuit has issued a ruling against Amarin Corp., 2-1, with the dissenting opinion not disagreeing with the ruling of the majority necessarily, but only in that the dissenting justice viewed the case to be outside the Federal Circuit’s jurisdiction, that the grounds upon which an appeal to the decision by the ITC weren’t even met. The Justice’s view pertains to a subtlety of law, whereby only the ITC’s “final determination” may be appealed, not their decision “not to institute an investigation.” The majority ruled that the ITC was correct in not instituting the investigation as the subject matter lied outside their jurisdiction. And, importantly, they (reiterating the ITC decision) left open the ability for Amarin to try their case again after the FDA issues a determination on whether or not the FD&C Act (FDCA) precludes ethyl ester forms, and forms that use a transesterification step, as drugs and not dietary ingredients. In essence, it was a majority verdict against Amarin.
Amarin may attempt to spend more resources appealing to the Supreme Court, but there is very little chance that their ruling differs.
 Douglas G Altman. Statistics notes: The normal distribution. BMJ 1995;310:298.
 Thomas Lumley, Paula Diehr, Scott Emerson, and Lu Chen. The Importance of the Normality Assumption in Large Public Health Data Sets. Annual Review of Public Health. 2002 23:1, 151-169
 Hui Jin, Xuejun Zhao. Transformation and Sample Size. Hogskolan Dalarna. 2009.
 DeCarlo LT. On the Meaning and Use of Kurtosis. APA. 1997.
 Feng C, Wang H, Lu N, et al. Log-transformation and its implications for data analysis. Shanghai Arch Psychiatry. 2014.
 Robert J. Glynn, Jean G. MacFadyen, Paul M Ridker, et al. Tracking of High-Sensitivity C-Reactive Protein after an Initially Elevated Concentration: The JUPITER Study. Clinical Chemistry. 2009.
 Geraci M1, Alston RD, Birch JM. Median percent change: a robust alternative for assessing temporal trends. Cancer Epidemiol. 2013.
 Bays HE, Ballantyne CM, Doyle RT, et al. Icosapent ethyl: Eicosapentaenoic acid concentration and triglyceride lowering effects across clinical studies. Elsevier. 2016.
 Tatsuno, Ichiro et al. Efficacy and safety of TAK-085 compared with eicosapentaenoic acid in Japanese subjects with hypertriglyceridemia undergoing lifestyle modification: The omega-3 fatty acids randomized double-blind (ORD) study. Journal of Clinical Lipidology. 2013.
 Maki KC1, Orloff DG, Nicholls SJ, et al. A highly bioavailable omega-3 free fatty acid formulation improves the cardiovascular risk profile in high-risk, statin-treated patients with residual hypertriglyceridemia (the ESPRIT trial). Clin Ther. 2013.
 Agrawal D, Manchanda SC, Sawhney JP, et al. To study the effect of high dose Atorvastatin 40 mg versus 80 mg in patients with dyslipidemia. IHJ. 2018.
 Bays HE, Ballantyne CM, Braeckman RA, Stirtan WG, Soni PN. Icosapent ethyl, a pure ethyl ester of eicosapentaenoic acid: effects on circulating markers of inflammation from the MARINE and ANCHOR studies. Am J Cardiovasc Drugs. 2013.
 Zamani B, Saatlo BB, Naghavi-Behzad M, Taqizadeh-Jahed M, Alikhah H, Abbasnezhad M. Effects of high versus low-dose atorvastatin on high sensitive C-reactive protein in acute coronary syndrome. Niger Med J. 2014.
 Scott Kinlay , Gregory G. Schwartz , Anders G. Olsson, et al. High-Dose Atorvastatin Enhances the Decline in Inflammatory Markers in Patients With Acute Coronary Syndromes in the MIRACL Study. Circulation. 2003.
 Bonnet J, McPherson R, Tedgui A, et al. Comparative effects of 10-mg versus 80-mg Atorvastatin on high-sensitivity C-reactive protein in patients with stable coronary artery disease: results of the CAP (Comparative Atorvastatin Pleiotropic effects) study. Clin Ther. 2008.